Organic and hybrid resistive switching materials and devices

Shuang Gao ab, Xiaohui Yi ab, Jie Shang ab, Gang Liu *ab and Run-Wei Li *ab
aCAS Key Laboratory of Magnetic Materials and Devices, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo, Zhejiang 315201, China. E-mail: liug@nimte.ac.cn; runweili@nimte.ac.cn
bZhejiang Province Key Laboratory of Magnetic Materials and Application Technology, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo, Zhejiang 315201, China

Received 31st July 2018

First published on 6th November 2018


The explosive increase in digital communications in the Big Data and internet of Things era spurs the development of universal memory that can run at high speed with high-density and nonvolatile storage capabilities, as well as demonstrating superior mechanical flexibility for wearable applications. Among various candidates for the next-generation information storage technology, resistive switching memories distinguish themselves with low power consumption, excellent downscaling potential, easy 3D stacking, and high CMOS compatibility, fulfilling key requirements for high-performance data storage. Employing organic and hybrid switching media in addition allows light weight and flexible integration of molecules with tunable device performance via molecular design-cum-synthesis strategy. In this review, we present a timely and comprehensive review of the recent advances in organic and hybrid resistive switching materials and devices, with particular attention on their design principles for electronic property tuning and flexible device performance. The current challenges posed with development of organic and hybrid resistive switching materials and flexible memory devices, together with their future perspectives, are also discussed.


image file: c8cs00614h-p1.tif

Shuang Gao

Shuang Gao is currently an Assistant Professor at the Ningbo Institute of Materials Technology and Engineering (NIMTE), Chinese Academy of Sciences (CAS). He received the BS degree from University of Science and Technology Beijing (USTB) in 2011 and then the PhD degree from Tsinghua University in 2016. His current research interests are mainly memristive materials and devices for wearable electronics and novel logic-in-memory as well as neuromorphic computing applications.

image file: c8cs00614h-p2.tif

Xiaohui Yi

Xiaohui Yi is currently an Associate Professor at the Ningbo Institute of Materials Technology and Engineering (NIMTE), CAS. He obtained his PhD degree from L’Institut National des Sciences Appliquées de Rennes, France at 2014. He then joined Prof. Run-Wei Li's group at the CAS Key Laboratory of Magnetic Materials and Devices of NIMTE as a postdoctoral researcher. His current research interests are focused on the electric, luminescent and magnetic properties on the basis of ligand–metal complex and their application for flexible memory.

image file: c8cs00614h-p3.tif

Jie Shang

Jie Shang is currently a Professor at Ningbo Institute of Materials Technology and Engineering (NIMTE), Chinese Academy of Sciences (CAS). After receiving the PhD degree from Kunming University of Science and Technology in 2010, he joined the CAS Key Laboratory of Magnetic Materials and Devices of NIMTE. His research work is focused on the design, preparation and engineering of flexible and elastic functional materials, as well as their applications in wearable devices.

image file: c8cs00614h-p4.tif

Gang Liu

Gang Liu is currently a full professor at the Ningbo Institute of Materials Technology and Engineering (NIMTE), CAS. After receiving his PhD degree from the National University of Singapore in 2010, he worked as a research associate at the Nanyang Technological University and then as a research fellow at the National University of Singapore from January 2010 to August 2012. In August 2012, he joined the CAS Key Laboratory of Magnetic Materials and Devices of NIMTE. His research interests include the design, preparation and engineering of polymer-based nanocomposite materials, as well as their applications in electronics and optoelectronics.

image file: c8cs00614h-p5.tif

Run-Wei Li

Run-Wei Li is currently a full professor at the Ningbo Institute of Materials Technology and Engineering (NIMTE), the Chinese Academy of Sciences (CAS) and the director of CAS Key Laboratory of Magnetic Materials and Devices. After receiving his PhD degree from the Institute of Physics, CAS in July 2002, he worked as a JSPS research fellow at the Osaka University. In September 2003, he moved to the Kaiserslautern University as an Alexander von Humboldt research fellow. Since March 2008, he has been “One Hundred Talents” professor of CAS. His research work is mainly focused on the functional materials and devices for new types of storage and sensors.


1. Introduction

The current Big Data era has witnessed a striking explosion in global digital information over the past decades, with huge volumes and at exponentially growing speed in the great diversity of civil defense and aerospace operations, market trend analysis, and consumer electronic gadgets.1 Over 40 trillion gigabytes of data will be generated by year 2020 and each person on this planet shares 5200 gigabytes.2 This raises a strong desire for a new “universal memory” device that can operate at high speed with high-density and nonvolatile storage capabilities, uniting the respective merits of dynamic random access memory (DRAM), hard-disk drive (HDD), and flash memory into a state-of-the-art information storage hierarchy.3,4 Marrying the recently rapid development of internet of Things technique on the other hand, the memory devices should also be mechanically flexible, thus allowing free contact to and conformal deformation with the complex curvilinear surface of human skin to deliver wearable and even implantable personal healthcare and smart medical equipment.5 However, conventional silicon-based semiconductor devices are encountering severe downscaling constraints in terms of data fidelity, heat death, and unaffordable manufacturing costs for ultra-large-scale implementation, as well as lack of intrinsic adaption to the soft deformation scenario by utilizing primarily brittle electronic components. Design and physical realization of alternative information storage devices that are built from new materials and run on completely different mechanisms therefore appear imperative tasks for academic and industrial communities to meet future microelectronics needs.

Accompanying the dimension shrinkage campaign of transistors as predicted by Moore's Law since the 1960s, four types of new memories, namely, magnetic random access memory (MRAM), phase-change random access memory (PcRAM), ferroelectric random access memory (FeRAM), and resistive random access memory (RRAM or resistive switching memory) have emerged in the new century and are listed as promising candidates for next-generation information storage technology by the International Technology Roadmap for Semiconductors (ITRS).6 In particular, RRAM that has a simple electrode/insulator/electrode sandwich-like structure, as shown in Fig. 1a, distinguishes itself extraordinary with versatile ranges of material selection and abundant possible switching mechanisms for flexible implementation.7–12 With localized, interfacial or bulk changes in the composition and conformation or charged state of the insulating layer, RRAM devices usually exhibit bistable write-once-read-many (WORM, Fig. 1b), unipolar (Fig. 1c), and bipolar (Fig. 1d) current–voltage (IV) hysteresis characteristics for binary information storage. Two-terminal (rather than three-terminals of transistors) structure with excellent miniaturization potential of several to tens of nanometers also enables RRAM to be readily integrated into crossbar arrays with three-dimensional stacking architecture for highest possible storage density.12 Resistive switching usually occurs within a few nanoseconds,13,14 which promises fast writing, reading, and erasing of data during operations. Together with the low energy consumption of sub-pico joule,15 high switching endurance over trillion cycles,16,17 and nonvolatile data retention exceeding ten years,18 resistive switching memories are recognized as an appealing player for “universal memory”. Supplementing the well explored bistable electrical behaviors, resistive switching memories have recently been extended further to the interesting concept of a memristor that uses the consecutive modulation of device resistance or conductance to emulate the physiological activity of the neural synapses or to perform logic-in-memory operations (Fig. 1e).19–23 This extended form of resistive switching memories is adopted as a star of tomorrow for the erection of a neuromorphic computing paradigm to solve the long lasting von Neumann bottle problems.


image file: c8cs00614h-f1.tif
Fig. 1 Schematic device structure and various current–voltage (I–V) characteristics of nonvolatile resistive switching memories, where ICC denotes the compliance current during set process to prevent the device from permanent breakdown.

Thanks to the great efforts all over the world since the initial discovery of resistive switching in amorphous oxides in the 1960s, a wide variety of organic, inorganic, and organic–inorganic hybrid materials have been found to display resistive switching features thus far.9,12,24,25 In comparison to metal oxides, chalcogenides, amorphous silicon, and other inorganic counterparts, organic and hybrid materials benefit a lot from atomic engineering that can fine-tune their electronic properties through molecular design-cum-synthesis strategy, and also exhibit light weight, low-cost fabrication with solution processing on various substrates, and most important of all, intrinsic softness through a carbon-based skeleton for flexible electronics.26 Specifically, flexible resistive switching memories based on organic and hybrid materials are core to constructing e-textiles, which can bring us closer to sci-fi scenarios where future electronic systems are totally integrated in our everyday outfits and help us in achieving a higher comfort level, for example, interacting readily with other digital devices such as smartphones or even with analog devices such as our brain/peripheral nervous system.27 Extensive studies have been conducted to date on the design, synthesis, and application of organic and hybrid information storage materials and devices. In this contribution, we present a timely and comprehensive review of the recent advances in the preparation, switching characteristics, and mechanisms of resistance-switchable organic small molecules, polymers, graphene oxides, metal–organic frameworks, and organic–inorganic hybrid perovskites towards the development of high-performance flexible memories and memristors that have recently inspired great research interests of material scientists and chemists.

2. Designing principle for flexible resistive switching memories

Mechanical flexibility and electrical reliability under deformation servitude establish the fundamental criteria for wearable and implantable electronic devices. For resistive switching memories with an electrode/insulator/electrode sandwich structure (Fig. 1a), both the electrode and insulator should maintain their complete morphology and stable conductive/switching characteristics when being bent, stretched, or even twisted. Many materials can be used as electrodes, including conductive metal oxides, metals, carbon (for instance, carbon nanotube networks or graphene thin films), and sometimes polymer conductors such as poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS). Among these, conductive metal oxides such as indium-tin oxide (ITO), fluorine-tin oxide (FTO), and aluminum-zinc oxide (AZO) are brittle in bulk but can become highly flexible in thin film form.28 As for the more widely used metals, carbon, and polymer conductors, they all have a certain degree of intrinsic mechanical flexibility and thus are not covered in the present review. Considering stress transmission across the out-of-plane direction of the entire sample under bending circumstances,29 the switching insulator should be as thin as possible to keep the strain distributed across the sample small and uniform, as well as carry similar (if not exactly the same) Young's modulus to that of a soft substrate such as poly(methyl methacrylate) (PMMA), polydimethylsiloxane (PDMS), and Ecoflex to avoid any possible delamination. In cases of stretching and twisting, the intrinsically molecular scale flexibility in addition to the structural flexibility would provide additional advantages to maintain the devices’ mechanical and electrical properties.30 Taking all these criteria into account, organic and organic–inorganic hybrid thin films are the first-rank choices for future molecular-scale memory applications with their attractive features of tunable electronic property, easy solution processability, and excellent mechanical flexibility.

Instead of encoding “0” and “1” as the amount of charges stored in conventional silicon-based devices, organic and hybrid resistive switching devices store information in a different manner, for instance, based on the low and high resistance (conductance) states in response to an external electric field.9,24 Since electrical conductivity is the product of charge carrier concentration and mobility, changes in either the carrier concentration or mobility, or both, can lead to changes in the device conductance. Many intrinsic and extrinsic switching mechanisms have been proposed and clarified in organic and hybrid devices, mainly categorized into thermochemical reaction,31–33 ion migration,34–39 interfacial reaction,40,41 charge trapping/de-trapping,42–45 charge transfer,46–49 electrochemical redox reaction,50–53 and conformational change.54,55 Among them, thermochemical reaction is dominated primarily by the effect of Joule heating, while all others are caused mainly by the effect of the electric field. A deep understanding of the structure–property relationship of each operating mechanism, which can be facilitated by the advanced in situ spectroscopic and microscopic analysis,34,39 as well as theoretical molecular simulation,48,49 is crucial for the rational design of the switching materials and consequently overall performance of the devices.

2.1 Thermochemical reaction

The first resistive switching behavior observed in polymer materials dates back to the 1970s, which were formed by a glow-discharge technique and work on a thermochemical reaction mechanism.31,32 Upon Joule heating-induced pyrolysis of the polymers, carbon-rich conducting filaments will be formed locally inside the insulating layer and switch the device from the initial high resistance state (HRS) to a low resistance state (LRS) (Fig. 2a). The pyrolysis procedure will also cause formation of many voids within the polymer layers. The as-formed carbon-rich filaments resemble fuses used in domestic electrical circuits and allow high current to flow through them. As such, the intensified Joule heating effect in the subsequent voltage sweep will rupture these fuse-like filaments and program the device back to HRS. The thermochemical reaction does not significantly depend on the molecular structure of the materials, while the direct observation of carbon-rich filaments is made possible with a cross-sectional transmission electron microscope (TEM) in a Cu/pEGDMA/ITO device.33
image file: c8cs00614h-f2.tif
Fig. 2 Schematic illustration of various switching mechanisms. Reproduced with permission from ref. 29, © 2015 American Chemical Society (a), from ref. 37, © 2016 Wiley-VCH (b), from ref. 38, © 2016 Wiley-VCH (c), from ref. 40, © 2014 Wiley-VCH (d), from ref. 45, © 2016 American Chemical Society (e), from ref. 47, © 2007 American Chemical Society (f), from ref. 50, © 2007 American Chemical Society (g), and from ref. 54, © 2008 American Chemical Society (h).

2.2 Ion migration

Mobile ions, either cations or anions, can migrate under an electric field across an insulator. For instance, an electrochemically active silver or copper electrode can be oxidized by an electric field to inject Ag+/Cu2+ ions into the resistive switching layer, which are then reduced and aggregate into metal filaments (Fig. 2b).34,35 The growth of metal filaments eventually connects the anode and cathode, generating a localized highly conductive pathway inside the insulator and switching the device into LRS. Under opposite voltage bias, the metal conductive filaments will be annihilated via electrochemical oxidization and turn the device back to HRS. Foreign cation migration-based resistive switching is extrinsic to the insulating layer, while native anions such as carboxyl (–COO) moieties in graphene oxides38 and I in organic–inorganic hybrid perovskites39 can also induce resistive switching phenomena (Fig. 2c). The detachment of carboxyl groups from the surfaces and edges of graphene oxides can partially restore carbon atoms from the insulating sp3 hybridization form to the more conductive sp2 hybridization form and switch on the device.

2.3 Interfacial reaction

In the case where a metal electrode can be passivated through an electric filed-induced redox reaction with the storage medium, interfacial reaction is considered to be the dominating switching mechanism, featured by a clear negative differential resistance (NDR) effect. One representative is the Ag/Rb-CD-MOF/Ag device,40 where the as-synthesized Rb-CD-MOF contains rich free hydroxy ions. Upon an external electric field, the hydroxy ions will migrate towards the anodic Ag electrode and then react with it, leading to the formation of an interfacial layer of silver hydroxide (AgOH) and silver oxide (AgOx). As such, the device resistance increases with external electric field, i.e., NDR effect. When the field polarization is inverted, the existing silver oxide layer will be reduced, and the other Ag electrode will be gradually oxidized (Fig. 2d), leading to a resistive switching behavior with symmetrical NDR effects. Similar resistive switching behavior recently has been observed in some polymers doped with metal oxide nanoparticles, where the interfacial reaction between oxygen-containing moieties in polymers and metal oxide nanoparticles is considered crucial.41

2.4 Charge trapping/de-trapping

Due to differences in their molecular orbital energy levels, metal nanoparticles,42,43 quantum dots,44,45 and semiconducting small molecules56,57 dispersed in an insulating polymer matrix can act as trapping centers for charge carriers that are injected from electrodes. With a sufficient amount of charge carriers trapped inside an insulating matrix, local percolation networks of the charged trapping centers will provide a continuous carrier hopping pathway in the bandgap of the polymer material and switch the device from initial HRS to LRS (Fig. 2e). Application of an electric field with reversed polarity will extract the trapped carriers out of the trapping centers, which consequently leads to rupture of existing transport pathways for charge carriers and thus the switching of related devices back to HRS.

2.5 Charge transfer

Under an external electric field, the charge transfer (CT) process can occur in electron donor–acceptor (D–A) systems, during which the electrons in donor moieties such as carbazole groups are partially transferred into acceptor (e.g., fullerene molecules) moieties, leaving positively charged holes residing on the donors (Fig. 2f).47–49 This process can lead to the formation of partially filled molecular orbitals and increases the concentration of free charge carriers with high mobility, thus switching the D–A systems into a more conductive state. The CT interaction may be reversible, wherein an opposite electric filed can drive electrons back to the donor group and force their recombination with holes, decreasing the free charge carrier concentration and returning the D–A systems back to HRS. Depending on the degree of electric field distribution uniformity inside the switching layer, the occurrence of charge transfer interaction may be either local or bulky in nature.

