catalytic combustion reactions during atomic layer deposition of ru studied using 18 ...

Catalytic Combustion Reactions During Atomic Layer Deposition ofRu Studied Using 18O2 Isotope LabelingN. Leick,† S. Agarwal,‡ A. J. M. Mackus,† S. E. Potts,† and W. M. M. Kessels*,†

†Department of Applied Physics, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands‡Department of Chemical and Biological Engineering, Colorado School of Mines, Golden, Colorado 80401, United States

*S Supporting Information

ABSTRACT: The surface reaction products liberated during the atomic layer deposition(ALD) of Ru from (C5H5)Ru(CO)2(C2H5) and

18O2 were quantitatively analyzed usingquadrupole mass spectrometry (QMS). The gas-phase reaction products during the Ruprecursor pulse were CO2, CO, H2, and H2O, while during the O2 pulse primarily CO2and CO were produced. Approximately 70% of the C atoms and ∼100% of the H atomscontained in the Ru precursor were released during the Ru pulse of the ALD process.From these observations, and on the basis of the surface science and catalysis literature, weconclude that the complex surface chemistry during Ru ALD can be described by catalyticcombustion reactions. These reactions consist of the dissociative chemisorption of the Ruprecursor’s ligands on the Ru surface during the metal pulse. These hydrocarbon ligandsundergo dehydrogenation and combustion reactions on the catalytic metal surface in thepresence of surface O formed due to the dissociation of O2 molecules in the previous O2pulse. We postulate that the carbon-rich species produced due to the dehydrogenation ofthe Ru precursor’s hydrocarbon ligands self-limit precursor chemisorption by blocking adsorption sites. The use of the 18O2isotope also unambiguously shows that a certain fraction of the precursor’s carbonyl ligands dissociate on the Ru surface.

■ INTRODUCTION

Thin films and nanostructures of Pt-group metals, such as Pt,Ru, Ir, and Pd, have been widely studied over the past decadefor a variety of applications, such as power generation,1

electricity and information storage,2−4 sensing,5−7 and controlof exhaust pollution.8−10 Atomic layer deposition (ALD) hasemerged as a promising technique to deposit these metallic thinfilms and nanostructures with accurate thickness control ontodemanding substrate topologies.11−13 In particular, Ru ALD isof interest in applications such as heterogeneous catalysis14,15

and memory devices.4,16−20 Similar to most other metal ALDprocesses, nucleation delays of 70−250 cycles have beenreported during ALD of Ru on a variety of substratesurfaces.21−23 Novel Ru ALD precursors24−29 have beendeveloped to reduce this long nucleation delay and to enhancethe growth per cycle (GPC), which is typically <1Å/cycle.19,21,30−33 Regardless of the type of precursoremployed, a better understanding of the relevant surfacereaction mechanisms during metal ALD is crucial for improvingthe deposition processes and tailoring the resulting propertiesof Ru and other Pt-group metal films.In one of the first studies of Ru ALD with ruthenocene

(RuCp2) and O2 gas, Aaltonen et al.,34 using quadrupole massspectrometry (QMS), identified CO2 and H2O as the primarysurface reaction products in both half-reaction cycles. Inaddition, similar observations were made for Pt ALD using(methylcyclopentadienyl)trimethylplatinum (MeCpPtMe3) andO2 gas.34−36 On the basis of these and several subsequentstudies,37,38 it has now been accepted that combustion-like

reactions govern the surface chemistry in both half-reactioncycles during ALD of Pt-group metals. Due to their catalyticnature, these Pt-group metal surfaces are able to break thebonds of adsorbed species,39−44 such as hydrocarbons and O2:the surface O, produced due to the dissociative chemisorptionof O2 gas, reacts with the precursor’s organic ligands to producecombustion-like products, such as CO2 and H2O.

