Dynamic random access memory (DRAM) is used as the main memory of every modern computer, due to its high density, high speed and efficient memory function. Each DRAM cell consists of one transistor, which functions as a switch for the stored charge, and one capacitor where the positive or negative electric charges corresponding to the digital 1 or 0 data are stored (see Fig. 1a). For successful operation of DRAM, a large cell capacitance ($25 fF) and low leakage current at the operation voltage (10 À7 A cm À2 or 1 fA/ cell) are required because of the following reasons; during the reading operation, stored charge is shared between the cell capacitor and bit line, which is connected to the sense amplifier. In modern DRAMs, hundreds of capacitors are connected to one bit-line so that the bit-line capacitance is usually a few ten times larger than that of the capacitor. Therefore, for a bit-line voltage variation of $100 mV through the charge sharing, which is the sensing margin of the circuit, at least $25 fF of cell capacitance is necessary. [1][2][3] The low leakage current is also essential to ensure a sufficient refresh time.In a traditional Si-based capacitor, the target cell capacitance has been achieved by increasing the surface area of the capacitor (semiconductor-insulator-semiconductor, SIS, in Fig. 1b) while the dielectric thickness is scaled down according to the design rules.[4] More recently, innovations have been made in the component materials. A metal electrode, TiN or Ru, and a dielectric material with a higher-k value (k is the relative dielectric constant) than that of the SiO 2 /Si 3 N 4 layer (k $ 6-7), such as HfO 2 (k $ 25), [5,6] ZrO 2 (k $ 40) [7] and Ta 2 O 5 (k $ 25-60) [8,9] are being explored in giga-bit scale DRAMs (metal-insulator-semiconductor, MIS, and metalinsulator-metal, MIM, in Fig. 1b). The ability of a dielectric film to store charge is conveniently represented by the equivalent oxide thickness (t ox , ¼ t phy  3.9/k, where t phy is the physical thickness of the film). The minimum achievable t ox is $0.7 nm for HfO 2 , ZrO 2 and Ta 2 O 5 which are currently being used in the DRAM industry. However, the technology road map for memory devices states that t ox less than 0.5 nm is necessary for the DRAMs with a design rule of <40 nm.[10] It is also noted that there are no known material solutions to serve this purpose. Reducing the thickness of the dielectric films with k values $20-30 to achieve the required t ox results in unacceptably high leakage currents. Therefore, a dielectric material with a higher k value is in demand. Perovskite-based dielectric films such as SrTiO 3 [11,12] and (Ba,Sr)TiO 3 [13] were reported to exhibit k values of several hundreds and therefore t ox of $0.24 nm is feasible with these materials. [14] However, growth of these films is extremely difficult with the atomiclayer-deposition (ALD) which is a method of choice for the growth of the dielectric films in microelectronic devices. A low thermal budget of 500-600 8C during the deposition and post-de...
The ever-shrinking dimensions of dynamic random access memory (DRAM) require a high quality dielectric film for capacitors with a sufficiently high growth-per-cycle (GPC) by atomic layer deposition (ALD). SrTiO 3 (STO) films are considered to be the appropriate dielectric films for DRAMs with the design rule of ∼20 nm, and previous studies showed that STO films grown by ALD have promising electrical performance. However, the ALD of STO films still suffers from much too slow GPC to be used in mass-production. Here, we accomplished a mass-production compatible ALD process of STO films using Ti(O-i Pr) 2 -(tmhd) 2 as a Ti-precursor for TiO 2 layers and Sr( i Pr 3 Cp) 2 as a Sr-precursor for SrO layers. O 3 and H 2 O were used as the oxygen sources for the TiO 2 and SrO layers, respectively. A highly improved GPC of 0.107 nm/unit-cycle (0.428 nm/supercycle) for stoichiometric STO films was obtained at a deposition temperature of 370 °C, which is ∼7 times higher than previously reported. The origin of such high GPC values in this STO films could be explained by the partial decomposition of the precursors used and the strong tendency of water adsorption onto the SrO layer in comparison to the TiO 2 layer. The STO film grown in this study also showed an excellent step coverage (∼95%) when deposited inside a deep capacitor hole with an aspect ratio of 10. Owing to the high bulk dielectric constant (∼ 146) of the STO film, an equivalent oxide thickness of 0.57 nm was achieved with a STO film of 10 nm. In addition, the leakage current density was sufficiently low (3 Â 10 À8 Acm À2 at þ0.8 V). This process is extremely promising for fabrication of the next generation DRAMs.
