Alternative Low-Pressure Surface Chemistry of TitaniumTetraisopropoxide on Oxidized MolybdenumAlexis M. Johnson† and Peter C. Stair*,†,‡

†Department of Chemistry, Northwestern University, Evanston, Illinois 60208, United States‡Chemical Sciences and Engineering Division, Argonne National Laboratory, Argonne, Illinois 60439, United States

*S Supporting Information

ABSTRACT: Titanium tetraisopropoxide (TTIP) is a precursor utilized in atomic layerdepositions (ALDs) for the growth of TiO2. The chemistry of TTIP deposition onto aslightly oxidized molybdenum substrate was explored under ultrahigh vacuum (UHV)conditions with X-ray photoelectron spectroscopy. Comparison of the Ti(2p) and C(1s)peak areas has been used to determine the surface chemistry for increasing substratetemperatures. TTIP at a gas-phase temperature of 373 K reacts with a MoOx substrate at373 K but not when the substrate is at 295 K, consistent with a reaction that proceeds via aLangmuir−Hinshelwood mechanism. Chemical vapor deposition was observed fordepositions at 473 K, below the thermal decomposition temperature of TTIP and withinthe ALD temperature window, suggesting an alternative reaction pathway competitive toALD. We propose that under conditions of low pressure and moderate substratetemperatures dehydration of the reacted precursor by nascent TiO2 becomes the dominantreaction pathway and leads to the CVD growth of TiO2 rather than a self-limiting ALDreaction. These results highlight the complexity of the chemistry of ALD precursors and demonstrate that changing the pressurecan drastically alter the surface chemistry.

■ INTRODUCTION

Atomic layer deposition (ALD) is a technique that relies ontwo sequential gas−surface reactions to grow materials in alayer-by-layer fashion. This method is employed in thesemiconductor industry and is most commonly used to depositmetal oxides.1 However, ALD has become an emergingtechnique for the fabrication of novel and complex catalyticsystems and in the process has been extended to the growth ofmetal films and nanoparticles on catalytic supports.2,3

ALD employs multiple cycles that are combined to buildmaterials with precise thickness control. One ALD cycle iscomposed of two half cycles that result in the regeneration ofthe initial surface groups, providing new reactive sites for thefollowing cycle. Given a finite number of surface reaction sitesand steric constraints imposed by the ligands, only a finitenumber of precursor molecules can react, resulting in the self-limiting nature of the technique.4 Reactive sites on oxidesinclude hydroxyl groups, bridging oxygen atoms, and defectsites.There has been a significant amount of work directed toward

elucidating ALD chemistry, utilizing techniques such as X-rayphotoelectron spectroscopy (XPS), mass spectrometry, quartzcrystal microbalance (QCM), temperature-programmed de-sorption (TPD), DFT, and ab initio calculations.5,6 The moststudied system is the deposition of Al2O3 from trimethylalu-minum (TMA) and H2O because it is widely used andconsidered an “ideal” deposition process. Significant work hasbeen done to establish a mechanism for the reaction, identifythe source of self-limiting behavior, and understand the reaction

kinetics.4,7,8 Many of these experiments are summarized in adedicated review by Puurunen.4 Studies of other ALD metalprecursors and corresponding growth processes of metal oxidefilms, such as Nb2O5, HfO2, ZrO2, Al2O5, and TiO2, have alsobeen performed.9−12 Additional experiments have explored theinfluence of different oxidizing precursors including H2O,ozone, and O2 plasma.

13−15

While TiCl4 is the common precursor for the growth ofTiO2, titanium tetraisopropoxide (TTIP) is an importantalternative. TTIP is preferable for materials intended forcatalytic applications since residual surface ligands can beremoved by calcination.16 In contrast with TiCl4, residualsurface chlorine induces Brønsted acidity that impartspotentially undesirable catalytic functionality. While the thermaldecomposition of TTIP for chemical vapor deposition (CVD)applications is the subject of a number of publications, a smallernumber have looked at the ALD chemistry of TTIP.5,17−19 Anenlightening study by Rahtu and Ritala combines massspectrometry, quartz crystal microbalance (QCM), and TPDexperiments to explore the reaction mechanism of TTIPdeposition onto both a predosed Ti(OCH(CH3)2)4−x coveredsurface and SiO2 for TPD and QCM experiments, respec-

Special Issue: John C. Hemminger Festschrift

Received: June 6, 2014Revised: October 13, 2014

Article

pubs.acs.org/JPCC

© XXXX American Chemical Society A dx.doi.org/10.1021/jp505653u | J. Phys. Chem. C XXXX, XXX, XXX−XXX

tively.17 The current understanding is that the TTIP reactionwith a metal oxide surface follows eqs 1a and 1b

− +

→ − − +

2 OH Ti(O Pr)

( O ) Ti(O Pr) 2(CH ) CHOH

i

i

(s) 4(g)

2 2(s) 3 2 (g) (1a)

− − +

→ − − − +

( O ) Ti(O Pr) 2H O

( O ) Ti( OH) 2(CH ) CHOH

i2 2(s) 2 (g)

2 2(s) 3 2 (g) (1b)

In this mechanism, half of the precursor ligands are removedupon the initial reaction with surface hydroxyl groups (eq 1a),and the other half are released with a subsequent water dose(eq 1b).Equations 1a and 1b can be generalized for many ALD

processes on metal oxide surfaces for which the majority of thereactive sites are surface hydroxyl groups. However, as ALDapplications are broadening more nontraditional substrates arebeing used, and the chemistry involved is less well understood.In particular, only a few studies have explored reactionmechanisms on noble metal surfaces, such as Cu and Pd, andthese studies predominantly involve the TMA/H2O proc-ess.20,21

In this study we use a slightly oxidized Mo(100) sample asour substrate to develop a better understanding of howdeposition proceeds on alternative surfaces compared todeposition on typical metal oxides such as Al2O3, TiO2, orSiO2. In addition, it is advantageous to use MoOx as the basesubstrate in order to minimize reactions with the alkoxideligands on TTIP. While molybdenum oxides can catalyzealcohol dehydration to alkenes through an alkoxide inter-mediate, recent studies have shown that an alkoxide on thesurface of MoO3 would be more likely to recombine with ahydrogen atom to desorb as the alcohol than dehydrate to thealkene.22,23 As the Mo oxidation state decreases below 6+dehydration of the alkoxide is even less probable. X-rayphotoelectron spectroscopy (XPS) was used to explore thenature of the chemistry during TiO2 deposition by TTIP ontoMoOx surfaces at different temperatures. The ratio of Ti(2p)and C(1s) XPS peak areas was used to determine whether theTiO2 deposition followed CVD or ALD mechanisms at eachtemperature. Surprisingly, we found that CVD can be thedominant reaction mechanism even at temperatures used instandard ALD growths that are below where TTIP decom-poses.

