RRAM-Based Neuromorphic Computing Systems
397
The second approach utilizes a thin insertion layer to either prevent unwanted
oxide formation at the active electrode/oxide interface or to obtain a better control
of cations injection and filament formation during device operation. Insertion of Ti
at the interface of Cu/TaO x -based devices reduced the cycle-to-cycle and device-todevice variation with significant improvement on device dynamic ratio (from ~10
to ~100) [55]. This was attributed to the formation of TiO x instead of CuO x at the
interface of Cu/Ti/TaO x structure. Insertion of thin TiW layer at the interface of
Cu/AlO x has also been shown to improve the overall performance of the device
[56]. This barrier layer helped to maintain the cell structural integrity up to BEOL
processing temperature of 400 °C. It also prevented gradual drifting of conductance
states due to parasitic diffusion effects, resulted in excellent cycling control. The
W\Al 2 O 3 \TiW\Cu RRAM cell fabricated on 90 nm W plug exhibited high voltagedisturb immunity with high dynamic ratio of >100 and fast switching operation of
~10 ns with <3 V pulse amplitude. The dynamic ratio was further enhanced to ~1000
by the insertion of WO x by thermal oxidation of the W plug at 500 °C. It was ascribed
to a filament constriction at WO x /Al 2 O 3 interface obtaining an hourglass conductive filament shape that enabled deeper RESET process. In Al/Cu/GeSe x /TaO x /W,
TaO x [57] insertion layer at the inert electrode side of the GeSe x switching layer
provided an additional layer with lower Cu mobility to alter the filament shape and
dimension during the switching. Improvement in switching stability was attributed
to nanofilament confinement within TaO x layer.
The third approach uses a mixture in the form of metal alloy as the source of
cations. The first example is copper tellurium (Cu x Te 1-x ) as active ions source. It
was first demonstrated on 180 nm CMOS technology in CuTe/GdO x /W structure
[49]. It was able to achieve excellent 2-bit memory property with excellent retention
under different compliance current levels. Dynamic ratio of ~1000 (10 M/10 k)
was achieved under programming parameters of 3 V, 110 μA, and 5 ns for SET
and −1.7 V, 125 μA, and 1 ns for RESET. The device also showed potential of
gradual RESET, but further optimization of pulse amplitude and width was required.
The effect of Cu and Te composition on the active electrode was investigated in
Cu x Te 1-x/ Al 2 O 3 /Si cells. It was found that the RRAM cells exhibit volatile switching
(SET), non-volatile switching with gradual RESET, and non-volatile switching with
abrupt RESET as the Te content decreases. This was associated with higher energy
barrier to inject Cu into Al 2 O 3 for Cu-Te phase compared to pure Cu. This provides a
very useful insight on how the cation source characteristic is able to tune the overall
RRAM cell property for specific applications.
An alternative implementation of the cation devices was proposed through
stochastic STDP learning rules. Instead of trying to precisely control the switching
operation to produce analog deterministic behavior, the binary probabilistic switching
nature of the device is being exploited under these learning rules. This approach
provides the equivalent system level functionalities to that of network utilizing the
analog synaptic devices under deterministic learning rules [58]. Supervised and
unsupervised NNs have been demonstrated using the binary probabilistic synapses
[59, 60]. The unsupervised NN was demonstrated using 1T1R and 1R system with
Ag/GST RRAM structure as synaptic device. Strong (pulse duration of ≥10 μs)
397
The second approach utilizes a thin insertion layer to either prevent unwanted
oxide formation at the active electrode/oxide interface or to obtain a better control
of cations injection and filament formation during device operation. Insertion of Ti
at the interface of Cu/TaO x -based devices reduced the cycle-to-cycle and device-todevice variation with significant improvement on device dynamic ratio (from ~10
to ~100) [55]. This was attributed to the formation of TiO x instead of CuO x at the
interface of Cu/Ti/TaO x structure. Insertion of thin TiW layer at the interface of
Cu/AlO x has also been shown to improve the overall performance of the device
[56]. This barrier layer helped to maintain the cell structural integrity up to BEOL
processing temperature of 400 °C. It also prevented gradual drifting of conductance
states due to parasitic diffusion effects, resulted in excellent cycling control. The
W\Al 2 O 3 \TiW\Cu RRAM cell fabricated on 90 nm W plug exhibited high voltagedisturb immunity with high dynamic ratio of >100 and fast switching operation of
~10 ns with <3 V pulse amplitude. The dynamic ratio was further enhanced to ~1000
by the insertion of WO x by thermal oxidation of the W plug at 500 °C. It was ascribed
to a filament constriction at WO x /Al 2 O 3 interface obtaining an hourglass conductive filament shape that enabled deeper RESET process. In Al/Cu/GeSe x /TaO x /W,
TaO x [57] insertion layer at the inert electrode side of the GeSe x switching layer
provided an additional layer with lower Cu mobility to alter the filament shape and
dimension during the switching. Improvement in switching stability was attributed
to nanofilament confinement within TaO x layer.
The third approach uses a mixture in the form of metal alloy as the source of
cations. The first example is copper tellurium (Cu x Te 1-x ) as active ions source. It
was first demonstrated on 180 nm CMOS technology in CuTe/GdO x /W structure
[49]. It was able to achieve excellent 2-bit memory property with excellent retention
under different compliance current levels. Dynamic ratio of ~1000 (10 M/10 k)
was achieved under programming parameters of 3 V, 110 μA, and 5 ns for SET
and −1.7 V, 125 μA, and 1 ns for RESET. The device also showed potential of
gradual RESET, but further optimization of pulse amplitude and width was required.
The effect of Cu and Te composition on the active electrode was investigated in
Cu x Te 1-x/ Al 2 O 3 /Si cells. It was found that the RRAM cells exhibit volatile switching
(SET), non-volatile switching with gradual RESET, and non-volatile switching with
abrupt RESET as the Te content decreases. This was associated with higher energy
barrier to inject Cu into Al 2 O 3 for Cu-Te phase compared to pure Cu. This provides a
very useful insight on how the cation source characteristic is able to tune the overall
RRAM cell property for specific applications.
An alternative implementation of the cation devices was proposed through
stochastic STDP learning rules. Instead of trying to precisely control the switching
operation to produce analog deterministic behavior, the binary probabilistic switching
nature of the device is being exploited under these learning rules. This approach
provides the equivalent system level functionalities to that of network utilizing the
analog synaptic devices under deterministic learning rules [58]. Supervised and
unsupervised NNs have been demonstrated using the binary probabilistic synapses
[59, 60]. The unsupervised NN was demonstrated using 1T1R and 1R system with
Ag/GST RRAM structure as synaptic device. Strong (pulse duration of ≥10 μs)