24.11.2012 Views

Al2O3 Atomic Layer Deposition with ... - TheEEStory

Al2O3 Atomic Layer Deposition with ... - TheEEStory

Al2O3 Atomic Layer Deposition with ... - TheEEStory

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

19530<br />

<strong>Al2O3</strong> <strong>Atomic</strong> <strong>Layer</strong> <strong>Deposition</strong> <strong>with</strong> Trimethylaluminum and Ozone Studied by in Situ<br />

Transmission FTIR Spectroscopy and Quadrupole Mass Spectrometry<br />

David N. Goldstein, † Jarod A. McCormick, ‡ and Steven M. George* ,†,‡<br />

Department of Chemistry and Biochemistry, and Department of Chemical and Biological Engineering,<br />

UniVersity of Colorado, Boulder, Colorad, 80309<br />

ReceiVed: May 14, 2008; ReVised Manuscript ReceiVed: September 25, 2008<br />

The atomic layer deposition (ALD) of <strong>Al2O3</strong> using sequential exposures of Al(CH3)3 and O3 was studied by<br />

in situ transmission Fourier transform infrared (FTIR) spectroscopy and quadrupole mass spectrometry (QMS).<br />

The FTIR spectroscopy investigations of the surface reactions occurring during <strong>Al2O3</strong> ALD were performed<br />

on ZrO2 particles for temperatures from 363 to 650 K. The FTIR spectra after Al(CH3)3 and ozone exposures<br />

showed that the ozone exposure removes surface AlCH3* species. The AlCH3* species were converted to<br />

AlOCH3* (methoxy), Al(OCHO)* (formate), Al(OCOOH)* (carbonate), and AlOH* (hydroxyl) species. The<br />

TMA exposure then removes these species and reestablishes the AlCH3* species. Repeating the TMA and O3<br />

exposures in a sequential reaction sequence progressively deposited the <strong>Al2O3</strong> ALD film as monitored by the<br />

increase in absorbance for bulk <strong>Al2O3</strong> infrared features. The identification of formate species was confirmed<br />

by separate formaldehyde adsorption experiments. The formate species were temperature dependent and were<br />

nearly absent at temperatures g650 K. QMS analysis of the gas phase species revealed that the TMA reaction<br />

produced CH4. The ozone reaction produced mainly CH4 <strong>with</strong> small amounts of C2H4 (ethylene), CO, and<br />

CO2. Transmission electron microscopy (TEM) was also used to examine the <strong>Al2O3</strong> ALD films deposited on<br />

the ZrO2 particles. These TEM images observed conformal <strong>Al2O3</strong> ALD films <strong>with</strong> thicknesses that were<br />

consistent <strong>with</strong> an <strong>Al2O3</strong> ALD growth rate of 1.1 Å/cycle. The surface species after the O3 exposures and the<br />

mass spectrometry results lead to a very different mechanism for <strong>Al2O3</strong> ALD growth using TMA and O3<br />

compared <strong>with</strong> <strong>Al2O3</strong> ALD using TMA and H2O.<br />

I. Introduction<br />

<strong>Atomic</strong> layer deposition (ALD) is an ideal technique to<br />

deposit ultrathin films <strong>with</strong> high conformality and precise<br />

thickness control. 1,2 Traditional methods to deposit <strong>Al2O3</strong> <strong>with</strong><br />

ALD involve sequential surface reactions of Al(CH3)3 (trimethylaluminum<br />

(TMA)) and water. 3-6 These sequential reactions<br />

allow conformal <strong>Al2O3</strong> film growth <strong>with</strong> thickness control<br />

onavarietyofsubstratesincludingpolymers, 7porousmembranes, 3,8<br />

and nanopowders. 9 The details of the <strong>Al2O3</strong> ALD reaction have<br />

been extensively studied by a variety of techniques, including<br />

the quartz crystal microbalance measurements, 10,11 Fourier<br />

transform infrared (FTIR) spectroscopy, 3,12 ellipsometry, 4,5 and<br />

X-ray photoelectron spectroscopy (XPS). 13 <strong>Al2O3</strong> ALD is a<br />

model system and serves as a reference point for other ALD<br />

systems.<br />

The semiconductor industry is interested in growing <strong>Al2O3</strong><br />

films <strong>with</strong> ozone instead of water as the oxygen source. <strong>Al2O3</strong><br />

is a high-k dielectric that is used as a dielectric film for both<br />

DRAM and MOS-FETs. 14 When ozone is used as the oxidant,<br />

the <strong>Al2O3</strong> ALD films can have leakage current densities that<br />

are reduced by two orders of magnitude in comparison <strong>with</strong><br />

<strong>Al2O3</strong> ALD films deposited <strong>with</strong> water. 15 This improvement and<br />

smaller flat band voltage shifts allow <strong>Al2O3</strong> ALD films grown<br />

using ozone to make better gate oxides. 15,16 There are also other<br />

advantages when replacing H2O <strong>with</strong> ozone. Water desorbs<br />

slowly from substrates and requires longer purge times. 10 Water<br />

can also leave unreacted hydroxyl groups in the <strong>Al2O3</strong> ALD<br />

* Corresponding author.<br />

† Department of Chemistry and Biochemistry.<br />

‡ Department of Chemical and Biological Engineering.<br />

J. Phys. Chem. C 2008, 112, 19530–19539<br />

films. 3 The unreacted hydroxyl groups in the films may affect<br />

the dielectric and material properties of the <strong>Al2O3</strong> ALD films.<br />

However, no change in equivalent oxide thickness (EOT) of<br />

<strong>Al2O3</strong> ALD films was observed when ozone was used as the<br />

oxidant. 15<br />

Previous research has been conducted on <strong>Al2O3</strong> ALD <strong>with</strong><br />

TMA and O3. A growth rate of ∼0.8 Å per cycle at 300-450 °C<br />

has been measured by several investigations. 13,14,17,18 XPS<br />

measurements revealed <strong>Al2O3</strong> ALD films that had lower carbon<br />

impurities <strong>with</strong> ozone compared <strong>with</strong> water. 13 <strong>Al2O3</strong> films grown<br />

<strong>with</strong> ozone also had a reduced percentage of Al-Al defects<br />

that degrade the electrical properties of the <strong>Al2O3</strong> ALD films.<br />

These defects have been characterized using XPS by the<br />

presence of a shoulder on the 72.5 eV Al 2p peak. 19 Time-offlight<br />

secondary ion mass spectrometer (TOF-SIMS) analysis<br />

has probed the bulk of <strong>Al2O3</strong> ALD films and revealed different<br />

impurity levels in <strong>Al2O3</strong> films grown <strong>with</strong> ozone compared <strong>with</strong><br />