34 Suchoxidation reactions of hydrocarbons are very important in theoverall surface reaction mechanism since the electrons that arereleased in these reactions are necessary to reduce the Ru or Ptatoms in the +2 and +4 oxidation state, respectively, to theirmetallic state (oxidation state 0).45

More recently, it has been shown that CO2 and H2O are notthe only surface reaction products during ALD of Pt-groupmetals.35,37,46−48 In our previous work on Ru ALD withCpRu(CO)2Et and O2 gas,

48 using QMS, H2 was identified as aprimary reaction product during the Ru precursor cycle, besidesCO2 and H2O. On the basis of the surface science and catalysisliterature,43,44,49 we concluded that H2 was released in the Ruprecursor cycle due to surface dehydrogenation reactions of theprecursor’s hydrocarbon ligands.48 In addition, during the ALDof Pt with MeCpPtMe3 and O2 gas, using gas-phase infraredspectroscopy and QMS, CH4 was identified as a surfacereaction product during the metal precursor pulse.35,36,47

Similarly, based on the catalysis literature, it was concluded

Received: June 19, 2013Revised: August 16, 2013Published: August 20, 2013

Article

pubs.acs.org/JPCC

© 2013 American Chemical Society 21320 dx.doi.org/10.1021/jp4060457 | J. Phys. Chem. C 2013, 117, 21320−21330

pubs.acs.org/JPCC

that surface dehydrogenation of the Me and MeCp ligands inthe Pt precursor was the most likely source of surface H, whichreacts with other Me ligands to form CH4.

46,47 Thus, theseobservations suggest that additional reactions of hydrocarbonligands, other than complete combustion to CO2 and H2O,must occur on Pt-group metal surfaces due to their catalyticnature. Since most Pt-group ALD processes, including Ru,utilize organometallic precursors and O2 gas,22,25,26,50 it isimportant to gain a deeper insight into the interaction of theseprecursors with catalytic metal surfaces.The surface interactions of adsorbates on Pt-group metals

have been extensively studied in the context of heterogeneouscatalysis and fuel cell applications.41,51−53 For example, theinteraction of CO and O2 with the Ru(0001) surface has beenstudied over a wide range of conditions from ultrahigh vacuum(∼10−10−10−9 Torr) to high pressure (∼103 Torr) and up totemperatures >500 K.54−56 The results of these studies aresummarized below, which are relevant to the pressure (∼1Torr) and temperature regime (∼600 K) used for Ru ALDprocesses.54−61 In particular, these studies suggest that thesurface oxygen coverage after the O2 pulse can significantlyaffect the surface reactions of organometallic Ru precursors.Figure 1 shows a summary of the literature on the Ru(0001)

surface O coverage, θO, as a function of O2 exposure.60 It has

been reported that O chemisorption extends over four differentordered overlayer phases on Ru(0001): p(2 × 2)-O at θO =0.25 monolayer (ML), p(2 × 1)-O at θO = 0.5 ML, p(2 × 2)-3O at θO = 0.75 ML, and p(1 × 1)-O at θO = 1 ML, where prefers to the primitive adlayer phase.60−63 According to thesestudies, Ru also has the ability to adsorb atomic O beyond 1ML, where atomic O dissolves into the subsurfaceregion.55,59,60,64,65 At coverages beyond θO = 3 ML,RuO2(110) microregions begin to form, which can coexistwith p(1 × 1)-O domains on Ru(0001).57,58,66 In fact, RuO2has been identified as the active phase in the oxidation of CO athigh pressures and temperatures.56,67,68 It is also known fromcatalysis that dehydrogenation reactions occur on catalyticmetal surfaces in which C−H bonds are broken, and the atomicH from the hydrocarbon is transferred onto the metal surface.Eventually, surface H atoms recombine and desorb as H2,leaving carbon-rich surface dehydrogenation products on themetal surface.40,41,69 This overall reaction can be written as