The effect of the O3 feeding time on the physical and electrical properties of TiO2 thin films on Ru electrodes was investigated. The density, composition, chemical state, and crystalline structure of the TiO2 films were almost identical, irrespective of the O3 feeding time, even when the TiO2 films were grown under subsaturated conditions with respect to the O3 feeding. However, increasing the O3 feeding time to more than 3s brought about surface roughening and severe local protrusion of the films, which significantly increased their leakage current density. The conduction band offset of the film surface was generally small and was not increased by increasing the O3 feeding time nor was the leakage current improved. Consequently, a minimum tox of 0.8nm with a leakage current density <∼1×10−7A∕cm2 at an applied voltage of 0.8V was achieved at an O3 feeding time of 2–3s.
HfO 2 thin films were grown by atomic layer deposition (ALD) using a novel heteroleptic precursor, tert-butoxytris(ethylmethylamido)hafnium [HfO t Bu(NEtMe) 3 ; BTEMAH] and ozone. The structure of BTEMAH is similar to that of tetrakis(ethylmethylamido)hafnium [Hf(NEtMe) 4 ; TEMAH] except that one of its four amido ligands is replaced with a tert-butoxy ligand. This heteroleptic structure largely improves the ALD growth rate (0.16 nm cycle À1 ) and Hf density (Hf mass per unit volume of HfO 2 film, 7.6 g cm À3 ) of the HfO 2 films. The self-regulated ALD growth behavior was confirmed at a growth temperature of 300 C. Higher Hf density induces anti-crystallization properties in the as-grown film. Consequently, the amorphous phase of a HfO 2 film is retained up to $15 nm during deposition at 300 C. The more amorphous-like nature and the higher Hf density of the HfO 2 film also retard crystallization during post-deposition annealing (PDA), which strongly enhances the thermal stability of the electrical performance. The capacitance equivalent thickness of the films with thicknesses ranging from 4 to 13 nm is relatively constant up to a PDA temperature of 1000 C.
This study examined the effect of a Ru buffer layer growth on a TiN electrode on the structural and electrical properties of a TiO 2 dielectric film grown by atomic layer deposition. The growth of a TiO 2 film directly on TiN resulted in the formation of a mixture of anatase and rutile TiO 2 with a dielectric constant of only 42. However, interposing a thin Ru layer altered the crystal structure of TiO 2 from a mixed phase to almost pure rutile, which was accompanied by an abrupt increase in the dielectric constant to ϳ130. This is a much better result compared with the TiO 2 film deposited on a bulk Ru film, which showed a dielectric constant of ϳ80. This improvement was attributed to a change in the preferred growth direction of a TiO 2 film on the Ru/TiN layer compared with that on a thicker Ru film.Among the various simple binary oxides, TiO 2 and Al-doped TiO 2 dielectric films have attracted considerable attention as high dielectric constant ͑k͒ materials for the capacitors in dynamic random access memory ͑DRAM͒ with sub-40 nm design rules because of their high k value. 1,2 TiO 2 exists in three main phases, rutile, anatase, and brookite. The k values of rutile along the a and c axes ͑90 and 170, respectively͒ are much larger than those of anatase ͑30-40͒ 3 and other comparable binary oxides such as ZrO 2 and HfO 2 . Al doping shifts the Fermi level of the metal to the midpoint of the bandgap of TiO 2 , resulting in a significant decrease in leakage current density. 2 It was recently reported that rutile ͑k value of ϳ80͒ and Al-doped TiO 2 thin films could be grown on a Ru electrode by atomic layer deposition ͑ALD͒ at temperatures as low as 250°C using O 3 as the oxidant, 1,2 even though they are the stable polymorphs at temperatures Ͼ ϳ 700°C. 4 This has been attributed to the local epitaxial relationship between the in situ formed RuO 2 on the Ru electrode and rutile TiO 2 . Rutile-structured RuO 2 induces the local epitaxial growth of TiO 2 with a lattice mismatch along the a and c axes of only 2.09 and 4.76%, respectively. 1,2 TiO 2 films showed a conformal growth behavior over the severe threedimensional capacitor hole structure due to the self-regulating nature of ALD. 5 The higher oxidation potential of O 3 than H 2 O is essential for achieving rutile TiO 2 because the in situ formation of RuO 2 on the Ru electrode during the ALD of TiO 2 is the prerequisite. 6,7 Although the growth of rutile and Al-doped TiO 2 films with a high k value and sufficiently low leakage current is quite promising, the use of Ru electrodes imposes a serious impediment to scaling up to a mass-production compatible process for DRAM fabrication due to the extremely high cost of Ru and metallorganic Ru precursors for ALD. 8 TiN is currently the most common electrode used in the fabrication of metal-insulator-metal ͑MIM͒ capacitors in DRAM. TiN electrodes are usually deposited by either chemical vapor deposition ͑CVD͒ or ALD using TiCl 4 and NH 3 as the Ti precursor and reactant, respectively. 9,10 The cost of TiCl 4 is almos...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.