Ozone-Based Atomic Layer Deposition of Alumina from TMA: Growth, Morphology, and Reaction Mechanism S. D. Elliott* Tyndall National Institute, Lee Maltings, Cork, Ireland G. Scarel, C. Wiemer, and M. Fanciulli CNR-INFM MDM National Laboratory, Via C. OliVetti 2, 20041 Agrate Brianza, Italy G. Pavia STMicroelectronics, Via C. OliVetti 2, 20041 Agrate Brianza, Italy ReceiVed April 18, 2006. ReVised Manuscript ReceiVed May 25, 2006 We examine the effect of growth temperature in the 150-300 °C range on the structural and morphological properties of Al 2 O 3 films deposited using atomic layer deposition, contrasting the effect of H 2 O and O 3 as oxygen sources. Trimethylaluminum (TMA) is the metal source. A mechanism for the O 3 reaction is investigated using ab initio calculations and provides an explanation for the observed temperature dependence. The simulations show that hydroxyl groups are produced at the surface by the oxidation of adsorbed methyl groups by O 3 . This is confirmed by the measured rates; both H 2 O and O 3 processes show molar growth rates per cycle that decrease with increasing reactor temperature, consistent with a decrease in hydroxyl coverage. At no temperature does the O 3 process deposit more Al 2 O 3 per cycle than the H 2 O process. Morphological characterization shows that O 3 as the oxygen source yields lower-quality films than H 2 O; the films are less dense and rougher, especially at low growth temperatures. The existence of voids correlates with the low film electronic density. This may indicate the low mobility of surface hydroxyl at low temperatures, an effect that is washed out by repeated exchange with the vapor phase in the H 2 O case. 1. Introduction The atomic layer deposition (ALD) of alumina (Al 2 O 3 ) using ozone (O 3 ) is studied over a wide temperature range and compared with the well-known H 2 O process, focusing on the growth rate, film density, roughness, and reaction mechanism. ALD is a type of chemical vapor deposition (CVD) in which precursors in gas form are admitted separately into the reactor in alternate pulses. Each precursor chemisorbs individually onto the substrate, rather than reacting in the gas phase. 1 Gas-phase reactions are avoided by purging with inert gas between each pulse. Under ideal conditions, substrate-precursor reactions are self-limiting and the surface is saturated with precursor fragments at the end of each pulse. Ligands in the fragments are eliminated by reaction with the other precursor during the next pulse. Because growth reactions occur only at the surface, deposition is slow, but with the advantage of atomic-level control of film thickness, as well as unparalleled conformality and uniformity. ALD is therefore used for the deposition of nanometer-thin films in high-aspect-ratio structures, such as alumina films in dynamic random access memory (DRAM) trenches, 2 read- write heads, 3 or in the interpoly dielectric stack of flash memories. 4 The use of ozone (O 3 ) as oxygen precursor in metal-oxide ALD has stemmed from its higher activity for ligand elimination relative to H 2 O. It is therefore needed for particularly stable metal (M) precursors, such as -diketo- nates. For instance, Y 2 O 3 has been deposited from yttrium -diketonate and O 3 , 5 and ZrO 2 has been deposited from various zirconium cyclopentadienyls and O 3 6 or H 2 O. 7 It is interesting that both of these examples show temperature windows over which the growth rate is approximately constant. Another reason for a preference for O 3 is that H 2 O adsorbs on reactor walls and is difficult to purge. Along with these processing advantages, higher-quality film is sometimes produced by O 3 ALD when compared with H 2 O. Much current research is directed toward the use of ALD of high permittivity films (e.g., ZrO 2 , 7 HfO 2 , 8,9,10 and Al 2 O 3 ) 11,12,13 in the microelectronics industry. For the deposi- * To whom correspondence should be addressed. E-mail: simon.elliott@ tyndall.ie. Tel: 353-21-4904392. Fax: 353-21-4270271. (1) Suntola, T. Mater. Sci. Rep. 1989, 4, 261-312. (2) Ha, D.; Shin, D.; Koh, D. H.; Lee, J.; Lee, S.; Ahn, Y. S.; Jeong, H.; Chung, T.; Kim, K. IEEE Trans. Electron DeVices 2000, 47, 1499. (3) Paranjpe, A.; Gopinath, S.; Omstead, T.; Bubber, R. J. Electrochem. Soc. 2001, 148, G465-G471. (4) Lee, T. P.; Jang, C.; Haselden, B.; Dong, M.; Park, S.; Bartholomew, L.; Chatham, H.; Senzaki, Y. J. Vac. Sci. Technol., B. 2004, 22, 2295- 2298. (5) Putkonen, M.; Sajavaara, T.; Johansson, L.-S.; Niinisto ¨ , L. Chem. Vap. Deposition 2001, 7, 44-50. (6) Putkonen, M.; Niinisto ¨, L. J. Mater. Chem. 2001, 11, 3141-3147. (7) Niinisto ¨ , J.; Putkonen, M.; Niinisto ¨ , L.; Kukli, K.; Ritala, M.; Leskela ¨, M. J. Appl. Phys. 2004, 95, 84-91. (8) Lee, S.-I.; Owyang, J. S.; Senzaki, Y.; Helms, A.; Kapkin, K. Solid State Technol. 2003, 46, 45-46. 3764 Chem. Mater. 2006, 18, 3764-3773 10.1021/cm0608903 CCC: $33.50 © 2006 American Chemical Society Published on Web 07/13/2006