CxHy* → CxHy−2* + H2 (gas), where * denotes surfacespecies.70 Thus, based on previous literature, we expect that,depending on the surface O coverage prior to the Ru precursorhalf-cycle, different surface reactions could occur: combustion-like products would be dominant for a high surface O coverage,and dehydrogenation reaction products would be more likelyfor a low surface O coverage.In this article, we present our study of the surface reaction

mechanisms during ALD of Ru from the heteroleptic precursorcyclopentadienylethyldi(carbonyl)ruthenium (CpRu(CO)2Et,see Figure 2a) and 18O2 using QMS. Isotope labeling enabled

us to differentiate between the 16O in the Ru precursor’scarbonyl ligands from the 18O delivered in the O2 half-reactioncycle. In addition, we were also able to distinguish betweenC18O and C2H4 in the QMS since both species can be releasedduring the surface reactions of the Ru precursor. To quantifythe gas-phase concentration of the surface reaction products,we have carefully calibrated the QMS ion current for differentmass-to-charge (m/z) ratios. On the basis of this analysis, weconclude that the Ru surface catalyzes the dissociation ofadsorbed specieshydrocarbons in the Ru-precursor half-cycleand 18O2 in the subsequent half-cyclethereby facilitating thecombustion of the fragmented precursor ligands. Such

Figure 1. O coverage as a function of O2 exposure in Langmuirs (1 L= 1 × 10−6 Torr·s) for a Ru(0001) surface at 600 K and the expectedsurface configurations of metastable O at coverages of 0.25, 0.5, 0.75,and 1 ML. Note, p refers to the primitive adlayer unit cell. The datahave been compiled from refs 59, 62, and 63.

Figure 2. (a) Molecular structure of CpRu(CO)2Et [C5H5Ru-(CO)2CH2CH3]. (b) Schematic overview of the ALD reactor. Thetop and bottom gate valves are used to isolate the chamber from theplasma source and pump, respectively, thereby trapping the surfacereaction products within a known volume. A differentially pumpedQMS is used to detect the gas-phase species. (c) Schematic of thepulsing sequence for the Ru ALD process in this study (2 sCpRu(CO)2Et and 18O2 pulses). The modified ALD processcomprised of additional steps in which the top and bottom valveswere closed for a 7 s period with 2 s CpRu(CO)2Et and

18O2 injectionperiods. After each half-cycle, the gas-phase products were pumped outof the reactor for 5 min, a duration that was sufficient to reach theQMS background signal intensity before each pulse.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp4060457 | J. Phys. Chem. C 2013, 117, 21320−2133021321

catalytically assisted oxidation reactions of hydrocarbons aregenerally referred to as catalytic combustion reactions in thefield of heterogeneous catalysis:71−73 herein, we show that theyalso play a significant role under the experimental conditionsfor Ru ALD.

■ EXPERIMENTAL DETAILSALD Reactor and Film Deposition. The Ru films were

deposited at 325 °C onto polished 150 mm diameter Si(001)wafers in an Oxford Instruments FlexAL ALD reactor74 (seereactor schematic in Figure 2b) suited for thermal and plasma-assisted ALD. The deposition chamber was evacuated by aturbomolecular pump, which provided a base pressure of ∼10−6Torr. The CpRu(CO)2Et (SAFC Hitech, purity >99.999%)precursor was kept at 90 °C and was vapor-drawn into thechamber through a stainless steel tube heated to 110 °C andpulsed by an ALD valve. In this chamber, 18O2 (Linde Gas,purity >99.8%) was delivered via a mass flow controller into thealumina tube of the inductively coupled plasma source that wasnot operated in the experiments described here. The pressurein the chamber increased to ∼30 and ∼120 mTorr during Ruand 18O2 pulsing, respectively. The chamber walls were heatedto 120 °C to avoid precursor condensation. The reactorsurfaces and the Si wafer were conditioned prior to the QMSstudy by depositing at least 30 nm of Ru. Since almost no filmdeposition was detected at substrate temperatures <200 °C, weexpected negligible Ru deposition on the reactor walls.However, the 225 mm diameter substrate heater can radiativelyheat the lower walls to a temperature at which Ru growth couldoccur, and therefore, it was difficult to accurately estimate thetotal surface area on which Ru was actually deposited. Inaddition, a significant part of the substrate stage was notcovered by the relatively smooth Si wafer and directly exposedto the ALD precursors, which could significantly increase thearea for deposition due to surface roughness.The Ru film properties deposited by this ALD process have