<strong>Al2O3</strong> films grown <strong>with</strong> water. 13 Hydrogen impurities were<br />

reduced in the ozone grown films. 13<br />

To understand the differences between <strong>Al2O3</strong> ALD <strong>with</strong> TMA<br />

and either H2O or ozone, this study employed in situ transmission<br />

FTIR spectroscopy to monitor the surface species formed<br />

and removed during the TMA and O3 exposures. The FTIR<br />

spectra also revealed the growth of <strong>Al2O3</strong> bulk vibrational modes<br />

versus the number of ALD reaction cycles. Additional experiments<br />

also monitored the gas phase products during O3 and<br />

TMA exposures using a quadrupole mass spectrometer (QMS).<br />

The resulting <strong>Al2O3</strong> ALD films on the ZrO2 particles were then<br />

analyzed <strong>with</strong> transmission electron microscopy (TEM) to obtain<br />

the <strong>Al2O3</strong> ALD growth per ALD cycle. These FTIR, QMS, and<br />

10.1021/jp804296a CCC: $40.75 © 2008 American Chemical Society<br />

Published on Web 11/13/2008


<strong>Al2O3</strong> <strong>Atomic</strong> <strong>Layer</strong> <strong>Deposition</strong> J. Phys. Chem. C, Vol. 112, No. 49, 2008 19531<br />

Figure 1. Schematic of W grid in the ALD reactor.<br />

Figure 2. Schematic of the inlet and outlet connections to the ALD<br />

reactor.<br />

TEM studies help to clarify the surface chemistry and thin film<br />

growth mechanism during <strong>Al2O3</strong> ALD <strong>with</strong> TMA and ozone.<br />

II. Experimental Section<br />

The surface chemistry and thin film growth during <strong>Al2O3</strong><br />

ALD was studied using sequential exposures of TMA and O3<br />

at various temperatures. <strong>Al2O3</strong> ALD films were grown on ZrO2<br />

particles in an ALD reactor designed for in situ FTIR spectroscopy<br />

studies. 12,20 Figure 1 presents a schematic of the ALD<br />

reactor. The reactor was a warm-wall reactor where the chamber<br />

walls were heated to 350 K while the sample could be<br />

independently heated to >900 K. Figure 2 shows a schematic<br />

of all the inlet and outlet connections to the ALD reactor. Two<br />

argon mass flow controllers (MFC) regulated the flow of argon<br />

through the reactor at 220 sccm (110 sccm per MFC). This flow<br />

established a base pressure of 1.30 Torr. An Alcatel 2012A<br />

rotary vane pump removed the argon and reaction byproducts<br />

from the reactor.<br />

Pneumatic leak valves <strong>with</strong> conductance metering valves<br />

allowed accurate exposure of the reactants. A Labview measurement<br />

system controlled the reactant exposures and integrated<br />

the area beneath the pressure transients that occurred during<br />

the reactant exposures. The reactant exposures were performed<br />

<strong>with</strong> use of micropulses that were less than the exposures<br />

required for the reactions to reach completion. The absolute<br />

reactant exposures were determined <strong>with</strong> no ZrO2 nanoparticles<br />

in the reactor after a sufficient number of micropulses for the<br />

reaction on the reactor walls to reach completion. Under these<br />

conditions, the reaction products do not interfere <strong>with</strong> the<br />

measurement of the absolute reactant exposure.<br />

The <strong>Al2O3</strong> ALD was coated onto ZrO2 nanopowders supported<br />

in a 2 × 3cm 2 tungsten grid. 12,20,21 Each W grid was 50<br />

µm thick and was photoetched to 100 grid lines per inch.<br />

Tantalum foil was spot-welded on the sides of the grid to<br />

improve current transfer through the grid. The entire grid was<br />

then attached to a copper clamp that was interfaced via an<br />

electrical feedthrough to a Hewlett-Packard 6268B power<br />

supply. Resistive heating was used to heat the sample. A Love<br />

Controls 16A3 PID controller interfaced to a type K thermocouple<br />

mounted on the sample grid provided temperature control<br />

of the sample. The feedback loop maintained the sample<br />

temperature at (2 °C.<br />

Preparation of the substrate involved pressing ZrO2 nanoparticles<br />

into the W grid. 12,20,21 Each grid was first sonicated in<br />

deionized water and methanol and then blown dry <strong>with</strong> ultrapure<br />

nitrogen. The grid was then placed into a stainless steel die and<br />

covered <strong>with</strong> an excess of nanopowders. Subsequently, a manual<br />

press forced the particles into the W grid until the particles made<br />

a dense matrix <strong>with</strong> very few pinholes in the sample. Excess<br />

nanopowders lying on the top of the grid were easily removed<br />

<strong>with</strong> a razor blade. The finished sample contained about 22 mg<br />

of ZrO2 powder. This quantity of ZrO2 powder is equivalent to<br />

a surface area of ∼0.44 m 2 . Finally, a type K thermocouple<br />

was attached to the top of the sample grid <strong>with</strong> Ceramabond<br />

571 Epoxy. This epoxy electrically isolated the thermocouple<br />

and kept the thermocouple firmly attached to the sample during<br />

the experiment.<br />

An infrared beam from a Nicolet Magna 560 FTIR spectrometer<br />

was externally aligned to pass through the W grid<br />

sample. The ZrO2 nanopowder substrates provided a large<br />

surface area and improved the signal-to-noise ratio for infrared<br />

absorption. The entire sample stage could be translated along<br />

the vertical z-axis direction to move the sample out of the FTIR<br />

beam. This displacement allowed the background reference<br />

spectra to be measured frequently over the course of these<br />

experiments. A liquid nitrogen cooled MCT-B (mercury cadmium<br />

telluride) detector allowed measurement of the infrared<br />

spectra from 400 to 4000 cm -1 . During the reactant exposures,<br />

the gate valves on the CsI windows were closed to prevent<br />

deposition on the windows. All FTIR spectra were obtained at<br />

4cm -1 resolution using 100 averaged scans and were referenced<br />

to the CsI window background. However, most of the FTIR<br />

spectra in this paper are presented as difference spectra.<br />

Mass spectrometry analysis was performed in a rotary reactor<br />

designed for ALD on nanoparticles. The design and operation<br />

of this reactor has been discussed in previous publications. 9,22<br />

To provide in situ quadrupole mass spectrometry analysis, a<br />

200 amu quadrupole mass spectrometer <strong>with</strong> a pressure reduc-


19532 J. Phys. Chem. C, Vol. 112, No. 49, 2008 Goldstein et al.<br />

tion system (PPR200, SRS Inc., Sunnyvale, CA) was attached<br />

to the reactor. During reactant exposures, the QMS scanned the<br />

mass range from 1-75 m/z <strong>with</strong> 0.1 m/z resolution. A Faraday<br />

cup was used as the detector <strong>with</strong> no electron multiplier. With<br />

these settings, about 5 s was required to scan the entire mass<br />

range.<br />

Micropulses of both TMA and O3 were used to determine<br />

the exposures required for the reactions to reach completion<br />

<strong>with</strong> 1.0 g of ZrO2 nanoparticles in the rotary reactor. 9 TMA<br />

was dosed into the rotary reactor to a pressure of 0.3 Torr above<br />

the baseline pressure. The O2/O3 mixture was dosed into the<br />

rotary reactor to a pressure of 0.5 Torr. Each reactant reacted<br />

for 60 s in the chamber and then was purged for 60 s before a<br />

final argon pulse flushed the chamber. The reactor then returned<br />

to base pressure before starting the next set of reactant<br />

micropulses. Approximately 20 micropulses of TMA and 60<br />

micropulses of O3 were required for each reaction to reach<br />

completion <strong>with</strong> 1.0 g of ZrO2 nanoparticles in the rotary reactor.<br />