2.6 Electrochemical redox reaction

Unpaired or lone pair electrons can be removed from electron-rich molecules (for example, ferrocene, thiophene, and triphenylamine) by an external electric field, which in turn introduces impurity energy levels into the bandgap of the insulating materials. As such, the conductance of a device can be significantly enhanced via intra- and inter-molecular hopping through these charged centers (Fig. 2g).50,51,53,58,59 The positive charges generated upon oxidation can be balanced by reduction of environmental oxygen in the atmosphere or additional counter electrode materials such as 4,4′-bipyridine salt (viologen).21 The use of a counter electrode may help maintain stability of the redox system and improve the endurance characteristics of the resistive switching devices.

2.7 Conformational change

Such a mechanism is mostly seen in polymers containing carbazole groups such as PVK-PF and iamP6.54,55 The initial HRS of these polymers is caused by the random orientation of carbazole groups that hinders the ordered π–π stacking and charge transport through the conjugated system (Fig. 2h). Upon application of an external electric field, the carbazole groups are able to be rearranged into a nearly face-to-face conformation, which facilitates carrier delocalization and transport and switches the polymers to LRS. Reversal in electric field polarity can break the ordered face-to-face conformation of carbazole groups, possibly due to thermal injecting at the electrode/polymer interface, and turn the device back to HRS.

According to the above mentioned switching mechanisms, special care should be taken when designing and synthesizing proper organic and organic–inorganic hybrid materials for flexible resistive switching memories, in particular when selecting functioning moieties that form the switching layer and are responsible for the resistive switching. Most of the attention focusing on organic and hybrid switching materials until now lay in the design and synthesis of molecules with enhanced electrical behavior, while efforts were only recently initialized to simultaneously improve their mechanical performance.30,60,61 In the following sections, recent advances on resistive switching materials of small molecules, polymers, graphene oxides, and organic–inorganic hybrid materials, are comprehensively summarized and discussed.

3. Organic small molecule switching materials

Due to the advantages of low cost, light weight, easy purification, and well-defined structures, organic small molecules have been of great research interest for resistive switching memory applications in the past two decades.9,62 The original devices were mostly composed of an Al/small molecule/metal nanoclusters/small molecule/Al five-layer structure processed by thermal evaporation, which was pioneered by Yang et al. from the University of California, Los Angeles.13,63 Many commercially available semiconducting small molecules had been explored, including 2-amino-4,5-imidazoledicarbonitrile (AIDCN), tris-(8-hydroxyquinoline) aluminum (Alq3), N,N′-di(naphthalen1yl)-N,N′-diphenyl-benzidine (NPB), pentacene, etc.13,63–68 As for the metal nanoclusters, various metals were found to be feasible, such as Al, Ag, Mg, Cr, and Ni.65,67 Both organic and metal layers were deposited by thermal evaporation, while the metal nanoclusters were formed by island growth that was terminated before the islands coalesced into a continuous film. Further analysis of the nanoclusters revealed that they each consisted of a metallic core and an oxide shell, which is due to the reaction of metal with small molecules and/or oxygen during the evaporation process.67 The as-fabricated devices usually show an OFF state due to the large barriers for carrier injection from the electrodes into small molecule layers. When a sufficient bias is applied, the metal–nanocluster layer becomes polarized, and opposite charges are subsequently induced (or stored) at the top and bottom interfaces between the middle metal–nanocluster layer and small molecule layers.63 The stored charges lower the interfacial resistance and thus switch the devices to the conducting ON state, while only a reverse bias can restore the devices to the initial OFF state.63,69–73 Promising switching performances have been demonstrated in these devices, such as high ON/OFF ratio of >104, fast operation speed of <10 ns, switching endurance of >106 cycles, and even multilevel switching with data retention of >105 s.13,65 It is noted, however, that the deposition of metal–nanocluster layers is often very hard to control and thus greatly complicates the device fabrication process. On the other hand, the switching characteristic has been found almost insensitive to the type of the metal–nanocluster or semiconducting charge transporting organic small molecules,67 which limits the tunability in switching characteristic and thus the application scope of these devices. Therefore, much attention since then has been paid to organic small molecules with intrinsic resistive switching properties, enabling the modulation of device performance via rational molecule design and a better understanding of the switching mechanism by theoretical simulation.

Generally, the incorporation of a donor–acceptor (D–A) structure in organic small molecules is necessary to realize the intrinsic resistive switching property. Donor groups with electron donating capability include triphenylamine (TPA), methoxy, diethylamino, thiophene, etc. while the acceptor groups are electron deficient or have a strong electron-withdrawing property such as imidazole-[4,5-b]phenazine (BIP), benzo[c][1,2,5]thiadiazole, 1,8-naphthalimide, 1,3,4-oxadiazole, sulfone, nitro, etc.49,74–77 Upon application of an external electric field, charge transfer from donor to acceptor groups will occur and then lead to the opening of a conducting channel in the small molecule film, thus switching the devices from OFF to ON state. Under an opposite electric field, the excited state can be either permanent or reversible, corresponding to WORM and rewritable resistive switching behaviors, respectively. The switching characteristics have been found very sensitive to the number of the incorporated donor/acceptor groups. For example, a serious of resistive switching small molecules with varied TPA/BIP ratios were reported by Zhang et al. from Nanyang Technological University.74–76 TPA–BIP with one donor group and one acceptor group, was synthesized through a one-step condensation between 1,4-bis((triisopropylsilyl)ethynyl)-2,3-diaminophenazine and 4-(diphenylamino)benzaldehyde. Due to the single-step charge transfer between the donor–acceptor pair, it shows a typical two-level bipolar resistive switching behavior (Fig. 3b).74 In contrast, the 2TPA–BIP with one more donor group, synthesized by condensing 1,4-bis[(triisopropylsilyl)ethynyl]-2,3-diaminophenazine and 4,4′-bis(N,N-diphenylamino)benzil, exhibits a clear three-level bipolar resistive switching behavior because of two-step charge transfer (Fig. 3a and d).75 However, no obvious change in switching behavior is found if adding one more acceptor group, i.e., the TPA–2BIP in Fig. 3c, which was synthesized via a one-step condensation between 1,4-bis[(triisopropylsilyl)ethynyl]-2,3-diaminophenazine and 4,4′-diformyltriphenylamine.76 This possibly resulted from the fact that only one of the two identical acceptor groups is active based on theoretical simulation (Fig. 3e). These results may serve as a guide for the memory behavior optimization of small molecules by tuning donor/acceptor ratios in future.


image file: c8cs00614h-f3.tif
Fig. 3 Chemical structures and I–V characteristics of (a) 2TPA–BIP, (b) TPA–BIP, and (c) TPA–2BIP. (d) Energy diagram and schematic switching process of 2TPA–BIP. (e) HOMO and LUMO levels of TPA–2BIP. Reproduced with permission from ref. 75, © 2015 Wiley-VCH (a and d), from ref. 74, © 2014 Wiley-VCH (b), and from ref. 76, © 2015 Wiley-VCH (c and e).

In order to increase the storage density of D–A systems through multilevel resistive switching, the most effective way has been demonstrated by Lu et al. from Soochow University49,77–81 to be the incorporation of different types of acceptor groups. For example, the (p,p′-bis(2-aryl-1,3,4-oxadiazol-5-yl)diphenyl sulfone (OZA-SO)) with two types of acceptor groups (namely, 3,4-oxadiazole and sulfone) was synthesized through a three-step reaction from p,p′-dicarboxydiphenyl sulfone. The as-obtained OZA-SO molecule is found to show a three-level WORM resistive switching behavior (Fig. 4a).77 More impressively, even a four-level WORM resistive switching behavior has been successfully observed in NONIBTDT (Fig. 4b), which was synthesized from thiophene, benzo[c][1,2,5]thiadiazole (BTD), 1,8-naphthalimide (NI) and nitro (NO) building blocks. Among them, only thiophene is a donor, while all the other three act as acceptors.49 Based on theoretical simulation, these multilevel switching phenomena are well explained by the sequential transfer of electrons to different acceptor groups with varied electron-withdrawing abilities. For instance, the four-level switching behavior of NONIBTDT is considered to originate from the transfer of electrons from donor thiophene to acceptor BTD, NI, and NO in sequence (Fig. 4c).49 Compared to the normal bistable I–V characteristics, the three- and four-level resistive switching behaviors can increase the storage states exponentially from 2n to 3n and 4n, respectively, which is highly desired in the current era of Big Data with information explosion.


image file: c8cs00614h-f4.tif
Fig. 4 Chemical structures and I–V characteristics of (a) OZA-SO and (b) NONIBTDT. (c) HOMO and LUMO levels of NONIBTDT. Reproduced with permission from ref. 77, © 2015 Wiley-VCH (a) and from ref. 49, © 2016 Wiley-VCH (b and c).

Other than varying the donor/acceptor group number and type, the switching characteristic of organic small molecules can be also effectively modulated by methods including single atom substitution,82,83 end group substitution,81 molecular planarity and length design,79,84,85 akyl chains length adjustment,86 modification in linkage group between donor and acceptor,87etc. For example, the substitution of a single O atom by S has been found able to upgrade the two-level WORM resistive switching behavior in TFMCU to the three-level in TFMCT (Fig. 5a),83 which is possibly caused by the low bandgap and easy polarization of TFMCT. Through replacing the nitro group in PTZ-NO2 by a trifluoromethyl group (Fig. 5b), the obtained PTZ-CF3 is found to have a better solubility, film-forming ability, and even hydrophobicity, thus leading to a lower operation voltage and higher long-term stability up to three months.81 An increase in molecular planarity (Fig. 5c),79 molecular length (Fig. 5d),85 or alkyl chains length (Fig. 5e)86 can normally result in more ordered and compact stacking of small molecules, which favors the transport of charge carrier between neighboring molecules. This can finally lower the operation voltage or upgrade the switching behavior from no switching or two-level WORM to three-level WORM. For instance, there is no resistive switching in MAzoAN with methyl chains, whereas clear two-level WORM resistive switching behavior is observed in EAzoAN with longer ethyl chains (Fig. 5e). Further, with even longer butyl chains, BAzoAN is found to show three-level WORM resistive switching behavior. Similar upgrading in switching behavior has also been confirmed feasible via changing the linkage group between donor and acceptor (Fig. 5f),88 which was explained by more ordered molecular stacking and crystalline orientation at the same time. Moreover, the change in linkage group between donor and acceptor has been found also to alter the distribution of electron density throughout the molecular backbone and thereby to change the resistive switching behavior, for example, from volatile to nonvolatile.87


image file: c8cs00614h-f5.tif
Fig. 5 Tuning the switching characteristic of organic small molecules via (a) single atom substitution, (b) end group substitution, (c) molecular planarity design, (d) molecular length design, (e) akyl chains length adjustment, and (f) change in linkage group between donor and acceptor. (c) Reproduced with permission from ref. 79, © 2012 Wiley-VCH.

Very recently, the flexibility of small molecule resistive switching memory has been initiatively studied by Lu et al.86 The detailed device structure is Al/BAzoAN/Al on a plastic poly(ethylene terephthalate) (PET) substrate, where all layers were deposited by thermal evaporation. The BAzoAN active layer and Al electrodes were designed to be 80 and 100 nm in thickness, respectively. The chemical structure of BAzoAN can been found in Fig. 5e, which was synthesized from 1,5-diamineanthraquinone and N,N-dibutylaniline with the help of sodium nitrite and an ice bath. With external voltage applied to the top Al electrode and in flat state, three-level WORM resistive switching behavior was observed in the device (Fig. 6a). The transition from OFF state to the first ON state (ON1) occurred at −1.79 V, and further increasing the voltage to −3.08 V triggered the transition from ON1 state to the second ON state (ON2). Theoretical simulation suggests that the azobenzene chromophore and anthraquinone moieties in BAzoAN can act as shallow and deep traps, respectively (Fig. 6b).89 As such, the transitions from OFF to ON1 and then from ON1 to ON2 should be caused by the filling of shallow traps from azobenzene chromophore moieties and deep traps from anthraquinone moieties in sequence. Under bending, no degradation was found even at a small radius of 6.5 mm after 500 bending cycles (Fig. 6c). However, further decreasing the bending radius to 5.6 mm resulted in device failure due to the generation of cracks in the BAzoAN film. Under an extreme bending condition with the 6.5 mm radius, all the three resistance states were able to remain unchanged under a constant reading voltage of −1 V over 3 hours (Fig. 6d). These results demonstrate the potential of small molecule memory for flexible applications.


image file: c8cs00614h-f6.tif
Fig. 6 (a) I–V characteristic, (b) energy level, (c) bending endurance, and (d) retention property of the BAzoAN memory device. Inset in (b): molecular electrostatic potential (ESP) of BAzoAN. Reproduced with permission from ref. 86, © 2016 Wiley-VCH (a, c and d) and from ref. 89, © 2012 American Chemical Society (b).

Although dozens of organic small molecules have been designed and synthesized for resistive switching and even flexible memory devices, it is worthy of mentioning that switching behaviors of the small molecules reported so far are normally WORM. Despite that stable two-, three-, and even four-level resistive switching behaviors are achieved via molecular design, their application scope is still limited because of being non-rewritable. Therefore, much attention deserves to be given to the design of small molecules with intrinsic rewritable switching behaviors. One possible direction may be indicated in the very recent work by Venkatesan et al. from National University of Singapore,17 where rewritable bipolar switching behavior with extreme endurance of over 1012 was observed in the compound mer-[Ru(L)3](PF6)2 with three bidentate ligands (L = 2(phenylazo)pyridine) each containing one azo (N[double bond, length as m-dash]N) functional group (Fig. 7). This compound was synthesized via the double decomposition of the corresponding perchlorate salt of the complex by NH4PF6. Based on in situ Raman and ultraviolet-visible spectroscopy as well as spectroelectrochemistry and quantum chemical calculations, they suggested that the redox state of the ligands determines the switching states of the devices whereas the counterions control the hysteresis. In principle, these ligands could be replaced by other similar species and the Ru is possibly substituted with other transition-metal centers, thus making this class of systems highly promising. Besides, the elaborate thermal evaporation deposition under vacuum conditions and easy crystallization nature of small molecules during film deposition may hinder their direct application for flexible memory, which needs to be intensely considered in future works.


image file: c8cs00614h-f7.tif
Fig. 7 (a) Molecule structure, (b) I–V characteristic, and (c) switching endurance of mer-[Ru(L)3](PF6)2. (d) In situ UV-vis spectra of the various redox states of the molecule in solution. In situ (e) UV-vis and (f) Raman spectra of the molecule in film. Reproduced with permission from ref. 17, © 2017 Springer Nature.

4. Polymer switching materials

Compared to small molecules, polymers seem to be more attractive for resistive switching memory applications due to their higher intrinsic flexibility and easy solution processability. In particular, the later allows polymer films to be deposited by various low-cost solution methods viz. dip-coating, drop-casting, spray-coating, spin-coating, roller-coating, blade casting and ink-jet printing, making the fabrication procedure of memory devices more economical without the involvement of vacuum deposition procedure.7,10,11 Various types of polymer materials, including single-component electroactive polymers, polymer blends or mixtures with small molecules and polymer matrix for nanoparticles or inorganic compounds have been investigated for resistive switching memories.

4.1 Single-component polymers

Solution-processable polymer materials that can provide the required resistive switching characteristics within a single macromolecule while still possessing good chemical, mechanical, and morphological properties are promising candidates for flexible information storage devices. Great efforts have been devoted to designing and synthesizing single polymer memory materials, and several comprehensive reviews have thoroughly discussed these themes over the past fifteen years.7,9–11,90 In this section, we will only review the latest progress made in the last 5–6 years; particularly, D–A polymers with potential charge transfer features are an effective approach to realize organic resistive switching memories and will be emphasized herein. To date, plenty of conjugated and non-conjugated polymers with donor and acceptor moieties, either incorporated in the polymer backbone, dangling at the end of a macromolecular chain, or linked as pendant groups, have been reported to exhibit resistive switching behavior.48,91–99 By regulating the strength and loading ratio, as well as spatial arrangement of the D/A moieties, versatile resistive switching characteristics ranging from volatile to nonvolatile memory behaviors can be realized and manipulated.

