Surface Reaction Mechanisms during Ozone and Oxygen Plasma

Jul 20, 2010 - Department of Chemical Engineering, Colorado School of Mines, Golden, ... control and abrupt interfaces on high-aspect-ratio nanostruct...
0 downloads 0 Views 886KB Size
pubs.acs.org/Langmuir © 2010 American Chemical Society

Surface Reaction Mechanisms during Ozone and Oxygen Plasma Assisted Atomic Layer Deposition of Aluminum Oxide Vikrant R. Rai,† Vincent Vandalon,‡ and Sumit Agarwal*,† †

Department of Chemical Engineering, Colorado School of Mines, Golden, Colorado 80401, and ‡Department of Applied Physics, Eindhoven University of Technology, 5600 MB Eindhoven, The Netherlands Received April 14, 2010. Revised Manuscript Received June 19, 2010

We have elucidated the reaction mechanism and the role of the reactive intermediates in the atomic layer deposition (ALD) of aluminum oxide from trimethyl aluminum in conjunction with O3 and an O2 plasma. In situ attenuated total reflection Fourier transform infrared spectroscopy data show that both -OH groups and carbonates are formed on the surface during the oxidation cycle. These carbonates, once formed on the surface, are stable to prolonged O3 exposure in the same cycle. However, in the case of plasma-assisted ALD, the carbonates decompose upon prolonged O2 plasma exposure via a series reaction kinetics of the type, A (CH3) f B (carbonates) f C (Al2O3). The ratio of -OH groups to carbonates on the surface strongly depends on the oxidizing agent, and also the duration of the oxidation cycle in plasma-assisted ALD. However, in both O3 and O2 plasma cycles, carbonates are a small fraction of the total number of reactive sites compared to the hydroxyl groups.

I. Introduction Atomic layer deposition (ALD) is an attractive method for depositing conformal films with precise thickness control and abrupt interfaces on high-aspect-ratio nanostructures.1,2 Thermal ALD processes operate in a surface-reaction-controlled regime where heterogeneous reactions of the precursor molecules with the surface species determine the film growth mechanism.3 However, in the case of plasma-assisted ALD, additional reactions may occur in the gas phase, which can strongly influence growth.4-6 For instance, the gas-phase reaction products such as H2O may be excited or dissociated due to the plasma leading to -OH groups on the surface.4 The underlying surface reactions for film growth during thermal ALD of various metal oxides and metal nitrides have been extensively studied in the literature using various surface and gas-phase diagnostic tools such as Fourier transform infrared spectroscopy,7,8 quartz crystal microbalance,9,10 mass spectrometry,10,11 and spectroscopic ellipsometry.12 Among the different metal oxides deposited from ALD, Al2O3 in particular has been extensively examined due to its application as a dielectric *To whom correspondence should be addressed. E-mail: sagarwal@ mines.edu. (1) George, S. M.; Ott, A. W.; Klaus, J. W. J. Phys. Chem. 1996, 100, 13121. (2) Leskela, M.; Ritala, M. Angew. Chem., Int. Ed. 2003, 42, 5548. (3) Puurunen, R. L. J. Appl. Phys. 2005, 97, 52. (4) Rai, V. R.; Agarwal, S. J. Phys. Chem. C 2009, 113, 12962. (5) Heil, S. B. S.; Kudlacek, P.; Langereis, E.; Engeln, R.; van de Sanden, M. C. M.; Kessels, W. M. M. Appl. Phys. Lett. 2006, 89, 3. (6) Langereis, E.; Keijmel, J.; van de Sanden, M. C. M.; Kessels, W. M. M. Appl. Phys. Lett. 2008, 92, 3. (7) Dillon, A. C.; Ott, A. W.; Way, J. D.; George, S. M. Surf. Sci. 1995, 322, 230. (8) Ferguson, J. D.; Weimer, A. W.; George, S. M. Thin Solid Films 2000, 371, 95. (9) Elam, J. W.; Groner, M. D.; George, S. M. Rev. Sci. Instrum. 2002, 73, 2981. (10) Rahtu, A.; Alaranta, T.; Ritala, M. Langmuir 2001, 17, 6506. (11) Matero, R.; Rahtu, A.; Ritala, M. Chem. Mater. 2001, 13, 4506. (12) Ott, A. W.; McCarley, K. C.; Klaus, J. W.; Way, J. D.; George, S. M. Appl. Surf. Sci. 1996, 107, 128. (13) Frank, M. M.; Wilk, G. D.; Starodub, D.; Gustafsson, T.; Garfunkel, E.; Chabal, Y. J.; Grazul, J.; Muller, D. A. Appl. Phys. Lett. 2005, 86, 3. (14) Garcia-Gutierrez, D. I.; Shahrjerdi, D.; Kaushik, V.; Banerjee, S. K. J. Vac. Sci. Technol., B 2009, 27, 2390. (15) Hoex, B.; Gielis, J. J. H.; de Sanden, M.; Kessels, W. M. M. J. Appl. Phys. 2008, 104, 7.