The ZrO2 particles were obtained from Nanomaterials Research<br />

Corporation (Longmont, CO). These ZrO2 particles were<br />

spherical <strong>with</strong> an average diameter of ∼50 nm and a surface<br />

area of ∼20.2 m 2 /g. TMA was obtained from Aldrich (Milwaukee,<br />

WI) and had a purity of 97%. The water was high<br />

performance liquid chromatography (HPLC) grade from Fisher<br />

Scientific (Pittsburgh, PA). Ozone was produced from UHP<br />

grade oxygen (99.9%) obtained from Airgas Ltd. (Cheyenne,<br />

WY). All chemicals were used as purchased, except for water,<br />

which was subjected to 3 freeze-pump-thaw cycles prior to<br />

use.<br />

Ozone was obtained from O2 by flowing 300 sccm of O2<br />

into a DelOzone LC-14 ozone generator (San Luis Obispo, CA).<br />

This flow produced a 6 psi pressure in the generating cell. At<br />

100% power, the ozone concentration at the outlet was 3.7%<br />

<strong>with</strong> the balance being O2. When the ozone was not going<br />

through the ALD reactor, the ozone was sent through a magnesia<br />

ozone destruct unit and the remaining O2 was evacuated <strong>with</strong> a<br />

separate rotary vane pump. In the rotary reactor, the O3 was<br />

generated <strong>with</strong> an Ozonia OZAT CFS-1A ozone generator<br />

(Duebendorf, Switzerland). The ozone generator ran <strong>with</strong> O2<br />

at a flow rate of 0.2 m 3 h -1 and power of 510 W. These<br />

conditions produced an O3 concentration of 12% by mass.<br />

TEM analysis was performed in the Department of Molecular<br />

and Cellular Biology at the University of Colorado at Boulder.<br />

The TEM results were obtained <strong>with</strong> a Philips CX11 highresolution<br />

transmission electron microscope <strong>with</strong> 80 kV beam<br />

potential. A Gatan slow scan charge-coupled device camera<br />

captured the TEM images. The TEM studies monitored the<br />

conformality and thickness of the <strong>Al2O3</strong> films on the ZrO2<br />

particles.<br />

III. Results and Discussion<br />

A. Fourier Transform Infrared Spectroscopy. Studying<br />

the surface chemistry of ALD processes requires a reliable<br />

starting surface. FTIR spectroscopy can determine the initial<br />

surface species on the ZrO2 nanopowders to ensure that they<br />

will be suitable for <strong>Al2O3</strong> ALD. A range of absorbances is<br />

observed on the ZrO2 nanoparticles including the following:<br />

O-H stretching vibrations at 3670-3780 cm-1 ;C-H stretching<br />

vibrations at 2850-3050 cm-1 ; and the bulk ZrO2 absorbance<br />

at frequencies


<strong>Al2O3</strong> <strong>Atomic</strong> <strong>Layer</strong> <strong>Deposition</strong> J. Phys. Chem. C, Vol. 112, No. 49, 2008 19533<br />

hydroxyl surface species. In addition, negative absorbance<br />

features are observed at 2920-2980 cm -1 corresponding to<br />

C-H stretching vibrations and at 1212 cm -1 corresponding to<br />

the Al-CH3 deformation mode. These negative absorbance<br />

features are both from the removal of AlCH3* surface species.<br />

Figure 3b shows the FTIR difference spectra for the next TMA<br />

exposure of 0.85 Torr · s at 450 K. This spectrum indicates that<br />

TMA removed the AlOH* surface species and added AlCH3*<br />

surface species. The FTIR difference spectra in Figure 3a,b are<br />

consistent <strong>with</strong> previous FTIR studies of <strong>Al2O3</strong> ALD <strong>with</strong> TMA<br />

and H2O. 3,12<br />

The surface species change dramatically after the first ozone<br />

exposure. Figure 3c shows a markedly different spectrum <strong>with</strong><br />

new positive absorbance features visible between 1200 and 1700<br />

cm -1 . The most prominent new features were observed at 1388,<br />

1404, and 1597 cm -1 . There were also smaller new features<br />

observed at 1320 and 1475 cm -1 and a shoulder at 1720 cm -1 .<br />

Not all of the C-H features were eliminated after long ozone<br />

exposures. There were small absorbance features at 2923 and<br />

3016 cm -1 for C-H stretching vibrations that partially result<br />

from slight frequency shifts. In addition, the absorbance for the<br />

O-H stretching vibrations at 3734-3778 cm -1 was reduced in<br />

intensity compared <strong>with</strong> the intensity observed in Figure 3a after<br />

H2O exposures. In addition, Figure 3d shows that the new<br />

absorbance features added by the O3 exposure were completely<br />

removed by the next TMA exposure. TMA exposures reform<br />

the absorbances from the C-H stretching vibrations at<br />

2820-2970 cm -1 and the methyl deformation at 1212 cm -1 .<br />

The absorbance from the O-H stretching vibrations at<br />

3734-3778 cm -1 was also removed by TMA doses.<br />

The new absorbance features appearing after the ozone<br />

exposure are very characteristic of formate and carbonate groups<br />

on the <strong>Al2O3</strong> ALD surface. We first reported these new formate<br />

features at the AVS Topical Conference for <strong>Atomic</strong> <strong>Layer</strong><br />

<strong>Deposition</strong> in 2006 (ALD2006). 24 Formate groups are very<br />

common on metal oxide surfaces and can be formed by exposing<br />

metal oxides to a variety of reagents including carbon monoxide,<br />

methanol, and formaldehyde. 25-27 The absorption features at<br />

1388 and 1597 cm -1 correspond to the symmetric and antisymmetric<br />

OCO modes of bound formate species. In addition,<br />

the 1404 cm -1 shoulder is attributed to the CH bend of formate<br />

species. In the C-H stretching region, the two peaks at 2923<br />

and 3016 cm -1 are identified as the Fermi resonance of the<br />

antisymmetric OCO mode mixing <strong>with</strong> the lone C-H stretching<br />

vibration in the formate surface group. 26 All of these values<br />

match literature values for formate on aluminum oxide. 26,28,29<br />

Formate features during <strong>Al2O3</strong> ALD <strong>with</strong> ozone have also<br />

been observed recently by FTIR studies on planar surfaces that<br />

were complemented by DFT calculations. 30 These studies<br />

observed the same methoxy modes at 1388 and 1475 cm -1 that<br />

were monitored in this study on the ZrO2 nanoparticles. In<br />

addition, the ratios of the absorbances for the methoxy and<br />

primary formate vibrational features were similar on the planar<br />

surfaces and the ZrO2 nanoparticles. This agreement also rules<br />

out the possibility that ozone decomposition may have prevented<br />

the ozone from reaching the interior surfaces of the ZrO2<br />

nanoparticle sample.<br />

The surface coordination of the formate features is described<br />

by the frequency difference between the symmetric and asymmetric<br />

OCO stretching vibrations. Our observed experimental<br />

difference of 212 cm -1 is greater than that of the free formate<br />

ion. Consequently, the formate species are doubly coordinated<br />

to aluminum sites on the surface. 28 These doubly coordinated<br />

surface species formed after ozone exposure may contribute to<br />

Figure 4. FTIR difference spectra after (a) reference ozone exposure,<br />

(b) HCOH (formaldehyde) exposure on hydroxyl-terminated surface,<br />

and (c) next TMA exposure after formaldehyde exposure. All exposures<br />

were conducted at 450 K. F ) formate.<br />

the oxygen-rich stoichiometry of <strong>Al2O3</strong> films deposited using<br />

