Fabrication of resistive switching memory devices

(a) Substrate and bottom electrode

Resistive memory devices based on HfOx capacitor structure were fabricated as follows. The p-type 4-inch Si (100) wafer was cleaned by the standard RCA cleaning process. Then, a 300-nm-thick SiO2 layer, served as an isolated layer to prevent the leakage current from Si substrate, was grown on the Si wafer by wet oxidation at 1000

°C. TiN metal as a bottom electrode was deposited onto the SiO2 layer by sputtering.

(b) HfOx thin film

Resistive memory devices were fabricated as follows. TiN metal as a bottom electrode was deposited on an 8 inch Si wafer by sputtering. Then, nonstoichiometric HfOx films were deposited by atomic layer deposition (ALD) (ASM, Polygon 8200) at temperature of 300 °C. Hafnium tetrachloride (HfCl4) and water (H2O) with a compositional ratio of 5:2 and 5:3 were both used as reactants for HfOx deposition.

The primitive chemical-formula is shown as follows.

HfCl4 + 4H2O →Hf(OH)4 + HCl ↑ (2-1) Hf(OH)4 heating → HfO2 + H2O (2-2)

(c) Top electrode

For a universal understanding of the switching properties of HfOx thin film, we not only fabricated various kinds of electrode materials but also varied different deposition condition during top electrode deposition.

First, we discuss the top electrode effect on the resistive switching properties as discussed in section 3.3.2. Various kinds of metals as top electrodes were deposited by dc sputtering using a shadow mask at room temperature, such as Pt, Pd, Cu, Ni, Ti, Al and etc. The thickness of metals was varied from 30 to 100 nm. The circular top electrodes were formed with radiuses of 100-800 μm.

In section 3.3.3, for a further investigation on the effect of the reactive metal Al, the stacked top electrodes Pt/Al were also fabricated with the thickness of Al varied from 0.5-10 nm and capping a 30 nm thick Pd electrode above. To maintain high quality of metal Pd/Al interface, we deposited the stacked film in the same chamber.

The pure Pd metal with 30 nm thick was also deposited for comparison. The RF sputtering power and deposition pressure were kept at 30 W and 3 mtorr, respectively.

Only argon gas (Ar, 99%) was used as the sputtering gas during deposition. Shadow mask with a diameter of 200 μm was used as the defined contact pad for measurement.

In section 3.3.5, we compared the process effect on the switching characteristics.

Dc magnetron sputtering and/or e-beam evaporation were both used for Pt deposition as top electrode through shadow mask with a diameter of 200 μm. The operating current, voltage and working pressure for e-beam evaporation were 0.4 mA, 3 V, and 9×10^-6 torr, respectively. Where sputtering was used on the Pt target (99.99%), minutes under the same gas composition as for the deposition to establish a steady state. The base pressure for sputtering the Pt top electrode was 2 ×10^-6 torr using a cryopump and the working pressure of Ar gas (99.99%) was kept at 7×10^-3 torr in all samples during the deposition process. The distance between the target and substrate

was kept at 13 cm. For the stacked electrode, the thin Pt layer, with a thickness of 5 nm or 15 nm, was deposited by e-beam evaporation first, followed by a 70 nm thick Pt electrode with sputtering power of 100 W, while keeping all the process parameters the same. Sputtering power was varied between 50 W (94 V, 0.53 A), 100 W (161 V, 0.62A), and 200 W (237 V, 0.84 A) to investigate the effect of different levels of sputtering power on the electrical characteristics. Pt metal was deposited by e-beam evaporation (Pt (E)) or dc sputtering (Pt (S)) and electrically analyzed. Furthermore, stacked pt electrode samples, consisting of a thin Pt layer with a thickness of 5 nm or 15 nm deposited by e-beam evaporation followed by a 70 nm-thick Pt layer deposited by dc sputtering, were denoted as Pt (E50+S) and Pt (E150+S), respectively.