13732 DOI: 10.1021/la101485a

material in electronic devices,13,14 as a passivation layer in Sibased solar cells,15 and as a water permeation barrier in organic electronics.16 Al2O3 is deposited by ALD using trimethyl aluminum [Al(CH3)3, TMA] and H2O1,7 as the oxidation source, where surface -OH groups generated during the H2O half-reaction cycle act as the reactive sites for TMA chemisorption in the subsequent cycle.1,7 Since H2O desorbs slowly in cold-wall reactors, H2O-based ALD processes require long purge times.7,17 Therefore, several nonaqueous oxidizing agents such as O3 and O2 plasma are being explored as an alternative to H2O.4,6,18-20 Advantages of these oxidizing agents include smaller purge times, relatively low deposition temperatures and contamination levels, and improved dielectric properties of the deposited metal oxide films.4,6,18,19,21,22 Although nonaqueous oxidizing agents have been previously investigated for the ALD of various metal oxides, the corresponding reaction mechanisms are not universal4 and, therefore, are not completely understood. For O3-based ALD of Al2O3, Goldstein et al. have proposed a reaction pathway where O inserts into Al-C and C-H bonds of chemisorbed TMA to primarily produce -OH groups and formates on the surface.18 On the other hand, for O2plasma-assisted ALD of Al2O3, Kessels and co-workers5,6 have proposed a combustion-like reaction mechanism where CO, CO2, and H2O are generated due to the combustion of the -CH3 ligands in chemisorbed TMA: only -OH groups were reported as the reactive sites, which were consumed during the subsequent TMA cycle, releasing CH4 into the gas phase.5,6 In this Letter, we have elucidated the reaction mechanisms during ALD of Al2O3 from (16) Groner, M. D.; George, S. M.; McLean, R. S.; Carcia, P. F. Appl. Phys. Lett. 2006, 88, 3. (17) Groner, M. D.; Fabreguette, F. H.; Elam, J. W.; George, S. M. Chem. Mater. 2004, 16, 639. (18) Goldstein, D. N.; McCormick, J. A.; George, S. M. J. Phys. Chem. C 2008, 112, 19530. (19) Rai, V. R.; Agarwal, S. J. Phys. Chem. C 2008, 112, 9552. (20) Rose, M.; Niinisto, J.; Endler, I.; Bartha, J. W.; Kucher, P.; Ritala, M. ACS Appl. Mater. Interfaces 2010, 2, 347. (21) Kim, S. K.; Lee, S. W.; Hwang, C. S.; Min, Y. S.; Won, J. Y.; Jeong, J. J. Electrochem. Soc. 2006, 153, F69. (22) Kim, S. K.; Hwang, C. S. J. Appl. Phys. 2004, 96, 2323.

Published on Web 07/20/2010

Langmuir 2010, 26(17), 13732–13735

Rai et al.

Letter

TMA using both O3 and an O2 plasma as oxidizers through realtime in situ attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy. We show that the reaction mechanism for both oxidizers is combustion-like, which produces -OH groups and carbonates as the reactive sites on the surface after the oxidation cycle. We further show that, in the case of an O2 plasma, carbonates are simply short-term intermediates, which decompose upon prolonged plasma exposure. The surface concentration of carbonates however does not affect TMA adsorption, since it readily chemisorbs on both reactive sites.