Polyazomethine (PAM) materials are a family of conjugated poly(Schiff base) that contain imine groups (C[double bond, length as m-dash]N) in the backbone and exhibit excellent thermal stability, good mechanical property, metal-chelating ability, and molecular doping-controlled electrical properties. Pan et al.48 successfully synthesized two D–A type PAM derivatives where triphenylamine (TPA), oxadiazole (OXD), and 3,3′-dinitro-diphenylsulfone (NPS) moieties are arranged in the conjugated polymer backbone (Fig. 8a). Sandwiched between two Pt electrodes, PAM-1 shows rewritable bipolar resistive switching behavior (Fig. 8c), whereas PAM-2 exhibits only WORM resistive switching behavior (Fig. 8d). Theoretical simulation indicates that the TPA and OXD moieties in PAM-1 act respectively as donor and acceptor, while in PAM-2 the donor and acceptor are OXD and NPS moieties, respectively (Fig. 8e). Therefore, the moderate electron withdrawing ability of the OXD acceptor makes the charge transfer interaction in the TPA–OXD D–A pair reversible, and thus rewritable memory behavior is observed in PAM-1. In contrast, the strong electron withdrawing ability of NPS moieties causes an irreversible charge transfer interaction in the OXD–NPS D–A pair and the resultant WORM switching behavior in PAM-2. On the other hand, hyperbranched polymers are usually characterized with unique three dimensional structures, good solubility, low melting temperature and solution viscosity, and excellent physiochemical properties. Given this point, a hyperbranched PAM with triphenylamine moieties (PAM-3, Fig. 8b) was synthesized and used as the functioning layers in Ta/PAM-3/Pt resistive switching memory devices.91 As shown in Fig. 8f and g, stable bipolar resistive switching behavior with highly uniform distribution of the HRS and LRS resistances was observed, differing significantly from the unstable switching behavior of a linear PAM (Fig. 8h). To further enhance thermal stability of organic resistive switching devices, poly(triphenylamine) (PTPA) was synthesized by Zhang et al.51 through a one-step oxidative coupling of the triphenylamine molecules, using ferric chloride as the oxidizing agent. In this polymer, the unpaired electrons on the nitrogen atoms of triphenylamine units can be easily removed to give a stable cationic conducting pathway. Sandwiched between Ta and Pt electrodes, this polymer was found to exhibit excellent non-volatile bistable resistive switching behavior with a long retention time over 8 × 103 s, superior ON/OFF ratio of 5 × 108, and particularly, a wide working temperature range of 30–390 K.


image file: c8cs00614h-f8.tif
Fig. 8 Chemical structures of (a) PAM-1, PAM-2 and (b) hyperbranched PAM-3 and linear PAM. (c and d) I–V characteristics of PAM-1 and PAM-2. (e) Charge transfer directions of PAM-1 and PAM-2. (f and g) I–V characteristic and switching endurance of PAM-3. (h) Switching endurance of the reference linear PAM. Reproduced with permission from ref. 48, © 2013 The Royal Society of Chemistry (a, c–e) and from ref. 91, © 2014 The Royal Society of Chemistry (b, f–h).

Instead of being placed in a conjugated backbone, the donor and acceptor moieties can also be arranged to the end of polymer backbones and pendants. For example, Chen et al. from National Taiwan University92 recently fabricated a flexible bipolar resistive switching memory consisting of a single-layer D–A conjugated polymer on plastic polyethylene naphthalate (PEN) substrate (Fig. 9a). The newly designed conjugated polymer PFT-PI with a main-chain donor of fluorene and thiophene and a side-chain acceptor of phenanthro[9,10-d]-imidazole was synthesized as the active memory material. The Al/PFT-PI/Al/PEN device was found to show stable bipolar resistive switching with large ON/OFF ratio of ∼104 not only in the flat state but also under bending with a small radius of 5 mm (Fig. 9b). Metal complexes also can be arranged to a side chain of the conjugate polymer for resistive switching memory applications. To this point, Huang et al. from Nanjing University of Posts & Telecommunications93,94 synthesized conjugated polymers containing Pt(II) complexes or Ir(II) complexes in pendant groups with different main-chains. Being employed as storage media, all the as-fabricated devices exhibited high ON/OFF current ratios and excellent retention stability. It was claimed that the excellent electron affinity and redox reversibility of the metal complex moieties benefit device stability. Moreover, it has been found that the threshold voltage and ON/OFF ratio can be tuned by changing the chemical structures of polymer main-chains.95 With ferrocene as a side chain moiety, four new conjugated polymers with triphenylamine, carbazole, or thiophene moieties in the main chain (PFcFE1–PFcFE4) have been designed and synthesized via a Sonogashira coupling reaction (Fig. 9c). Sandwiched between Al and ITO, PFcFE1, PFcFE2, and PFcFE3 exhibited rewritable bistable behaviors with varied threshold voltages and ON/OFF ratios, whereas only a WORM memory effect was observed in PFcFE4. Notably, the triphenylamine-based PFcFE1 presented the best performance with a high ON/OFF ratio of over 103 (Fig. 9d), which provides great potential for further memory applications.


image file: c8cs00614h-f9.tif
Fig. 9 (a and b) Chemical structure and IV characteristic of PFT-PI. (c) Chemical structure of PFcFE and (d) IV characteristic of PFcFE1. Reproduced with permission from ref. 92, © 2012 The Royal Society of Chemistry (a and b) and from ref. 95, © 2016 The Royal Society of Chemistry (c and d).

In comparison to the conjugated ones with a rigid backbone, non-conjugated polymers usually have advantages of environmental stability and mechanical flexibility. Incorporation of the D–A structure into non-conjugated polymers such as functional polyimides also enables resistive switching behaviors. For example, Chen et al.96 developed triphenylamine–pyrene containing D–A polyimides for resistive switching memory applications with Al electrodes on a plastic PEN substrate. The polyimides were prepared from 4,4′-diamino-4′′-methyltriphenylamine (AMTPA) or N,N-bis(4-aminophenyl)aminopyrene (APAP) and 4,4′-(hexafluoroisopropylidene)diphthalic anhydride (6FDA) units. Through tuning the molar content of APAP and AMTPA units, volatile, rewritable, and WORM memory behavior can be achieved. More importantly, superior memory performance and high mechanical durability were demonstrated in these polyimides. On the other hand, Wang et al.97 reported a series of non-conjugated homopolymers and random copolymers containing a triphenylamine moiety as the donor and (2,5-biphenyl)-1,3,4-oxadiazole (OXD) moiety as the acceptor, which are connected to the vinyl backbone through an amide linkage. By increasing the D/A loading ratios from 0% to 100% continuously, the Al/polymer/ITO devices demonstrate insulator, threshold switching, nonvolatile rewritable memory, WORM memory, and insulator behaviors, respectively.

Stretchable resistive switching memory is also possible to be achieved through the D–A strategy in non-conjugated polymers. For example, Chen et al.98 developed a series of rod-coil diblock copolymers, PF14-b-Pison, containing electron-donating poly[2,7-(9,9-dihexylfluorene)] (PF) rods and electron-withdrawing poly(pendent isoindigo) (Piso) coils (Fig. 10a). The PF and Piso blocks were first synthesized through chain-growth Suzuki–Miyaura coupling polymerization and atom transfer radical polymerization methods, respectively, and then the target diblock copolymers were obtained via a click coupling reaction. The common pre-strain method was adopted to fabricate memory devices with an Al/PF14-b-Pison/carbon nanotube (CNT) sandwich structure on a soft PDMS substrate (Fig. 10b). With n = 10, volatile threshold switching behavior that can be maintained under a tensile strain of 40% was observed (Fig. 10c). In contrast, with n = 60 the copolymer was found to show WORM resistive switching behavior with a high ON/OFF ratio of ∼105 under the same tensile strain (Fig. 10d). Recently, the same group reported a novel carbohydrate-block-polyisoprene (MH-b-PIn) block copolymer-based memory device.99 Oligosaccharides were employed because the hydroxyl moiety could trap electrons and serve as the charge-trapping element while the PI in a block copolymer system accounts for the excellent stretchability. The flexibility of the as-prepared MH-b-PIn thin film can be effectively improved by increasing the flexible PI component. Especially in the case of MH-b-PI12.6k thin film, no obvious crack was observed even under 100% strain. As such, stable volatile resistive switching with an excellent ON/OFF ratio of >106 was successfully obtained under 0–100% strain.


image file: c8cs00614h-f10.tif
Fig. 10 (a) Chemical structure of PF14-b-Pison. (b) Device configuration of the PF14-b-Pison-based stretchable memory. I–V characteristics of (c) PF14-b-Piso10 and (d) PF14-b-Piso60 under 40% tensile strain after 200 stretch/release cycles. Reproduced with permission from ref. 98 under the Creative Commons.

Instead of designing electroactive polymers with a D–A structure, another effective approach to obtain reliable resistive switching behavior is to employ single-component insulating polymers as the information storage media in which local conducting filaments are formed. One type is associated with the conducting carbon-rich filaments formed by local thermal degradation of polymers.31–33 For example, Lee et al.33 recently reported the direct observation of carbon-rich filaments using TEM in the poly(ethylene glycol dimethacrylate) (pEGDMA) resistive switching layer sandwiched between Cu and ITO electrodes on a plastic PET substrate (Fig. 11a–c). The pEGDMA herein was deposited by an initiated chemical vapor deposition (iCVD) process, which is an all-dry vapor-phase technique for creating polymer films with high conformability, good purity, and also strong adhesion. With the mechanism of carbon-rich filaments, the Cu/pEGDMA/ITO device showed no degradation in the ON/OFF ratio after being bended to a small radius of 4 mm (Fig. 11d) or even after being submerged in water for 27 hours (Fig. 11e), demonstrating a great application potential of the device in the coming IoT era.


image file: c8cs00614h-f11.tif
Fig. 11 (a) Cross-sectional TEM image of the Cu/pEGDMA/ITO memory device. TEM Images of (b) a complete carbon-rich filament and (c) a ruptured one. (d) Bending performance and (e) water-resistant property of the device. Reproduced with permission from ref. 33, © 2015 American Chemical Society.

The other type is related to metal filaments that are formed through the diffusion of metal ions/atoms from electrochemically active electrodes such as Ag and Cu on an electric field.34–37 For example, direct observation of Ag filaments using TEM has been reported by Cho et al.34 in the Ag/WPF-BT-FEO/p-Si resistive switching memory. Later on, Gao et al.35,36 confirmed the existence of various growth modes of metal filaments in organic resistive switching memories and suggested a cation mobility-controlled filament growth model. Recently, Choi et al.100 reported a flexible Cu/poly(1,3,5-trimethyl-1,3,5-trivinyl cyclotrisiloxane) (pV3D3)/Al resistive switching device with the mechanism of Cu filaments on a plastic polyethersulfone (PES) substrate (Fig. 12a and b). The pV3D3 herein was deposited also by an iCVD process and was selected due to its outstanding chemical stability, which allows the Cu/pV3D3/Al devices to be fabricated at the microscale using conventional photolithography techniques. The obtained memory cells with a size of 5 × 5 μm2 in a 8 × 8 array show good device-to-device uniformity (Fig. 12c). The high ON/OFF ratio of >107 can be maintained until the radius of bending reaches 3.8 mm for both directions (Fig. 12d). Bending further to a 3 mm radius will cause device failure, which was found to be induced by crack formation in the bottom Al electrodes rather than in the pV3D3 layer. At the extreme bending radius of 3.8 mm, no degradation was found after 103 bending cycles in both directions (Fig. 12e), thus indicating the high potential of pV3D3 with a metal filament mechanism for flexible memory applications. More impressively, Huang et al. from Peking University101 developed a polychloro-para-xylylene (parylene-C)-based flexible resistive switching memory array with an Al filament mechanism, wherein the parylene-C is a Food and Drug Administration (FDA)-approved material safe for use within a human body and the device can be fabricated through a fully CMOS-compatible process. A high ON/OFF ratio of >104 and good bending endurance of >500 cycles with a 10 mm radius have been demonstrated. In addition, highly reproducible quantum conductance behavior has been very recently found in metal filaments of organic resistive switching devices,102,103 which may provide a new opportunity for developing multilevel nonvolatile memory devices.


image file: c8cs00614h-f12.tif
Fig. 12 (a) Schematic structure of the Cu/pV3D3/Al memory device with Cu filament mechanism. (b) Demonstration of filamentary mechanism by conductive atomic force microscopy (C-AFM). (c) Device-to-device uniformity, (d) bending performance, and (e) bending endurance at 3.8 mm radius of the device. Reproduced with permission from ref. 100, © 2016 American Chemical Society.

Besides normal single-component filaments, Huang et al.104 recently reported the observation of spliced filaments consisting of Al and oxygen vacancy dual conductive channels growing through carbazole groups in Al/PVK/ITO resistive switching devices (Fig. 13a–c). Based on such understanding, they further developed a low-temperature solution method to deposit porous PVK film with randomly distributed nano pores on the surface. That is, chloroform was used as the solvent for spin-coating the film in damp air with a humidity of 30–40%. These nano pores showed a cone shape with the tip pointing towards ITO but not through the whole PVK layer. As such, they were able to act as a template for formation of Al nano cones during Al electrode deposition (Fig. 13d). Originating from the confined formation of oxygen vacancy filaments under Al nano cones, the device with porous PVK film showed a forming-free resistive switching behavior with lower operation voltage and higher ON/OFF ratio than the reference planar device (Fig. 13e). Moreover, the porous device showed better endurance performance of over 45 successive cycles, possibly due to the good capability of heat dissipation of the porous structures.


image file: c8cs00614h-f13.tif
Fig. 13 (a) Cross-sectional TEM image, (b) I–V characteristic, and (c) switching mechanism of the planar Al/PVK/ITO memory device. (d) Surface morphology and (e) ON/OFF ratio of the porous Al/PVK/ITO memory device. Reproduced with permission from ref. 104, © 2017 Wiley-VCH.

Moreover, it is interesting as recently found by Chiolerio et al.,105 that a polyaniline (PANI)-based resistive switching device can exhibit simultaneously a programmable nanobattery effect. Herein, the PANI was obtained via oxidative polymerization of aniline using ferric chloride as the oxidizing agent. With Al electrodes, the device can output 110 mV after being programmed with a 7 V/10 s pulse, but it vanishes after a −7 V/10 s pulse. Such behavior is possibly caused by electric filed-induced charge accumulation and may find novel applications such as a programmable voltage regulator.

4.2 Polymer blends or mixtures with organic small molecules

Polymer blends or mixtures with small molecules, wherein partial or all the components are electroactive, can be employed as the active layer in resistive switching memories.9–11 The obtained polymer composites may benefit from advantages of both components and thus feature improved physical properties including electronic property, processability, and mechanical flexibility. Poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) is one representative where the addition of PSS acid into PEDOT can not only improve the conductivity of the conjugated PEDOT polymer by proton doping but also makes the PEDOT water-soluble. Many efforts have been paid to the use of PEDOT:PSS for resistive switching memory applications and two switching mechanisms have been confirmed, including a redox reaction of PEDOT:PSS and rearrangement of the PSS component.106–108 For example, Forreset et al. from Princeton University106 first reported a fused memory pixel consisting of a conductive PEDOT:PSS, layered onto the surface of a thin-film Si p–i–n diode deposited onto a stainless steel substrate. The reduction of PEDOT+ into PEDOT0 upon an electric field was confirmed to give rise to the WORM memory effect from low resistance state to high resistance state. In contrast, Alshareef et al.108 recently reported a flexible and transparent all-PEDOT:PSS resistive switching memory on plastic PET substrate, which consists of an as-received PEDOT:PSS thin film (conductivity: 1–3 S cm−1) as the active layer sandwiched between two highly conducting PEDOT:PSS (m-PEDOT, conductivity: >900 S cm−1) electrodes (Fig. 14a and b). The m-PEDOT was prepared through addition of 4 wt% dimethyl sulfoxide (DMSO), which caused the excess PSS to phase-segregate into PSS-rich domains while the PEDOT grains merged together to form an interconnected network of conducting fibers. When a sufficiently high electric field was applied, the device was irreversibly switched from the initial ON state to a final OFF state with a large ON/OFF ratio of >103 (Fig. 14c). Such transition was attributed to the aggregation of the insulating PSS chains near the electrode/polymer interface upon high current density. Moreover, a similar switching behavior was observed even after 3 months (Fig. 14c), indicating a high stability of the all-PEDOT:PSS memory device.
image file: c8cs00614h-f14.tif
Fig. 14 (a) Schematic structure and (b) transparent property of the all-PEDOT:PSS memory devices. I–V characteristics of the device (c) in fresh state and (d) after storage for 3 months. Reproduced with permission from ref. 108, © 2013 American Chemical Society.

