Physical mechanisms involved in the formation and operation of memory devices based on a monolayer of gold nanoparticle-polythiophene hybrid materials

Abstract

Understanding the physical and chemical mechanisms occurring during the forming process and operation of an organic resistive memory device is a requisite for better performance. Various mechanisms were suggested in vertically stacked memory structures, but the analysis remains indirect and needs destructive characterization (e.g. analysis of the cross-section to access the organic layers sandwiched between electrodes). Here, we report a study on a planar, monolayer thick, hybrid nanoparticle/molecule device (10 nm gold nanoparticles embedded in an electro-generated poly(2-thienyl-3,4-(ethylenedioxy)thiophene) layer), combining in situ physical (scanning electron microscopy, physicochemical thermogravimetry and mass spectroscopy, and Raman spectroscopy) and electrical (temperature dependent current-voltage) characterization on the same device. We demonstrate that the forming process causes an increase in the gold particle size, almost 4 times larger than the starting nanoparticles, and that the organic layer undergoes a significant chemical rearrangement from an sp³ to sp² amorphous carbon material. Temperature dependent electrical characterization of this nonvolatile memory confirms that the charge transport mechanism in the device is consistent with a trap-filled space charge limited current in the off state, with the sp² amorphous carbon material containing many electrically active defects.

Fig. 1. (a) Typical voltage sequence for the memory behavior and (b) the corresponding I–V curves for the ON-to-OFF and OFF-to-ON switches of a device after the forming process with a length L = 500 nm and width W = 1 mm. The current–voltage I–V traces are numbered according to the voltage sequences shown in (a). Inset, SEM image of a network of functionalized-GNP monolayer on a nanogap platinum electrodes (electrode spacing L = 200 nm and electrode width W = 100 nm); bar scale represents 30 nm; and schematic magnified view of a GNP interspace from ref. 3. Reprinted from ref. 3, copyright American Chemical Society 2017.

Fig. 2. SEM images of a pTEDOT-C10-S-GNP monolayer in a 200 nm channel length (GNP size 10 nm) before (a) and after (b) the forming process consisting of 3–4 voltage sweeps from 0 to 20 V (corresponding to an electric field maximum of 100 MV m⁻¹).

Fig. 3. Top: (a) SEM image of the pTEDOT-C10-S-GNP monolayer (GNP size 10 nm). (b) SEM image of a formed-pTEDOT-C10-S-GNP monolayer after the forming process. Bottom: corresponding HAADF-STEM images of the cross section between the electrode along the dotted line in SEM images, (c) for device before the forming process (scale bar 10 nm) and (d) after the forming process (scale bar 50 nm). The two devices have the same channel length of 500 nm.

Fig. 4. SEM images of a pTEDOT-C10-S-GNP monolayer after thermal annealing at different temperatures from 423 to 623 K with 50 K steps. Images were acquired in the gap electrode.

Fig. 5. Thermogravimetry analysis (TGA) coupled with mass spectrometry (MS) of (a) a brittle-porous sample of TEDOT-C10-S-GNPs with 2 nm diameter GNPs (scheme in the inset) (b) the thin p(TEDOT) film on Si/Pt substrate (p(TEDOT) structure in the inset).

Fig. 6. Raman spectroscopy acquired from p(TEDOT) polymer deposited on Pt-silicon substrate before and after a thermal annealing in helium at 940 K.

Fig. 7. Arrhenius plot of the current in the ON and OFF states in the temperature range from 80 to 300 K read at 1 V for a formed-pTEDOT-C10-S-GNP monolayer in a 500 nm × 1000 μm channel device.

Fig. 8. Typical current I versus voltage V curves (log–log scale) for the memory behavior measured on a formed-pTEDOT-C10-S-GNP device with a 1 μm gap length. Dots correspond to experimental data and red lines to the fitting adjustments before switching regions; i.e. in region 1 (0.1 to 1 V) and region 2 (1 to 10 V) for the OFF state, and from 0.1 to 7 V for the ON state. From these adjustments at low and high voltage and with eqn (1) and eqn (2), we estimate the density of trapped charge n_t ∼ 1.6 × 10¹⁴ cm⁻³.