II. Experimental Section Ligand-exchange reactions during the O3 and O2-plasmaassisted ALD of Al2O3 were studied in a cold-wall reactor equipped with an in situ ATR-FTIR spectroscopy19,23 setup and a remote inductively coupled plasma (ICP) source. A background pressure of ∼90 mTorr was maintained in the chamber by flowing O2, which also served as the purge gas, since it does not react with TMA under the process conditions.5,6,24 TMA was transported into the chamber without any carrier gas from a bubbler maintained at room temperature. The vapor delivery lines were heated to ∼65 °C to prevent precursor condensation. O3 was delivered by passing O2 through an in-line corona-discharge-based O3 generator, which produced ∼6 wt % O3. An O2 discharge was created by flowing O2 gas through the ICP source operated at 100 W radio frequency power at 13.56 MHz. TMA was pulsed into the chamber using a solenoid valve, which was controlled along with the ICP source using Labview. TMA was pulsed for 15 s followed by a 180 s background O2 purge. The TMA saturation dose was the same for both O3 and O2-plasma-assisted ALD. The TMAchemisorbed surface was then exposed either to O3 for 300 s in several small doses by turning on and off the O3 generator or to an O2 plasma for 70 s also in small intervals. The chamber was again purged with the background O2 for 120 s. The maximum chamber pressure during TMA, O3, and O2 plasma cycles was 10, 85, 100 mTorr, respectively. The intermittent exposure of oxidizers is likely to increase the exposure times due to the additional time required to fill the chamber again to the maximum pressure after every incremental exposure. However, this was necessary to investigate the kinetics of the different surface reactions involved. Al2O3 films were deposited on ZnSe internal reflection crystals. In between each half-reaction cycle, the surface species were detected using in situ ATR-FTIR spectroscopy. This setup was also sensitive to the gas-phase species due to the ∼25-cm-long infrared (IR) beam path through the vacuum chamber. The substrate temperature was maintained at 150 °C. All the IR spectra were recorded as difference spectra; that is, a new background spectrum was collected before each half-reaction cycle.19 Each spectrum herein was collected after repeating 10-20 ALD cycles at any given process condition, averaged over 1000 scans with a 4 cm-1 resolution, and reported without baseline correction, including the -OH stretching region (3800-3200 cm-1).

III. Results and Discussion Different surface species detected during TMA-O3 and TMAO2 plasma half-reaction cycles at 150 °C are shown in the IR difference spectra in Figure 1. A one-to-one exchange of surface species during TMA and oxidizer (O3 and O2 plasma) halfreaction cycles can be directly interpreted from these IR spectra. In these difference spectra, an increase in absorbance is due to freshly chemisorbed surface species, and a decrease in absorbance is due to the reaction of a surface species with the incoming (23) Harrick, N. J. Internal Reflection Spectroscopy; John Wiley: New York, 1967; p 179. (24) Szymanski, S. F.; Seman, M. T.; Wolden, C. A. Surf. Coat. Technol. 2007, 201, 8991.

Langmuir 2010, 26(17), 13732–13735

Figure 1. IR difference spectra showing the different surface species due to ligand-exchange reactions at 150 °C: (a) TMA and (b) O3 half-reaction cycles; (c) TMA and (d) O2 plasma half-reaction cycles; (e) O2 plasma exposure of surface in (d) for a longer duration. The exposure times of TMA, O3, and O2 plasma half cycles for surface saturation were 15, 300, and 70 s, respectively.

precursor. After TMA chemisorption, there was an increase in absorbance in the 3000-2800 and 1250-1150 cm-1 regions corresponding to CH3 stretching and deformation modes, respectively (see Figure 1a,c). Specific vibrational frequencies were assigned to various structural units on the surface. The vibrational modes at 2942, 2896, 2831, and 1211 cm-1 were assigned to the CH3 antisymmetric stretching,6,25-27 CH3 symmetric stretching,6,26,27 bending overtone,6,26,27 and CH3 deformation mode,6,25,26 respectively in TMA chemisorbed on the Al2O3 surface. The decrease in IR absorbance in the 3800-3200 cm-1 region was due to -OH groups,6,28-30 while the decrease in absorbance in the 18001350 cm-1 region can be assigned to the symmetric and antisymmetric stretching modes of metal-bicarbonates,31-35 carbonates,31-34 or formates.36 Bicarbonates are formed due to a simultaneous reaction between CO2 and H2O on a metal oxide surface and decompose at the temperature of the experiment into carbonates,35 while formates are formed due to the reaction of CO with surface -OH groups or H2O.36 Although the IR absorbance in the 1800-1350 cm-1 region did not change upon prolonged O3 exposure, it was completely diminished upon prolonged O2 plasma exposure (see Figure 1e). Thus, these species were short-term intermediates only during plasma-assisted ALD. The broad -OH stretching vibration in Figure 1b,d,e originates from several overlapping IR bands attributed to isolated -OH groups28-30 at 3740 cm-1 and (25) Frank, M. M.; Chabal, Y. J.; Wilk, G. D. Appl. Phys. Lett. 2003, 82, 4758. (26) Soto, C.; Tysoe, W. T. J. Vac. Sci. Technol., A 1991, 9, 2686. (27) Kunawicz, J.; Jones, P.; Hockey, J. A. Trans. Faraday Soc. 1971, 67, 848. (28) Peri, J. B. J. Phys. Chem. 1965, 69, 211. (29) Knozinger, H.; Ratnasamy, P. Catal. Rev. - Sci. Eng. 1978, 17, 31. (30) Morterra, C.; Magnacca, G. Catal. Today 1996, 27, 497. (31) Baltrusaitis, J.; Jensen, J. H.; Grassian, V. H. J. Phys. Chem. B 2006, 110, 12005. (32) Pokrovski, K.; Jung, K. T.; Bell, A. T. Langmuir 2001, 17, 4297. (33) Primet, M.; Pichat, P.; Mathieu, M. V. J. Phys. Chem. 1971, 75, 1221. (34) Yates, D. J. C. J. Phys. Chem. 1961, 65, 746. (35) Henderson, M. A. Surf. Sci. 1998, 400, 203. (36) Busca, G.; Lorenzelli, V. Mater. Chem. 1982, 7, 89.