Molecules with good semiconducting behavior such as organic dyes,109,110 polycyclic aromatic compounds,111 and oxadiazole derivatives56,57,112 can be selected as guest additives into polymers for generating resistive switching characteristics through donor–acceptor charge transfer. For example, Chen et al.111 reported two memory devices based on poly[4,4′-diamino-4′′-methyltriphenylamine-hexafluoroisopropylidene-diphthalimide] (PI(AMTPA)) blended with polycyclic aromatic compounds, namely, coronene and N,N-bis[4-(2-octyldodecyloxy)phenyl]-3,4,9,10-perylenetetracarboxylic diimide (PDI-DO) (Fig. 15a). The blend films were spin-coated with well-mixed solutions containing PI(AMTPA) and coronene or PDI-DO in tetrahydrofuran or chloroform. The calculated molecular energy levels indicate the feasibility of charge transfer between PI(AMTPA) and coronene or PDI-DO. As such, with an increase in loading ratio of coronene or PDI-DO, a variation in switching behavior from volatile to nonvolatile rewritable and finally to WORM was observed (Fig. 15b). Meanwhile, due to not only the close π–π interaction of heterocyclic aromatic PDI-DO structures but also the high electron affinity of the PDI-DO acceptors, the initiation of nonvolatile rewritable and WORM switching behaviors in the blend of PI(AMTPA):PDI-DO was found to occur at a lower loading ratio. Moreover, the nonvolatile rewritable switching behavior of PI(AMTPA):PDI-DO(3%) was able to be maintained even under bending with a 5 mm radius (Fig. 15c), thus highly possible for applications in future flexible electronics. Similarly, Wen et al.56,112 observed stable nonvolatile rewritable switching behaviors after adding 2-(4-tert-butylphenyl)-5-(4-biphenylyl)-1,3,4-oxadiazole (PBD) into electroactive polymers such as PVK and polyurethane, which were all caused by the electric field-induced charge transfer between the polymer donor and the PBD acceptor.


image file: c8cs00614h-f15.tif
Fig. 15 (a) Structure and involved molecules of the memory devices based on PI(AMTPA) blends. (b) Switching behavior variation of the memory devices. (c) Bending performance of the device with PI(AMTPA):PDI-DO(3%). Reproduced with permission from ref. 111, © 2013 American Chemical Society.

Besides acting as electron accepters, small molecules also have been demonstrated as feasible for tuning the conduction property of conjugated polymers by proton doping. To this point, Hu et al.113 designed a novel resistive switching composite of p-toluenesulfonic acid-doped polyazomethine (PA-TsOH, Fig. 16a). Sandwiched between two Pt electrodes, the composite exhibited highly uniform bipolar resistive switching behavior with an ON/OFF ratio of 600 and self-rectifying ratio of 10–100 (Fig. 16b). A small dispersive index of less than 5% for both the switching voltage and the ON/OFF ratio in over 1200 switching cycles was demonstrated. Also, the gradual reset process allowed the composite to exhibit highly stable multilevel storage behavior by setting various reset voltages (Fig. 16c). Such switching behavior was attributed to the controllable doping/de-doping of the PA by TsOH under electrical fields, which effectively tunes the energy band diagram and thus conductivity of the composite. These results suggest that electric-field-controllable doping of polymers is an effective approach for tuning the resistive-switching effect and provides an important opportunity for development of high-performance organic nonvolatile memory devices.


image file: c8cs00614h-f16.tif
Fig. 16 (a) Device structure, (b) I–V characteristic, and (c) multilevel switching behavior of the Pt/PA-TsOH/Pt memory device. Reproduced with permission from ref. 113, © 2012 American Chemical Society.

As introduced in the previous section, the oxidation of a polymer can give rise to resistive switching behavior in a single component polymer system.50,58 The positively charged constituent induced by the electric field is usually neutralized by oxygen from air and thus causes an unstable memory effect. To avoid this drawback, blends of polymers with redox-active molecules have been employed to construct a double-layer system with the redox-active polymer and to offer a redox counter-reaction. For instance, McCreery et al.59 reported a redox memory device with a “buried” three-electrode geometry, in which poly(3,3′′′-didodecylquaterthiophene) (PQT) and “ethyl viologen diperchlorate + polyethylene oxide” (EV(ClO4)2 + PEO) were employed as the functional layers. The EV(ClO4)2 + PEO layer could provide a redox counter-reaction to accompany PQT oxidation and also mobile ions to compensate the positive charge of the conducting polarons, thus leading to a highly enhanced electrical hysteresis. Inspired by this work, Liu et al.21 subsequently developed a EV(ClO4)2/BTPA-F bilayer redox memory with a simpler two-terminal structure (Fig. 17a and b), where the BTPA-F was a triphenylamine-containing polymer synthesized via a Suzuki coupling polymerization reaction. Interestingly, typical history-dependent memristive behavior was observed in the device, based on which a series of synaptic behaviors were successfully emulated, such as spike-rate-dependent plasticity and the “learning-forgetting-relearning” process (Fig. 17c and d). Recently, they further fabricated a EV(ClO4)2/TPA-PI (TPA-PI = triphenylamine-based polyimide) bilayer memristor on plastic PET substrate.114 Notably, it still exhibited synaptic potentiation and depression characteristics even after being bended 100 times with a bending radius of no less than 10 mm. These findings demonstrate the feasibility of in-memory-computing in organic resistive switching devices and thus provide a promising method for the development of high-efficiency neuromorphic computing systems that can resolve the long lasting von Neumann bottle problems.


image file: c8cs00614h-f17.tif
Fig. 17 (a) Schematic illustration of the Ta/EV(ClO4)2/BTPA-F/Pt memristor and the biological synapse. (b) Chemical structures of BTPA-F and EV(ClO4)2. Demonstration of (c) the spike-rate-dependent plasticity and (d) the “learning-forgetting-relearning” process. Reproduced with permission from ref. 21, © 2016 Wiley-VCH.

4.3 Polymer matrix for nanoparticles and inorganic compounds

Other than interacting with the guest components, polymer materials can also be used as an inert matrix for nanoparticles and inorganic compounds in resistive switching memories, which was initiated by Yang et al.46 in 2004 with the hope of developing a single-layer, solution-processible, and rewritable two-terminal memory device. They mixed 1-dodecanethiol-protected Au nanoparticles (Au–DT NPs) and 8-hydroxyquinoline (8HQ) in an insulating polystyrene thin film and sandwiched them between two Al electrodes to form the Al/Au–DT + 8HQ + PS/Al structure by spin-coating a 1,2-dichlorobenzenic solution of 0.4 wt% Au–DT NPs, 0.4 wt% 8HQ, and 1.2 wt% PS onto the substrate. The obtained device exhibited a stable bipolar resistive switching behavior with Vset of 2.7 V, Vreset of −1.8 V, and high ON/OFF ratio of >103, which was attributed to the electric-field-induced charge transfer between Au–DT NPs and 8HQ. Inspired by this work, a large number of polymer–inorganic composites have been subsequently explored for resistive switching memory applications, which can be classified into three groups based on the dimensionality of the included components.

Zero-dimensional (0D) nanoparticles (or quantum dots) form the first dopant groups for polymer–inorganic composites. Started with Au nanoparticles, plenty of metal and semiconducting nanoparticles have been explored as charge trapping centers, including Ag,42,115–119 fullerene and its derivatives,120–123 Cu2O,124 TiO2,125 ZnO,126,127 CdS,128etc. When being incorporated into insulating polymer matrixes such as polystyrene,123 polyimide,121 poly(vinyl alcohol) (PVA),129 polyvinylpyrrolidone (PVP),120 poly(ethylene-co-vinyl acetate) (PEVA)130 and poly(methyl methacrylate) (PMMA),42 these nanoparticles can trap the injected charge carriers from electrodes and thus lead to resistive switching behaviors. For example, using PMMA as insulating matrix and Ag nanoparticles as trapping centers, Kim et al.42 observed distinct and reliable bipolar resistive switching behavior with an ON/OFF ratio of >103 and switching endurance of >104 cycles in an Al/PMMA–Ag/ITO device. To develop flexible resistive switching memories, the famous fullerene derivative 6-phenyl-C61 butyric acid methyl ester (PCBM) has been frequently used as trapping centers. Lee et al.121 explored a mixture of polyimide (PI):PCBM, wherein PI was selected due to its thermal stability, good chemical resistance, and excellent mechanical properties. The fabricated Al/PI:PCBM/Al devices on a plastic PET substrate were found to exhibit stable unipolar resistive switching behavior with a high ON/OFF ratio of >104 not only in flat state but also under bending with 9 mm radius. On the other hand, Kim et al.122 incorporated PCBM into the nanostructure of self-assembled poly(styrene-b-methyl methacrylate) (PS10-b-PMMA130) diblock copolymer. The favorable miscibility of PCBM molecules with PS allows them to be confined to 10 nm PS nanospheres surrounded by a PMMA matrix and thus can prevent the aggregation of PCBM. The obtained PS10-b-PMMA130:PCBM nanocomposite exhibited stable bipolar resistive switching behavior with no significant degradation before and after bending. Recently, taking advantage of the excellent charge trapping property of PCBM, Ji et al.123 successfully demonstrated an ultraviolet-patterned and vertically stacked 64-bit one-diode-one-resistor (1D-1R) memory array on a plastic PEN substrate (Fig. 18a). In these devices, P3HT was selected for the active layer of diodes and a mixture of PS and PCBM was selected for the active layer of resistive switching memory cells. Stable unipolar resistive switching behavior with a high ON/OFF ratio of ∼103 and endurance of >100 cycles was confirmed in these devices in flat, bended, twisted, and even rolled conditions (Fig. 18b and c). Meanwhile, these devices could work normally with the high ON/OFF ratio of ∼103 until an extreme strain of 6.28%, thus being a highly promising flexible resistive switching memory candidate.


image file: c8cs00614h-f18.tif
Fig. 18 (a) Structure, (b) flexible performance, and (c) switching endurance of the PS:PCBM memory device with P3HT diode. Reproduced with permission from ref. 123, © 2013 Springer Nature.

Besides nanoparticles, quantum dots (QDs) of layered materials such as graphene,131–133 black phosphorus (BP),44,134 MoSe2,135 WS2135 and NbSe2135 have also been found feasible as charge trapping centers, wherein excellent switching performances were demonstrated in related devices. One representative work was done by Han et al.134 in which uniform and high-quality BP QDs with an average size of 2.7 nm were first synthesized via liquid exfoliation assisted by sonication and then used as charge trapping centers in an Al/PMMA/BP QDs/PMMA/Al device on a plastic PET substrate (Fig. 19a–d). The PMMA layers were deposited by spin-coating, whereas the BP QDs layer was obtained by drop-casting with a 2 min self-assembly process before blowing away the excessive suspension with a rubber suction bulb in one direction. Bipolar resistive switching behavior with an ultra-high ON/OFF ratio of >107 was observed (Fig. 19e). Reliable multilevel resistive switching behavior controlled by setting various compliance currents has also been confirmed (Fig. 19f). Moreover, even after being bent for 500 times with a 15 mm radius, almost no degradation was found in the ON/OFF ratio (Fig. 19g).


image file: c8cs00614h-f19.tif
Fig. 19 (a) Low-magnification TEM image, (b) statistical size analysis, and (c) high-resolution TEM image of the adopted BP QDs. (d) Schematic structure, (e) I–V characteristic, and (f) multilevel switching behavior of the Al/PMMA/BP QDs/PMMA/Al memory device. (g) Bending endurance of the device at 15 mm radius. Reproduced from ref. 134 under the Creative Commons License.

One-dimensional (1D) inorganic nanotubes or nanowires represent the second group of components used for the formation of polymer–inorganic composites.136–141 Particularly, a carbon nanotube (CNT) with excellent carrier mobility, mechanical flexibility, compatibility with solution process, and even a tunable electrical property, is the most representative and versatile inorganic 1D nano dopant. In 2009, Liu et al.136 reported the dependence of resistive switching behavior on CNT doping content in a PVK matrix. Within the Al/PVK–CNT/ITO device (Fig. 20a), the PVK–CNT composite layer was deposited by spin-coating a toluene solution of PVK (10 mg mL−1, containing 0–5% of CNT). It was found that doping PVK with 0.2% CNT induced only an increase in electrical conductivity (Fig. 20b). This is because the large separation between CNTs can prevent the charge carriers from inter-CNT hopping. After the CNT content was increased to 1%, the device exhibited a remarkable WORM characteristic with an ON/OFF ratio of >104 (Fig. 20c). Further, increasing the CNT content to 2% resulted in stable bipolar switching behavior with an ON/OFF ratio of >103 (Fig. 20d). In these conditions, the CNT contents are large enough to ensure inter-CNT hopping of trapped carriers, and thus the device was able to be switched from the OFF to ON state. Under a reverse bias, the applied electric field in the device with 1% CNT was opposed by the build-in electric field associated with the space-charge layer in PVK, which prevented the trapped electrons from being neutralized or extracted, i.e., WORM switching behavior. In contrast, for the device with 2% CNT, some CNTs can come into contact with the electrode, thus eliminating the space-charge layer and making the device feasible to be switched back into its OFF state, i.e., rewritable bipolar switching behavior. However, if the CNT doping content was too large, such as 3%, then continuous π-conjugated networks were formed in the film due to the close stacking of CNTs, leading to only a single conducting state in the device (Fig. 20e). A similar dependence trend of the resistive switching behavior on dopant content was also reported by Pandurangan et al.139 in a Al/PVA–CNT/p-Si memory device.


image file: c8cs00614h-f20.tif
Fig. 20 (a) Schematic structure of the Al/PVK–CNT/ITO memory device and its I–V characteristics with (b) 0.2%, (c) 1%, (d) 2%, and (e) 3% CNT. Reproduced with permission from ref. 136, © 2009 American Chemical Society (a–e).

On the other hand, chemical doping of CNT has been proven as an effective route to modulate the resistive switching behavior of polymer–CNT composite memory devices. For instance, Kim et al.137 observed significant enhancement in the ON/OFF ratio and retention property of the Al/PS–CNT/Al resistive switching device after replacing undoped CNTs by B- or N-doped ones (BCNTs and NCNTs), which was attributed to the higher dispersibility and resultant better charge trapping capacity of doped CNTs. More interestingly, by the use of a stacked PS–NCNT/PS–BCNT bilayer, the device was found to exhibit a tri-level resistive switching behavior with stable data retention property for each state. This phenomenon was explained by the coexistence of two distinct charge trapping levels from BCNTs and NCNTs and has great potential for high-density storage applications. Ag and ZnO nanowires have also been explored for polymer/inorganics composite memory applications, with Ag nanowires contributing to the resistive switching behavior by forming silver filaments between adjacent nanowire clusters whereas ZnO nanowires acted as charge trapping centers like CNTs.140,141

Polymers mixed by two-dimensional (2D) inorganic nanosheets (e.g., graphene and MoS2) are the third group of polymer–inorganic resistive switching composites.142–146 Both metallic graphene and semiconducting MoS2 nanosheets can act as trapping centers in a polymer matrix. In 2012, Shang et al.142 reported a resistive switching memory with PVK as matrix and graphene nanosheets as electron trapping centers, which was found to exhibit WORM and bipolar resistive switching behaviors with 2 and 4 wt% graphene nanosheets, respectively. Compared to metallic graphene, semiconducting MoS2 has been considered more suitable as trapping centers due to its appropriate quantum confinement and available energy states. To demonstrate this, Choi et al. from Jeju National University145 fabricated a resistive switching memory with a MoS2 nanosheet-doped PVA as the active layer and Ag as both top and bottom electrodes, i.e., Ag/PVA–MoS2/Ag, on a plastic PET substrate (Fig. 21a). PVA was selected due to its branched structure which is highly suitable for coating a 2D material to form uniform thin films. PVA is also cheap, non-toxic, non-hazardous, biocompatible, and environmental-friendly. The composite layer was deposited by electrohydrodynamic atomization. Under flat state, the device was found to exhibit a bipolar resistive switching behavior with Vset of 2.5 V, Vreset of −2.6 V, and a high ON/OFF ratio of >102 (Fig. 21b). Good switching endurance of >1000 cycles and stable data retention of >105 s had been confirmed. Even after being bent for up to 1500 times with a small radius of 5 mm, both ON and OFF states remained unaffected (Fig. 21c). The extreme bending radius of the device is as small as ∼2 mm and further reducing the bending radius will result in an open circuit due to the formation of obvious cracks in the Ag electrodes (Fig. 21d). These demonstrated good reliability and flexibility of the Ag/PVA–MoS2/Ag resistive switching memory on the PET substrate.


image file: c8cs00614h-f21.tif
Fig. 21 (a) Schematic structure, (b) I–V characteristic, (c) bending endurance at 5 mm radius, and (d) bending performance of the Ag/PVA–MoS2/Ag memory device. Reproduced with permission from ref. 145 under the Creative Commons License.