■ EXPERIMENTAL METHODS

Instrumentation and Sample Mounting. All experi-ments were performed in a previously described three-levelultrahigh vacuum (UHV) apparatus with a base pressure thatranged from 5 × 10−10 to 2 × 10−9 Torr.24 Briefly, the upperlevel includes an Extrel quadrupole mass spectrometer, lowenergy electron diffraction (LEED), ion sputtering gun, heateddosing lines, and a dual-anode Thermo/VG Microtech X-rayphotoelectron spectrometer with a hemispherical electronenergy analyzer. The middle level houses a high-resolutionelectron energy loss spectrometer, and the bottom levelcontains a high-pressure cell (HPC). Recently, an ArradianceGemstar-6 ALD system was interfaced to the top level of theUHV apparatus. A 360° rotating XYZ manipulator with a 1 mZ-translator was used to position the sample for sputtering, gas

dosing, transferring into the HPC, and performing analyticalmeasurements.The sample was a 1 cm diameter Mo(100) single-crystal disk

that was cut and mechanically polished to within 1° of the 100plane. It was initially cleaned in a separate UHV chamber byheating to 1500 K in 5 × 10−7 Torr of oxygen for 48 h toremove bulk carbon and sulfur. Two 0.25 mm Mo wires wereused to suspend the crystal between two thermally isolatedcopper leads at the bottom of the manipulator shaft. Thesample was resistively heated, and the temperature wasmeasured with a K-type thermocouple spot-welded to theback of the crystal. Between TiO2 deposition experiments thecrystal was cleaned with argon ion sputtering using 2 keV ionsat normal incidence followed by brief annealing to 1073 K. Thisprocedure was repeated until the crystal was mostly free oftitanium. The crystal was subsequently annealed at 773 K under600 L of O2 to remove carbide species that had formed duringthe cleaning process. XPS was used to verify the chemical stateand composition of the crystal surface. Peak areas werecalculated with the appropriate sensitivity factor for comparisonbetween elements, and key elemental ratios are tabulated foreach experiment in the Supporting Information.25

TTIP Deposition. Titanium tetraisopropoxide (TTIP,Sigma-Aldrich 97 and 99.999%) was purified using severalfreeze−pump−thaw cycles before it was deposited onto theMo(100) single crystal in the UHV chamber by one of twoexperimental methods: (1) dosing lines and (2) the Gemstar-6ALD system. 97% pure TTIP was used in the dosing lines,while 99.999% purity was used in the Gemstar-6 system;however, both systems produced similar deposition chemistryand are detailed below.26 Once a dose was complete, thechamber was evacuated for 2−3 h until the pressure reached10−9 Torr for analysis with XPS. A Mo(3d) satellite peakoverlaps with the C(1s) region and was removed throughsubtraction of spectra that were taken of the base substrateprior to precursor dosing.

Gas Delivery via Dosing Lines. The dosing lines wereused for experiments with the substrate heated to 573 and 473K. The dosing line is evacuated with a mechanical pump to ∼10mTorr and is comprised of a precursor vessel, a leak valve, anda sapphire tube. The precursor vessel is isolated from thedosing line by a needle valve. A leak valve (Granville-Philips)separates the low vacuum side of the line from the UHV systemand is used to introduce gaseous precursors into the chamber.The end of the dosing line is a sapphire tube with a 0.25 mmtantalum wire coiled around it for resistive heating to controlthe temperature of the gas entering the system. Prior todeposition, the head space of the precursor vessels and thedosing lines were evacuated and purged several times with N2gas before directed dosing with either TTIP or H2O to apressure of 1 × 10−6 Torr. All precursor vessels and dosing lineswere kept at room temperature, ∼295 K, during the duration ofthese experiments, while the sapphire tube was heated to 383K. The vapor pressures of TTIP and water at room temperatureare 0.4 and 21 Torr, respectively.27,28 After each dose, the linesand the chamber were purged with 1 × 10−5 Torr of N2 for aminimum of 15 min. The QMS was used to verify the presenceof the intact precursor during the dose and the disappearance ofthe precursor during the N2 purge.

Gas Delivery via the Gemstar-6. A commercial ALDsystem (Arradiance Gemstar-6) was interfaced to the UHVchamber through an auxiliary port on the system. One end of a1/4″ OD stainless steel tube is inserted into the oven of the

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp505653u | J. Phys. Chem. C XXXX, XXX, XXX−XXXB

ALD system, and a fraction of the gas stream is directed intothe UHV system toward the surface of the Mo substrate. AVCR gasket at the end of the stainless steel tube with a 75 μmdiameter laser-drilled hole reduces the pressure from 1 to 10Torr to ∼10−7−10−6 Torr. The connecting line could beevacuated with a turbo pump (Pfeiffer-Balzer 070) and heatedto 383 K. The temperatures of the TTIP and water containerswere 353 and 295 K, respectively. The corresponding vaporpressure for TTIP is ∼2.2 Torr.28 The head space of theprecursor bottle was evacuated before dosing. Doses wereperformed by first closing the isolation valve between the ALDsystem and the mechanical pump evacuating it. The precursorwas then dosed into the ALD reaction chamber followed byminimizing the carrier N2 flow to prevent large changes to thepartial pressures within the reaction chamber over the course ofthe dose. After 15 s of equilibration time the gas was introducedinto the UHV chamber. After each dose was complete theconnecting line was purged with N2 for a minimum of 20 min,and the concentration of precursor remaining in the chamberwas monitored with the QMS. The partial pressure of TTIPand H2O entering the chamber was less than 3 × 10−7 and 3 ×10−6 Torr, respectively, based on the changes in the basepressure read by an ion gauge (Granville-Philips). Thisexperimental setup was used for the set of depositionexperiments in which the surface of the Mo crystal was heatedto 373 K.