DOI: 10.1021/la101485a

13733

Letter

hydrogen-bonded -OH groups below 3680 cm-1 on γ-Al2O3.28-30 These different types of -OH groups were observed during both O3 and O2 plasma half-reaction cycles (see Figure 1b,d,e). Thus, we conclude that -OH groups along with carbonates or formates are the reactive sites for TMA chemisorption during both O3-based and O2-plasma-assisted ALD of Al2O3. Hydroxyl groups are previously known to be the reactive sites in H2O-based ALD processes,6,7 while carbonates and formates have been reported in O2-plasma-assisted4 and O3-based ALD.18,19,37 The stretching modes for carbonates and formates completely overlap in the 1800-1350 cm-1 region and, therefore, are difficult to distinguish. Kwon et al.37 and Goldstein et al.18 studied the ALD of Al2O3 from TMA and O3 also using IR spectroscopy and assigned the bands at 1597, 1388, and 1404 cm-1 to antisymmetric and symmetric OCO stretching and CH bending vibrations, respectively, in formates.18,37 The broad IR band at 1470 cm-1 was assigned to methoxy groups (AlOCH3) on the surface.18,37 In addition, in the same experiments, Goldstein et al. observed bands at 1720 and 1320 cm-1, which were assigned to antisymmetric and symmetric OCO stretching modes of carbonates18 on Al2O3. Goldstein et al. reported that these formates were stable to long O3 exposure but decomposed above 300 °C by releasing CO into the gas phase: this decomposition reaction was proposed as the mechanism for the creation of -OH groups on the surface.18,37 In our experiments, we observe two strong IR absorption bands at 1602 and 1404 cm-1 after the O3 (Figure 1b) and O2 plasma (Figure 1d) cycles. The absorbance due to these bands did not change under prolonged O3 exposure, which is consistent with ref 18. However, under prolonged O2 plasma exposure (see Figure 1e), these species decomposed completely. Based on the surface reaction products detected in the gas-phase during both O3 and O2-plasma-based ALD, we have assigned these bands to surface carbonates. Other weak bands in the 1800-1350 cm-1 region (1737 and 1470 cm-1) were also assigned to surface carbonates, since these species can be present as mono-, bi-, and polydentates, which have different vibrational frequencies.32,36 In the experiment in Figure 1, when a TMA-chemisorbed surface was exposed to O3 or O2 plasma (see Figure 1b, d, and e), the surface reaction products released into the gas phase were not directly observed in IR spectra because their concentration was below the detection limit of the setup. Therefore, a controlled set of experiments were performed to increase the residence time of the reaction products to enhance their concentration in the chamber. In the first experiment, the surface was exposed to a 10 s TMA cycle while the chamber was isolated from the vacuum pump with a gate valve. The reaction products were trapped inside the chamber instead of being continuously pumped, thus enabling their detection in the gas phase along with the surface species. An IR spectrum collected under this condition is shown in Figure 2a. In this IR spectrum, in addition to the surface species, we also observe absorption bands due to gasphase TMA and CH4, and weakly adsorbed CO2 on Al2O3. In Figure 2a, CH3 stretching in the 3000-2800 cm-1 region contains both chemisorbed and gas-phase TMA. The sharp IR bands at 3016 and 1306 cm-1 correspond to CH stretching and bending modes of gas-phase CH4.26,38,39 These frequencies were confirmed by collecting an IR spectrum with only CH4 in the chamber, which is shown in the top panel in Figure 2. This CH4 was produced due to the reaction between TMA and the surface -OH groups as can be inferred by the decrease in the -OH stretching absorption in the 3800-3200 cm-1 (37) Kwon, J. H.; Dai, M.; Halls, M. D.; Chabal, Y. J. Chem. Mater. 2008, 20, 3248. (38) Dannison, D. M. Rev. Mod. Phys. 1940, 12, 175. (39) Scarano, D.; Bertarione, S.; Spoto, G.; Zecchina, A.; Arean, C. O. Thin Solid Films 2001, 400, 50.

