The same measurements were performed for Ag/Ag-PEO/Pt, Ag/PEO/Pt, and Pt/
Ag-PEO/Pt planar devices with gaps of 6–9 μm. Depending on the electrode material
and salt inclusion in the PEO matrix, different filament growth behaviors as well as
different current responses were observed. Ag included PEO-based devices
exhibited similar current behavior as the Ag/Ag-PEO/Ag device (Fig. 4a), whereas
PEO-based devices showed that the current never reach the compliance level due to
the large gap. The symmetric electrode configuration (Ag/Ag) induced gradual and
random growth of conducting filaments, but the asymmetric electrode configuration
(Ag/Pt) promoted unidirectional filament growth.
3.2 Impacts of Device Configuration and Experimental
Parameters
The observation of a planar device enabled us to investigate how redox reactions and
subsequent filament formations take place under certain bias conditions. However,
for gaps of μm scale, the devices exhibit no switching behavior even after the
forming process is clearly observed. This indicates that planar devices with μm
gaps are difficult to correlate the filament growth processes to the actual switching
behavior. Hence, we used EB lithography processes to fabricate planar devices.
Owing to the decreased distances between the electrodes, stable switching behavior
could be observed, and the subsequent SEM observations allowed us to investigate
the filament growth processes affected by the device and experimental parameters
such as the gap between electrodes, salt inclusion, and I CC [19]. Most of this
subsection is reproduced with permission from Ref. 19. Copyright 2016 the Royal
Society of Chemistry.
Figure 5a–c show SEM images of Ag/Ag-PEO/Pt devices with gaps of 2, 1, and
0.5 μm, taken after the forming process was done under positive bias sweeping with
a constant sweep rate and a certain I CC . All the SEM images show dendritic filament
morphologies, in which one of the filaments connects between the electrodes. For the
smallest gap (0.5 μm), narrower filaments grow randomly from the Pt electrode. As
the gap increases, the filaments grow further with changing to dendritic morphologies. This indicates that the dendritic growth can be enhanced in devices with larger
gaps, although the filament growth is driven by the same dynamics. The forming
voltage is reduced for decreased gaps, suggesting that the number of Ag
+ ions
required to form a filament is reduced for smaller gaps. This is evidenced by the
decreased void area in the Ag electrode for reduced gaps.
As shown in Fig. 5d–f, the impact of the gap between electrodes on the filament
growth processes is more pronounced for Ag/PEO/Pt devices. No completed filament formation occurred for the largest gap (2 μm), even after biasing up to ~9 V,
although some precipitations appeared on the Pt electrode (Fig. 6d). This means that
the device cannot be turned on for such a gap range under the bias condition used.
However, when the gap was reduced, very narrow filaments, consisting of small Ag
Solid-Polymer-Electrolyte-Based Atomic Switches
147
Ag-PEO/Pt planar devices with gaps of 6–9 μm. Depending on the electrode material
and salt inclusion in the PEO matrix, different filament growth behaviors as well as
different current responses were observed. Ag included PEO-based devices
exhibited similar current behavior as the Ag/Ag-PEO/Ag device (Fig. 4a), whereas
PEO-based devices showed that the current never reach the compliance level due to
the large gap. The symmetric electrode configuration (Ag/Ag) induced gradual and
random growth of conducting filaments, but the asymmetric electrode configuration
(Ag/Pt) promoted unidirectional filament growth.
3.2 Impacts of Device Configuration and Experimental
Parameters
The observation of a planar device enabled us to investigate how redox reactions and
subsequent filament formations take place under certain bias conditions. However,
for gaps of μm scale, the devices exhibit no switching behavior even after the
forming process is clearly observed. This indicates that planar devices with μm
gaps are difficult to correlate the filament growth processes to the actual switching
behavior. Hence, we used EB lithography processes to fabricate planar devices.
Owing to the decreased distances between the electrodes, stable switching behavior
could be observed, and the subsequent SEM observations allowed us to investigate
the filament growth processes affected by the device and experimental parameters
such as the gap between electrodes, salt inclusion, and I CC [19]. Most of this
subsection is reproduced with permission from Ref. 19. Copyright 2016 the Royal
Society of Chemistry.
Figure 5a–c show SEM images of Ag/Ag-PEO/Pt devices with gaps of 2, 1, and
0.5 μm, taken after the forming process was done under positive bias sweeping with
a constant sweep rate and a certain I CC . All the SEM images show dendritic filament
morphologies, in which one of the filaments connects between the electrodes. For the
smallest gap (0.5 μm), narrower filaments grow randomly from the Pt electrode. As
the gap increases, the filaments grow further with changing to dendritic morphologies. This indicates that the dendritic growth can be enhanced in devices with larger
gaps, although the filament growth is driven by the same dynamics. The forming
voltage is reduced for decreased gaps, suggesting that the number of Ag
+ ions
required to form a filament is reduced for smaller gaps. This is evidenced by the
decreased void area in the Ag electrode for reduced gaps.
As shown in Fig. 5d–f, the impact of the gap between electrodes on the filament
growth processes is more pronounced for Ag/PEO/Pt devices. No completed filament formation occurred for the largest gap (2 μm), even after biasing up to ~9 V,
although some precipitations appeared on the Pt electrode (Fig. 6d). This means that
the device cannot be turned on for such a gap range under the bias condition used.
However, when the gap was reduced, very narrow filaments, consisting of small Ag
Solid-Polymer-Electrolyte-Based Atomic Switches
147