■ RESULTSBase MoOx Surface. A clean Mo (100) surface is extremely

reactive and will dissociatively adsorb gases at room temper-ature until a passivation layer is formed.29 The annealingtemperatures used in the study temporarily removed thecarbon, but were not high enough to completely desorb all ofthe oxygen. The base Mo surface in these experiments waspassivated by both carbon and oxygen, as shown in therepresentative XPS data in Figure 1 for the base substrate usedin the 573 K experiments. The integrated peaks from panels a,b and c corresponding to Mo(3d), O(1s) and C(1s) spectra,respectively, give a Mo:O:C ratio of 11:7:1. The bindingenergies for the Mo(3d5/2) peak from the base substrates in thiswork ranged between 227.9 and 228.0 eV with a full width half-maximum (fwhm) of ∼1.1 eV, Figure 1a. The binding energiesfor the O(1s) peak range from 530.4 to 530.7 eV with acorresponding range of fwhm of 1.9−3.3 eV, Figure 1b. AllO(1s) spectra are well fit with two Gaussians, with peaks at530.5 and 531.4 eV, consistent with O(1s) spectra frommolybdenum oxides found in the literature.30 The C(1s) peakhas a binding energy of 285.2 eV and a fwhm of 2.4 eV,consistent with adventitious carbon.After a brief anneal to 573 K the C(1s) signal disappeared

from the base MoOx surface but grew back to the concentrationshown in Figure 1c during the experiment. It is difficult with thecurrent XPS orientation, θ = 0°, to determine the exactelemental ratio of the top few layers of the substrate. Previouswork from our laboratory has explored the effects ofchemisorbed oxygen on the structure, oxidation state, andacidity of Mo(100).29,31−34 These early experiments used COdissociation to calibrate the O(1s) signal to establish a metricfor the expected intensity ratio of the Mo(3d5/2) to the O(1s)peaks for increasing oxygen coverage ranging from 0.3 to 2monolayers on a Mo(100) surface.35 For this work, the O/Moratios are 0.06, 0.15, and 0.14 for the base substrate in the 573,473, and 373 K experiments, respectively. These ratios

correspond to oxygen surface coverages of 0.7, 1.7, and 1.5monolayers. For oxygen coverages above 1.5 monolayers, thetwo higher coverages in our case, it has been reported that thesurface has chemical and physical properties similar to MoO2.Oxygen on surfaces with less than 1.5 monolayers ischemisorbed.36,37 Because the surface structure and hence thestoichiometry is not known we refer to the surface as MoOx.The presence of chemisorbed oxygen implies that the Mo(100)surface is passivated by a layer of oxygen atoms onto which theadventitious carbon becomes adsorbed.

TTIP Deposition onto a 573 K MoOx Surface. Theresults of TTIP deposition onto a 573 K base MoOx surface in300 L (1 Langmuir = 1 × 10−6 Torr s) intervals are shown inFigure 2. 573 K is above the reported TTIP thermaldecomposition temperature and would be expected to producechemical vapor deposition, whereby the growth of TiO2 isdependent only on the integrated precursor exposure. Panels aand b show the Ti(2p) and C(1s) spectra from the base surface(black) and after three 300 L doses of TTIP (red). TheTi(2p3/2) peak from the base Mo surface is insignificant, butafter TTIP exposure there is a large peak with a binding energyof 459.4 eV and a peak splitting of 5.7 eV which is indicative ofTiO2 formation; however, there is charging in the sample whichleads to slightly higher binding energies.25 There is a persistentC(1s) peak from adventitious carbon present on the basesurface, but this set of experiments also resulted in acontribution at lower binding energies from carbidic carbon.In Figure 2b, the C(1s) spectrum from the base Mo surface andthe total C(1s) spectrum taken after 900 L of TTIP are shownin black and red. The C(1s) spectrum from the base surfacewas subtracted from the total C(1s) spectrum after 900 L of

Figure 1. Representative (a) Mo(3d), (b) O(1s), and (c) C(1s)spectra from the base Mo substrate.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp505653u | J. Phys. Chem. C XXXX, XXX, XXX−XXXC

TTIP to isolate the C(1s) signal due to the depositions alone.The reported C(1s) spectra have been corrected in thismanner. The resultant C(1s) peak shape (Figure 2c) is bimodalwith similar peak ratios to that of the spectrum from anisopropoxide ligand (blue) obtained in a previous study fromcondensed vanadyl triisopropoxide.38

All elemental ratios and surface densities are summarized inTable S1 (Supporting Information), with an error associatedwith the integration and background subtraction process of∼8%. The Ti and C surface densities for the base substratewere 0.2 and 1.8 atoms/nm2, respectively. Figure 2d displaysthe change in Ti (black squares) and C (open blue circles)surface density above the base substrate value as a function ofprecursor exposure time. As demonstrated by the graph, the Tisurface density increases linearly with exposure time, while theC surface density is approximately constant. The ratio of peakareas, C(1s):Ti(2p), from the spectra after 900 L of TTIP is∼1:4.9. The Ti(2p):O(1s) ratio indicates that at 573 K thefilms are near stoichiometric. The reported density of ALDgrown TiO2 films at 573 K is 4.1 g/cm3, resulting in a Ti surfacedensity of 9.9 atoms/nm2.39 This is a conservative guide todetermine when a monolayer of TiO2 has been deposited. For asubstrate temperature of 573 K, after the second 300 L TTIPdose a monolayer of TiO2 has been grown with 11.3 Ti atoms/nm2.TTIP and Water Deposition onto a 473 K MoOx

Surface. TTIP was deposited onto the base MoOx surfaceheld at 473 K. The results shown in Figure 3 are from foursequential 300 L doses (1200 total L) of TTIP followed by twosequential water doses of 300 and 600 L. The temperature ofthe substrate is within the ALD window for TTIP deposition

and is commonly used when growing TiO2 inside an ALDreactor using TTIP as the metal precursor.13 Panels a and bshow the respective Ti(2p) and C(1s) spectra for the baseMoOx (black) and after four sequential 300 L of TTIP (red).While the peak area is significantly lower at this temperaturethan at 573 K, there is still a significant Ti(2p) signal fromdeposited TiO2 which has a slightly lower binding energy of459.0 eV, likely from reduced sample charging on a thinnerdeposited metal oxide film. The C(1s) spectrum from the basesurface (black line in Figure 3b) is similar to the base signal inFigure 2b but is lacking a contribution from carbidic carbon.The red line represents the increase in C(1s) signal after four300 L doses. The spectrum is bimodal and resembles the C(1s)spectrum of condensed isopropoxide ligands, similar to Figure2c. While the water doses reduced the intensity of the C(1s)signal, the shape of the peak remained the same, indicating theloss of exchangeable isopropoxide ligands.The elemental ratios and surface densities calculated after

each dose are shown in Table S2 of the SupportingInformation. The base substrate had a Ti and C surfacedensity of 0.4 and 2.0 atoms/nm2. Figure 3c shows the changesin Ti (black squares) and C (open blue circles) surfacedensities above the base substrate value after each dose. The Csurface density grows in a manner similar to the case of the 573K MoOx surface, in which it remains fairly constant withincreasing TTIP exposure. On the other hand, the Ti surfacedensity is almost linear with exposure time. These films areslightly deficient in oxygen, which is consistent with previousstudies of TiO2 ALD with TTIP onto another clean surface,Si(001), at 373 K.18 The Ti atom surface densities suggest that40% of a full monolayer of titanium oxide was grown by the