Depending on the arrangement of its S atoms, single-layered MoS2 has two phases at room temperature, namely, the stable semiconducting 2H (trigonal prismatic D3h) phase and the metastable metallic 1T (octahedral Oh) phase. It was observed that by using MoS2 with different phases, the resistive switching behavior transited from bipolar in the Al/PVA–2H-MoS2/ITO device to WORM in the Al/PVA–1T@2H-MoS2/ITO device (Fig. 22).147 The 1T@2H-MoS2 nanosheets herein refer to 2H-MoS2 nanosheets with small 1T regions, which were obtained by dispersing 2H-MoS2 nanosheets in ethanol solution and then maintaining at 220 °C for 8 h in a Teflon-lined stainless autoclave. The transition in switching behavior was attributed to the higher charge trapping capacity of 1T@2H-MoS2 nanosheets with abundant S vacancies. Recently, hexagonal boron nitride (hBN) nanosheets have also been explored as dopants in a polymer–inorganics composite for memory applications.146 The Ag/polyvinyl alcohol (PVOH)–hBN/ITO sandwich-structured device showed typical bipolar resistive switching behavior with a high ON/OFF ratio of >103 and good switching endurance of >1000 cycles. Possibly due to the insulating nature of hBN nanosheets, the as observed resistive switching behavior in the PVOH–hBN composite was attributed to a cation migration-based conductive filament mechanism, rather than charge trapping-induced resistive switching in graphene or MoS2 nanosheets containing polymer matrix.


image file: c8cs00614h-f22.tif
Fig. 22 I–V characteristic of the Al/PVK-MoS2/ITO memory device with (a) 2H-MoS2 and (b) 1T@2H-MoS2 nanosheets. Reproduced with permission from ref. 147, © 2016 Wiley-VCH.

Although the polymer blends, mixtures with organic small molecules, and composites with nanoparticles and/or inorganic compounds provide a simple yet useful strategy of modulating the electrical characteristics and memory performance of resistive switching memories, it should be pointed out that doping or mixing of polymers with guest components may not always give rise to uniform dispersion of the constituting material. The resultant phase separation and/or ion aggregation are unfavorable to device performance, and thus special care should be paid to controlling the composition and microstructure of the composite switching layer. Chemical modifications to the dopants may be an effective approach to improve compatibility between the polymers and doping agents with different chemical natures and geometrical dimensionality.

5. Graphene oxide switching materials

In addition to the 0D fullerenes and 1D carbon nanotubes, graphene is another elemental carbon allotrope with sp2 configuration of a honeycomb crystal lattice in a single-layer sheet and has been extensively studied for electronic, optoelectronic, and spintronic devices. Its 2D single-layer nature also makes graphene an ideal material for flexible applications. Generally, graphene material can be generated in large scale by electrochemical exfoliation of graphite in the form of graphene oxide (GO), which is decorated with various oxygen containing hydroxyl and epoxy groups on the basal plane and meanwhile smaller amounts of carboxy, carbonyl, etc. at the sheet edges.148,149 It has been demonstrated that the electric property of GOs can be easily and remarkably tuned by changing its oxygen content. The presence of polar oxygen functional groups in GOs nevertheless allows the preparation of uniform GO film through solution processing such as drop-casting, spin-coating, Langmuir–Blodgett (LB) deposition, and vacuum filtration.

In 2009, He et al.150 for the first time demonstrated GO-based resistive switching memory with a Cu/GO/Pt sandwich structure, where the GO film was ∼30 nm in thickness as deposited by a vacuum filtration method. When external voltage was applied to the top Cu electrode, the device exhibited bipolar resistive switching behavior with Vset of 0.8 V, Vreset of −0.75 V, and an ON/OFF ratio of ∼20 (Fig. 23a). A forming process was not required to active the GO device, which is favorable for practical applications with low power consumption and simplified circuit structure. Acceptable fluctuations in critical switching parameters were observed during continuous 100 switching cycles or 104 s continuous read operation. Also, multilevel switching behavior can be realized by setting various compliance currents during the set process (Fig. 23b).151 Temperature-dependent electrical measurements suggested that the LRS device showed metallic conduction behavior (e.g., device resistance increased with temperature). The temperature coefficient of resistance α at 300 K was calculated to be 1.7 × 10−3 K−1, which is close to that of high-purity Cu nanowires and thus confirms the formation of Cu filaments via the diffusion of top electrode accounts for the observed resistive switching in GO devices.151


image file: c8cs00614h-f23.tif
Fig. 23 (a) Bistable and (b) multi-level resistive switching behaviors of the Cu/GO/Pt memory device. Reproduced with permission from ref. 150, © 2009 AIP Publishing (a) and from ref. 151, © 2011 Elsevier (b).

To develop low-cost, large-area and low-power flexible nonvolatile memory, Jeong et al.152 fabricated a GO-based resistive switching memory with an Al/GO/Al sandwich structure on a plastic PES substrate (Fig. 24a). The GO film herein was ∼15 nm in thickness and deposited by a simple and scalable spin-casting process. This procedure was intentionally adopted because it can ensure large-area uniform GO films and is also feasible for device integration in conjunction with a standard CMOS process. Under flat condition with external voltage applied on the top Al electrode, the device showed bipolar resistive switching behavior with Vset of ∼−3 V, Vreset of ∼2 V, and a high ON/OFF ratio of >102 (Fig. 24b). No noticeable degradation in the ON/OFF ratio was observed after the device was bent up to 103 times (Fig. 24c) or down to a small radius of 8 mm (Fig. 24d). With the help of high-resolution spherical aberration-corrected TEM observation, the resistive switching behavior was attributed to the electric field-induced reversible oxygen exchange between the top interface layer and the GO layer (Fig. 24e), which modified the interfacial AlOx layer thickness and height of insulating barrier for charge carrier injection.


image file: c8cs00614h-f24.tif
Fig. 24 (a) Schematic structure, (b) IV characteristic, (c and d) bending performance, and (e) switching mechanism of the Al/GO/Al memory device. Reproduced with permission from ref. 152, © 2010 American Chemical Society.

Full solution-processed all-GO flexible resistive switching memory was also demonstrated by Zhang et al. from Nanyang Technological University153 with the device structure of hrGO/lrGO/hrGO on PET substrate (Fig. 25a). The conductive highly reduced graphene oxide (hrGO) was obtained by high-temperature annealing of the normally insulating GO nanosheets at 1000 °C and used as electrodes, while the lightly reduced graphene oxide (lrGO) storage medium was made by low-temperature annealing and UV irradiation of GO nanosheets. The as-fabricated hrGO/lrGO/hrGO device demonstrated a nonvolatile WORM memory behavior with the writing voltage of ∼−13 V and a moderate ON/OFF ratio of >102 (Fig. 25b), possibly arising from the irreversible migration of oxygen in the lrGO active layer and the use of relatively inert hrGO electrode. Both ON and OFF states showed acceptable variation in resistance under a test period of 103 s at ambient condition (Fig. 25c) and can be maintained during continuous 103 bending cycles at a small radius of 5 mm (tensile strain: ∼2.9%, Fig. 25d). By using carbon nanotube fibers as electrodes, they further found that the ON/OFF ratio of GO-based WORM memory can be increased by one order of magnitude.154


image file: c8cs00614h-f25.tif
Fig. 25 (a) Schematic structure, (b) IV characteristic, (c) retention property, and (d) bending endurance at a small radius of 5 mm (tensile strain: ∼2.9%) of the hrGO/lrGO/hrGO memory device. Reproduced with permission from ref. 153, © 2013 Wiley-VCH.

Given the fact that GO contains many carbonyl and carboxyl groups at the plane edge, the graft of polymer onto it is certainly a feasible way to enhance its solution dispersibility and processability, as well as to modulate its resistive switching behavior. To demonstrate this, Kang et al. from National University of Singapore155–160 synthesized a series of GO–polymer nanocomposites through a grafting method (Fig. 26a). The adopted electroactive polymers include PVK, PANI, polythiophene (PTh), triphenylamine-based polyazomethine (TPAPAM), etc. As expected, highly improved dispersibility was confirmed in GO–polymer nanocomposites, making them easy for solution processing (Fig. 26b). All the resultant Al/GO–polymer/ITO sandwich-structured devices with spin-coated GO–polymer layers were found to show good memory behaviors. For example, a typical bipolar resistive switching behavior with a high ON/OFF ratio of >103 was observed in GO–PVK (Fig. 26c).155 Such behavior has actually combined the rewritable feature of GO memory devices with the metal filament mechanism and the high ON/OFF ratio of GO-based WORM memory with carbon nanotube fiber electrodes. Also, both memory states remained almost intact for at least more than 3 hours or 100 million read cycles. Assisted by theoretical simulation, the observed switching behavior was reasonably attributed to the electric field-induced reversible charge transfer between polymer donor and GO acceptor (Fig. 26d). These results can demonstrate the great potential of GO–polymer nanocomposites for nonvolatile memory applications, with their flexibility remaining to be explored in the future.


image file: c8cs00614h-f26.tif
Fig. 26 (a) Chemical structures of GO–PANI, GO–PVK, GO–TPAPAM, and GO–PTh nanocomposites. (b) Images of GO and GO–PTh nanocomposite dispersions in DMF. From left to right: GO and GO–PTh samples prepared at 70 °C over 24 h, 12 h, 4 h, and prepared at 50 °C and room temperature over 24 h, respectively. (c and d) Memory behavior and switching mechanism of the GO–polymer nanocomposites. Reproduced with permission from ref. 159, Copyright (2014) Wiley-VCH (a, b and d) and from ref. 155, © 2009 AIP Publishing (c).

6. Organic–inorganic hybrid resistive switching materials

Although great efforts have been devoted so far to the development of soft organic materials for flexible electronics, their poor resistance to changing environments of atmospheric moisture and oxygen severely deteriorates device stability and reliability and hinders direct applications at the moment. On the other hand, inorganic materials usually demonstrate established electronic properties yet lack adaptability to large strains or deformations. Therefore, resistive switching memories that juggle both electrical and mechanical stability seem a technical dilemma for material scientists and electrical engineers. One possible solution to this issue is the employment of organic–inorganic hybrid materials, such as metal–organic frameworks (MOFs)161 and organic–inorganic hybrid perovskites (perovskites for short),162 which inherit the unique features of both the parenting inorganic and organic building blocks. With this concern, much attention has recently been addressed to the design and synthesis of resistance switchable MOF and perovskite materials, with the hope to implement nonvolatile memory devices with superior electromechanical performance.

6.1 Resistance switchable metal–organic frameworks

MOFs are a unique class of hybrid crystalline porous material with supermolecular structures composed of metal cations (or clusters) and multitopic soft organic bridging ligands, which demonstrate great versatility in their physicochemical properties and three dimensional periodicity in functionalities achieved through mature and rich coordination synthesis chemistry.161,163 The synergetic interplay between the metal nodes and the organic linkers, as well as the host–guest interactions between the absorbed guest molecules and solid framework along the one, two, or three dimensional nanopores also endow deliberate tuning of electronic and magnetic properties of the MOFs. Due to the presence of soft organic linkers, their moderate coordinating bond strength with the metal species, and thus the signature deformable nanoporosity, MOF materials provide an alternative opportunity for mounting intrinsically flexible electronic and spintronic devices.

Initial attempts to develop MOF-based resistive switching memories were conducted individually yet simultaneously by Grzybowski et al. at Northwestern University40 and Pan et al. at the Chinese Academy of Sciences.164 Nevertheless, they were running at completely different tracks. For instance, the US group used a cubic-shaped millimeter size Rb-CD-MOF bulk crystal as the switching material (Fig. 27a), wherein the electric field-induced self-limiting oxidation of the anodic Ag metal can lead to a clear hysteresis in the I–V loops (Fig. 27b). The as observed resistive switching characteristics are exotic to the use of MOF materials, and the conductance of the ON state decayed quickly with time (Fig. 27c). On the other hand, Pan et al. utilized a micrometer scale hexagonal prism RSMOF-1 crystal with the chemical formula [InC16H11N2O8]·1.5H2O for resistive switching memory applications (Fig. 27d). The W/RSMOF-1/Pt structure device exhibited a highly uniform resistive switching behavior with stable data retention over 6000 s (Fig. 27e and f). Further electrical measurements and first principle molecular dynamics simulation revealed that the observed resistive switching behavior originated from the ferroelectric transition of the N⋯H–O⋯H–N bridge-structured dipoles of the guest water molecules and the amino-tethered MOF nanochannel.


image file: c8cs00614h-f27.tif
Fig. 27 Chemical structures, I–V characteristics, and retention properties of (a–c) Rb-CD-MOF and (d–f) RSMOF-1 bulk materials. Reproduced with permission from ref. 40, © 2014 Wiley-VCH (a–c) and from ref. 164, © 2014 American Chemical Society (d–f).

The above results preliminarily confirmed the possibility of using organic–inorganic hybrids to build resistive switching memories, although bulk crystals are difficult to be integrated into thin film devices. To solve this issue, Pan et al.165 further extended their interest to second generation MOF thin films. They developed a modified laboratory-scale liquid-phase epitaxy (LPE) facility, through which high quality thin MOF films of HKUST-1 ((Cu3(BTC)2), BTC = benzene-1,3,5-tricarboxylic acid, Fig. 28a) with thickness of ∼130 nm and roughness of ∼4 nm was successfully prepared on Cu-coated PET substrates. The entire synthetic process was conducted in a N2 atmosphere continuously without being exposed to ambient conditions, thus avoiding possible contamination from air. After covering the top Au electrodes, resistive switching devices with an Au/HKUST-1/Au sandwich structure were obtained, which demonstrated a very stable bipolar resistive switching behavior during a dynamic bending process (Fig. 28b) or even being bent to a small radius of 3.2 mm that corresponded to a tensile strain of 2.8% (Fig. 28c). Under the strain level of 2.0%, these devices showed a highly uniform bipolar resistive switching behavior during continuous 300 operation cycles (Fig. 28d) or 160 bending cycles (Fig. 28e), with excellent endurance >107 cycles (Fig. 28f) and stable data retention of >104 s (Fig. 28g), demonstrating the great application potential of MOF resistive switching memory in future flexible electronics. Through conductive atomic force microscopic (C-AFM) and depth-profiling X-ray photoelectron spectroscopic (XPS) analysis, the authors suggested that electric field-induced migration of the Cu2+ ions, which may lead to subsequent pyrolysis of the trimesic acid linkers and thus the formation of highly conducting filaments, could be the possible origin for the observed uniform resistance switching in HKUST-1 nanofilms (Fig. 28h). Lee et al.166 recently reported a flexible MOF-based resistive switching memory with an Al/ZIF-8/Au sandwich structure on a PET substrate, where the ZIF-8 film was deposited by dip-coating and showed stable bipolar resistive switching behavior with a high ON/OFF ratio of >103 that was maintained for 100 bending cycles at a bending radius of 1.35 cm.


image file: c8cs00614h-f28.tif
Fig. 28 (a) Chemical structure of HKUST-1. (b) I–V characteristic under dynamic bending and (c) static bending performance of the Au/HKUST-1/Au memory device. (d) I–V characteristic, (e) bending endurance, (f) switching endurance, and (g) retention property of the device under bending with 4.5 mm radius. (h) Schematic switching mechanism of the device. Reproduced with permission from ref. 163, © 1999 AAAS (a) and from ref. 165, © 2015 Wiley-VCH (b–h).

With their intrinsic high porosity, MOFs are also feasible for constructing novel memory devices with the potential to overcome the gap between chemical information and electrical properties. As a proof of concept, Chen et al.167 designed and fabricated a MOF-based resistive switching memory with an Ag/ZIF-8/Si sandwich structure (Fig. 29a). The ZIF-8 film was deposited by immersing the Si substrate in a fresh mixture of 5 mL Zn(NO3)2 stock solution and 5 mL 2-methylimidazole stock solution for 30 minutes at room temperature, followed by washing with methanol and then drying with a nitrogen flow. Under ambient conditions with external voltage applied on the top Ag electrode, a nonvolatile bipolar resistive switching behavior with a high ON/OFF ratio of >106 was observed, as depicted by the solid IV curve in Fig. 29b. This behavior was explained by the combined action of the formation of Ag nanoparticles in the insulator layer through electro-migration and the electron hopping effect in between the nanoparticles that leads to LRS, which would be fractured by application of a reversed electric field in the reset process, causing the resistance state of the memory device to switch back to HRS. Interestingly, after the device was put in saturated methanol vapor conditions, its HRS resistance decreased dramatically and thus resulted in a much lower ON/OFF ratio of ∼104, as shown by the dotted IV curve in Fig. 29b. This phenomenon was highly reversible and robust (Fig. 29c) and confirmed the great potential of MOF-based resistive switching memory for achieving chemically mediated multilevel and environment-responsive information storage. Molecular dynamics simulations revealed that the ordered packing mode and the hydrogen bonding system of the guest molecules adsorbed in MOF crystals accounted for the observed switching characteristics in ZIF-8 thin film devices (Fig. 29d).


image file: c8cs00614h-f29.tif
Fig. 29 (a) Schematic structure of the Ag/ZIF-8/Si memory device. (b) I–V characteristics of the device in air and saturated methanol vapor. (c) Repeated cycles of the device in air and saturated methanol vapor. (d) Schematic mechanism of the alcohol-mediated HRS resistance. Reproduced with permission from ref. 167, © 2016 Wiley-VCH.

