Skip to content


  • Nano Express
  • Open Access

Atomic Layer Deposited Hf0.5Zr0.5O2-based Flexible Memristor with Short/Long-Term Synaptic Plasticity

Nanoscale Research Letters201914:102

  • Received: 15 December 2018
  • Accepted: 8 March 2019
  • Published:


Artificial synapses are the fundamental of building a neuron network for neuromorphic computing to overcome the bottleneck of the von Neumann system. Based on a low-temperature atomic layer deposition process, a flexible electrical synapse was proposed and showed bipolar resistive switching characteristics. With the formation and rupture of ions conductive filaments path, the conductance was modulated gradually. Under a series of pre-synaptic spikes, the device successfully emulated remarkable short-term plasticity, long-term plasticity, and forgetting behaviors. Therefore, memory and learning ability were integrated to the single flexible memristor, which are promising for the next-generation of artificial neuromorphic computing systems.


  • Atomic layer deposition
  • Low-temperature process
  • Flexible electronics
  • Synaptic plasticity


The classical von Neumann computing scheme is suffering a bottleneck of information transfer between the processing center and storage units [1]. Through emulating biological brains, neuromorphic computing has become an attractive candidate with the ability of learning and memory in one single system [2, 3]. Electronic synapses, with the ability of mimicking bio-synaptic behavior, are the foundation of neuromorphic systems. Recently, bio-synaptic behaviors have been emulated by various memristors, including two-terminal devices and novel three-terminal synaptic transistors based on ionic defects [4, 5]. With history-dependent conductance, memristors were reported to simulate the long-term depression (LTD) or potentiation (LTP), pair-pulse fluctuation (PPF), paired-pulse depression (PPD), and spike-timing-dependent plasticity (STDP) [68]. Especially, LTP/LTD is vital for face classification, digital recognition, and other artificial intelligence applications based on synaptic weight modification [911]. Originating from immediate post-synaptic current response, STP is widely used for information filtering and instantaneous signal transmission [12].

A variety of material systems were studied for artificial synapses with bio-synaptic plasticity, including HfO2, ZnO, WOx, TaOx, InGaZnO, organic polymers, and 2D transition-metal dichalcogenides (TMDCs) [1319]. Among them, Hf0.5Zr0.5O2 (HZO) is one of the novel high-k materials and compatible with the process of complementary metal oxide semiconductor (CMOS) [20]. Although HZO-based artifical synapstic devices have been reported, the high-temperature preparation process is hard to aviod [2123].

On the other hand, flexible artificial synaptic devices were widely studied to satisfy the rising need for wearable artificial intelligence applications [24, 25]. However, the high-temperature preparation process is an impediment to the application of a flexible substrate. Although a transfer process was proposed to solve the problem, the high failure rate and wrinkle defects caused by transfer hinder the large-scale use of this method [26, 27]. It is worth noting that low-temperature processing has no damage to flexible substrates, which is an effective way of developing large-scale wearable synaptic arrays.

In this work, a low-temperature ALD technique for HZO-based memristor (PET/ITO/HZO/Ag) was developed. Gradual conductance switching process was demonstrated in this memristor. Based on gradual resistance switching characteristics, typical synaptic plasticity was emulated, including LTP/LTD, STP, PPF, and forgetting curves. With the function of biological synapses, the flexible HZO-based memristor is attractive for future applications in a neuromorphic computing system.


The flexible synaptic device was prepared on ITO-coated polyethylene terephthalate (PET) substrate, which was cleaned in acetone, isopropanol, and deionized water and dried by N2 flow. A 10-nm-thick HZO film was deposited on PET/ITO substrate by ALD with the carrier gas of N2. The precursors were tetrakis (ethylmethylamino) hafnium (TEMAH), tetrakis (ethylmethylamino) zirconium (TEMAZ), and H2O, and the growth temperature of the ALD chamber was maintained at 130 °C. Then, a 50-nm Ag top electrode (TE) layer with an area of 100 × 100 μm2 was deposited by physical vapor deposition (PVD) followed by photolithography and lift-off process. The structure of PET/ITO/ HZO/Ag was shown in Fig. 1. The top electrode of Ag and bottom electrode of ITO are corresponding to pre- and post-synaptic neuron in biological synapse.
Fig. 1
Fig. 1

Schematic illustration of biological synapse between neurons and artificial electrical synapses. A bio-synapse was composed of pre-synaptic neuron, synaptic cleft, and post-synaptic neuron. The HZO-based flexible electrical synapse was fabricated with the structure of ITO/HZO/Ag on the plastic substrate at low temperature

The electrical characteristics were performed using a semiconductor parameter analyzer (Agilent B1500A) in the atmospheric environment at room temperature. The bottom electrode was grounded while the programming bias was applied to the top electrode.

