An electrically modifiable synapse array of resistive switching memory

Choi, Hyejung; Jung, Heesoo; Lee, Joonmyoung; Yoon, Jaesik; Park, Jubong; Seong, Dong-jun; Lee, Wootae; Hasan, Musarrat; Jung, Gun‐Young; Hwang, Hyunsang

doi:10.1088/0957-4484/20/34/345201

Cited by 133 publications

(77 citation statements)

References 19 publications

(23 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

Section: Introductionmentioning

confidence: 99%

“…Moreover, memristors can function as stateful Boolean logic gates via the material implication operation [15]. In addition, memristors can also be used for neuromorphic computing [16,17] because of their analog switching, and some hybrid circuits due to their ease of stacking [18,19].Among all the kinds of switching materials that have been reported, oxides are the most extensively studied [4]. The interfaces between the metal electrodes and the oxide play a crucial role, especially for bipolar switches [5,6,20].…”

mentioning

confidence: 99%

See 1 more Smart Citation

Metal/TiO2 interfaces for memristive switches

et al. 2011

View full text Add to dashboard Cite

The interfaces between metal electrodes and the oxide in TiO 2 -based memristive switches play a key role in the switching as well as in the I -V characteristics of the devices in different resistance states. We demonstrate here that the work function of the metal electrode has a surprisingly minor effect in determining the electronic barrier at the interface. In contrast, Ti oxides can be readily reduced by most electrode metals. The amount of oxygen vacancies created by these chemical reactions essentially determines the electronic barrier at the device interfaces.The memristor, the fourth fundamental passive circuit element [1][2][3], has a wide variety of potential applications based on its promising device properties [1][2][3][4][5][6][7], including non-volatility, fast switching (<10 ns), low energy (∼1 pJ/operation), multiple-state operation, scalability and stackability. As suggested by the name, memristors can be used for information storage [8][9][10][11][12][13][14]. Moreover, memristors can function as stateful Boolean logic gates via the material implication operation [15]. In addition, memristors can also be used for neuromorphic computing [16,17] because of their analog switching, and some hybrid circuits due to their ease of stacking [18,19].Among all the kinds of switching materials that have been reported, oxides are the most extensively studied [4]. The interfaces between the metal electrodes and the oxide play a crucial role, especially for bipolar switches [5,6,20]. Under an applied electric field, oxygen vacancies can drift into the interface region, reducing the electronic barrier and resulting in a low-resistance state. Under an electric field with the opposite polarity, the oxygen vacancies are repelled away from the interface region, recovering the electronic barrier to regain the high resistance state [6,21,22]. A family of nanodevices with different switching and currentvoltage (I -V ) characteristics has been demonstrated by manipulating the elemental composition at the two interfaces of a simple metal/oxide/metal device stack [23]. For a crossbar array of memristors in a storage/memory circuit, sneak path currents [24] can be minimized by using memristors with a rectifying I -V characteristic, especially when they are in the low-resistance state. Therefore, interface engineering is critical to obtain the desired switching behavior and electrical properties for memristive devices.From a semiconductor physics point of view, the work function of the electrode metal might be an important factor in determining the electronic barrier at the metal/oxide interface [25]. This barrier in our thin film devices is not a traditional Schottky barrier because the oxide film is amorphous with a large concentration of defects and likely thinner than the depletion region of the semiconducting oxide. We prepared a set of 5 µm × 5 µm cross-point devices with six different top electrode metals (Au, Pt, Ag, Ni, W and Ti) to examine the effect of these contacts on the I -V characteristics, as show...

show abstract

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

Metal/TiO2 interfaces for memristive switches

et al. 2011

View full text Add to dashboard Cite

show abstract

“…The recent milestones for metal-oxide memristors, which are compatible with existing silicon technology, include demonstrations of sub-10-nm devices 16 , 10 12 -cycle endurance 17 , pico-Joule 18 and sub-ns switching 19 , and the monolithic integration of several memristive crossbar layers 20 . These milestones, in turn, revived interest in the development of memristor-based ANNs and led to numerous few-or single-memristor demonstrations of synaptic functionality and simple associative memory [21][22][23][24][25][26][27][28] , as well as the theoretical modelling of large-scale networks 10,[29][30][31][32][33] . Despite significant progress in memristor crossbar memories 20,[34][35][36][37] , memristor-based ANNs have proven to be significantly more challenging and have yet to be demonstrated.…”

mentioning

confidence: 99%

Pattern classification by memristive crossbar circuits using ex situ and in situ training

2013

View full text Add to dashboard Cite

Memristors are memory resistors that promise the efficient implementation of synaptic weights in artificial neural networks. Whereas demonstrations of the synaptic operation of memristors already exist, the implementation of even simple networks is more challenging and has yet to be reported. Here we demonstrate pattern classification using a single-layer perceptron network implemented with a memrisitive crossbar circuit and trained using the perceptron learning rule by ex situ and in situ methods. In the first case, synaptic weights, which are realized as conductances of titanium dioxide memristors, are calculated on a precursor software-based network and then imported sequentially into the crossbar circuit. In the second case, training is implemented in situ, so the weights are adjusted in parallel. Both methods work satisfactorily despite significant variations in the switching behaviour of the memristors. These results give hope for the anticipated efficient implementation of artificial neuromorphic networks and pave the way for dense, high-performance information processing systems.

show abstract

“…Fig. 7 depicts an I-V curve on its log-log scale [30]. Table 2 summarizes the comparison between different available models.…”

Section: B Proposed Model For Memristor Dynamic Behaviormentioning

confidence: 99%