Moon, K.
(Pohang University of Science and Technology (POSTECH))
,
Lim, S.
(Pohang University of Science and Technology (POSTECH))
,
Park, J.
(Pohang University of Science and Technology (POSTECH))
,
Sung, C.
(Pohang University of Science and Technology (POSTECH))
,
Oh, S.
(Pohang University of Science and Technology (POSTECH))
,
Woo, J.
(Pohang University of Science and Technology (POSTECH))
,
Lee, J.
(Pohang University of Science and Technology (POSTECH))
,
Hwang, H.
(Pohang University of Science and Technology (POSTECH))
Hardware artificial neural network (ANN) systems with high density synapse array devices can perform massive parallel computing for pattern recognition with low power consumption. To implement a neuromorphic system with on-chip training capability, we need to develop an ideal synapse device with var...
Hardware artificial neural network (ANN) systems with high density synapse array devices can perform massive parallel computing for pattern recognition with low power consumption. To implement a neuromorphic system with on-chip training capability, we need to develop an ideal synapse device with various device requirements, such as scalability, MLC characteristics, low power operation, data retention, and symmetric/linear conductance changes under potentiation/depression modes. Although various devices have been proposed for synapse applications, they have limitations for application in neuromorphic systems. In this paper, we will cover various RRAM synapse devices, such as filamentary switching RRAM (HfOx, TaOx, Cu-CBRAM) and analog RRAM devices, based on interface resistive switching (Pr0.7Ca0.3MnOx and TiOx) and ferroelectric polarization (HfZrOx). By optimizing potentiation/depression conditions, we could improve the conductance linearity and MLC characteristics of filamentary synapse devices. Interface RRAM has better MLC characteristics with limited retention and conductance linearity. By controlling the reactivity of metal electrodes and the oxygen concentration in oxides, we can modulate the synapse characteristics. Metal-Ferroelectric-Insulator-Semiconductor (MFIS) FET devices exhibit good retention characteristics and analog memory characteristics due to polarization. Based on various synapse device characteristics, we have estimated the pattern recognition accuracy of MNIST handwritten digits and CIFAR-10 datasets. We have confirmed that synapse device characteristics directly affect the pattern recognition accuracy of ANNs. In order to simultaneously satisfy all the requirements of synapse devices, it is necessary to develop new technology capable of controlling the movement of oxygen vacancies and metal ions at the atomic scale. Considering the limited synapse characteristics of current 2-terminal RRAM devices, hardware ANNs capable of only off-chip training can be constructed by optimizing the current RRAM devices by limiting the bit number. A 3-terminal synapse device or a device based on a new operation principle should be developed as an alternative for on-chip training applications.Graphic AbstractWe demonstrated a proton-based 3-terminal synapse device which shows symmetric conductance change characteristics. Using the optimized device, we successfully confirmed the improved classification accuracy of neural networks for on-chip training.
Hardware artificial neural network (ANN) systems with high density synapse array devices can perform massive parallel computing for pattern recognition with low power consumption. To implement a neuromorphic system with on-chip training capability, we need to develop an ideal synapse device with various device requirements, such as scalability, MLC characteristics, low power operation, data retention, and symmetric/linear conductance changes under potentiation/depression modes. Although various devices have been proposed for synapse applications, they have limitations for application in neuromorphic systems. In this paper, we will cover various RRAM synapse devices, such as filamentary switching RRAM (HfOx, TaOx, Cu-CBRAM) and analog RRAM devices, based on interface resistive switching (Pr0.7Ca0.3MnOx and TiOx) and ferroelectric polarization (HfZrOx). By optimizing potentiation/depression conditions, we could improve the conductance linearity and MLC characteristics of filamentary synapse devices. Interface RRAM has better MLC characteristics with limited retention and conductance linearity. By controlling the reactivity of metal electrodes and the oxygen concentration in oxides, we can modulate the synapse characteristics. Metal-Ferroelectric-Insulator-Semiconductor (MFIS) FET devices exhibit good retention characteristics and analog memory characteristics due to polarization. Based on various synapse device characteristics, we have estimated the pattern recognition accuracy of MNIST handwritten digits and CIFAR-10 datasets. We have confirmed that synapse device characteristics directly affect the pattern recognition accuracy of ANNs. In order to simultaneously satisfy all the requirements of synapse devices, it is necessary to develop new technology capable of controlling the movement of oxygen vacancies and metal ions at the atomic scale. Considering the limited synapse characteristics of current 2-terminal RRAM devices, hardware ANNs capable of only off-chip training can be constructed by optimizing the current RRAM devices by limiting the bit number. A 3-terminal synapse device or a device based on a new operation principle should be developed as an alternative for on-chip training applications.Graphic AbstractWe demonstrated a proton-based 3-terminal synapse device which shows symmetric conductance change characteristics. Using the optimized device, we successfully confirmed the improved classification accuracy of neural networks for on-chip training.
Proc. Symp. VLSI Technol. Goux 2013 L. Goux , et al., Understanding of the intrinsic characteristics and memory trade-offs of sub-μA filamentary RRAM operation , in Proc. Symp. VLSI Technol. , 2013 , pp. T162-T163
IEEE Trans. Electron Devices Burr 62 11 3498 2015 10.1109/TED.2015.2439635
Proc. IEEE Conf. Nanotechnol. Huang 2008 10.1109/NANO.2008.163 Q. Huang , et al., Electrical properties of Cu nanowires in Proc. IEEE Conf. Nanotechnol. , 2008 , pp. 549-552 , 10.1109/NANO.2008.163
IEEE Electron Device Lett. Kim 38 9 1228 2017 10.1109/LED.2017.2730959
Proc. IEEE Int. Rel. Phys. Symp. (IRPS) Celano 2015 10.1109/IRPS. 2015.7112813 U. Celano , et al., Conductive filaments multiplicity as a variability factor in CBRAM in Proc. IEEE Int. Rel. Phys. Symp. (IRPS) , 2015 , pp. MY.11.1-MY.11.3 , 10.1109/IRPS. 2015.7112813
IEDM Tech. Dig. Park 3.7.1 2011 10.1109/IEDM.2011.6131484
IEEE Electron Device Lett. Park 32 1 63 2011 10.1109/LED.2010.2084560
Proc. NANOARCH Bürger 2014 10.1109/NANOARCH.2014.6880493 J. Bürger and C.Teuscher , Volatile memristive devices as short-term memory in a neuromorphic learning architecture in Proc. NANOARCH , 2014 , pp. 104-109 , 10.1109/NANOARCH.2014.6880493
Proc. IEEE VLSI Technol. Fujii 2016 S. Fujii et al., First demonstration and performance improvement of ferroelectric HfO2-based resistive switch with low operation current and intrinsic diode property in Proc. IEEE VLSI Technol. , 2016 , pp. 148-149
Appl. Phys. Lett. Dat 64 20 2673 1994 10.1063/1.111488
※ AI-Helper는 부적절한 답변을 할 수 있습니다.