Figure 2. Results for increasing 300 L doses of TTIP onto a 573 K Mo substrate. (a) Ti(2p) spectra taken before (black) and after (red) a total of900 L of dosing time, (b) C(1s) spectra taken of the base surface (black) and the total C(1s) signal (red) after a total of 900 L of TTIP, (c) theC(1s) signal added from 900 L of TTIP dosing (red) and the C(1s) spectrum from the isopropoxide in condensed vanadyl triisopropoxide (VOTP),and (d) Ti (black squares) and C (open blue circles) surface densities as a function of TTIP exposure.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp505653u | J. Phys. Chem. C XXXX, XXX, XXX−XXXD

end of the final dose. The C surface density is greater than theTi surface density for the first two doses with a C(1s):Ti(2p)ratio of 1.3:1. This ratio implies that an average of 1.3 C atomsare deposited for every Ti atom. After 1200 L of TTIP exposurethe Ti surface density is larger than that of C with a ratio of1:2.3.The red squares and red circles in Figure 3c represent the Ti

and C surface densities, respectively, after 300 L and then 600 Lof H2O. There is a 47% decrease in the C density upon the first300 L of H2O, but there was no impact on the C density fromthe subsequent 600 L H2O dose. As expected, the Ti surfacedensity increases only slightly due to partial removal of thecarbon overlayer.

Reactions on MoOx at 295 and 373 K. TTIP was dosedinto the UHV chamber and onto a 373 K base Mo surface usingthe Gemstar-6 system in three 20 L intervals followed by two70 L intervals. After a total of 200 L of TTIP exposure thesurface was dosed with 360 L of water followed by another 20 Ldose of TTIP. The results are shown in Figure 4. The substrate

temperature is well below reported decomposition temper-atures for TTIP. Figure 4a shows the Ti(2p) spectra from abase Mo surface and after 200 L of TTIP in black and red,respectively. The binding energy, 458.9 eV, and peak splitting,5.9 eV, are similar to the results displayed in Figure 3 andconsistent with a Ti−O containing species on the surface.

Figure 3. Results for increasing 300 L doses of TTIP onto a 473 K Mosubstrate. (a) Ti(2p) spectra taken before (black) and after (red) atotal of 1200 L of TTIP, (b) C(1s) spectra taken before (black) andafter (red) a total of 1200 L of TTIP, and (c) Ti and C surfacedensities as a function of TTIP (black squares and blue circles) andwater (red squares and circles) exposure.

Figure 4. Results for increasing three 20 L and two 70 L doses ofTTIP onto a 473 K Mo substrate. (a) Ti(2p) spectra taken before(black) and after (red) a total of 200 L of TTIP, (b) C(1s) spectrataken before (black) and after (red) a total of 200 L of TTIP, and (c)Ti (black squares) and C (open blue circles) surface densities as afunction of precursor exposure. Surface density changes from H2Odoses are shown in red.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp505653u | J. Phys. Chem. C XXXX, XXX, XXX−XXXE

Figure 4b displays the C(1s) peak from the base surface whichhas a similar shape to previous adventitious carbon spectra. Thered line in Figure 4b represents the increase in C(1s) signal onthe surface after 200 L of TTIP dosing. While the peak shapeand binding energies are very similar to the spectra in Figures 2and 3, the signal-to-noise ratio is too low for further peak shapeanalysis.Elemental ratios and Ti atom surface densities are reported

in Table S3 in the Supporting Information. Panel c depicts thechanges in the Ti and C surface densities (black squares andopen blue circles) above the base substrate values as a functionof precursor exposure time. Unlike the previous experiment at473 K, the C surface density was consistently higher than thecorresponding Ti surface density. The C(1s):Ti(2p) ratio afterthe first 20 L TTIP dose was approximately 5.8:1 and droppedto 2.8:1 after a total of 200 L of TTIP exposure. Following 200L of TTIP exposure, the surface was dosed with 360 L of water,and the results are represented by the red markers in Figure 4c.The Ti surface density remained the same, while the C surfacedensity decreased by ∼33%. The remaining signal had the sameshape as shown in Figure 4b, consistent with the loss ofisopropoxide from the surface. After the water another 20 Ldose of TTIP was applied which increased both the Ti(2p) andthe C(1s) signals. Films grown at 373 K are significantly oxygendeficient. The Ti atom surface densities indicate that 17% of afull monolayer of titanium oxide was grown by the end of thefinal dose.Two experiments with the surface at 295 K were also

performed. In the first experiment, 40 L of 388 K TTIP wasdosed onto a base room temperature MoOx substrate. Notitanium or carbon was deposited. The second experimentinvolved dosing a room temperature TTIP-terminated surfacewith 720 L of 388 K water. No detectable changes in the C(1s)spectra were observed.

■ DISCUSSIONSeveral studies have demonstrated that ALD can be carried outunder UHV conditions (10−6 Torr) with pressures that are 6−7orders of magnitude lower than in traditional ALD reactors(0.1−10 Torr).12,18,40,41 However, it has been suggested thatthe pressure difference between UHV conditions and tradi-tional ALD can result in different chemistry.42,43 XPS, massspectrometry, secondary ion mass spectrometry (SIMS),temperature-programmed desorption (TPD), IR spectroscopy,and electron energy loss spectroscopy (EELS) are some of themethods that have been used to study and verify self-limitinggrowth on flat surfaces at lower pressures.5,8,12,40,44,45 The moststudied process is the growth of Al2O3 by deposition oftrimethylaluminum and water.4 IR and EELS experimentsconducted by Soto et al. and mass spectrometry experimentsperformed by Juppo et al. have successfully observed ALD-likebehavior from lower-pressure dosing.8,43 A few other ALDprocesses have been explored under UHV conditions, includingXPS studies by Roy et al. on the deposition of ZrO2 andcoupled XPS and SIMS experiments by Lemonds et al. toinvestigate the chemistry of TaF5 and Si2H6 half-cyclereactions.40,41 In both cases self-limiting deposition wasobserved with substrate temperatures up to 523 K. Finally,our own experiments investigating VOx deposition onto metaloxide surfaces have observed ALD-like growth of vanadia in anUHV system.38

Room Temperature TTIP Chemistry. Unlike traditionalALD reactors, our UHV setup allows us to independently

control the temperatures of the gaseous precursors and thesubstrate and therefore study substrate temperature effects inmore detail. In the current study we have explored the changesin the chemistry of TTIP with increasing substrate temper-atures while keeping the precursor gas temperature at 373 K.On the time scale of our experiments, there is no evidence ofreaction with either water or TTIP when the substrate is held atroom temperature (295 K). However, deposition was observedwhen the substrate temperature was increased to 373 K. Ourresults indicate that thermal activation by the substrate iscritical as increasing the gas-phase temperature alone to 373 Kis insufficient for achieving growth. This means that theprecursor first accommodates to the surface temperature priorto reaction and implies a Langmuir−Hinshelwood rather thanan Eley−Rideal mechanism for the surface chemistry.