6.2 Organic–inorganic hybrid perovskites

Organic–inorganic hybrid perovskites MAPbX3 (MA = CH3NH3+; X = I, Br, Cl) are comprised of an extended framework of corner-sharing PbX6 octahedra with the methylammonium cation CH3NH3+ occupying the central A site and surrounded by 12 nearest-neighbour halide ions (Fig. 30a).168 These materials have strong optical absorption, tunable band gap, ambipolar charge transport, and long electron–hole diffusion length, and thus are suitable for various applications, for example, solar cells, light-emitting diodes, and thin-film transistors.162 Recently, they have also been explored for nonvolatile resistive switching memory applications.
image file: c8cs00614h-f30.tif
Fig. 30 (a) Chemical structure of the MAPbX3. (b) I–V characteristic of the Au/CH3NH3PbI3−xClx/FTO memory device. (c and d) I–V characteristics under various bending conditions and switching mechanism of the Au/CH3NH3PbI3/ITO memory device. Reproduced with permission from ref. 168, © 2016 Wiley-VCH (a), from ref. 169, © 2015 Wiley-VCH (b), and from ref. 170, © 2016 American Chemical Society (c and d).

In 2015, Choi et al. from Sejong University169 reported the first MAPbX3-based resistive switching memory with an Au/CH3NH3PbI3−xClx/FTO sandwich structure, which exhibited bipolar resistive switching behavior with an ON/OFF ratio of >3 over 100 operation cycles (Fig. 30b). Later on, given the easy solution-processable feature of MAPbX3 with high crystallinity at a low temperature of ∼100 °C, Lee et al.170 explored the possibility of using MAPbX3 for flexible nonvolatile memory applications by fabricating resistive switching devices with an Au/CH3NH3PbI3/ITO sandwich structure on a plastic PET substrate. The CH3NH3PbI3 herein was ∼270 nm in thickness and deposited by spin-coating, during which toluene was quickly dropped onto the center of the film to increase its uniformity. Under flat condition with external voltage applied to the top Au electrode, the Au/CH3NH3PbI3/ITO device showed bipolar resistive switching behavior with an operation voltage of <1 V and ON/OFF ratio of >10, as shown by the black curve in Fig. 30c. More importantly, almost no degradation in switching behavior was observed when the device was under tensile or compressive bending with radius of 15 mm (the blue and red curves in Fig. 30c). Also, both the HRS and LRS were able to sustain continuous bending of over 100 cycles. These results indicate that MAPbX3-based resistive switching memory has good electrical reliability as well as mechanical stability, thus is very promising for applications in future flexible electronics. Based on a detailed conduction property analysis and the easy migration of iodide ions in MAPbX3, the authors suggested that the observed resistive switching behavior was caused by the reversible formation and rupture of iodide vacancy filaments (Fig. 30d). Such a switching mechanism accords well with the observed low operation voltages and also has been solidly confirmed via energy-dispersive X-ray spectroscopy (EDX) analysis in a recent work by Zhu et al.39 Moreover, through intentional addition of hydroiodic acid in a precursor solution of perovskite, the CH3NH3PbI3 film was found to exhibit highly reduced grain size and surface roughness, which enabled the flexible Ag/CH3NH3PbI3/Pt memory device to work under bending with a small radius of 5 mm.171

In order to fabricate high-density and large-area MAPbX3-based memory for practical applications, vapor-based CMOS-compatible deposition of MAPbX3 films is highly desired and was preliminarily explored by Lee et al.172 on wafers perforated with 250 nm via-holes (Fig. 31a and b). The CH3NH3PbI3 resistive switching layer of the Au/CH3NH3PbI3/Pt memory device was deposited by sequential deposition of the PbI2 and CH3NH3I constituents via sublimation into the via-holes pre-defined by SiO2 on a Pt-coated silicon wafer, until the film color changed from yellow to brown which indicated complete formation of a CH3NH3PbI3 layer. With external voltage applied to the top Au electrode, the fabricated nanoscale Au/CH3NH3PbI3/Pt devices exhibited bipolar resistive switching behavior with a Vset of 1 V, Vreset of −1 V, and a high ON/OFF ratio of >103 (Fig. 31c). No degradation in ON/OFF ratio was observed during continuous 500 switching cycles or in a retention test period of >105 s. Meanwhile, a fast pulse operation of <200 ns was successfully demonstrated (Fig. 31d and e). Moreover, clear resistive switching behavior also was observed when the vapor-deposited CH3NH3PbI3 layer was integrated into a 16 × 16 cross-point memory array with line width of 300 μm.


image file: c8cs00614h-f31.tif
Fig. 31 (a and b) Fabrication procedure and schematic structure of the Au/CH3NH3PbI3/Pt memory device with 250 nm via-hole architecture via vapor method. (c) I–V characteristic and (d and e) pulse operations of the device. Reproduced with permission from ref. 172, © 2017 Wiley-VCH.

Given the outstanding photoresponsive characteristic of organic–inorganic perovskite materials, many efforts have been made to develop MAPbX3-based optoelectronic memories.173,174 Theoretically, such devices can integrate light sensing, data storage, and data processing functions into one single device, and thus will exhibit a much wider application scope. To this point, Chai et al.174 designed a vertical Au/CH3NH3PbI3−xClx/FTO device and found that its writing voltage could be greatly lowered under light illumination (Fig. 32a and b). They attribute the observed lowering of switching voltage to light-assisted hole injection into the CH3NH3PbI3−xClx/Au interfacial hole trapping centers. By defining light and voltage pulses with proper intensity and duration as the optical and electrical logic input “1” and “0”, nonvolatile AND and OR logic was successfully realized with HRS and LRS of the device defined as output logic “0” and output logic “1” (Fig. 32c and d). Optically modulated volatile resistive switching behavior was also observed by Zhu et al.173 in Ag/CH3NH3PbI3/Ag devices (Fig. 32e and f). With an increase in the optical illumination intensity, switching the device to LRS became harder during the forward positive voltage sweep, while the restoration of the device to HRS during the backward voltage sweep became easier. This phenomenon was attributed to the optical inhibition of the electric field-induced generation of iodine vacancies as well as subsequent facilitation in their annihilation, which could be used to emulate the optogenetic processes in biological synapses. As illustrated in Fig. 32g, synaptic plasticity polarities of the device were successfully changed after the introduction of light illumination, suggesting that MAPbX3-based optoelectronic memories are promising building elements for neuromorphic computing applications.


image file: c8cs00614h-f32.tif
Fig. 32 (a) Schematic structure, (b) I–V characteristic, and (c and d) nonvolatile logic applications of the vertical Au/CH3NH3PbI3−xClx/FTO optoelectronic memory device. (e) Schematic structure, (f) I–V characteristic, and (g) light-induced change in synaptic functions of the planar Ag/CH3NH3PbI3/Ag optoelectronic memory device. Reproduced with permission from ref. 174, © 2018 Wiley-VCH (a–d) and ref. 173, © 2018 American Chemical Society (e–g).

7. Conclusions and outlook remarks

As a rule of thumb that describes the revolutionary evolution of the first crude home computer made during the 1970s into the sophisticated machines of the 1990s decade, as well as the invention of the 21st century high speed internet, smartphones and auto-pilot mobile cars, Moore's Law precisely predicted the performance-guided development of the microelectronic industry for over fifty years.175 Yet not for much longer after the new millennium arrived, the device dimension campaign started to falter. The downscaling of transistors beyond the 2–3 nanometer limit will not be possible with silicon when quantum effects (e.g., quantum uncertainties) become dominant. The severe leaking of the stored charges through thinner gate insulators also makes transistors hopelessly unreliable. When more and more electronic components are jammed into a single piece of integrated circuits of the same small area and the electrons move faster than ever before between the separated memory and processor, the large amount of heat generated on microchips will make mobile phones too hot for consumers to sustain, in addition to major refinements in photolithography with more complexity and higher prices of the final products. Obviously, recently initialized low-power computation-in-memory devices and neuromorphic paradigms may greatly help solve Moore's Law limitation and von Neumann bottle problems.176 In 2016, the eighteen years old International Technology Roadmap for Semiconductors (ITRS) was for the first time renamed as International Roadmap for Devices and Systems (IRDS),177 with the hope of converting industrial attention from performance-guided innovations into an application-centered revolution. Facing the coming IoT century, flexible, low-power, and multifunctional devices will be future fashions for customized consumer electronics.

Resistive switching memories, which normally utilize atoms or ions to memorize digital data, embrace a completely new idea for information storage that was previously dominated by charge-based technologies. This may not only overcome quantum uncertainty and leaking problems during the dimension shrinking, but also engenders the possibility of working as a universal memory to replace the entire memory hierarchy with ultrafast, ultrahigh-density, and nonvolatile storage via multilevel switching55,168 or even conductance quantization.103,178,179 The recent extension of resistive switching memory to a widened definition of memristor nevertheless enables the realization of long-time dreaming computation-in-memory algorithm,19,22,180 which provides a promising solution to the von Neumann bottle neck of computation efficiency and power consumption.

Over the past few decades, great achievements have been made with flexible organic electronic devices, ranging from transistors and photovoltaics (PVs) to light emitting diodes (LEDs). Commercial products such as OLED, PLED, and QLED televisions with a curved screen are also available on the market nowadays. Plenty of experiences gathered during this success, including the rational design and synthesis of high-performance electroactive materials and their underlying electronic processes, can be adopted to the development of flexible resistive switching memories. In particular, both OPV and OLED are working on the external field-induced manipulation (separation, migration and recombination) of free charge carriers in electron donor–acceptor (D–A) systems. Based on this idea, the pioneers demonstrated at an early stage organic and polymer resistive switching memories with a charge-transfer mechanism,46,47 wherein the field-induced separation of electrons and holes can greatly enhance the charge carrier concentration inside the organic layer and thus overall conductance of the memory device. However, due to the lack of direct physical evidence to confirm that the charge transfer state can last for hours, days, and even years (transient spectral measurements on an OPV device usually demonstrate ∼μs lifetime for CT complex), suspicions are rising that the previously observed resistive switching in organic D–A systems may be results of artificial effects such as the involvement of chemically active metal electrodes. As such, people turn to focus on other established mechanisms of redox reactions, ion migration, etc. which can be visualized directly by in situ fluorescence, XPS, and high-resolution TEM measurements.17,34,165 Benefiting from the deep understanding of these switching mechanisms and the discovery of novel functional materials by rational molecular design with clearer structure–property relationships, rapid progress has been made in organic resistive switching memories with versatile selections of small molecules, polymers, macromolecular biomaterials, and carbon nanomaterials.

In comparison to their inorganic counterparts, the stability of organic memory devices is readily influenced by environmental moisture and oxygen gas. Continuous electrical stressing also fatigues organic devices quickly. There is still a long way for them to go before being practically used. Hybridizing the organic and inorganic species may solve this problem, wherein metal–organic frameworks and perovskite materials are already confirmed to be capable of exhibiting stable resistive switching characteristics under mechanical deformation.165,171 Two-dimensionalization of these hybrid materials into monolayer or few-layer single crystals can further attenuate the influence of a sample's thickness on stress transfer and thus significantly improve their mechanical flexibility and deformability. Combining the intrinsic stretchability and even twistability through an alterable coordination bond angle with a tunable electronic structure through molecular design, organic–inorganic hybrid materials prove to be outstanding candidates for soft memory devices.

Emergent electronic and optoelectronic devices based on organic and hybrid materials have advanced rapidly over the past half century, which greatly ameliorate the daily life of human beings all around the world. Comparatively, the development of organic and hybrid memory devices did not get enough attention nor formed into a complete set of academic system until now. This situation, nonetheless, grants chemists, materials scientists, and electrical engineers’ infinite possibilities in the upcoming epoch of artificial intelligence with organic and hybrid electronic materials.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors acknowledge the financial supports from the National Key R&D Program of China (2017YFB0405604), National Natural Science Foundation of China (61704178, 61722407, 61774161, 61674153, 61504154, 11474295 and 51525103), Natural Science Foundation of Zhejiang Province (LR17E020001), Public Welfare Technical Applied Research Project of Zhejiang Province (2017C31100), Ningbo Science and Technology Innovation Team (2015B11001), and Ningbo Natural Science Foundation (2017A610093).