13734 DOI: 10.1021/la101485a

Rai et al.

Figure 2. IR difference spectra for a single TMA-O2 plasma ALD cycle showing various gas-phase reaction products and surface species. (a) Spectrum was collected after 10 s of TMA exposure with the chamber isolated from the pump. (b) Chamber was evacuated to the base pressure followed by a 120 s inert-gas purge. (c) TMA-chemisorbed surface in (b) exposed to an O2 plasma for 30 s with the chamber isolated from the pump. (d) IR spectrum collected after the chamber was evacuated to the base pressure followed by a 200 s O2 plasma exposure. The top two panels show the gas-phase CH4 and TMA absorbance for comparison with the gas-phase peaks in (a).

region, a reaction similar to H2O-based ALD.7,26 The broad vibrational bands at 2360 and 2333 cm-1 were assigned to undissociated weakly chemisorbed CO2 molecules on O and Al sites, respectively.40,41 We hypothesize that this CO2 was released from the surface carbonates upon chemisorption of TMA, similar to our previous results for O3 and O2-plasma-assisted ALD of TiO2.4,19 This suggests that the species in the 1800-1350 cm-1 region of the IR spectra in Figures 1 and 2 are carbonates and not formates, which would release CO upon decomposition. However, it may be possible that formates are intermediates in the formation of carbonates from methyl groups, and are formed initially upon very small doses of the oxidizers (O3 and O2 plasma).18 After the TMA cycle, the chamber was then evacuated to the base pressure of ∼10-3 Torr to remove gas-phase TMA, CH4, and weakly adsorbed CO2 followed by a 180 s O2 purge. The IR spectrum in Figure 2b shows absorbance only due to chemisorbed TMA while all other species were removed. In the second experiment to detect the gas-phase reaction products, the -CH3-terminated surface was exposed to an O2 plasma for 30 s with the chamber again isolated from the vacuum pump. Similar to Figure 2a, two vibrational bands were observed for CO2 in the IR difference spectrum shown in Figure 2c: this CO2 was formed due to the combustion of the surface -CH3 ligands by O atoms from the O2 plasma. No IR absorption band was observed for gas-phase CO2 at 2349 cm-1,42 gas-phase CO at 2149 cm-1,43 and surface chemisorbed CO on Al2O3 which (40) Anderson, J. A. J. Chem. Soc., Faraday Trans. 1992, 88, 1197. (41) Rege, S. U.; Yang, R. T. Chem. Eng. Sci. 2001, 56, 3781. (42) Wong, J. C. S.; Linsebigler, A.; Lu, G. Q.; Fan, J. F.; Yates, J. T. J. Phys. Chem. 1995, 99, 335. (43) Dutta, P. K.; Ginwalla, A.; Hogg, B.; Patton, B. R.; Chwieroth, B.; Liang, Z.; Gouma, P.; Mills, M.; Akbar, S. J. Phys. Chem. B 1999, 103, 4412.

Langmuir 2010, 26(17), 13732–13735

Rai et al.

Figure 3. Temporal evolution of the integrated absorbance change for CH3 (O), carbonate (b), and OH (0) stretching vibrations, and Al-O-Al phonons (9) during O2 plasma exposure of a TMAchemisorbed surface at 150 °C. The solid lines are fits to the experimental data. The maximum integrated absorbance change for each species was normalized to unity.