Temperature Dependence of TTIP Deposition ontoMoOx. The growth mode and chemistry of TTIP depositiononto a base MoOx surface was studied with XPS as a functionof substrate temperature. The shapes of the Ti(2p) and C(1s)spectra, ratio of their peak areas, and Ti and C surface densitiesafter TTIP or water dosing are used to determine the type ofgrowth.Atomic layer deposition and chemical vapor deposition are

related processes for the growth of materials. As mentionedpreviously, ALD is a process that is composed of two sequentialself-limiting gas−surface reactions. Alternatively, CVD is thegrowth of materials from chemical reactions of gas-phaseprecursors. CVD can arise from a variety of mechanisms,including gas-phase bimolecular reactions, thermal decom-position of the precursor or surface reactions that regeneratereactive sites. Since CVD and ALD are similar, it can be hard todistinguish between the two. One critical difference is self-limiting growth in ALD.Chemical vapor deposition through thermal decomposition

of TTIP has been well studied.17,28,46−52 CVD products caninclude TiO2, H2O, propene, H2, isopropanol, acetone, anddiisopropyl ether, depending on the reaction condi-tions.17,47,48,53,54 A common thermal decomposition mecha-nism includes propene as a product, as represented in eq 2.

→ + +Ti(OCH(CH ) ) TiO 4C H 2H O3 2 4(g) 2(s) 3 6(g) 2 (g)

(2)

For ALD of TTIP on a metal oxide surface, the reaction in thefirst half-cycle is eq 1a. It is important to note that the reactionterminates with some number of carbon-containing ligandsremaining on the surface, covering potential reactive sites. Thisis responsible for the self-limiting behavior of ALD. Comparingeq 1a and 2, one critical difference between CVD and ALD isthe lack of carbon expected to remain on the surface after CVD.In our studies, we use this difference to determine the growthmode. Not only do we observe surface carbon species but alsothe shape of the C(1s) spectra is similar to the C(1s) spectrumof condensed vanadyl triisopropoxide observed in previousstudies performed in the same apparatus. This indicates that thecarbon signal is predominantly composed of isopropoxideligands remaining on the surface after the reaction. Althoughthe Ti(2p) or Mo(3d) spectra cannot be used to distinguishwhether the isopropoxide ligands are bound to Ti or Mo atoms,the decreases in C(1s) intensity after water dosing indicate thatexchangeable ligands are present.The more informative XPS results are the C(1s) to Ti(2p)

peak area ratio and corresponding surface densities whichshould indicate the ratio of ligands to titanium atoms remaining

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp505653u | J. Phys. Chem. C XXXX, XXX, XXX−XXXF

after each dose. For ALD-like reactions, we expect that thereshould be, on average, 1−2 isopropoxide ligands per Ti atomremaining on the surface if there is less than one monolayercoverage. This would be reflected in the XPS spectra as aC(1s):Ti(2p) ratio that is greater than or equal to 3:1. Instead,for all doses of TTIP onto a 573 K MoOx surface the Ti(2p)signal is greater than the C(1s) signal (Figure 2d), and the Tisurface density increases linearly with TTIP exposure, thereforeindicating that TiO2 is being formed by CVD. This observationis consistent with previous thermal decomposition stud-ies.17,19,46,47,49,51

Chemical Vapor Deposition of TTIP at 473 K. At 473 Kall doses of TTIP also resulted in low C(1s):Ti(2p) atomicratios which decreased from 1.3:1 to 1:2.3 as the TTIPexposure increased. This indicates that CVD also occurs at 473K even though this is within the observed ALD temperaturewindow.13,55 Moreover, while thermal decomposition is acommon mechanism of CVD, 473 K is below the reportedTTIP thermal decomposition temperatures. We have consid-ered the possibility of gas-phase reactions from residual water inthe chamber regenerating reactive sites from ligand exchangecausing uncontrolled growth, especially during the longevacuation and data collection periods. However, this can beruled out by our experiment showing the absence of reactionbetween dosed water and the 295 K TTIP-covered surface.One possible CVD mechanism which explains our

observations is that, under UHV conditions at 473 K, we areobserving TiO2-catalyzed dehydration of TTIP molecules. Asfar as we know, this reaction mechanism has not been proposedbefore; however, it is well known that TiO2 is a dehydrationcatalyst that is active for alcohol dehydration above 500 K.56−60

Recent TPD experiments of isopropanol on TiO2(110)reported a low-temperature dehydration pathway around 300K.58,59 It has also been established that the intermediate for thereaction is the alkoxide which then undergoes β-hydrideelimination to form the alkene.56−58,60 While the adsorbedspecies from TTIP deposition onto TiO2 is not exactlyequivalent to the reaction intermediate for isopropanoldehydration on TiO2, the same mechanism and products,H2O and C3H6, have been reported for TTIP decomposi-tion.17,46,47,53,54 It is reasonable that the alkoxide ligandchemistry of a deposited TTIP is similar to the chemistry ofthe intermediate alkoxide formed during alcohol dehydration.Thus, during low-pressure deposition, initially formed TiO2species catalyze dehydration of subsequent TTIP molecules, toform more TiO2 and regenerate reactive sites rather thanterminating the deposition reaction by saturating the surfacewith isopropoxide ligands. This alternative dehydration path-way is always possible during TTIP deposition onto/or nearTiO2 but is likely suppressed under the higher-pressure ALDconditions because the increased flux of TTIP molecules hittingthe surface leads to a more rapid surface saturation covering theTiO2 sites active for dehydration. We propose that under theexperimental conditions of low flux and moderate temperaturesthe ALD reaction kinetics are slow enough that dehydrationbecomes the dominant pathway.Lowering the MoOx substrate temperature to 373 K causes a

significant change in the growth mode. The integrated C(1s)signal is larger than for the Ti(2p) signal for all TTIP and waterdoses. After the first TTIP dose the calculated C(1s):Ti(2p)ratio is 5.3:1 which means that on average 1−2 ligands areremaining on the surface per reacted TTIP molecule, consistentwith ALD chemistry for the first dose. After the second TTIP

dose, the total C(1s):Ti(2p) ratio immediately decreases to2.8:1, indicating that less than one ligand remains on thesurface per reacted TTIP molecule and there was an overall lossof ligands from the first dose to the second. The ratio remainsrelatively constant through 200 L of TTIP exposure. Theseratios are more suggestive of CVD-like growth of TiO2compared to the first dose. These results suggest that ALDchemistry takes place during the first dose because TiO2 is notyet present on the surface, but following doses approach CVDbehavior as the TiO2 concentration builds up and thedehydration reaction pathway becomes possible.We have considered the possible role of the MoOx surface in

this alternative reaction pathway. MoO3 and to a lesser degreeMoO2 are also alcohol dehydration catalysts; however, recentstudies have shown that the surface alkoxides preferentiallydesorb as the alcohol rather than dehydrate to the alkene.22,23