been reported elsewhere,75 and only the properties relevant forthis study are briefly summarized below. The Ru films had amass density of 12.2 ± 0.5 g·cm−3, which is close to the Ru bulkdensity (12.4 g·cm−3). From Rutherford back-scattering (RBS)and X-ray reflectivity, it was determined that 7.1 × 1014 cm−2 ofRu atoms were deposited in one ALD cycle.75 Furthermore, theaverage areal density of Ru atoms on the surface based on thisRBS data is 1.7 × 1015 cm−2, which is very similar to the valuereported on a Ru(0001) surface.60 X-ray diffraction (XRD)showed that the films were polycrystalline with no preferredorientation. RBS and time-of-flight secondary ion massspectrometry detected traces of oxygen in the films; however,no RuOx peak was observed in XRD. The electrical resistivity ofthe films was as low as 16 μΩ·cm for a film thickness of 30 nm.The surface roughness measured by atomic force microscopywas ∼5 nm for a ∼15 nm thick film. This roughness value ischaracteristic for Ru ALD films using O2 gas,

25,76,77 although itmay be too high for several applications of Ru ALD.Quadrupole Mass Spectrometry. A differentially pumped

QMS (MKS Vision 2000C) with a mass range of 300 amusampled the gases in the process chamber via an orifice. Tobalance the sensitivity with the level of fragmentation of themolecules in the QMS ionizer, an electron energy of 40 eV wasselected. The unit cycle of the ALD process reportedpreviously75 was modified to accurately identify and quantifythe surface reaction products using the QMS. Thesemodifications did not affect the GPC and the material

properties of the Ru films. Specifically, in the modified cycles,a 5 min evacuation step separated the Ru and the 18O2 pulses,during which all of the remaining gas-phase species werepumped out of the chamber. This was necessary to establish aconstant baseline for the QMS ion current at any given m/zvalue. In addition, the top and bottom gate valves of thevacuum chamber were closed for 7 s (see Figure 2c) duringwhich time we introduced 2 s CpRu(CO)2Et and

18O2 pulses.Since the gate valves were closed, the reaction products weretrapped within the chamber. The QMS signal at a given m/zwas thus proportional to the mole fraction of the correspondingneutral species in the reaction chamber.During ALD, the gas-phase concentration of the surface

reaction products can be significantly lower than that of theexcess precursor in the chamber. In the QMS, the precursor isdissociatively ionized by the electrons in the ionizer leading toionic fragments whose m/z values can overlap with those of thereaction products. To select the relevant m/z values that resultfrom fragmentation of the reaction products, and not the parentprecursor molecule, we first performed a mass scan over therange of m/z = 1−60 with a low time resolution during areference measurement (CpRu(CO)2Et pulsing only) andduring the ALD process (data shown in SupportingInformation).78 On the basis of the difference in intensity inthese mass spectra (see Figure S.1, Supporting Information), nosignal from reaction products could be detected above the noiselevel beyond m/z = 48. By comparing the fragmentationpatterns of the measured m/z values with those published inthe NIST database,79 we could assign the different peaks tospecific parent molecules produced during surface reactions.We also checked for potential artifacts in the signal intensitydue to the limited resolution of the QMS and, if present, tookthis into consideration in the assignment of the m/z values toparticular gas-phase species. For example, the QMS signal at agiven m/z value has an asymmetric shape with a tail toward thelower m/z value.80 Thus, when the peak intensity at a certainm/z is very high compared to the intensity at the neighboringlower m/z, it can result in an erroneous determination of theQMS signal intensity at this lower m/z value. In such cases, wecarefully analyzed the peak shape to ascertain the true origin ofthe signal. Following the identification of the relevant m/zvalues to track different surface reaction products, we recordedthe data in a multiplexing mode where the QMS signal intensityfor the relevant m/z values was tracked simultaneously duringthe ALD cycles.To perform a quantitative analysis of the QMS data, it was

necessary to calibrate the QMS signal intensities. The QMSsignal S for any parent molecule XY is related to its gas-phasenumber density nXY as