TMA and ozone. 31 The remaining shoulders at 1320 and 1720<br />

cm -1 are the symmetric and asymmetric OCO stretching<br />

vibrations of carbonate groups bound to alumina. Some of the<br />

hydroxyl features observed in the difference spectra could also<br />

result from the C-OH group atop surface carbonate species.<br />

These carbonate species can be formed by further oxidation of<br />

the formate species. Carbonate species can be prepared by<br />

reacting CO2 <strong>with</strong> alumina surfaces. The vibrational frequencies<br />

observed in this study match closely <strong>with</strong> literature values for<br />

carbonate. 25<br />

Control experiments confirm the identity of the formate<br />

species formed during ozone exposures. For these control<br />

experiments, a fresh <strong>Al2O3</strong> ALD film was grown at 450 K and<br />

then exposed to 1.0 Torr · s of formalin solution. Formalin is a<br />

solution of 37% formaldehyde, 10% methanol, and 53% water.<br />

Exposing aluminum oxide to formaldehyde will produce surface<br />

formate groups. 27,32 For reference, Figure 4a shows the FTIR<br />

difference spectrum after an ozone exposure on the <strong>Al2O3</strong> ALD<br />

surface. This spectrum is identical to the spectrum shown in<br />

Figure 3c. Figure 4b displays the FTIR difference spectrum after<br />

a formalin exposure on an <strong>Al2O3</strong> ALD surface at 450 K. The<br />

formaldehyde leads to the same major absorbance features at<br />

1388, 1404, and 1597 cm -1 observed in Figure 4a. The<br />

differences between these spectra are the peaks at 1098, 1320,<br />

and 1720 cm -1 . The first two of these peaks are attributed to<br />

carbonate groups bound on the surface. The last feature is likely<br />

absorbance from C-O stretching vibrations of surface methoxy<br />

groups. 25,26<br />

The formate species on the <strong>Al2O3</strong> ALD surface formed by<br />

the formaldehyde exposure was then exposed to TMA at 450<br />

K. The FTIR difference spectrum after the TMA exposure is<br />

shown in Figure 4c. The negative absorbances at 1404 and 1597<br />

cm -1 indicate that TMA removes the formate species from the<br />

<strong>Al2O3</strong> surface. In addition, the TMA removes additional AlOH*<br />

species as shown by the negative infrared absorbance features<br />

at 3730 and 3770 cm -1 . The positive absorbance features from<br />

2920 to 2980 cm -1 in the C-H stretching region and at 1212<br />

cm -1 for the Al-CH3 deformation mode indicate that TMA<br />

has repopulated the <strong>Al2O3</strong> surface <strong>with</strong> AlCH3* species.<br />

The surface chemistry during <strong>Al2O3</strong> ALD <strong>with</strong> TMA and<br />

ozone may depend on the substrate temperature. Figure 5 shows


19534 J. Phys. Chem. C, Vol. 112, No. 49, 2008 Goldstein et al.<br />

Figure 5. FTIR difference spectra after (a) third TMA exposure, (b)<br />

third ozone exposure, (c) fourth TMA exposure, and (d) fourth ozone<br />

exposure. All exposures were performed at 550 K. F ) formate.<br />

FTIR difference spectra after the third and fourth sequential<br />

TMA and ozone exposures at 550 K. The reference spectrum<br />

for each difference spectrum is the FTIR spectrum after the<br />

previous exposure. The absorbance features observed after the<br />

TMA and ozone exposures are very similar for the third and<br />

fourth cycle and also correspond closely to the absorbance<br />

features monitored at 450 K. This close correspondence suggests<br />

that the <strong>Al2O3</strong> ALD reaction mechanism is similar at 450 and<br />

550 K.<br />

The FTIR difference spectra in Figure 5 also consistently<br />

show the disappearance and appearance of a small absorbance<br />

feature at 2280 cm -1 <strong>with</strong> TMA and O3 exposures, respectively.<br />

This feature is attributed to weakly bound CO coordinating to<br />

Al 3+ centers on an alumina surface. CO can be formed whenever<br />

the formate species (OCHO) decompose on the alumina<br />

surface. 25 This absorbance feature becomes more prominent as<br />

more formate species undergo decomposition. To check this<br />

hypothesis, formate decomposition was examined during the<br />

formaldehyde control experiment. The formate features were<br />

observed to disappear slowly at 550 K. The intensity of the<br />

formate features was reduced by one-third in 22 h. In contrast,<br />

when the temperature was reduced to 473 K, the formate<br />

remained constant over 8hofscanning. This behavior may<br />

explain why little CO was observed at 450 K. Not enough<br />

formate is decomposing at 450 K to produce CO on the alumina<br />

surface.<br />

Figure 5 also shows the decrease and increase of absorbance<br />

for the bulk <strong>Al2O3</strong> absorbance at ∼900-1000 cm -1 for TMA<br />

and ozone exposures, respectively. 33 The increase of this<br />

absorbance during the O3 exposures is always larger than the<br />

decrease of absorbance during the TMA exposures. As a result,<br />

the absorbance of this feature increases gradually versus the<br />

number of AB cycles. This behavior is consistent <strong>with</strong> the<br />

growth of the <strong>Al2O3</strong> ALD film.<br />

The self-limiting nature of the surface reactions during <strong>Al2O3</strong><br />