The presence of MoO3 was undetectable by XPS, and itspresence is not expected on Mo surfaces treated as described.Therefore, substrate-catalyzed dehydration is not expected tocontribute to the overall dehydration reaction. To furtheraddress the possibility of molybdenum oxide catalyzeddehydration TTIP was deposited onto a 473 K MoOx substratewith an oxygen coverage of 0.5 monolayer, a significantly loweroxygen concentration than in the previous experiment.Vibrational spectroscopy and surface polarizability studieshave shown that oxygen overlayers with this concentration donot have the structure or electronic properties of molybdenumoxide but rather behave as chemisorbed oxygen.34−36 Thissurface will have even less activity than MoO2 for dehydrationof the alkoxide ligand. The elemental ratios, Ti atom surfacedensities, and a description of the experiment are detailed inTable S4 of the Supporting Information. Due to the lowreactivity of TTIP only 45% of a complete TiO2 monolayer isdeposited after a total of 2040 L of TTIP exposure. However,we observe low C(1s):Ti(2p) ratios, indicating CVD-likegrowth which is consistent with the results from the moreoxidized substrate. This suggests that the base MoOx substratedoes not play a significant role in the alternative reactionpathway.

Implications for Atomic Layer Deposition Processes.This alternative reaction pathway to ALD chemistry for TTIPdeposition is not restricted to MoOx substrates because theactive dehydration site is predominantly TiO2. However, it isimprobable that the dehydration reaction pathway plays a majorrole during ALD processes in traditional reactors because theprecursor flux on the surface is 106−107 higher. Because of thehigher flux, all active dehydration sites are covered by depositedTTIP molecules, and dehydration is quenched. This fluxdependence has possible implications for substrate architectureswith mass transport limitations, such as long nanoscale pores.Our results suggest that CVD, instead of ALD, occurs at thebottom of these pores because the flux of TTIP molecules maybe insufficient to inhibit the dehydration pathway. The CVDreactions would then lead to a filling of the pores rather than athin conformal coating.

■ CONCLUSIONThe chemistry of TTIP deposition on to a MoOx surface wasexplored at a series of substrate temperatures in UHVconditions. Compared to successful depositions onto a 373 KMoOx surface, the lack of reactivity of 373 K TTIP with a 295K surface indicates that the surface temperature is a criticalparameter. TTIP molecules must accommodate to the surface

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp505653u | J. Phys. Chem. C XXXX, XXX, XXX−XXXG

prior to reaction and follow a Langmuir−Hinshelwoodmechanism.CVD of TiO2 from TTIP was observed for substrate

temperatures of 573 and 473 K, which are, respectively, withinand below the reported thermal decomposition range for TTIP.It is believed that, under the conditions of low pressure (10−6

Torr) and moderate temperatures (473 K), dehydration ofTTIP molecules by nascent TiO2 not only becomescompetitive with the ALD reaction but actually becomes thedominant reaction. At lower temperature, 373 K, ALD-likechemistry is observed during the first TTIP dose, but XPSresults suggest that for subsequent doses the growth modetransitions into CVD. In all cases, the influence of the MoOxsubstrate appears to be negligible. While these results may notdirectly impact traditional ALD processes, this type ofalternative chemistry may be relevant to high surface areasubstrates with long nanopores, where mass transport to thebottom of the pore is limited and CVD occurs in preference totypical ALD reactions.

■ ASSOCIATED CONTENT*S Supporting InformationDetails on elemental ratios and surface densities and effects ofthe oxygen coverage on the base substrate on TTIP deposition.This material is available free of charge via the Internet athttp://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: pstair@northwestern.edu.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis material is based upon work supported by the NationalScience Foundation under Grant No. CHE-1058835. Theauthors would also like to thank Kunlun Ding, Miles Tan,Charlie Campbell, and Bruce Rosen for helpful discussions.

■ REFERENCES(1) George, S. M. Atomic Layer Deposition: An Overview. Chem.Rev. 2010, 110, 111−131.(2) Lu, J. L.; Elam, J. W.; Stair, P. C. Synthesis and Stabilization ofSupported Metal Catalysts by Atomic Layer Deposition. Acc. Chem.Res. 2013, 46, 1806−1815.(3) Stair, P. C. Synthesis of Supported Catalysts by Atomic LayerDeposition. Top. Catal. 2012, 55, 93−98.(4) Puurunen, R. L. Surface Chemistry of Atomic Layer Deposition:A case study for the Trimethylaluminum/Water Process. J. Appl. Phys.2005, 97, 52.(5) Knapas, K.; Ritala, M. In Situ Studies on Reaction Mechanisms inAtomic Layer Deposition. Crit. Rev. Solid State Mater. Sci. 2013, 38,167−202.(6) Lei, Y.; Lu, J.; Zhao, H.; Liu, B.; Low, K.-B.; Wu, T.; Libera, J. A.;Greeley, J. P.; Chupas, P. J.; Miller, J. T.; et al. Resolving PrecursorDeligation, Surface Species Evolution, and Nanoparticle Nucleationduring Palladium Atomic Layer Deposition. J. Phys. Chem. C 2013,117, 11141−11148.(7) Rahtu, A.; Alaranta, T.; Ritala, M. In Situ Quartz CrystalMicrobalance and Quadrupole Mass Spectrometry Studies of AtomicLayer Deposition of Aluminum Oxide from Trimethylaluminum andWater. Langmuir 2001, 17, 6506−6509.(8) Soto, C.; Tysoe, W. T. The Reaction Pathway for the Growth ofAlumina on High Surface Area Alumina and in Ultrahigh Vacuum by a