Results and Discussion

Figure 2a shows the typical bipolar resistive switching curve of the memristor with the current compliance of 500 uA. The sweeping voltage was applied in a sequence of 0 → 2 V → 0 V for the set process, and the resistance turned from high-resistance state (HRS) to low-resistance state (LRS). In contrast, a negative voltage was applied from 0 V to − 2 V and returned to 0 V for the reset process. The gradual switching characteristic in positive and negative bias sweeps indicates the potential of HZO-based memristor emulating synaptic behaviors. The cumulative probability of operating voltages in the set and reset process during consecutive sweep cycles are shown in Fig. 2. The means (μ) of the set voltage and reset voltage are 0. 99 V and − 1. 33 V, respectively, which showed the average level of operating voltage. The standard deviation (σ) of the operating voltage (0.245 for set process and 0.566 for reset process) indicated the degree of deviation from the center. The relative fluctuation of data could be described as a coefficient of variance (σ/μ). Superior uniformity was obtained in the set process while the variation of HRS resistance and reset voltage are remarkable, which could attribute to the formation and rupture process of conductive filament (CF) of Ag atoms. During the process of set operation, the size or number of CFs would increase. The current level of device is almost linearly proportional to the increment of CFs. During the reset process, the CFs would break and decrease. While the current level of device is exponentially dependent on the breaking length of CFs [28]. A small change of CFs during the reset process could result to obvious changes of resistance and reset voltage. The ON/OFF ratio of μ in HZO-based device was larger than 300, as shown in Fig. 2c.
Fig. 2
Fig. 2

a Resistive switching characteristics of HZO-based device measured by DC sweep. b Distribution of the set and reset voltages extracted from DC sweep cycles in flexible device. c Statistical data of HRS and LRS, where resistance were measured at a read voltage of 0.1 V

Besides gradually resistance switching behaviors in DC sweep, the device with modulated conductance could be programed by a sequence of consecutive pulses. As shown in Fig. 3a, the conductance could be modulated gradually to emulate LTP and LTD with 400 consecutive programing pulses, indicating the potential of the synaptic device for neuromorphic computing. With 200 consecutive positive pulses (0 .8V, 20 ms) and 200 negative pulses (− 0 .5V, 20 ms), the conductance of synaptic device potentiated and depressed gradually. The conductance state was obtained under a read voltage of 0.1 V after each consecutive pulse. Forgetting is one of the common phenomena in human brains, which could be simulated by the relaxation of post-synaptic current in electrical synapses. After a series of pulses, the post-synaptic current (PSC) decayed and turned to an intermediate state over time, as shown in Fig. 3b. The forgetting curve could be fitted with the Kohlrausch equation that was frequently used in psychology:
$$ I(t)={I}_0+A\exp \left(-t/\tau \right) $$
where I(t) is the PSC at the time of t, I0 is the stabilized current, A is a prefactor, and τ is a relaxation time constant. In the artificial synaptic device, the constant τ was 57 s which was used to evaluate the forgetting characteristics.
Fig. 3
Fig. 3

a Gradual conductance modulation for LTP and LTD in the artificial flexible synapse, where the post-synaptic current was obtained at a read voltage of 0.1 V. b Forgetting behaviors after 100 consecutive programing pulses (1 V, 50 ms) and fitted curves of the electrical synapse

To better understand the work mechanism of the HZO-based synaptic device, the conductive filaments (CF) in different states were shown in Fig. 4. The formation and rupture of the CFs were due to the migration of Ag atoms and mobile Ag+. When the positive programing stimulus was applied to the top electrode, atoms of top electrode were oxided to Ag+, which were accumulated in the bottom electrode and reduced to Ag atoms. In Fig. 4a–c, the thickness and diameter of CF increased slightly from state I to state III, which induced the conductance increase [29]. In contrast, the bridge of Ag atoms ruptured with a weak effect on the conductance after applying a series of negative spikes in memristor, as shown in Fig. 4d–f. Typical LTP and LTD behaviors in this HZO-based artificial synaptic device were organized from the CF gradual formation and rupture, respectively.
Fig. 4
Fig. 4

ac The schematic diagrams of forming Ag cations conductive path under consecutive positive pulses in LTP. df Rupture of the conductive filament after consecutive negative pulses in LTD