81

σ= · ·→ +S k nXY XY XY (1)

where S is the measured ion current after subtracting thebackground (see Supporting Information); σXY→XY

+ is theelectron-energy-dependent direct ionization cross-section ofthe molecule XY available in the literature;82 and the factor kaccounts for the m/z-dependent sensitivity of the QMS. Toeliminate k and σXY→XY

+, the QMS signal intensity can bedirectly related to the number density of the species using thefollowing expression

=S S n n/ /XY XYALD ref ALD ref

(2)

with Sref being the QMS reference signal for species XY at aknown reference pressure and nXY

ref the associated number



density of that species. However, eq 2 requires QMS referencemeasurements for every species identified as a reaction product.This was not possible for practical reasons, especially with theuse of 18O2, which results in reaction products that contain theisotope. Therefore, the factor k was determined by calibratingthe QMS signal over the range of 18 < m/z < 40 using severalgases (see Figure S.2 in Supporting Information) and taking theappropriate cross sections into account.82 The factor k for anym/z value within or slightly outside this range was interpolatedor extrapolated, and eq 1 was used to relate the QMS signal ofeach reaction product to its gas-phase number density.However, H2 generally does not follow this trend due to theinefficient pumping of H2 by turbomolecular pumps, andtherefore eq 2 was used for H2.To check our quantification procedure for the QMS signal,

first we estimated the number of gas-phase C and H atoms perunit volume, Nx (x = C or H). Assuming that all C and Hcontained in CpRu(CO)2Et is converted to gas-phase reactionproducts after one complete ALD cycle, Nx can be expressed as

σ=

· ·N

m AVx

x Ru d(3)

where mx is the number of atoms of x (x = C or H) inCpRu(CO)2Et (mC = 9, mH = 10); σRu = 7.1 × 1014 cm−2 is thenumber of Ru atoms deposited in one ALD cycle; Ad is thesurface area for Ru deposition (Ad = 398 cm2 based on a 225mm diameter heated area); and V is the volume of the ALDchamber with the gate valve to the pump closed (V = 21 ± 1L). From these values, we obtain NC = 1.2 × 1014 cm−3 and NH= 1.3 × 1014 cm−3. Second, we accounted for the numberdensity of all C- and H-containing reaction products identifiedin the gas phase with the QMS, and using the correspondingstoichiometry of these species, we determined Nx (x = C or H):the values obtained via the quantification of the QMS signals(see Table 1) were overestimated by a factor of ∼4. The largesterror probably lies in the underestimation of the active surfacearea used in eq 3 above, which was simply based on the area ofthe heated substrate. The actual surface area in the Ru ALDprocess was probably greater than that calculated from thediameter of the substrate heater due to two reasons: radiative

heating of the surrounding surfaces and the surface roughnessof the substrate heater.