Reaction between Trimethyl Aluminum and Water. J. Vac. Sci.Technol., A 1991, 9, 2686−2695.(9) Aarik, J.; Aidla, A.; Sammelselg, V.; Uustare, T.; Ritala, M.;Leskela, M. Characterization of Titanium Dioxide Atomic LayerGrowth from Titanium Ethoxide and Water. Thin Solid Films 2000,370, 163−172.(10) Kukli, K.; Forsgren, K.; Aarik, J.; Uustare, T.; Aidla, A.;Niskanen, A.; Ritala, M.; Leskela, M.; Harsta, A. Atomic LayerDeposition of Zirconium Oxide from Zirconium Tetraiodide, Waterand Hydrogen Peroxide. J. Cryst. Growth 2001, 231, 262−272.(11) Kukli, K.; Ritala, M.; Sundqvist, J.; Aarik, J.; Lu, J.; Sajavaara, T.;Leskela, M.; Harsta, A. Properties of Hafnium Oxide Films Grown byAtomic Layer Deposition from Hafnium Tetraiodide and Oxygen. J.Appl. Phys. 2002, 92, 5698−5703.(12) Rahtu, A.; Kukli, K.; Ritala, M. In Situ Mass Spectrometry Studyon Atomic Layer Deposition from Metal (Ti, Ta, and Nb) Ethoxidesand Water. Chem. Mater. 2001, 13, 817−823.(13) Ritala, M.; Leskela, M.; Niinisto, L.; Haussalo, P. TitaniumIsopropoxide as a Precursor in Atomic Layer Epitaxy of TitaniumDioxide Thin Films. Chem. Mater. 1993, 5, 1174−1181.(14) Rai, V. R.; Vandalon, V.; Agarwal, S. Influence of SurfaceTemperature on the Mechanism of Atomic Layer Deposition ofAluminum Oxide Using an Oxygen Plasma and Ozone. Langmuir2012, 28, 350−357.(15) Rai, V. R.; Agarwal, S. Surface Reaction Mechanisms duringOzone-Based Atomic Layer Deposition of Titanium Dioxide. J. Phys.Chem. C 2008, 112, 9552−9554.(16) Lu, J. L.; Kosuda, K. M.; Van Duyne, R. P.; Stair, P. C. SurfaceAcidity and Properties of TiO2/SiO2 Catalysts Prepared by AtomicLayer Deposition: UV-visible Diffuse Reflectance, DRIFTS, andVisible Raman Spectroscopy Studies. J. Phys. Chem. C 2009, 113,12412−12418.(17) Rahtu, A.; Ritala, M. Reaction mechanism studies on titaniumisopropoxide-water atomic layer deposition process. Chem. Vap.Deposition 2002, 8, 21−28.(18) Lee, S. Y.; Jeon, C.; Kim, S. H.; Kim, Y.; Jung, W.; An, K.-S.;Park, C.-Y. In-Situ X-ray Photoemission Spectroscopy Study of AtomicLayer Deposition of TiO2 on Silicon Substrate. Jpn. J. Appl. Phys.2012, 51, 031102/1−031102/3.(19) Cho, S.-I.; Chung, C.-H.; Moon, S. H. Temperature-Programmed Desorption Study on the Decomposition Mechanismof Ti(OC3H7)4 on Si(100). J. Electrochem. Soc. 2001, 148, C599−C603.(20) O’Neill, B. J.; Jackson, D. H. K.; Crisci, A. J.; Farberow, C. A.;Shi, F. Y.; Alba-Rubio, A. C.; Lu, J. L.; Dietrich, P. J.; Gu, X. K.;Marshall, C. L.; et al. Stabilization of Copper Catalysts for Liquid-Phase Reactions by Atomic Layer Deposition. Angew. Chem., Int. Ed.2013, 52, 13808−13812.(21) Lu, J.; Liu, B.; Greeley, J. P.; Feng, Z.; Libera, J. A.; Lei, Y.;Bedzyk, M. J.; Stair, P. C.; Elam, J. W. Porous Alumina ProtectiveCoatings on Palladium Nanoparticles by Self-Poisoned Atomic LayerDeposition. Chem. Mater. 2012, 24, 2047−2055.(22) Seman, M.; Kondo, J. N.; Domen, K.; Reed, C.; Oyama, S. T.Direct Observation of Surface Intermediates Formed by SelectiveOxidation of Alcohols on Silica-Supported Molybdenum Oxide. J.Phys. Chem. B 2004, 108, 3231−3239.(23) Li, Z.; Fang, Z.; Kelley, M. S.; Kay, B. D.; Rousseau, R.;Dohnalek, Z.; Dixon, D. A. Ethanol Conversion on Cyclic (MO3)3(M = Mo, W) Clusters. J. Phys. Chem. C 2014, 118, 4869−4877.(24) Parker, B. R.; Jenkins, J. F.; Stair, P. C. The Adsorption ofMethyl Radicals on Oxygen-Modified Mo(100) Studied by HREELS.Surf. Sci. 1997, 372, 185−192.(25) Briggs, D.; Seah, M. P.; Eds., Practical Surface Analysis by Augerand X-ray Photoelectron Spectroscopy; John Wiley and Sons: New York,1983; p 533.(26) The dosing lines and the Gemstar-6 system were both used todeposit TTIP onto a MoOx substrate at 473 K. It was challenging toquantitatively compare the total flux from each system as the dosingmethods are significantly different. In the Gemstar-6 the TTIP had to

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp505653u | J. Phys. Chem. C XXXX, XXX, XXX−XXXH

be heated to 353 K; there is a carrier gas (N2); and the aperturerestricts the pressure of TTIP into the chamber to 3 × 10−7 Torr.Most importantly, both methods resulted in TiO2 deposition wherethe ratio of the Ti(2p) peak area was greater than the C(1s) peak areaindicating CVD. In addition, both methods did not result in TiO2

deposition with the substrate at 295 K, verifying that the chemistry istaking place on the MoOx surface and is independent of which dosingtechnique was used.(27) Wagner, W.; Pruss, A. The IAPWS Formulation 1995 for theThermodynamic Properties of Ordinary Water Substance for Generaland Scientific Use. J. Phys. Chem. Ref. Data 2002, 31, 387−535.(28) Siefering, K. L.; Griffin, G. L. Growth Kinetics of CVD TiO2-Influence of Carrier Gas. J. Electrochem. Soc. 1990, 137, 1206−1208.(29) Henry, R. M.; Walker, B. W.; Stair, P. C. Adsorption Selectivityof Lewis-Acids and Bases on an Oxidized Mo(100) Surface Studied byLEED, Auger, and XPS. Surf. Sci. 1985, 155, 732−750.(30) Bianchi, C. L.; Cattania, M. G.; Villa, P. XPS Characterization ofNi and Mo Oxides Before and After “In Situ” Treatments. Appl. Surf.Sci. 1993, 70−71 (Part 1), 211−216.(31) Grant, J. L.; Fryberger, T. B.; Stair, P. C. ESCA Measurement ofSurface Atom Oxidation-states on Chemically Modified Mo(100)Surfaces. Appl. Surf. Sci. 1986, 26, 472−487.(32) Overbury, S. H.; Stair, P. C. The Structure of Atomic Oxygenand Carbon Overlayers on the Mo(001) Surface Studied by Low-Energy Ion-Scattering. J. Vac. Sci. Technol., A 1983, 1, 1055−1058.(33) Walker, B. W.; Stair, P. C. Chemical Evidence for the Formationof an Acidic Mo(100) Surface by Adsorbed Oxygen. Surf. Sci. 1980, 91,L40−L44.(34) Kim, S. H.; Stair, P. C. The Structure of Oxygen Adsorbed onMo(100) Studied by High-Resolution Electron Eenergy-Loss Spec-troscopy. Surf. Sci. 2000, 457, L347−L353.(35) Fryberger, T. B. Studies of Chemically Modified Mo(100) Surfacesusing Photoelectron Spectroscopies and Physisorbed Xenon as a Probe;Northwestern: Illinois, 1986.(36) Smudde, G. H.; Stair, P. C. The Oxidation of Mo(100) Studiedby XPS and Surface Raman-Spectroscopy - The Onset of MoO2