References

  1. C. L. Philip Chen and C.-Y. Zhang, Inf. Sci., 2014, 275, 314–347 CrossRef .
  2. J. Gantz and D. Reinsel, https://www.emc-technology.com/collateral/analyst-reports/idc-the-digital-universe-in-2020.pdf, accessed 25th July 2018.
  3. G. I. Meijer, Science, 2008, 319, 1625–1626 CrossRef CAS PubMed .
  4. H. S. P. Wong and S. Salahuddin, Nat. Nanotechnol., 2015, 10, 191–194 CrossRef CAS PubMed .
  5. M. Ghoneim and M. Hussain, Electronics, 2015, 4, 424–479 CrossRef CAS .
  6. http://www.itrs2.net/, accessed 25th July 2018.
  7. Q.-D. Ling, D.-J. Liaw, C. Zhu, D. S.-H. Chan, E.-T. Kang and K.-G. Neoh, Prog. Polym. Sci., 2008, 33, 917–978 CrossRef CAS .
  8. R. Waser, R. Dittmann, G. Staikov and K. Szot, Adv. Mater., 2009, 21, 2632–2663 CrossRef CAS .
  9. B. Cho, S. Song, Y. Ji, T.-W. Kim and T. Lee, Adv. Funct. Mater., 2011, 21, 2806–2829 CrossRef CAS .
  10. W.-P. Lin, S.-J. Liu, T. Gong, Q. Zhao and W. Huang, Adv. Mater., 2014, 26, 570–606 CrossRef CAS PubMed .
  11. Y. Chen, G. Liu, C. Wang, W. Zhang, R.-W. Li and L. Wang, Mater. Horiz., 2014, 1, 489–506 RSC .
  12. F. Pan, S. Gao, C. Chen, C. Song and F. Zeng, Mater. Sci. Eng., R, 2014, 83, 1–59 CrossRef .
  13. L. P. Ma, J. Liu and Y. Yang, Appl. Phys. Lett., 2002, 80, 2997–2999 CrossRef CAS .
  14. Y. C. Yang, F. Pan, Q. Liu, M. Liu and F. Zeng, Nano Lett., 2009, 9, 1636–1643 CrossRef CAS PubMed .
  15. C.-L. Tsai, F. Xiong, E. Pop and M. Shim, ACS Nano, 2013, 7, 5360–5366 CrossRef CAS PubMed .
  16. M.-J. Lee, C. B. Lee, D. Lee, S. R. Lee, M. Chang, J. H. Hur, Y.-B. Kim, C.-J. Kim, D. H. Seo, S. Seo, U. I. Chung, I.-K. Yoo and K. Kim, Nat. Mater., 2011, 10, 625–630 CrossRef CAS PubMed .
  17. S. Goswami, A. J. Matula, S. P. Rath, S. Hedström, S. Saha, M. Annamalai, D. Sengupta, A. Patra, S. Ghosh, H. Jani, S. Sarkar, M. R. Motapothula, C. A. Nijhuis, J. Martin, S. Goswami, V. S. Batista and T. Venkatesan, Nat. Mater., 2017, 16, 1216–1224 CrossRef CAS PubMed .
  18. C. Chen, S. Gao, G. Tang, C. Song, F. Zeng and F. Pan, IEEE Electron Device Lett., 2012, 33, 1711–1713 CAS .
  19. D. B. Strukov, G. S. Snider, D. R. Stewart and R. S. Williams, Nature, 2008, 453, 80–83 CrossRef CAS PubMed .
  20. S. H. Jo, T. Chang, I. Ebong, B. B. Bhadviya, P. Mazumder and W. Lu, Nano Lett., 2010, 10, 1297–1301 CrossRef CAS PubMed .
  21. G. Liu, C. Wang, W. Zhang, L. Pan, C. Zhang, X. Yang, F. Fan, Y. Chen and R.-W. Li, Adv. Electron. Mater., 2016, 2, 1500298 CrossRef .
  22. J. Borghetti, G. S. Snider, P. J. Kuekes, J. J. Yang, D. R. Stewart and R. S. Williams, Nature, 2010, 464, 873–876 CrossRef CAS PubMed .
  23. S. Gao, G. Liu, Q. Chen, W. Xue, H. Yang, J. Shang, B. Chen, F. Zeng, C. Song, F. Pan and R.-W. Li, ACS Appl. Mater. Interfaces, 2018, 10, 6453–6462 CrossRef CAS PubMed .
  24. J. Choi, S. H. Ji, K. Hong, Y. K. Soo and W. J. Ho, Adv. Mater., 2018, 30, 1704002 CrossRef PubMed .
  25. T. W. Kim, Y. Yang, F. Li and W. L. Kwan, NPG Asia Mater., 2012, 4, e18 CrossRef .
  26. N. Gergel-Hackett, J. L. Tedesco and C. A. Richter, Proc. IEEE, 2012, 100, 1971–1978 CAS .
  27. K. Rajan, E. Garofalo and A. Chiolerio, Sensors, 2018, 18, 367 CrossRef PubMed .
  28. M. Laurenti, D. Perrone, A. Verna, C. Pirri and A. Chiolerio, Micromachines, 2015, 6, 1729–1744 CrossRef .
  29. S.-T. Han, Y. Zhou and V. A. L. Roy, Adv. Mater., 2013, 25, 5425–5449 CrossRef CAS PubMed .
  30. S. Wang, J. Xu, W. Wang, G.-J. N. Wang, R. Rastak, F. Molina-Lopez, J. W. Chung, S. Niu, V. R. Feig, J. Lopez, T. Lei, S.-K. Kwon, Y. Kim, A. M. Foudeh, A. Ehrlich, A. Gasperini, Y. Yun, B. Murmann, J. B. H. Tok and Z. Bao, Nature, 2018, 555, 83–88 CrossRef CAS PubMed .
  31. L. F. Pender and R. J. Fleming, J. Appl. Phys., 1975, 46, 3426–3431 CrossRef CAS .
  32. Y. Segui, B. Ai and H. Carchano, J. Appl. Phys., 1976, 47, 140–143 CrossRef CAS .
  33. B.-H. Lee, H. Bae, H. Seong, D.-I. Lee, H. Park, Y. J. Choi, S.-G. Im, S. O. Kim and Y.-K. Choi, ACS Nano, 2015, 9, 7306–7313 CrossRef CAS PubMed .
  34. B. Cho, J.-M. Yun, S. Song, Y. Ji, D.-Y. Kim and T. Lee, Adv. Funct. Mater., 2011, 21, 3976–3981 CrossRef CAS .
  35. S. Gao, C. Song, C. Chen, F. Zeng and F. Pan, J. Phys. Chem. C, 2012, 116, 17955–17959 CrossRef CAS .
  36. S. Gao, C. Song, C. Chen, F. Zeng and F. Pan, Appl. Phys. Lett., 2013, 102, 141606 CrossRef .
  37. K. Krishnan, T. Tsuruoka, C. Mannequin and M. Aono, Adv. Mater., 2016, 28, 640–648 CrossRef CAS PubMed .
  38. S. K. Kim, J. Y. Kim, B. C. Jang, M. S. Cho, S.-Y. Choi, J. Y. Lee and H. Y. Jeong, Adv. Funct. Mater., 2016, 26, 7406–7414 CrossRef CAS .
  39. X. Zhu, J. Lee and W. D. Lu, Adv. Mater., 2017, 29, 1700527 CrossRef PubMed .
  40. S. M. Yoon, S. C. Warren and B. A. Grzybowski, Angew. Chem., Int. Ed., 2014, 53, 4437–4441 CrossRef CAS PubMed .
  41. A. Chiolerio, I. Roppolo, K. Bejtka, A. Asvarov and C. F. Pirri, RSC Adv., 2016, 6, 56661–56667 RSC .
  42. W. T. Kim, J. H. Jung, T. W. Kim and D. I. Son, Appl. Phys. Lett., 2010, 96, 253301 CrossRef .
  43. D. I. Son, D. H. Park, J. B. Kim, J.-W. Choi, T. W. Kim, B. Angadi, Y. Yi and W. K. Choi, J. Phys. Chem. C, 2011, 115, 2341–2348 CrossRef CAS .
  44. X. Zhang, H. Xie, Z. Liu, C. Tan, Z. Luo, H. Li, J. Lin, L. Sun, W. Chen, Z. Xu, L. Xie, W. Huang and H. Zhang, Angew. Chem., Int. Ed., 2015, 54, 3653–3657 CrossRef CAS PubMed .
  45. P. Zhang, B. Xu, C. Gao, G. Chen and M. Gao, ACS Appl. Mater. Interfaces, 2016, 8, 30336–30343 CrossRef CAS PubMed .
  46. J. Ouyang, C.-W. Chu, C. R. Szmanda, L. Ma and Y. Yang, Nat. Mater., 2004, 3, 918–922 CrossRef CAS PubMed .
  47. Q.-D. Ling, S.-L. Lim, Y. Song, C.-X. Zhu, D. S.-H. Chan, E.-T. Kang and K.-G. Neoh, Langmuir, 2007, 23, 312–319 CrossRef CAS PubMed .
  48. L. Pan, B. Hu, X. Zhu, X. Chen, J. Shang, H. Tan, W. Xue, Y. Zhu, G. Liu and R.-W. Li, J. Mater. Chem. C, 2013, 1, 4556–4564 RSC .
  49. Q. Zhang, J. He, H. Zhuang, H. Li, N. Li, Q. Xu, D. Chen and J. Lu, Adv. Funct. Mater., 2016, 26, 146–154 CrossRef CAS .
  50. T.-L. Choi, K.-H. Lee, W.-J. Joo, S. Lee, T.-W. Lee and M. Y. Chae, J. Am. Chem. Soc., 2007, 129, 9842–9843 CrossRef CAS PubMed .
  51. W. Zhang, C. Wang, G. Liu, X. Zhu, X. Chen, L. Pan, H. Tan, W. Xue, Z. Ji, J. Wang, Y. Chen and R.-W. Li, Chem. Commun., 2014, 50, 11856–11858 RSC .
  52. G. Liu, C. Wang, W. Zhang, L. Pan, C. Zhang, X. Yang, F. Fan, Y. Chen and R.-W. Li, Adv. Electron. Mater., 2015, 2, 1500298 CrossRef .
  53. C. Zhang, J. Shang, W. Xue, H. Tan, L. Pan, X. Yang, S. Guo, J. Hao, G. Liu and R.-W. Li, Chem. Commun., 2016, 52, 4828–4831 RSC .
  54. L.-H. Xie, Q.-D. Ling, X.-Y. Hou and W. Huang, J. Am. Chem. Soc., 2008, 130, 2120–2121 CrossRef CAS PubMed .
  55. S.-J. Liu, P. Wang, Q. Zhao, H.-Y. Yang, J. Wong, H.-B. Sun, X.-C. Dong, W.-P. Lin and W. Huang, Adv. Mater., 2012, 24, 2901–2905 CrossRef CAS PubMed .
  56. Y. Sun, L. Li, D. Wen and X. Bai, Org. Electron., 2015, 25, 283–288 CrossRef CAS .
  57. Y. Sun, F. Miao, R. Li and D. Wen, Phys. Chem. Chem. Phys., 2015, 17, 29978–29984 RSC .
  58. A. Bandyopadhyay, S. Sahu and M. Higuchi, J. Am. Chem. Soc., 2011, 133, 1168–1171 CrossRef CAS PubMed .
  59. R. Kumar, R. G. Pillai, N. Pekas, Y. Wu and R. L. McCreery, J. Am. Chem. Soc., 2012, 134, 14869–14876 CrossRef CAS PubMed .
  60. J. Y. Oh, S. Rondeau-Gagné, Y.-C. Chiu, A. Chortos, F. Lissel, G.-J. N. Wang, B. C. Schroeder, T. Kurosawa, J. Lopez, T. Katsumata, J. Xu, C. Zhu, X. Gu, W.-G. Bae, Y. Kim, L. Jin, J. W. Chung, J. B. H. Tok and Z. Bao, Nature, 2016, 539, 411–415 CrossRef CAS PubMed .
  61. J. Xu, S. Wang, G.-J. N. Wang, C. Zhu, S. Luo, L. Jin, X. Gu, S. Chen, V. R. Feig, J. W. F. To, S. Rondeau-Gagné, J. Park, B. C. Schroeder, C. Lu, J. Y. Oh, Y. Wang, Y.-H. Kim, H. Yan, R. Sinclair, D. Zhou, G. Xue, B. Murmann, C. Linder, W. Cai, J. B.-H. Tok, J. W. Chung and Z. Bao, Science, 2017, 355, 59–64 CrossRef CAS PubMed .
  62. J. C. Scott and L. D. Bozano, Adv. Mater., 2007, 19, 1452–1463 CrossRef CAS .
  63. L. Ma, S. Pyo, J. Ouyang, Q. Xu and Y. Yang, Appl. Phys. Lett., 2003, 82, 1419–1421 CrossRef CAS .
  64. L. D. Bozano, B. W. Kean, V. R. Deline, J. R. Salem and J. C. Scott, Appl. Phys. Lett., 2004, 84, 607–609 CrossRef CAS .
  65. J.-G. Park, W.-S. Nam, S.-H. Seo, Y.-G. Kim, Y.-H. Oh, G.-S. Lee and U.-G. Paik, Nano Lett., 2009, 9, 1713–1719 CrossRef CAS PubMed .
  66. D. Tondelier, K. Lmimouni, D. Vuillaume, C. Fery and G. Haas, Appl. Phys. Lett., 2004, 85, 5763–5765 CrossRef CAS .
  67. L. D. Bozano, B. W. Kean, M. Beinhoff, K. R. Carter, P. M. Rice and J. C. Scott, Adv. Funct. Mater., 2005, 15, 1933–1939 CrossRef CAS .
  68. Y. Yang, J. Ouyang, L. Ma, R. J. H. Tseng and C. W. Chu, Adv. Funct. Mater., 2006, 16, 1001–1014 CrossRef CAS .
  69. J. Chen and D. Ma, Appl. Phys. Lett., 2005, 87, 023505 CrossRef .
  70. J. Chen, L. Xu, J. Lin, Y. Geng, L. Wang and D. Ma, Appl. Phys. Lett., 2006, 89, 083514 CrossRef .
  71. J. Chen and D. Ma, J. Appl. Phys., 2006, 100, 034512 CrossRef .
  72. J. Chen, L. Xu, J. Lin, Y. Geng, L. Wang and D. Ma, Semicond. Sci. Technol., 2006, 21, 1121–1124 CrossRef CAS .
  73. J. Lin and D. Ma, Org. Electron., 2009, 10, 275–279 CrossRef CAS .
  74. C. Wang, J. Wang, P.-Z. Li, J. Gao, S. Y. Tan, W.-W. Xiong, B. Hu, P. S. Lee, Y. Zhao and Q. Zhang, Chem. – Asian J., 2014, 9, 779–783 CrossRef CAS PubMed .
  75. C. Wang, M. Yamashita, B. Hu, Y. Zhou, J. Wang, J. Wu, F. Huo, P. S. Lee, N. Aratani, H. Yamada and Q. Zhang, Chem. – Asian J., 2015, 10, 116–119 CrossRef PubMed .
  76. C. Wang, B. Hu, J. Wang, J. Gao, G. Li, W.-W. Xiong, B. Zou, M. Suzuki, N. Aratani, H. Yamada, F. Huo, P. S. Lee and Q. Zhang, Asian J. Org. Chem., 2015, 4, 646–651 CrossRef CAS .
  77. Q.-F. Gu, J.-H. He, D.-Y. Chen, H.-L. Dong, Y.-Y. Li, H. Li, Q.-F. Xu and J.-M. Lu, Adv. Mater., 2015, 27, 5968–5973 CrossRef CAS PubMed .
  78. H. Li, Q. Xu, N. Li, R. Sun, J. Ge, J. Lu, H. Gu and F. Yan, J. Am. Chem. Soc., 2010, 132, 5542–5543 CrossRef CAS PubMed .
  79. S. Miao, H. Li, Q. Xu, Y. Li, S. Ji, N. Li, L. Wang, J. Zheng and J. Lu, Adv. Mater., 2012, 24, 6210–6215 CrossRef CAS PubMed .
  80. P.-Y. Gu, J. Gao, C.-J. Lu, W. Chen, C. Wang, G. Li, F. Zhou, Q.-F. Xu, J.-M. Lu and Q. Zhang, Mater. Horiz., 2014, 1, 446–451 RSC .
  81. Z. Liu, J. He, H. Li, Q. Xu, N. Li, D. Chen, L. Wang, X. Chen, K. Zhang and J. Lu, Adv. Electron. Mater., 2016, 2, 1500474 CrossRef .
  82. Z. Liu, J. He, H. Zhuang, H. Li, N. Li, D. Chen, Q. Xu, J. Lu, K. Zhang and L. Wang, J. Mater. Chem. C, 2015, 3, 9145–9153 RSC .
  83. X.-F. Cheng, E.-B. Shi, X. Hou, S.-G. Xia, J.-H. He, Q.-F. Xu, H. Li, N.-J. Li, D.-Y. Chen and J.-M. Lu, Chem. – Asian J., 2017, 12, 45–51 CrossRef CAS PubMed .
  84. Y. Li, H. Li, H. Chen, Y. Wan, N. Li, Q. Xu, J. He, D. Chen, L. Wang and J. Lu, Adv. Funct. Mater., 2015, 25, 4246–4254 CrossRef CAS .
  85. S. Miao, H. Li, Q. Xu, N. Li, J. Zheng, R. Sun, J. Lu and C. M. Li, J. Mater. Chem., 2012, 22, 16582–16589 RSC .
  86. Q. Zhang, J. He, H. Zhuang, H. Li, N. Li, Q. Xu, D. Chen and J. Lu, Chem. – Asian J., 2016, 11, 1624–1630 CrossRef CAS PubMed .
  87. G. Wang, S. Miao, Q. Zhang, H. Liu, H. Li, N. Li, Q. Xu, J. Lu and L. Wang, Chem. Commun., 2013, 49, 9470–9472 RSC .
  88. Q. Zhang, G. Wang, H. Li, F. Zhu, H. Li and J. Lu, Adv. Funct. Mater., 2018, 28, 1800568 CrossRef .
  89. Y. Zhang, H. Zhuang, Y. Yang, X. Xu, Q. Bao, N. Li, H. Li, Q. Xu, J. Lu and L. Wang, J. Phys. Chem. C, 2012, 116, 22832–22839 CrossRef CAS .
  90. C.-L. Liu and W.-C. Chen, Polym. Chem., 2011, 2, 2169–2174 RSC .