■ RESULTSIdentification and Quantification of the Reaction

Products. For clarity, the QMS data shown in Figure 3 wereseparated into two panels to distinguish between the C- and H-containing reaction products. During the Ru pulse, thecontributions of the reaction products formed due to ALDsurface reactions were isolated by first recording the m/z valuesof interest during the pulsing of CpRu(CO)2Et several timeswithout pulsing 18O2 in between. Consequently, the data shownin Figure 3a,b provide a reference signal level for each of the m/z values in the absence of ALD surface reactions. This referencesignal was recorded by interrupting a sequence of normal ALDcycles consisting of alternating exposure of the surface toCpRu(CO)2Et and

18O2 to ensure that the surface during thereference measurement was representative of ALD conditions.The QMS signal for a regular ALD cycle during the Ru pulse

is shown in Figure 3c,d. For m/z = 2, 18, 20, 30, 44, 46, and 48,the signal is much higher than the corresponding reference levelin Figure 3a,b, indicating that these ions originate frommolecules produced via surface reactions during the Ru pulse.Consequently, the difference between the signals in Figure 3a,band Figure 3c,d for each m/z value directly provides the QMSion current measured due to the surface reaction productsproduced during the Ru precursor pulse.78 The signal at m/z =2 was assigned to H2

+, and m/z = 18 and 20 represent H216O+

and H218O+, respectively. We also considered the possibility

that m/z = 18 corresponds to 18O+ produced from thedissociative ionization of 18O2,

18O16O, or H218O during the Ru

precursor pulse (Figure 3c,d). However, the concentration ofthese species is expected to be negligible since no QMS signalfor m/z = 36 (18O2

+) or 34 (18O16O+) was detected, and 18O+

only accounts for <1% of the cracking pattern of H218O.83 The

signal at m/z = 30 was assigned to C18O+, and m/z = 44, 46,and 48 were assigned to C16O2

+, C16O18O+, and C18O2+,

respectively. The ionized carbonyl and ethyl ligands can bothcontribute to m/z = 28 (C16O+ or C2H4

+). In Figure 3d, m/z =28 is lower than the reference level in Figure 3b. This can beattributed to the fact that the signal at m/z = 28 mainly arisesfrom cracking products of the precursor which is consumed bysurface reactions during the ALD cycle. The contribution ofreaction products to m/z = 28 in Figure 3d is apparently nothigh enough to compensate for this precursor consumption,and therefore, the overall signal at m/z = 28 is lower during theALD process. Because m/z = 28 could not be attributed to asingle parent molecule, no clear conclusion concerning C16Ocould be drawn as a reaction product during the Ru pulse.However, the isotope labeling enables the distinction betweenthe C2H4

+ and C18O+, as m/z = 30 can clearly be assigned toC18O+ only.The reference measurement for the 18O2 gas was performed

using the same procedure as for the CpRu(CO)2Et half-cycle,but this reference measurement did not lead to any significantsignal for the m/z values scanned, other than 18O2

+ and 18O+.Therefore, the QMS signal prior to 18O2 injection (i.e., t < 4 sin Figure 3e,f) was taken as the reference level for each m/zvalue measured, and m/z = 20, 28, 30, 44, 46, and 48 wereidentified as the species contributing to the reaction productsdue to ALD surface reactions. During the 18O2 pulse, the signalat m/z = 20 represents H2

18O+ (similar to the Ru pulse), butunlike during the Ru pulse, the signal at m/z = 18 can now be

Table 1. Gas-Phase Number Densities of CO2, CO, H2, andH2O in the ALD Reactor Calculated from the QMS DataMeasured during the Ru Precursor and 18O2 Pulses

a

densities (1014 cm−3)

Ru pulse 18O2 pulse total

CO2 C18O2 m/z = 48 1.9 0.6 2.5C16O18O m/z = 46 1.6 0.2 1.8C16O2 m/z = 44 negligible 0.3 0.3

CO C18O m/z = 30 0.27 0.18 0.45C16O m/z = 28 unknown 0.11 0.11

Total C 3.77 1.39 5.16H2 m/z = 2 2 negligible 2H2O H2

18O m/z = 20 0.1 negligible 0.1H2

16O m/z = 18 0.26 negligible 0.26Total H 4.72 negligible 4.72

aThe uncertainty in the number densities is estimated as ∼10%. Thesignal at m/z = 28 for the Ru precursor pulsing-only conditions ALDwas lower than during Ru precursor pulsing only conditions, andtherefore, no number density is included in the table.