ALD <strong>with</strong> TMA and ozone can be monitored using the<br />

integrated infrared absorbance for various surface species. If<br />

an ALD reaction is self-limiting, then the surface coverage will<br />

not increase after a certain reactant exposure. For these<br />

experiments, the integrated absorbance was defined for the C-H<br />

stretching vibrations at 2820-2980 cm -1 , the Al-CH3 deformation<br />

mode at 1175-1250 cm -1 , the O-H stretching vibra-<br />

Figure 6. (a) Normalized integrated absorbances during (a) TMA<br />

exposure and (b) ozone exposure at 550 K showing C-H stretch, O-H<br />

stretch, asymmetric OCO stretch, symmetric OCO stretch, and CH3<br />

deformation.<br />

tions at 3600-3800 cm -1 , and the two major formate vibrations:<br />

the antisymmetric OCO band at 1575-1625 cm -1 and the<br />

symmetric OCO band between 1350 and 1425 cm -1 .<br />

Figure 6a compares the normalized integrated absorbance of<br />

the surface species versus TMA exposure at 550 K. The<br />

absorbances of the O-H stretching vibration and the symmetric<br />

and asymmetric OCO stretching vibrations for the formate<br />

species decrease versus TMA exposure. Hydroxyls react much<br />

more rapidly than the formate features since the absorbance for<br />

the O-H stretching vibrations is reduced before the absorbance<br />

for the formate features. In close correspondence, the absorbance<br />

of the C-H stretching vibrations and the Al-CH3 deformation<br />

mode for the AlCH3* species concurrently increase versus TMA<br />

exposure. The measurements indicate that TMA exposures of<br />

0.9 Torr · s are sufficient for the TMA surface reaction to reach<br />

completion. However, the absorbance for the antisymmetric<br />

OCO stretch is not completely extinguished even after 2.0 Torr · s<br />

of TMA exposure. This behavior indicates that some formate<br />

groups do not react <strong>with</strong> TMA at 550 K.<br />

The normalized integrated absorbance of the surface species<br />

during the ozone exposure is presented in Figure 6b. The<br />

absorbance for the various surface species again shows the characteristic<br />

signature for self-limiting surface reactions. The<br />

absorbances of the O-H stretching vibration and the symmetric<br />

and asymmetric OCO stretching vibrations for the formate<br />

species increase versus O3 exposure. In close correspondence,<br />

the absorbances of the C-H stretching vibrations and the<br />

Al-CH3 deformation mode for the AlCH3* species concurrently<br />

decrease versus O3 exposure. Fermi resonances from the formate


<strong>Al2O3</strong> <strong>Atomic</strong> <strong>Layer</strong> <strong>Deposition</strong> J. Phys. Chem. C, Vol. 112, No. 49, 2008 19535<br />

Figure 7. FTIR difference spectra in the region from 900 to 1900<br />

cm -1 recorded after the second ozone exposure at (a) 363, (b) 450, (c)<br />

550, and (d) 650 K. M ) methoxy, C ) carbonate, F ) formate.<br />

groups leave residual absorbances in the 2820-2980 cm -1 range<br />

and prevent the C-H stretching features from being completely<br />

removed. These measurements indicate that O3 exposures of<br />

1.0 Torr · s are sufficient for the O3 surface reaction to reach<br />

completion.<br />

The formate species are dependent on the surface temperature.<br />

Figure 7 displays FTIR spectra that were recorded after the<br />

second ozone exposure at temperatures between 363 and 650<br />

K. Below 650 K, the peaks associated <strong>with</strong> formate and<br />

carbonate groups were visible between 1200 and 1700 cm -1 .<br />

The formate features were obscured when the temperature was<br />

raised to 650 K. This disappearance may result from the<br />

decomposition of the formate and carbonate species into CO<br />

and CO2. At all temperatures, surface methyl groups were<br />

removed based on the negative absorbance of the methyl<br />

deformation feature at 1212 cm -1 . The bulk infrared absorbance<br />

of <strong>Al2O3</strong> was visible in all the spectra at ∼900-1000 cm -1 .<br />

The strongest bulk infrared absorbance was observed at the<br />

higher temperatures.<br />

Transient species that could produce formate species were<br />

also resolved during the temperature studies. The low-temperature<br />

experiments revealed new absorbances at 1089 and 1456<br />

cm -1 that can be attributed to the C-O stretching vibration and<br />

antisymmetric CH3 deformation of methoxy species. The C-O<br />

feature was obscured at higher temperatures by the broad Al-O<br />

bulk absorption mode. In addition, the symmetric CH3 deformation<br />

of surface methoxy groups was obscured by the symmetric<br />

deformation of AlCH3* surface species. Methoxy groups are a<br />

potential intermediate to surface formate groups. 26 In agreement,<br />

features attributed to methoxy groups decrease and the formate<br />

features increase above 363 K.<br />

The 2000-4000 cm -1 region after the second ozone exposure<br />

at a variety of temperatures is shown in Figure 8. The hydroxyl<br />

region at 3650-3770 cm -1 revealed a greater proportion of<br />

higher frequency peaks at higher temperatures. The hydroxyl<br />

vibrations after ozone exposures have two prominent absorbances<br />

at 3718 and 3778 cm -1 . In comparison, the hydroxyls<br />

observed during <strong>Al2O3</strong> ALD <strong>with</strong> H2O have a diffuse band<br />

ranging from 3670 to 3730 cm -1 . 12 This contrast is consistent<br />

<strong>with</strong> a difference in the basicity of the hydroxyls on the alumina<br />

surface. The hydroxyl vibrations are directly correlated to the<br />

basicity of surface hydroxyl groups. 34,35 A higher ν(OH)<br />

Figure 8. FTIR difference spectra in the region from 2000 to 4000<br />

cm -1 recorded after the second ozone exposure at (a) 363, (b) 450, (c)<br />

550, and (d) 650 K. M ) methoxy.<br />

vibrational frequency indicates increased basicity of surface<br />

hydroxyl groups. The FTIR spectra suggest that <strong>Al2O3</strong> ALD<br />

grown at higher temperatures <strong>with</strong> ozone produces a larger<br />

proportion of strongly basic hydroxyls than <strong>Al2O3</strong> ALD grown<br />

<strong>with</strong> water.<br />

The C-H stretching features are consistent <strong>with</strong> the removal<br />

of AlCH3* species and production of new C-H stretching<br />

vibrations corresponding to methoxy species. The absorbance<br />

features around 2850 cm -1 are associated <strong>with</strong> the CH3<br />

symmetric stretching vibration of methoxy species. These<br />

methoxy features are lost at higher temperatures. As the<br />

temperature is raised to 550 K, Figure 8c reveals an absorbance<br />

at 2280 cm -1 that has already been identified as CO from the<br />

decomposition of formate. In addition, a peak at 2324 cm -1<br />

appeared when the temperature was increased to 650 K. This<br />

peak is attributed to weakly bound CO2 that coordinates to Lewis<br />

base sites on the alumina surface. 25 A correlation of the<br />

vibrations of the observed surface species, their referenced<br />

literature values, and observed experimental values is given in<br />

Table 1.<br />

Another control experiment was performed at temperatures<br />

between 450 and 650 K to distinguish the effects of ozone from<br />

oxygen on a surface covered <strong>with</strong> AlCH3* species following<br />

the TMA reaction. Oxygen was dosed into the reactor chamber<br />

through the ozone generator <strong>with</strong> the same exposure as ozone<br />

for each growth temperature. The control reactions were<br />

performed one week after the final ozone experiment to ensure<br />

that no residual ozone was left in the ozone generator. Direct<br />

comparison of the control reactions in Figure 9 shows that<br />

oxygen reacts only slightly up to 550 K because very little of<br />

the absorbance for the A1–CH3 deformation mode at 1212 cm -1<br />

was removed from the spectrum. At 650 K, oxygen reacts <strong>with</strong><br />

