Reply to “Comment on ‘Dynamic Processes of Resistive Switching in Metallic Filament-Based Organic Memory Devices’”

Abstract

J. Phys. Chem. C 2012, 116 (33), 17955−17959. DOI: 10.1021/jp305482c. J. Phys. Chem. C 2013, 10.1021/jp4014306. I our recent publication, we attributed the bipolar resistive switching phenomenon in the Cu/poly(3-hexylthiophene): [6,6]-phenyl C61-butyric acid methyl ester/indium−tin oxide (Cu/P3HT:PCBM/ITO in short) structure to the Cu filament, which was proposed to grow from the anode (Cu) to the cathode (ITO) and to rupture initially near the cathode (ITO). The same switching kinetics has also been demonstrated directly by the use of in situ transmission electron microscope (TEM) in Ag/α-Si/Pt, Cu(Ag)/ZrO2/Pt, 3 and Au/ZnO/Au sandwich structures. The switching kinetics in our work is completely opposite to that in memory cells based on traditional solid electrolytes (e.g., Cu-doped GeTe) with extremely high cation mobility. This contradictory situation somehow verifies that the filament growth mode is profoundly dependent on cation mobility in the storage media. Moreover, in our very recent work, Ag filaments have been confirmed to nucleate initially at the middle region (neither near the anode nor near the cathode) in the Ag/poly(3,4ethylene-dioxythiophene):poly(styrenesulfonate)/Pt planar device, further supporting the theory that the filament growth mode is dependent on cation mobility. Valov and Waser made a comment, in which they declared that switching kinetics in our work is violating the laws of physical chemistry and electrochemistry. In their opinion, if the Cu filament (irrespective of its dimensions) is mechanically (and therefore electrically) connected to the Cu electrode, only dissolution (growth) of the Cu filament is possible in the case that the Cu electrode is positively (negatively) charged. First of all, we strongly agree with Valov and Waser about the basic laws of physical chemistry and electrochemistry described in the comment. Unfortunately, they ignore the core issue in our paper that the organic material adopted has very low cation mobility, which is in sharp contrast to traditional solid electrolytes with extremely high cation mobility. Figure 1 shows a more detailed schematic of the Cu filament growth mode. According to this figure, the Cu ions originate initially from anode dissolution and then move toward the cathode under an external electric field (Figure 1a,b). However, due to very low mobility, the ions can only migrate an extremely short distance before becoming reduced by the oncoming electrons (Figure 1c). The precipitated Cu atoms in the vicinity of the anode share almost the same electrostatic potential as the anode, acting as an extension of the anode. Subsequently, the extended anode tip will be dissolved again and reduced nearby, depicted in Figure 1d,e, thus resulting in the phenomenon that the Cu filament grows from the anode to the cathode. This process means that the growth of metallic filaments is not a simple ′′growth′′ mode. Instead, it includes growth, dissolution, and the repetition of these two steps. Such a growth mode is reasonable from the viewpoint of kinetics and can not be explained only by the classic laws of physical chemistry and electrochemistry. Our concept is also supported by ref 2 in which the filaments in storage medium are composed of a string of particles, instead of perfect nanowires. We admit that Figure 5b in our recent paper is just a brief illustration of Figure 1 in this response. To directly observe the Cu filament, we fabricated a planar Cu/P3HT:PCBM/Cu device via a two-step ultraviolet lithography process followed by spin-coating. Figure 2a shows the schematic of the planar device and measurement configuration. A scanning electron microscope (SEM) image of the pristine planar device is displayed in Figure 2b. In this figure, one can see that the distance between these two Cu electrodes is about 200 nm. Under positive voltage sweep, a sharp current jump appears at ∼6.3 V (Figure 2c), indicating