Formation and The Formation of Surface Polymolybdate. Surf. Sci.1994, 317, 65−72.(37) Zhang, C.; Vanhove, M. A.; Somorjai, G. A. The Interaction ofOxygen with the Mo(100) and Mo(111) Single-Crystal Surfaces -Chemisorption and Oxidation at High-Temperatures. Surf. Sci. 1985,149, 326−340.(38) Johnson, A. M.; Quezada, B. R.; Marks, L. D.; Stair, P. C.Influence of the Metal Oxide Substrate Structure on Vanadium OxideMonomer Formation. Top. Catal. 2014, 57, 177−187.(39) Rausch, N.; Burte, E. P. Thin TiO2 Films Prepared by Low-Pressure Chemical Vapor-Deposition. J. Electrochem. Soc. 1993, 140,145−149.(40) Lemonds, A. M.; White, J. M.; Ekerdt, J. G. Surface ScienceInvestigations of Atomic Layer Deposition Half-Reactions using TaF5and Si2H6. Surf. Sci. 2003, 538, 191−203.(41) Roy, P. C.; Jeong, H. S.; Doh, W. H.; Kim, C. M. Atomic LayerDeposition (ALD) of ZrO2 in Ultrahigh Vacuum (UHV). Bull. KoreanChem. Soc. 2013, 34, 1221−1224.(42) Ritala, M.; Leskela, M. In Atomic Layer Deposition; AcademicPress: New York, 2002; pp 103−159.(43) Juppo, M.; Rahtu, A.; Ritala, M.; Leskelae, M. In Situ MassSpectrometry Study on Surface Reactions in Atomic Layer Depositionof Al2O3 Thin Films from Trimethylaluminum and Water. Langmuir2000, 16, 4034−4039.(44) Matero, R.; Rahtu, A.; Ritala, M. In Situ Quadrupole MassSpectrometry and Quartz Crystal Microbalance Studies on the AtomicLayer Deposition of Titanium Dioxide from Titanium Tetrachlorideand Water. Chem. Mater. 2001, 13, 4506−4511.(45) Zaera, F. Mechanisms of Surface Reactions in Thin Solid FilmChemical Deposition Processes. Coord. Chem. Rev. 2013, 257, 3177−3191.

(46) Siefering, K. L.; Griffin, G. L. Kinetics of Low-Pressure ChemicalVapor Deposition of TiO2 from Titanium Tetraisopropoxide. J.Electrochem. Soc. 1990, 137, 814−818.(47) Wu, Y. M.; Bradley, D. C.; Nix, R. M. Studies of TitaniumDioxide Film Growth from Titanium Tetraisopropoxide. Appl. Surf.Sci. 1993, 64, 21−28.(48) Komiyama, H.; Kanai, T.; Inoue, H. Preparation of Porous,Amorphous, and Ultrafine Titanium Dioxide Particles by ChemicalVapor Deposition. Chem. Lett. 1984, 1283−6.(49) Karlsson, P. G.; Richter, J. H.; Andersson, M. P.; Johansson, M.K. J.; Blomquist, J.; Uvdal, P.; Sandell, A. TiO2 Chemical VaporDeposition on Si(111) in Ultrahigh Vacuum: Transition fromInterfacial Phase to Crystalline Phase in the Reaction Limited Regime.Surf. Sci. 2011, 605, 1147−1156.(50) Taylor, C. J.; Gilmer, D. C.; Colombo, D. G.; Wilk, G. D.;Campbell, S. A.; Roberts, J.; Gladfelter, W. L. Does Chemistry ReallyMatter in the Chemical Vapor Deposition of Titanium Dioxide?Precursor and Kinetic Effects on the Microstructure of PolycrystallineFilms. J. Am. Chem. Soc. 1999, 121, 5220−5229.(51) Chen, Z. X.; Derking, A. TiO2 Thin Films by Chemical VaporDeposition: Control of the Deposition Process and Film Character-ization. J. Mater. Chem. 1993, 3, 1137−40.(52) Battiston, G. A.; Gerbasi, R.; Porchia, M.; Marigo, A. Influenceof Substrate on Structural Properties of TiO2 Thin Films Obtained viaMOCVD. Thin Solid Films 1994, 239, 186−91.(53) Fictorie, C. P.; Evans, J. F.; Gladfelter, W. L. Kinetic andMechanistic Study of the Chemical Vapor Deposition of TitaniumDioxide Thin Films using Tetrakis(isopropoxo)titanium(IV). J. Vac.Sci. Technol., A 1994, 12, 1108−13.(54) Chen, S.; Mason, M. G.; Gysling, H. J.; Paz-Pujalt, G. R.;Blanton, T. N.; Castro, T.; Chen, K. M.; Fictorie, C. P.; Gladfelter, W.L.; Franciosi, A.; et al. Ultrahigh Vacuum Metalorganic ChemicalVapor Deposition Growth and In Situ Characterization of EpitaxialTitania Films. J. Vac. Sci. Technol., A 1993, 11, 2419−29.(55) Ritala, M.; Leskela, M.; Nykanen, E.; Soininen, P.; Niinisto, L.Growth of Titanium Dioxide Thin Films by Atomic Layer Epitaxy.Thin Solid Films 1993, 225, 288−295.(56) Kim, K. S.; Barteau, M. A. Reactions of Aliphatic Alcohols onthe {011}-Facetted Titanium Dioxide (001) Surface. J. Mol. Catal.1990, 63, 103−17.(57) Gamble, L.; Jung, L. S.; Campbell, C. T. Decomposition andProtonation of Surface Ethoxys on TiO2(110). Surf. Sci. 1996, 348, 1−16.(58) Bondarchuk, O.; Kim, Y. K.; White, J. M.; Kim, J.; Kay, B. D.;Dohnalek, Z. Surface Chemistry of 2-Propanol on TiO2(110): Low-and High-Temperature Dehydration, Isotope Effects, and Influence ofLocal Surface Structure. J. Phys. Chem. C 2007, 111, 11059−11067.(59) Kim, Y. K.; Kay, B. D.; White, J. M.; Dohnalek, Z. AlcoholChemistry on Rutile TiO2(110): The Influence of Alkyl Substituentson Reactivity and Selectivity. J. Phys. Chem. C 2007, 111, 18236−18242.(60) Farfan-Arribas, E.; Madix, R. J. Role of Defects in theAdsorption of Aliphatic Alcohols on the TiO2(110) Surface. J. Phys.Chem. B 2002, 106, 10680−10692.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp505653u | J. Phys. Chem. C XXXX, XXX, XXX−XXXI