Short-term synaptic plasticity is crucial to both excitatory and inhibitory bio-synapses, which is considered to play important roles in treatment of temporal information [30, 31]. The PPF and PPD behaviors are typical short-term phenomenon organized from two consecutive synaptic spikes with a short interval. Such plasticity was also successfully mimicked in our flexible HZO-based synaptic device. The PPF function was short-term enhancement of synaptic weights trigged by a pair of spikes (2 V, 10 ms) with an interval of 60 ms, as shown in Fig. 5a. In contrast, the response current of second spike is smaller than that of previous spike, which is described as PPD and simulated by two negative pulses (− 1 .5V, 10 ms) with an interval of 60 ms.
Fig. 5
Fig. 5

a Typical PPF behavior induced by a pair of pre-synaptic spikes (2 V, 10 ms). b The PPD phenomenon of the artificial flexible synapse under inhibited spikes (− 1 .5V, 10 ms)

To demonstrate the reliability of long-term plasticity in our synaptic device, retention characteristics were measured for over 1000 s. As shown in Fig. 6, the PSC in excitatory and inhibitory states were read at a bias of 0.1 V after a single pre-synaptic spike. The long-term retention behavior of our HZO-based device shows the potential of storage, and the consecutive modulated conductance paves the way for memory function, which could be integrated into a system.
Fig. 6
Fig. 6

a The retention characteristics of electrical synapse under positive programing pulse, indicating the long-term potential behaviors. b In LTD process, the post-synaptic current could be inhibited under a single negative pulse (− 0 .5V, 20 ms) and the conductance state could stay stable for over 1000 s


In summary, a flexible HZO-based artificial synaptic device was proposed based on low-temperature ALD. Typical bipolar resistive switching characteristics were demonstrated in this flexible memristor. By applying consecutive pulses in the top electrode, long-term plasticity and short-term plasticity were simulated by the electrical synapse, including LTP, LTD, PPF, PPD, and forgetting behaviors. Gradually modulated conductance could be attributed to controllable Ag ions conductive filament path. The flexible electrical synapse becomes one of the promising candidates for hardware implementation of neuromorphic circuits.



Atomic layer deposition


High-resistance state


Low-resistance state


Long-term depression


Long-term potentiation


Short-term plasticity



This work was supported by the NSFC (61704030 and 61522404), the 02 State Key Project (2017ZX02315005), the Program of Shanghai Subject Chief Scientist (18XD1402800), the Support Plans for the Youth Top-Notch Talents of China, and the “Chen Guang” project supported by the Shanghai Municipal Education Commission and Shanghai Education Development Foundation.

Availability of Data and Materials

All data are fully available without restriction.

Authors’ Contributions

T-YW prepared the HZO-based flexible artificial synaptic devices. T-YW and J-LM designed the electrical measurements method of synaptic plasticity. T-YW and Z-Y H carried out the bipolar resistive switching characteristics of the memristor. HZ and S-JD revised the manuscript. LC, Q-QS, and D-WZ supervised the whole work. All authors critically read and approved the final manuscript.

Competing Interests

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors’ Affiliations

State Key Laboratory of ASIC and System, School of Microelectronics, Fudan University, Shanghai, 200433, China
Frontier Science Center for Synthetic Biology and Key Laboratory of Systems Bioengineering (MOE), and School of Chemical Engineering and Technology, Tianjin University, Tianjin, 300072, China