appears at 2190, 2050, and 1870 cm-1.44 Since there was no CO present in the chamber (see Figure 2c), this experiment further confirms that the IR absorbance in the 1800-1350 cm-1 region must be due to carbonates and not formates. The H2O released due to the combustion of methyl ligands was not observed in the IR spectrum in Figure 2c, which may be lost to the cold walls of the reactor, similar to our previous results for plasma-assisted ALD of TiO2.4 CO2 may also be lost to the metal-oxide-coated walls of the reactor since we observed physisorbed CO2 on the heated substrate, but this effect was less pronounced since the sticking coefficient for H2O is significantly higher than that for CO2. However, H2O must be released upon combustion of the ligands since we detected a weak absorbance due to gas-phase CH4 at 3016 and 1306 cm-1 produced due to the reaction of gasphase H2O with the uncombusted methyl ligands. Some CH4 may also be formed upon direct reaction between uncombusted methyl ligands with the adjacent -OH groups. The gas-phase CH4 and weakly chemisorbed CO2 were removed when the chamber was evacuated to the base pressure. The surface was then exposed to a 120 s O2 plasma to completely combust the TMA ligands followed by a 120 s O2 purge. The integrated absorbance in the CH3 stretching region in Figure 2a and d was similar, indicating a complete one-to-one exchange of the ligands and other surface species (see Figure S1 in the Supporting Information for a schematic of the overall surface reaction process). Figure 3 shows the temporal evolution of the integrated absorbance change for the CH3 (O), carbonate (b), and OH (0) stretching regions along with the Al-O-Al phonon modes (9) during O2 plasma exposure of a TMA-chemisorbed surface at 150 °C. The solid lines represent the best fits to the experimental data. The fits in Figure 3 show that the combustion of -CH3 ligands and the formation of Al2O3 and -OH groups during O2 plasma exposure follow pseudo-first-order kinetics. On the other (44) Primet, M.; Basset, J. M.; Mathieu, M. V.; Prettre, M. J. Catal. 1973, 29, 213.

Langmuir 2010, 26(17), 13732–13735

Letter

hand, the surface carbonate concentration first increases, reaches a maximum, and then decreases until almost all the carbonates are removed upon prolonged O2 plasma exposure. The kinetic behavior of the surface carbonates suggests a series reaction of the type A (CH3) f B (CO32-) f C (Al-O-Al), where the carbonates (species B) are the short-term intermediates formed upon combustion of -CH3 (species A), which eventually get converted to Al-O-Al bonds (species C) via the elimination of CO2.45 In the above reaction pathway, in addition to the observed temporal evolution of carbonates (species B), absorbance due to the Al-O-Al phonon modes (species C) was expected to increase continuously as the carbonates were decomposed.45 However, the time constant for the Al-O-Al phonon absorbance change was almost identical to that for CH3 absorbance change, which suggests that the -OH groups contribute as the primary sites for the formation of Al-O-Al bonds, and the carbonates are only a very small fraction the total reactive sites on the surface. Therefore, absorbance due to Al-O-Al phonon modes was not affected by the decomposition of the surface carbonates and simply reflects the combustion of -CH3 groups. This was further confirmed through experiments where a surface with no carbonates (long O2 plasma exposure, ∼60 s) and a surface with the maximum carbonate concentration (intermediate O2 plasma exposure, ∼5 s) were exposed to TMA in the subsequent ALD cycle. We observed an almost identical integrated absorbance increase for surface -CH3 groups in both experiments, confirming that -OH groups are the primary reactive surface sites.

IV. Conclusions The ALD of Al2O3 from TMA in conjunction with O3 and an O2 plasma was investigated using real-time, in situ ATR-FTIR spectroscopy. For both O3- and O2 plasma-assisted ALD, the mechanism during the oxidation cycle was combustion-like, leading to -OH groups and carbonates on the surface. In the case of plasma-assisted ALD, carbonates were merely short-term intermediates, which decomposed upon prolonged O2 plasma exposure, further contributing to Al2O3 growth. Combustion of methyl ligands and formation of -OH groups and Al2O3followed pseudofirst-order kinetics. On the other hand, the kinetic behavior of carbonates, which were present in a very small concentration on the surface compared to -OH groups, suggests a series reaction of the type A (CH3) f B (carbonates) f C (Al2O3). The ratio of -OH groups to carbonates on the surface was strongly dependent on the type of oxidizing agent and the duration of the oxidation cycle in the case of plasma-assisted ALD. Acknowledgment. We gratefully acknowledge support from the NSF CAREER program (Grant No. CBET-0846923) and the NSF Renewable Energy MRSEC program at the Colorado School of Mines (Grant No. DMR-0820518). Supporting Information Available: Schematic showing the mechanism for O3 and O2 plasma-assisted ALD of Al2O3. This material is available free of charge via the Internet at http://pubs.acs.org. (45) Levenspiel, O. Chemical Reaction Engineering, 3rd ed.; John Wiley: New York, 1998; p 181.

DOI: 10.1021/la101485a

13735