AlCH3* species but does not produce any absorbance for O-H<br />

stretching vibrations from hydroxyl species. <strong>Al2O3</strong> ALD growth<br />

may also be occurring at 650 K because the bulk <strong>Al2O3</strong><br />

absorbance increases after the oxygen exposure. However,<br />

multiple sequential TMA and O2 exposures were not performed<br />

to confirm <strong>Al2O3</strong> ALD growth.<br />

The bulk Al-O absorbance feature can be used to monitor<br />

the growth of the <strong>Al2O3</strong> ALD film on the ZrO2 nanopowders.<br />

An <strong>Al2O3</strong> ALD film was grown using 40 sequential exposures<br />

of 0.9 Torr · s TMA and 1.0 Torr · s of ozone at 550 K. A 90 s<br />

purge separated the reactants to minimize possible CVD and to


19536 J. Phys. Chem. C, Vol. 112, No. 49, 2008 Goldstein et al.<br />

Figure 9. FTIR difference spectra after an O2 exposure equivalent to<br />

the O2/O3 exposures at (a) 450, (b) 550, and (c) 650 K. (d) FTIR<br />

difference spectra after an O3 exposure at 650 K.<br />

TABLE 1: Vibrational Assignments Comparing Literature<br />

and Observed Experimental Values<br />

assignment lit. (cm-1 ) exptl (cm-1 )<br />

Al-O phonon 900-1000<br />

ν(s) C-O (m) 110026,29,30,32,38 1089<br />

Sym. Al-CH3 def 11973,12 1212<br />

ν (s) OCO (c) 131525 1320<br />

ν (s) OCO (f) 138026,29,30,32,38 1388<br />

δCH (f, m) 139526,29,30,32,38 1404<br />

asym CH3 def (m) 145026,29,30,32,38 1475<br />

ν (as) OCO (f) 159526,29,30,32,38 1597<br />

ν (a) OCO (c) 171025 1720<br />

ν (s) CH3 (m) 284626,29,30,32,38 2850<br />

ν (as) CH3 (m) 295026,29,30,32,38 2923<br />

ν (s) OCO + δCH (f) 297026,29,30,32,38 3016<br />

allow desorbing species to be swept from the nanopowders.<br />

FTIR scans were recorded at regular intervals to monitor the<br />

<strong>Al2O3</strong> ALD film growth. The absolute FTIR spectra shown in<br />

Figure 10 were acquired after the O3 exposures and are<br />

referenced to the CsI windows. There is a continuous increase<br />

of the bulk Al-O absorbance mode versus number of AB<br />

cycles.<br />

Figure 10. Absolute FTIR spectra showing the growth of the bulk<br />

Al-O bulk absorbance feature after various numbers of TMA and O3<br />

reaction cycles.<br />

Figure 11. TEM image of ZrO2 particles coated <strong>with</strong> <strong>Al2O3</strong> after 40<br />

reaction cycles of TMA and ozone.<br />

After the 40 AB reaction cycles at 550 K, the sample was<br />

removed from the reactor and TEM was performed on the coated<br />

ZrO2 nanopowders to determine the film conformality and ALD<br />

growth rate. The bright field TEM image recorded at 130 000×<br />

is shown in Figure 11. The ZrO2 nanopowders have a conformal<br />

coating <strong>with</strong> a thickness of 65 Å. This thickness is consistent<br />

<strong>with</strong> a growth rate of 1.1 Å/cycle for <strong>Al2O3</strong> ALD <strong>with</strong> ozone at<br />

550 K. This value was determined by subtracting the estimated<br />

12 Å base layer of <strong>Al2O3</strong> ALD grown using TMA and H2O. A<br />

second deposition experiment at 650 K provided the same<br />

growth rate and also showed conformality of the <strong>Al2O3</strong> ALD<br />

film.<br />

B. Quadrupole Mass Spectrometry. Identification of the<br />

gas phase species formed after O3 exposures can help determine<br />

how the new surface species were formed on the <strong>Al2O3</strong> surface.<br />

The mass spectrum recorded during the first O3 micropulse at<br />

550 K is shown in Figure 12a. This spectrum is shown using a<br />

log intensity scale because the reaction products are very small<br />

compared <strong>with</strong> the O2 fragmentation pattern. The large peak at<br />

m/z 32 is attributed to a mixture of O2 and O3. Hydrocarbons<br />

evolved in this first ozone micropulse are identified as C2H4<br />

(ethylene) by the three peaks from m/z 26 to 28 and CH4<br />

(methane) by the peaks from m/z 12 to 16. Ethylene peaks are<br />

observed in the QMS scans until the 10th micropulse of ozone<br />

when their signals reach the noise level of


<strong>Al2O3</strong> <strong>Atomic</strong> <strong>Layer</strong> <strong>Deposition</strong> J. Phys. Chem. C, Vol. 112, No. 49, 2008 19537<br />

Figure 12. (a) Mass spectrum during the first O3 micropulse showing<br />

CH4 and C2H4 at 550 K. (b) Mass spectrum during the 55th O3<br />

micropulse showing CH4, CO, and CO2 at 550 K.<br />

TMA micropulse. No peaks from TMA are observed during<br />

this first micropulse. The only major reaction product is CH4<br />

<strong>with</strong> its characteristic peaks between m/z 12 and 16. In addition,<br />

a slight amount of CO is detected at m/z 28. The appearance of<br />

CO reaction product suggests that formate and carbonate species<br />

are still decomposing on the surface at 550 K.<br />

Figure 13b shows the mass spectrum after the 55th TMA<br />

micropulse. The TMA reaction has reached completion on the<br />

surface and the mass spectrum is consistent <strong>with</strong> unreacted TMA<br />

<strong>with</strong> its primary mass cracking fragments centered at m/z 57<br />

and 42. The peaks between m/z 12 and 16 are mass cracking<br />

fragments for TMA. In addition, slight amounts of CO and CO2<br />

are detected at m/z 28 and 44. The appearance of these reaction<br />

products suggests that formate and carbonate species are still<br />

decomposing on the surface at 550 K. Gas phase CO and CO2<br />

are present in the mass spectrum after subtracting the cracking<br />

patterns from both CH4 and TMA. Analysis of the QMS data<br />

shows that the quantity of CO and CO2 steadily decreases versus<br />

time.<br />

C. Mechanism of <strong>Al2O3</strong> ALD <strong>with</strong> Ozone. The FTIR<br />

spectra and the QMS data allow a mechanism to be proposed<br />

for <strong>Al2O3</strong> ALD <strong>with</strong> TMA and ozone. The mechanism during<br />

the ozone reaction is presented in Figure 14. The initial surface<br />

is covered <strong>with</strong> AlCH3* species resulting from the TMA reaction<br />

on a hydroxylated initial surface. There are two potential<br />

pathways for ozone to react <strong>with</strong> AlCH3* species. One reaction<br />

pathway involves oxygen insertion into the AlC-H bond. The<br />

second reaction pathway involves oxygen insertion into the<br />

Al-C bond.<br />

Figure 13. (a) Mass spectrum during the first TMA micropulse<br />

showing CH4 and CO at 550 K. (b) Mass spectrum during the 55th<br />

TMA micropulse showing TMA at 550 K.<br />

Figure 14. Proposed mechanism for O3 reaction during <strong>Al2O3</strong> ALD<br />