predominantly attributed to 18O+ originating from thedissociative ionization of 18O2, rather than H2

16O+. The signalat m/z = 28 and 30 can be attributed to C16O+ and C18O+,respectively and m/z = 44, 46, and 48 to C16O2

+, C16O18O+,and C18O2

+, respectively. We will show later that, since nearlyall H is released during the Ru pulse, detection of C2H4

+ at m/z= 28 is unlikely.Summarizing, we can state that from the assigned m/z values

it is possible to identify the main gas-phase reaction productsduring each ALD half-cycle: during the Ru pulse, H2, H2

18O,H2

18O, C16O, C18O2, and C18O were released as the gas-phasereaction products, whereas C16O2, C

16O18O, C18O2, C16O, and

C18O were measured as the most dominant reaction productsduring the O2 pulse. Table 1 gives an overview of the gas-phasenumber densities of each reaction product identified above,including isotopes, using the method described in theExperimental Section.On the basis of this overview in Table 1, we can compare the

type and amount of reaction products liberated during the Ruand O2 pulses. Approximately 70% of the carbon in the metalprecursor was released during the Ru pulse in the form of CO2

and CO and ∼30% during the O2 pulse. In contrast, ∼100% ofthe hydrogen was released during the Ru precursor pulse as H2(85%) and H2O (15%). Previously, Aaltonen et al.34 studiedthe ALD of Ru using RuCp2 and O2 gas and reported thatduring the RuCp2 pulse 50% of the carbon was released in theform of CO2 and ∼90% of hydrogen was liberated in the formof H2O. Thus, the reaction products observed in our ALDprocess are in contrast with those reported by Aaltonen et al.Further analysis of the QMS data reveals that a certain

fraction of the CO ligands in the Ru precursor dissociated onthe Ru surface during the metal precursor pulse. Specifically,species containing 16O, such as H2

16O (m/z = 18) and C16O2(m/z = 44), were measured during the Ru pulse. Due to the useof the 18O2 isotope, the main source for 16O in this ALDprocess is the Ru precursor’s carbonyl ligands (C16O). Thus,the most likely pathway for the formation of H2

16O and C16O2during the Ru precursor cycle is via combustion-like reaction ofthe hydrocarbon ligands with surface 16O atoms produced viathe dissociative chemisorption of the C16O ligands.We have also estimated the surface O coverage (θO) on Ru

after the 18O2 pulse. To obtain θO, first, we assume all 18O

Figure 3. QMS ion currents for m/z ratios relevant to Ru ALD. (a), (b) Correspond to a typical CpRu(CO)2Et dose with no preceding O2 dose;(c), (d) correspond to the CpRu(CO)2Et dose, and (e), (f) correspond to the O2 dose during a typical ALD cycle. The figures in the upper panels,(a), (c), and (e), show H-containing species: m/z = 2 (H2

+), 18 (H216O+ or 18O+), and 20 (H2

18O+). The lower panels, (b), (d), and (f), show the C-containing species: m/z = 28 (C2H4

+ and C16O+), 30 (C18O+), 36 (18O2+), 44 (C16O2

+), 46 (C16O18O+), and 48 (C18O2+). The reference signals for

the m/z values measured during the Ru precursor pulse in (c) and (d) were taken from (a) and (b), respectively.



atoms incorporated on the Ru surface after the 18O2 cycle reactwith the Ru precursor ligands in the subsequent Ru precursorcycle. Second, we have determined the gas-phase numberdensity of each 18O-containing species (C18O, C18O2, C

18O16O,and H2

18O) released during the Ru precursor cycle using thesame procedure described in the Experimental Section (see eq1). Third, using the stoichiometry of each 18O-containingspecies, we can also obtain the number of 18O atoms per unitvolume released into the gas phase, NO = 5.77 × 1014 cm−3.Finally, using a similar expression as eq 3, we obtained an 18Oareal density on Ru, σO = ∼30.4 × 1015 cm−2. However, asmentioned in the Experimental Section, the values obtained viathe quantification of the QMS signals might be overestimatedby a factor of up to ∼4. Taking this correction factor and theareal density of our Ru films (1.7 × 1015 cm−2), the estimated18O coverage on Ru was ∼4.5 ML.