  1. Jeong DS, Kim KM, Kim S, Choi BJ, Hwang CS (2016) Memristors for energy-efficient new computing paradigms. Adv Electron Mater 2:1600090View ArticleGoogle Scholar
  2. Kim M-K, Lee J-S (2018) Short-term plasticity and long-term potentiation in artificial biosynapses with diffusive dynamics. ACS Nano 12:1680–1687View ArticleGoogle Scholar
  3. Li C, Hu M, Li Y, Jiang H, Ge N, Montgomery E, Zhang J, Song W, Dávila N, Graves CE, Li Z, Strachan JP, Lin P, Wang Z, Barnell M, Wu Q, Williams RS, Yang JJ, Xia Q (2018) Analogue signal and image processing with large memristor crossbars. Nat Electron 1:52–59View ArticleGoogle Scholar
  4. Shi J, Ha SD, Zhou Y, Schoofs F, Ramanathan S (2013) A correlated nickelate synaptic transistor. Nat Commun 4:2676View ArticleGoogle Scholar
  5. Fuller EJ, Gabaly FE, Léonard F, Agarwal S, Plimpton SJ, Jacobs-Gedrim RB, James CD, Marinella MJ, Talin AA (2017) Li-ion synaptic transistor for low power analog computing. Adv Mater 29:1604310View ArticleGoogle Scholar
  6. Yan X, Zhang L, Chen H, Li X, Wang J, Liu Q, Lu C, Chen J, Wu H, Zhou P (2018) Graphene oxide quantum dots based memristors with progressive conduction tuning for artificial synaptic learning. Adv Funct Mater 28:1803728View ArticleGoogle Scholar
  7. Wu C, Kim TW, Choi HY, Strukov DB, Yang JJ (2017) Flexible three-dimensional artificial synapse networks with correlated learning and trainable memory capability. Nat Commun 8:752View ArticleGoogle Scholar
  8. He H-K, Yang R, Zhou W, Huang H-M, Xiong J, Gan L, Zhai T-Y, Guo X (2018) Photonic potentiation and electric habituation in ultrathin memristive synapses based on monolayer MoS2. Small 14:1800079View ArticleGoogle Scholar
  9. Wang T-Y, He Z-Y, Liu H, Chen L, Zhu H, Sun Q-Q, Ding S-J, Zhou P, Zhang DW (2018) Flexible electronic synapses for face recognition application with multimodulated conductance states. ACS Appl Mater Interfaces 10:37345–37352View ArticleGoogle Scholar
  10. Kim S, Choi B, Lim M, Yoon J, Lee J, Kim H-D, Choi S-J (2017) Pattern recognition using carbon nanotube synaptic transistors with an adjustable weight update protocol. ACS Nano 11:2814–2822View ArticleGoogle Scholar
  11. Yao P, Wu H, Gao B, Eryilmaz SB, Huang X, Zhang W, Zhang Q, Deng N, Shi L, Wong HSP, Qian H (2017) Face classification using electronic synapses. Nat Commun 8:15199View ArticleGoogle Scholar
  12. Li B, Liu Y, Wan C, Liu Z, Wang M, Qi D, Yu J, Cai P, Xiao M, Zeng Y, Chen X (2018) Mediating short-term plasticity in an artificial memristive synapse by the orientation of silica mesopores. Adv Mater 30:1706395View ArticleGoogle Scholar
  13. Wang L-G, Zhang W, Chen Y, Cao Y-Q, Li A-D, Wu D (2017) Synaptic plasticity and learning behaviors mimicked in single inorganic synapses of Pt/HfOx/ZnOx/TiN memristive system. Nanoscale Res Lett 12:65View ArticleGoogle Scholar
  14. Yan X, Wang J, Zhao M, Li X, Wang H, Zhang L, Lu C, Ren D (2018) Artificial electronic synapse characteristics of a Ta/Ta2O5-x/Al2O3/InGaZnO4 memristor device on flexible stainless steel substrate. Appl Phys Lett 113:013503View ArticleGoogle Scholar
  15. Yang CS, Shang DS, Liu N, Shi G, Shen X, Yu RC, Li YQ, Sun Y (2017) A synaptic transistor based on quasi-2D molybdenum oxide. Adv Mater 29:1700906View ArticleGoogle Scholar
  16. Tan Z-H, Yang R, Terabe K, Yin X-B, Zhang X-D, Guo X (2015) Synaptic metaplasticity realized in oxide memristive devices. Adv Mater 28:377–384View ArticleGoogle Scholar