using TMA and O3.<br />

Oxygen insertion into the AlC-H bond has been predicted<br />

by quantum mechanical simulations and is believed to produce<br />

ethylene as a reaction product. 6,17,37 Oxygen insertion into the<br />

AlC-H bond produces AlCH2OH* species. The presence of<br />

these AlCH2OH* species are not ruled out by the FTIR spectra<br />

because the O-H stretching vibration for AlCH2OH* species<br />

would be difficult to distinguish from that of AlOH* species.<br />

These AlCH2OH* species could also be identified by the<br />

presence of methylene C-H stretching vibrations or CH2<br />

rocking modes. However, the intense formate vibrations and<br />

Fermi resonances obscure these vibrations. The transient nature<br />

of these species may also prevent these species from being<br />

isolated by using time- or temperature-dependent measurements.<br />

Adjacent AlCH2OH* species then eliminate ethylene and<br />

leave behind two AlOH* species as shown in Figure 14. This


19538 J. Phys. Chem. C, Vol. 112, No. 49, 2008 Goldstein et al.<br />

reaction pathway is supported by the immediate production of<br />

hydroxyl groups after small ozone doses in Figure 6b and the<br />

ethylene observed by the QMS results as shown in Figure 12a.<br />

The combination of two AlCH2OH* species to produce two<br />

AlOH* species is important because this decomposition pathway<br />

regenerates hydroxyl groups that are needed for the subsequent<br />

TMA reaction.<br />

Oxygen atom insertion into the Al-C bond produces a<br />

methoxy group that is only stable below 473 K according to<br />

the FTIR spectra. Adjacent methoxy groups then combine to<br />

form formate species, releasing CH4 and H2 into the gas phase.<br />

This reaction has been well-characterized in the surface<br />

chemistry literature. 29 This reaction is also the likely source of<br />

gaseous CH4 observed by the QMS measurements in Figure<br />

12. Further oxidation of the formate groups yields carbonate<br />

groups. These carbonate features are in a minority as suggested<br />

by their weak infrared absorbances. Oxygen insertion into the<br />

Al-C bond is likely the primary pathway for ozone activation<br />

based on the large amount of CH4 observed in the QMS<br />

compared <strong>with</strong> the small C2H4 signals that were only monitored<br />

during the initial O3 micropulses. These minority C2H4 reaction<br />

products may only be observed during the initial O3 micropulses<br />

because of higher AlCH3* surface coverages.<br />

As the reaction progresses, the formate and carbonate groups<br />

decompose into CO and CO2, respectively. The CO and CO2<br />

gas phase products were both observed by the QMS measurements<br />

during the later O3 micropulses in Figure 12b and during<br />

the TMA micropulses in Figure 13. The formate and carbonate<br />

decomposition leaves behind one hydroxyl group on alumina<br />

per decomposed moiety. This hydroxyl group may be the<br />

isolated hydroxyl observed by the FTIR spectra at 3778 cm -1 .<br />

CH4 could also be produced when these surface hydroxyl groups<br />

react <strong>with</strong> nearby AlCH3* species.<br />

The TMA reaction is believed to be very similar to the<br />

reported TMA reaction for <strong>Al2O3</strong> ALD <strong>with</strong> TMA and water. 3<br />

AlOH* species formed during ethylene elimination or formate<br />

or carbonate decomposition in Figure 14 can easily react <strong>with</strong><br />

TMA. This reaction reforms the AlCH3* species that are<br />

observed in the FTIR spectra. The efficiency of this TMA<br />

reaction is illustrated in Figure 6a and requires similar exposures<br />

of TMA as the TMA reaction during <strong>Al2O3</strong> ALD <strong>with</strong> TMA<br />

and water. The hydroxyl species are reduced more rapidly than<br />

the formate species during the TMA reaction.<br />

Figure 6a also shows that TMA is unable to remove all the<br />

formate species from the surface. This observation has been<br />

noted by other infrared studies of the TMA reaction during<br />

<strong>Al2O3</strong> ALD <strong>with</strong> ozone. 30 On the basis of our results, the<br />

complete removal of the formate species is dependent on the<br />

decomposition of the formate species. If these formate species<br />

are not removed, then carbon may build up in the <strong>Al2O3</strong> ALD<br />

films. These residual formate species that are not removed by<br />

TMA may explain the high carbon concentration observed in<br />

<strong>Al2O3</strong> ALD films grown <strong>with</strong> TMA and ozone at lower<br />

temperatures. 13<br />

IV. Conclusions<br />

In situ transmission FTIR spectroscopy and QMS were used<br />

to study <strong>Al2O3</strong> ALD <strong>with</strong> sequential exposures of Al(CH3)3 and<br />

O3. The FTIR spectroscopy studies of the surface reactions<br />

occurring during <strong>Al2O3</strong> ALD were performed at temperatures<br />

from 363 to 650 K. The FTIR spectra were recorded after<br />

Al(CH3)3 and ozone exposures. These FTIR spectra showed that<br />

the ozone exposure removes surface AlCH3* species and<br />

produces AlOCH3* (methoxy), Al(OCHO)* (formate), Al(O-<br />

COOH)* (carbonate), and AlOH* (hydroxyl) species. The TMA<br />

exposure then removes the methoxy, formate, carbonate, and<br />

hydroxyl species and reestablishes the AlCH3* species. The<br />

identification of formate species in the FTIR spectra was<br />

confirmed by separate formaldehyde adsorption experiments on<br />

<strong>Al2O3</strong> ALD surfaces. The formate species were temperature<br />

dependent and were nearly absent from the FTIR spectra at<br />

g650 K.<br />

Repeating the TMA and O3 exposures in a sequential reaction<br />

sequence progressively deposited an <strong>Al2O3</strong> ALD film as<br />

monitored by the absorbance increase for bulk <strong>Al2O3</strong> infrared<br />

features. TEM was used to examine the <strong>Al2O3</strong> ALD films<br />

deposited on the ZrO2 particles. The <strong>Al2O3</strong> ALD films were<br />

very conformal to the underlying ZrO2 particles. These TEM<br />

images were consistent <strong>with</strong> an <strong>Al2O3</strong> ALD growth rate of 1.1<br />

Å/cycle at 550 K. QMS analysis of the gas phase species<br />

revealed that the TMA reaction produced CH4. The ozone<br />

reaction then produced mainly CH4 <strong>with</strong> a small amount of C2H4<br />

(ethylene), CO, and CO2.<br />

On the basis of the surface species observed by the FTIR<br />

studies and the gas phase species monitored by the QMS<br />

investigations, a very different mechanism is suggested for <strong>Al2O3</strong><br />