In section 3.3.6, we used the transparent electrode ITO as top electrode to investigate the impact of the UV laser on the switching properties. ITO with a thickness of 100 nm was deposited by rf-magnetron sputtering using a shadow mask with a 200 μm diameter at RT. The ITO film directly deposited on the glass substrate was also experimented on to measure the transmittance. The flow rate of the Ar sputtering gas was maintained at 15 sccm during deposition. Deposition power of 100 W and process pressure of 5 mtorr were used. An average intensity of 80 mW/cm² at

= 365 nm of UV light from a mercury lamp was used in our experiments. The UV-light source was 8 cm away from the samples. UV-light exposes on the HfOx/TiN sample was experimented inside glove box, which requires oxygen and moisture concentration below 1 ppm. Exposure time of 30, 120, 300 and 600 seconds was adopted to perform the experiments.

(d) Treatment

After HfOx film deposition, some samples were thermally annealed in a rapid thermal annealing (RTA) system at 400 or 500 °C for 30, 60, 90, and 120 seconds.

Argon (Ar) and oxygen were used as annealing gas for the annealing systems. Then, nickel (Ni) as a top electrode was deposited by e-beam evaporation using a shadow mask at RT. The thickness of Ni and diameter of top electrode was 50 nm and 500 μm, respectively. The topic is discussed in section 3.3.4.

2.5.2 Deposition of Gd2O3 RRAM devices

(a) Experimental details of Ti/Gd2O3/Pt RRAM devices

For the Si based Gd2O3 RRAM, the p-type 4-inch Si (100) wafer was cleaned by the standard RCA cleaning process. Then, a 300-nm-thick SiO2 layer, served as an isolated layer to prevent the leakage current from Si substrate, was grown on the Si wafer by wet oxidation at 1000 °C. A bilayer metal Pt/Ti was deposited on SiO2 layer as a bottom electrode by e-beam evaporation. Thin metal Ti was used to enhance the adhesion of Pt electrode on SiO2 layer. A 25 nm-thick Gd2O3 thin film was then deposited on Pt/Ti/SiO2/Si substrates at room temperature (RT) by pulsed laser deposition (PLD) with oxygen pressure of 0.01 Pa using a metal Gd target. Finally, the bilayer electrode Pt/Ti was deposited on Gd2O3 as the top electrode. Shadow mask with a diameter of 200 μm was used to define contact pad for measurement.

(b) Experimental details of the ITO/Gd2O3/ITO T-RRAM devices

For the transparent RRAM, we use commercial ITO glass with as our bottom electrode and substrate. Then, we use PLD to deposit Gd2O3 using ceramic Gd2O3

target as the resistive layer of TRRAM device at room temperature (RT), 200 °C, and 300 °C. Lambda physics of excimer layser was furnished by λ=248 nm, pulse duration of 25 ns, repetition rate of 3 Hz, and the laser energy of 500 mJ. The process parameter of oxygen pressure is 10 mtorr and the thickness is controlled to 25 nm. At last, we also use ITO as a top electrode, a 100 nm-thick ITO thin film is deposited by

dc sputter system using a shadow mask with a diameter of 1200 μm.

2.5.3 Deposition of LAO T-RRAM devices

The commercial ITO (Corning 1737) glass substrates deposited with 300-nm-thick transparent conducting indium tin oxide (ITO) thin films prepared as the bottom electrode. Then, PLD was used to deposit the LaAlO using LaAlO target as the resistive layer of T-RRAM device. A KrF excimer laser (λ=248 nm) was used as the light source of PLD with a pulse duration of 25 ns, a repetitio n rate of 3Hz, and a laser energy of 500 mJ. The distance from the target to the substrate was 10 cm and the target rotation rate was 5 rpm. The ambient oxygen pressure, which is the most important process parameter of PLD, was varied with 7×10^-3, 1.4×10^-2, and 2.8×10^-2 torr. At last, we use ITO as a top electrode, a 100 nm-thick ITO thin film is deposited by dc sputter system using a shadow mask with a diameter of 1200 μm. Electrical characteristics were performed on Agilent 4156C semiconductor parameter analyzer (SPA) at room temperature. Current flow from the top to the bottom electrode was defined as positive sweep.