  17. van de Burgt Y, Lubberman E, Fuller EJ, Keene ST, Faria GC, Agarwal S, Marinella MJ, Alec Talin A, Salleo A (2017) A non-volatile organic electrochemical device as a low-voltage artificial synapse for neuromorphic computing. Nat Mater 16:414View ArticleGoogle Scholar
  18. Jiang J, Guo J, Wan X, Yang Y, Xie H, Niu D, Yang J, He J, Gao Y, Wan Q (2017) 2D MoS2 neuromorphic devices for brain-like computational systems. Small 13:1700933View ArticleGoogle Scholar
  19. Sun J, Oh S, Choi Y, Seo S, Oh MJ, Lee M, Lee WB, Yoo PJ, Cho JH, Park J-H (2018) Optoelectronic synapse based on IGZO-alkylated graphene oxide hybrid structure. Adv Funct Mater 28:1804397View ArticleGoogle Scholar
  20. Wang T, Yu L, Chen L, Liu H, Zhu H, Sun Q, Ding S, Zhou P, Zhang DW (2017) Atomic layer deposited Hf0.5Zr0.5O2-based flexible RRAM, 2017 IEEE 12th International Conference on ASIC (ASICON), pp 203–206Google Scholar
  21. Chen L, Wang T-Y, Dai Y-W, Cha M-Y, Zhu H, Sun Q-Q, Ding S-J, Zhou P, Chua L, Zhang DW (2018) Ultra-low power Hf0.5Zr0.5O2 based ferroelectric tunnel junction synapses for hardware neural network applications. Nanoscale 10:15826–15833View ArticleGoogle Scholar
  22. Oh S, Kim T, Kwak M, Song J, Woo J, Jeon S, Yoo IK, Hwang H (2017) HfZrOx-based ferroelectric synapse device with 32 levels of conductance states for neuromorphic applications. IEEE Electron Device Lett 38:732–735View ArticleGoogle Scholar
  23. Yoong HY, Wu H, Zhao J, Wang H, Guo R, Xiao J, Zhang B, Yang P, Pennycook SJ, Deng N, Yan X, Chen J (2018) Epitaxial ferroelectric Hf0.5Zr0.5O2 thin films and their implementations in memristors for brain-inspired computing. Adv Funct Mater 28:1806037View ArticleGoogle Scholar
  24. Lin Y, Zeng T, Xu H, Wang Z, Zhao X, Liu W, Ma J, Liu Y (2018) Transferable and flexible artificial memristive synapse based on WOx Schottky junction on arbitrary substrates. Adv Electron Mater 4:1800373View ArticleGoogle Scholar
  25. John RA, Ko J, Kulkarni MR, Tiwari N, Chien NA, Ing NG, Leong WL, Mathews N (2017) Flexible ionic-electronic hybrid oxide synaptic TFTs with programmable dynamic plasticity for brain-inspired neuromorphic computing. Small 13:1701193View ArticleGoogle Scholar
  26. Dai S, Wu X, Liu D, Chu Y, Wang K, Yang B, Huang J (2018) Light-stimulated synaptic devices utilizing interfacial effect of organic field-effect transistors. ACS Appl Mater Interfaces 10:21472–21480View ArticleGoogle Scholar
  27. Wang H, Zhu B, Wang H, Ma X, Hao Y, Chen X (2016) Ultra-lightweight resistive switching memory devices based on silk fibroin. Small 12:3360–3365View ArticleGoogle Scholar
  28. Yu S, Guan X, Wong HP (2012) On the switching parameter variation of metal oxide RRAM—part II: model corroboration and device design strategy. IEEE Trans Electron Devices 59:1183–1188View ArticleGoogle Scholar
  29. Wu Q, Wang H, Luo Q, Banerjee W, Cao J, Zhang X, Wu F, Liu Q, Li L, Liu M (2018) Full imitation of synaptic metaplasticity based on memristor devices. Nanoscale 10:5875–5881View ArticleGoogle Scholar
  30. Abbott LF, Regehr WG (2004) Synaptic computation. Nature 431:796View ArticleGoogle Scholar
  31. Zhang X, Liu S, Zhao X, Wu F, Wu Q, Wang W, Cao R, Fang Y, Lv H, Long S, Liu Q, Liu M (2017) Emulating short-term and long-term plasticity of bio-synapse based on cu/a-Si/Pt memristor. IEEE Electron Device Lett 38:1208–1211View ArticleGoogle Scholar


© The Author(s). 2019