  91. W. Zhang, C. Wang, G. Liu, J. Wang, Y. Chen and R.-W. Li, Chem. Commun., 2014, 50, 11496–11499 RSC .
  92. H.-C. Wu, A.-D. Yu, W.-Y. Lee, C.-L. Liu and W.-C. Chen, Chem. Commun., 2012, 48, 9135–9137 RSC .
  93. P. Wang, S.-J. Liu, Z.-H. Lin, X.-C. Dong, Q. Zhao, W.-P. Lin, M.-D. Yi, S.-H. Ye, C.-X. Zhu and W. Huang, J. Mater. Chem., 2012, 22, 9576–9583 RSC .
  94. S.-J. Liu, W.-P. Lin, M.-D. Yi, W.-J. Xu, C. Tang, Q. Zhao, S.-H. Ye, X.-M. Liu and W. Huang, J. Mater. Chem., 2012, 22, 22964–22970 RSC .
  95. J. Xiang, T.-K. Wang, Q. Zhao, W. Huang, C.-L. Ho and W.-Y. Wong, J. Mater. Chem. C, 2016, 4, 921–928 RSC .
  96. A.-D. Yu, T. Kurosawa, Y.-C. Lai, T. Higashihara, M. Ueda, C.-L. Liu and W.-C. Chen, J. Mater. Chem., 2012, 22, 20754–20763 RSC .
  97. K.-L. Wang, G. Liu, P.-H. Chen, L. Pan and H.-L. Tsai, Org. Electron., 2014, 15, 322–336 CrossRef CAS .
  98. J.-T. Wang, K. Saito, H.-C. Wu, H.-S. Sun, C.-C. Hung, Y. Chen, T. Isono, T. Kakuchi, T. Satoh and W.-C. Chen, NPG Asia Mater., 2016, 8, e298 CrossRef .
  99. C. C. Hung, Y.-C. Chiu, H.-C. Wu, C. Lu, C. Bouilhac, I. Otsuka, S. Halila, R. Borsali, S.-H. Tung and W.-C. Chen, Adv. Funct. Mater., 2017, 27, 1606161 CrossRef .
  100. B. C. Jang, H. Seong, S. K. Kim, J. Y. Kim, B. J. Koo, J. Choi, S. Y. Yang, S. G. Im and S.-Y. Choi, ACS Appl. Mater. Interfaces, 2016, 8, 12951–12958 CrossRef CAS PubMed .
  101. Y. Cai, J. Tan, L. YeFan, M. Lin and R. Huang, Nanotechnology, 2016, 27, 275206 CrossRef PubMed .
  102. S. Gao, F. Zeng, C. Chen, G. Tang, Y. Lin, Z. Zheng, C. Song and F. Pan, Nanotechnology, 2013, 24, 335201 CrossRef PubMed .
  103. K. Krishnan, M. Muruganathan, T. Tsuruoka, H. Mizuta and M. Aono, Adv. Funct. Mater., 2017, 27, 1605104 CrossRef .
  104. H. Ling, M. Yi, M. Nagai, L. Xie, L. Wang, B. Hu and W. Huang, Adv. Mater., 2017, 29, 1701333 CrossRef PubMed .
  105. A. Chiolerio, S. Bocchini, M. Crepaldi, K. Bejtka and C. F. Pirri, Synth. Met., 2017, 229, 72–81 CrossRef CAS .
  106. S. Möller, C. Perlov, W. Jackson, C. Taussig and S. R. Forrest, Nature, 2003, 426, 166–169 CrossRef PubMed .
  107. X. Xu, R. A. Register and S. R. Forrest, Appl. Phys. Lett., 2006, 89, 142109 CrossRef .
  108. U. S. Bhansali, M. A. Khan, D. Cha, M. N. AlMadhoun, R. Li, L. Chen, A. Amassian, I. N. Odeh and H. N. Alshareef, ACS Nano, 2013, 7, 10518–10524 CrossRef CAS PubMed .
  109. J. Liu, P. Gu, F. Zhou, Q. Xu, J. Lu, H. Li and L. Wang, J. Mater. Chem. C, 2013, 1, 3947–3954 RSC .
  110. A. Bandyopadhyay and A. J. Pal, Adv. Mater., 2003, 15, 1949–1952 CrossRef CAS .
  111. A.-D. Yu, T. Kurosawa, Y.-H. Chou, K. Aoyagi, Y. Shoji, T. Higashihara, M. Ueda, C.-L. Liu and W.-C. Chen, ACS Appl. Mater. Interfaces, 2013, 5, 4921–4929 CrossRef CAS PubMed .
  112. Y. Sun, L. Li, D. Wen and X. Bai, J. Phys. Chem. C, 2015, 119, 19520–19525 CrossRef CAS .
  113. B. Hu, X. Zhu, X. Chen, L. Pan, S. Peng, Y. Wu, J. Shang, G. Liu, Q. Yan and R.-W. Li, J. Am. Chem. Soc., 2012, 134, 17408–17411 CrossRef CAS PubMed .
  114. C. Zhang, Y.-T. Tai, J. Shang, G. Liu, K.-L. Wang, C. Hsu, X. Yi, X. Yang, W. Xue, H. Tan, S. Guo, L. Pan and R.-W. Li, J. Mater. Chem. C, 2016, 4, 3217–3223 RSC .
  115. K. Rajan, I. Roppolo, K. Bejtka, A. Chiappone, S. Bocchini, D. Perrone, C. F. Pirri, C. Ricciardi and A. Chiolerio, Appl. Surf. Sci., 2018, 443, 475–483 CrossRef CAS .
  116. I. Roppolo, M. Castellino, K. Bejtka, G. Rizza, D. Perrone, P.-E. Coulon, A. Chiappone, K. Rajan, S. Bocchini, C. Ricciardi, C. F. Pirri and A. Chiolerio, J. Phys. Chem. C, 2017, 121, 14285–14295 CrossRef CAS .
  117. K. Rajan, K. Bejtka, S. Bocchini, D. Perrone, A. Chiappone, I. Roppolo, C. F. Pirri, C. Ricciardi and A. Chiolerio, J. Mater. Chem. C, 2017, 5, 6144–6155 RSC .
  118. K. Rajan, S. Bocchini, A. Chiappone, I. Roppolo, D. Perrone, M. Castellino, K. Bejtka, M. Lorusso, C. Ricciardi, C. F. Pirri and A. Chiolerio, Flexible Printed Electron., 2017, 2, 024002 CrossRef .
  119. K. Rajan, S. Bocchini, A. Chiappone, I. Roppolo, D. Perrone, K. Bejtka, C. Ricciardi, C. F. Pirri and A. Chiolerio, Microelectron. Eng., 2017, 168, 27–31 CrossRef CAS .
  120. S. Paul, A. Kanwal and M. Chhowalla, Nanotechnology, 2006, 17, 145–151 CrossRef CAS .
  121. Y. Ji, S. Lee, B. Cho, S. Song and T. Lee, ACS Nano, 2011, 5, 5995–6000 CrossRef CAS PubMed .
  122. J. Ko, J. A. Lim, H. J. Chang and Y. S. Kim, Macromol. Rapid Commun., 2013, 34, 355–361 CrossRef PubMed .
  123. Y. Ji, D. F. Zeigler, D. S. Lee, H. Choi, A. K. Y. Jen, H. C. Ko and T.-W. Kim, Nat. Commun., 2013, 4, 2707 CrossRef PubMed .
  124. J. H. Jung, J.-H. Kim, T. W. Kim, M. S. Song, Y.-H. Kim and S. Jin, Appl. Phys. Lett., 2006, 89, 122110 CrossRef .
  125. B. Cho, T.-W. Kim, M. Choe, G. Wang, S. Song and T. Lee, Org. Electron., 2009, 10, 473–477 CrossRef CAS .
  126. K. Onlaor, T. Thiwawong and B. Tunhoo, Org. Electron., 2014, 15, 1254–1262 CrossRef CAS .
  127. Y. Sun, D. Wen and X. Bai, Phys. Chem. Chem. Phys., 2018, 20, 5771–5779 RSC .
  128. B. C. Das and A. J. Pal, ACS Nano, 2008, 2, 1930–1938 CrossRef CAS PubMed .
  129. L. Meng, M. Lan, L. Guo, L. Xie, W. Hui, J. Ge, W. Liu, Y. Wang and P. Wang, RSC Adv., 2015, 5, 26886–26890 RSC .
  130. K. Onlaor, T. Thiwawong and B. Tunhoo, Org. Electron., 2016, 31, 19–24 CrossRef CAS .
  131. S. Ali, J. Bae, C. H. Lee, K. H. Choi and Y. H. Doh, Org. Electron., 2015, 25, 225–231 CrossRef CAS .
  132. J. Ali, G.-U.-D. Siddiqui, Y. J. Yang, K. T. Lee, K. Um and K. H. Choi, RSC Adv., 2016, 6, 5068–5078 RSC .
  133. D. H. Kim, W. K. Kim, S. J. Woo, C. Wu and T. W. Kim, Org. Electron., 2017, 51, 156–161 CrossRef CAS .
  134. S.-T. Han, L. Hu, X. Wang, Y. Zhou, Y.-J. Zeng, S. Ruan, C. Pan and Z. Peng, Adv. Sci., 2017, 4, 1600435 CrossRef PubMed .
  135. X. Zhang, Z. Lai, Z. Liu, C. Tan, Y. Huang, B. Li, M. Zhao, L. Xie, W. Huang and H. Zhang, Angew. Chem., 2015, 127, 5515–5518 CrossRef .
  136. G. Liu, Q.-D. Ling, E. Y. H. Teo, C.-X. Zhu, D. S.-H. Chan, K.-G. Neoh and E.-T. Kang, ACS Nano, 2009, 3, 1929–1937 CrossRef CAS PubMed .
  137. S. K. Hwang, J. M. Lee, S. Kim, J. S. Park, H. I. Park, C. W. Ahn, K. J. Lee, T. Lee and S. O. Kim, Nano Lett., 2012, 12, 2217–2221 CrossRef CAS PubMed .
  138. S. K. Hwang, J. R. Choi, I. Bae, I. Hwang, S. M. Cho, J. Huh and C. Park, Small, 2013, 9, 831–837 CrossRef CAS PubMed .
  139. S. ChandraKishore and A. Pandurangan, RSC Adv., 2014, 4, 9905–9911 RSC .
  140. S. I. White, P. M. Vora, J. M. Kikkawa and K. I. Winey, Adv. Funct. Mater., 2011, 21, 233–240 CrossRef CAS .
  141. J. Jang, W. Park, K. Cho, H. Song and T. Lee, Curr. Appl. Phys., 2013, 13, 1237–1240 CrossRef .
  142. Q. Zhang, J. Pan, X. Yi, L. Li and S. Shang, Org. Electron., 2012, 13, 1289–1295 CrossRef CAS .
  143. J. Liu, Z. Zeng, X. Cao, G. Lu, L.-H. Wang, Q.-L. Fan, W. Huang and H. Zhang, Small, 2012, 8, 3517–3522 CrossRef CAS PubMed .
  144. P. Zhang, C. Gao, B. Xu, L. Qi, C. Jiang, M. Gao and D. Xue, Small, 2016, 12, 2077–2084 CrossRef CAS PubMed .
  145. M. M. Rehman, G. U. Siddiqui, J. Z. Gul, S.-W. Kim, J. H. Lim and K. H. Choi, Sci. Rep., 2016, 6, 36195 CrossRef CAS PubMed .
  146. G. U. Siddiqui, M. M. Rehman, Y.-J. Yang and K. H. Choi, J. Mater. Chem. C, 2017, 5, 862–871 RSC .
  147. P. Zhang, C. Gao, B. Xu, L. Qi, C. Jiang, M. Gao and D. Xue, Small, 2016, 12, 2077–2084 CrossRef CAS PubMed .
  148. D. Chen, H. Feng and J. Li, Chem. Rev., 2012, 112, 6027–6053 CrossRef CAS PubMed .
  149. S. Porro, E. Accornero, C. F. Pirri and C. Ricciardi, Carbon, 2015, 85, 383–396 CrossRef CAS .
  150. C. L. He, F. Zhuge, X. F. Zhou, M. Li, G. C. Zhou, Y. W. Liu, J. Z. Wang, B. Chen, W. J. Su, Z. P. Liu, Y. H. Wu, P. Cui and R.-W. Li, Appl. Phys. Lett., 2009, 95, 232101 CrossRef .
  151. F. Zhuge, B. Hu, C. He, X. Zhou, Z. Liu and R.-W. Li, Carbon, 2011, 49, 3796–3802 CrossRef CAS .
  152. H. Y. Jeong, J. Y. Kim, J. W. Kim, J. O. Hwang, J.-E. Kim, J. Y. Lee, T. H. Yoon, B. J. Cho, S. O. Kim, R. S. Ruoff and S.-Y. Choi, Nano Lett., 2010, 10, 4381–4386 CrossRef CAS PubMed .
  153. J. Liu, Z. Yin, X. Cao, F. Zhao, L. Wang, W. Huang and H. Zhang, Adv. Mater., 2013, 25, 233–238 CrossRef PubMed .
  154. G. Sun, J. Liu, L. Zheng, W. Huang and H. Zhang, Angew. Chem., Int. Ed., 2013, 125, 13593–13597 CrossRef .
  155. G. Liu, X. Zhuang, Y. Chen, B. Zhang, J. Zhu, C.-X. Zhu, K.-G. Neoh and E.-T. Kang, Appl. Phys. Lett., 2009, 95, 253301 CrossRef .
  156. X.-D. Zhuang, Y. Chen, G. Liu, P.-P. Li, C.-X. Zhu, E.-T. Kang, K.-G. Noeh, B. Zhang, J.-H. Zhu and Y.-X. Li, Adv. Mater., 2010, 22, 1731–1735 CrossRef CAS PubMed .
  157. B. Zhang, Y.-L. Liu, Y. Chen, K.-G. Neoh, Y.-X. Li, C.-X. Zhu, E.-S. Tok and E.-T. Kang, Chem. – Eur. J., 2011, 17, 10304–10311 CrossRef CAS PubMed .
  158. B. Zhang, Y. Chen, Y. Ren, L.-Q. Xu, G. Liu, E.-T. Kang, C. Wang, C.-X. Zhu and K.-G. Neoh, Chem. – Eur. J., 2013, 19, 6265–6273 CrossRef CAS PubMed .
  159. G. Liu, Y. Chen, R.-W. Li, B. Zhang, E.-T. Kang, C. Wang and X. Zhuang, ChemElectroChem, 2014, 1, 514–519 CrossRef .
  160. Y. Chen, B. Zhang, G. Liu, X. Zhuang and E.-T. Kang, Chem. Soc. Rev., 2012, 41, 4688–4707 RSC .
  161. H. Furukawa, K. E. Cordova, M. O’Keeffe and O. M. Yaghi, Science, 2013, 341, 1230444 CrossRef PubMed .
  162. Y. Zhao and K. Zhu, Chem. Soc. Rev., 2016, 45, 655–689 RSC .
  163. S. S.-Y. Chui, S. M.-F. Lo, J. P. H. Charmant, A. G. Orpen and I. D. Williams, Science, 1999, 283, 1148–1150 CrossRef CAS PubMed .
  164. L. Pan, G. Liu, H. Li, S. Meng, L. Han, J. Shang, B. Chen, A. E. Platero-Prats, W. Lu, X. Zou and R.-W. Li, J. Am. Chem. Soc., 2014, 136, 17477–17483 CrossRef CAS PubMed .
  165. L. Pan, Z. Ji, X. Yi, X. Zhu, X. Chen, J. Shang, G. Liu and R.-W. Li, Adv. Funct. Mater., 2015, 25, 2677–2685 CrossRef CAS .
  166. M.-J. Park and J.-S. Lee, RSC Adv., 2017, 7, 21045–21049 RSC .
  167. Y. Liu, H. Wang, W. Shi, W. Zhang, J. Yu, B. K. Chandran, C. Cui, B. Zhu, Z. Liu, B. Li, C. Xu, Z. Xu, S. Li, W. Huang, F. Huo and X. Chen, Angew. Chem., Int. Ed., 2016, 55, 8884–8888 CrossRef CAS PubMed .
  168. J. Choi, S. Park, J. Lee, K. Hong, D.-H. Kim, C. W. Moon, G. D. Park, J. Suh, J. Hwang, S. Y. Kim, H. S. Jung, N.-G. Park, S. Han, K. T. Nam and H. W. Jang, Adv. Mater., 2016, 28, 6562–6567 CrossRef CAS PubMed .
  169. E. J. Yoo, M. Lyu, J.-H. Yun, C. J. Kang, Y. J. Choi and L. Wang, Adv. Mater., 2015, 27, 6170–6175 CrossRef CAS PubMed .
  170. C. Gu and J.-S. Lee, ACS Nano, 2016, 10, 5413–5418 CrossRef CAS PubMed .
  171. J. Choi, Q. V. Le, K. Hong, C. W. Moon, J. S. Han, K. C. Kwon, P.-R. Cha, Y. Kwon, S. Y. Kim and H. W. Jang, ACS Appl. Mater. Interfaces, 2017, 9, 30764–30771 CrossRef CAS PubMed .
  172. B. Hwang and J.-S. Lee, Adv. Mater., 2017, 29, 1701048 CrossRef PubMed .
  173. X. Zhu and W. D. Lu, ACS Nano, 2018, 12, 1242–1249 CrossRef CAS PubMed .
  174. F. Zhou, Y. Liu, X. Shen, M. Wang, F. Yuan and Y. Chai, Adv. Funct. Mater., 2018, 28, 1800080 CrossRef .
  175. M. M. Waldrop, Nature, 2016, 530, 144–148 CrossRef CAS PubMed .
  176. D. S. Jeong, K. M. Kim, S. Kim, B. J. Choi and C. S. Hwang, Adv. Electron. Mater., 2016, 2, 1600090 CrossRef .
  177. https://irds.ieee.org/, accessed 25th July 2018.
  178. X. Zhu, W. Su, Y. Liu, B. Hu, L. Pan, W. Lu, J. Zhang and R.-W. Li, Adv. Mater., 2012, 24, 3941–3946 CrossRef CAS PubMed .
  179. S. R. Nandakumar, M. Minvielle, S. Nagar, C. Dubourdieu and B. Rajendran, Nano Lett., 2016, 16, 1602–1608 CrossRef CAS PubMed .
  180. J. J. Yang, D. B. Strukov and D. R. Stewart, Nat. Nanotechnol., 2013, 8, 13–24 CrossRef CAS PubMed .

This journal is © The Royal Society of Chemistry 2019