ALD growth using TMA and O3 compared <strong>with</strong> <strong>Al2O3</strong> ALD<br />

using TMA and H2O. During <strong>Al2O3</strong> ALD using ozone, both<br />

Al-C and C-H bond insertion occurs when O3 reacts <strong>with</strong><br />

AlCH3* species to create methoxy and AlCH2OH* species. The<br />

methoxy species decompose to formate and carbonate species.<br />

The formate and carbonate species release CO and CO2 to<br />

produce AlOH* species. Two AlCH2OH* species also eliminate<br />

C2H4 and produce two AlOH* species.<br />

Acknowledgment. This work was supported by the National<br />

Science Foundation (CHE-0715552). Some of the equipment<br />

used in this research was provided by the Air Force Office of<br />

Scientific Research. The authors thank Dr. Thomas Giddings<br />

for assistance <strong>with</strong> the TEM analysis.<br />

References and Notes<br />

(1) George, S. M.; Ott, A. W.; Klaus, J. W. J. Phys. Chem. 1996, 100,<br />

13121.<br />

(2) Ritala, M.; Leskelä, M. In Handbook of Thin Film Materials; Nalwa,<br />

H. S., Ed.; Academic Press: San Diego, CA, 2001; Vol. 1, p 129.<br />

(3) Dillon, A. C.; Ott, A. W.; Way, J. D.; George, S. M. Surf. Sci.<br />

1995, 322, 230.<br />

(4) Ott, A. W.; Klaus, J. W.; Johnson, J. M.; George, S. M. Thin Solid<br />

Films 1997, 292, 135.<br />

(5) Ott, A. W.; McCarley, K. C.; Klaus, J. W.; Way, J. D.; George,<br />

S. M. Appl. Surf. Sci. 1996, 107, 128.<br />

(6) Puurunen, R. L. J. Appl. Phys. 2005, 97, 121301.<br />

(7) Wilson, C. A.; Grubbs, R. K.; George, S. M. Chem. Mater. 2005,<br />

17, 5625.<br />

(8) Elam, J. W.; Routkevitch, D.; Mardilovich, P. P.; George, S. M.<br />

Chem. Mater. 2003, 15, 3507.<br />

(9) McCormick, J. A.; Rice, K. P.; Paul, D. F.; Weimer, A. W.; George,<br />

S. M. Chem. Vap. <strong>Deposition</strong> 2007, 13, 491.<br />

(10) Groner, M. D.; Fabreguette, F. H.; Elam, J. W.; George, S. M.<br />

Chem. Mater. 2004, 16, 639.<br />

(11) Elam, J. W.; Groner, M. D.; George, S. M. ReV. Sci. Instrum. 2002,<br />

73, 2981.<br />

(12) Ferguson, J. D.; Weimer, A. W.; George, S. M. Thin Solid Films<br />

2000, 371, 95.<br />

(13) Kim, S. K.; Lee, S. W.; Hwang, C. S.; Min, Y. S.; Won, J. Y.;<br />

Jeong, J. J. Electrochem. Soc. 2006, 153, F69.<br />

(14) Kim, S. K.; Hwang, C. S. J. Appl. Phys. 2004, 96, 2323.<br />

(15) Kim, J. B.; Kwon, D. R.; Chakrabarti, K.; Lee, C.; Oh, K. Y.; Lee,<br />

J. H. J. Appl. Phys. 2002, 92, 6739.<br />

(16) Avice, M.; Grossner, U.; Monakhov, E. V.; Grillenberger, J.; Nilsen,<br />

O.; Fjellvag, H.; Svensson, B. G. Mater. Sci. Forum 2005, 483-485, 705.<br />

(17) Elliott, S. D.; Scarel, G.; Wiemer, C.; Fanciulli, M.; Pavia, G. Chem.<br />

Mater. 2006, 18, 3764.<br />

(18) Lee, T.-P.; Jang, C.; Haselden, B.; Dong, M. J. Vac. Sci. Technol.<br />

B 2004, 22, 2295.


<strong>Al2O3</strong> <strong>Atomic</strong> <strong>Layer</strong> <strong>Deposition</strong> J. Phys. Chem. C, Vol. 112, No. 49, 2008 19539<br />

(19) Jakschik, S.; Schroeder, U.; Hecht, T.; Krueger, D.; Dollinger, G.;<br />

Bergmaier, A.; Luhmann, C.; Bartha, J. W. Appl. Surf. Sci. 2003, 211, 352.<br />

(20) Ferguson, J. D.; Weimer, A. W.; George, S. M. J. Vac. Sci. Technol.<br />

A 2004, 22, 118.<br />

(21) Ballinger, T. H.; Wong, J. C. S.; Yates, J. T. Langmuir 1992, 8,<br />

1676.<br />

(22) McCormick, J. A.; Cloutier, B. L.; Weimer, A. W.; George, S. M.<br />

J. Vac. Sci. Technol. A 2007, 25, 67.<br />

(23) Benfer, S.; Knozinger, E. J. Mater. Chem. 1999, 9, 1203.<br />

(24) Goldstein, D. N.; George, S. M. 6th International Conference on<br />

<strong>Atomic</strong> <strong>Layer</strong> <strong>Deposition</strong>, (ALD2006), Seoul, Korea, July 25, 2006.<br />

(25) Busca, G.; Lorenzelli, V. Mater. Chem. 1982, 7, 89.<br />

(26) Greenler, R. G. J. Chem. Phys. 1962, 37, 2094.<br />

(27) Carlos-Cuellar, S.; Li, P.; Christensen, A. P.; Krueger, B. J.;<br />

Burrichter, C.; Grassian, V. H. J. Phys. Chem. A 2003, 107, 4250.<br />

(28) Busca, G. Catal. Today 1996, 27, 323.<br />

(29) McInroy, A. R.; Lundie, D. T.; Winfield, J. M.; Dudman, C. C.;<br />

Jones, P.; Lennon, D. Langmuir 2005, 21, 11092.<br />

(30) Kwon, J. H.; Dai, M.; Halls, M. D.; Chabal, Y. J. Chem. Mater.<br />

2008, 20, 3248.<br />

(31) Kim, S. K.; Hwang, C. S. J. Appl. Phys. 2004, 96, 2323.<br />

(32) Busca, G.; Lamotte, J.; Lavalley, J. C.; Lorenzelli, V. J. Am. Chem.<br />

Soc. 1987, 109, 5197.<br />

(33) Frank, M.; Wolter, K.; Magg, N.; Heemeier, M.; Kuhnemuth, R.;<br />

Baumer, M.; Freund, H. J. Surf. Sci. 2001, 492, 270.<br />

(34) Lundie, D. T.; McInroy, A. R.; Marshall, R.; Winfield, J. M.; Jones,<br />

P.; Dudman, C. C.; Parker, S. F.; Mitchell, C.; Lennon, D. J. Phys. Chem.<br />

B 2005, 109, 11592.<br />

(35) Chesters, M. A.; McCash, E. M. Spectrochim. Acta A 1987, 43,<br />

1625.<br />

(36) Juppo, M.; Rahtu, A.; Ritala, M.; Leskela, M. Langmuir 2000, 16,<br />

4034.<br />

(37) Prechtl, G.; Kersch, A.; Icking-Konert, G. S.; Jacobs, W.; Hect,<br />

T.; Boubekeur, H.; Schroder, U. Technical Digest of the IEEE International<br />

Electron DeVices Meeting (IEDM); IEEE: Piscataway, NJ, 2003; p 9.6.1.<br />

(38) Gopal, P. G.; Schneider, R. L.; Watters, K. L. J. Catal. 1987, 105,<br />

366.<br />

JP804296A

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!