■ DISCUSSIONDehydrogenation and Combustion Reactions. In our

previous work,48 we showed that surface dehydrogenationreactions of the Et and Cp ligands in the Ru precursor occur onthe catalytic Ru surface, and nearly all of the precursor’shydrogen was released during the Ru pulse as H2. Thisobservation was in agreement with experimentally observeddehydrogenation reactions of hydrocarbon molecules con-ducted on Ru surfaces under similar conditions.40,41,43,49,84

The fraction of CO2 and H2O produced during the Ruprecursor pulse from the combustion of the organic precursorligands with the surface 18O atoms is higher than that reportedfor the metal precursor pulses in Pt and Ir ALD processes.During Pt ALD from MeCpPtMe3, the CO2 produced duringthe Pt pulse was only ∼6% of the total amount of CO2produced, with the remaining ∼94% being released duringthe O2 pulse.

36 Similarly, for Ir ALD using Ir(acac)3 and O2 gas,Knapas et al. measured 14% of the CO2 and 57% of the H2Obeing released into the gas phase during the Ir pulse.38 Usingthe same precursors, Christensen et al. measured 31% of theCO2 during the Ir pulse.35 According to surface scienceinvestigations, the saturation coverage of O on Pt(111) and onIr(110) surfaces is ∼0.25 ML (at 300 °C) and ∼1 ML (at 300°C), respectively.85,86 As mentioned in the Introduction, Ru isable to accommodate a higher O coverage, even >1 ML, underthe ALD conditions used in this study. Therefore, in this RuALD process, it is likely that the high amounts of CO2 and H2Oformed during the Ru pulse are due to the higher surface Ocoverage on Ru compared to Pt and Ir. In line with thesefindings, Aaltonen et al.34 showed that the CO2 and H2Omeasured in the gas phase during the RuCp2 pulse wereequivalent to an O coverage of ∼3 ML on their Ru surfacewhich is close to our established 4.5 ML of O coverage.Catalytic Combustion Reactions. Dehydrogenation and

combustion reactions can be regarded as two different kinds ofchemical reactions independent of each other, which is veryillustrative, but in reality (catalysis or ALD) these processesmay be harder to separate. According to the catalysis literature,dehydrogenation and combustion reactions are both inter-mediate steps in so-called catalytic combustion reactions. Thisterm, generally used in the field of heterogeneous catalysis,refers to catalytically assisted oxidation reactions of gaseoushydrocarbon species on Ru surfaces.71−73 In practical catalysisapplications, the temperature (300 K < T < 1000 K) andpressure (750 Torr < p < 7500 Torr) ranges used are higherthan those used in ALD, but laboratory-scale experiments

nevertheless cover the pressure and temperature regimerelevant to ALD. The catalytic combustion reactions arecomprised of two processes that can occur simultaneously: (1)the dissociation of incoming species on the catalytic surface and(2) the oxidation of C- and H-rich species. Usually, the catalyticcombustion reactions are divided into two distinct regimes: thefuel-rich combustion regime and the fuel-lean combustionregime. Analogous to these regimes, the metal precursor pulseduring noble metal ALD is characterized by an excess of metalprecursor (with its organic ligands) and can therefore beregarded as the fuel-rich regime, whereas the second half-cycleis best represented by the fuel-lean conditions since O2 is thepredominant species compared to the carbon-rich species onthe metal surface.Applying the concept of catalytic combustion reactions to Ru

ALD means that in each ALD half-cycle species dissociativelychemisorb onto the surface, thereby facilitating the combustionreactions. The reaction mechanism for Ru ALD presented inthis study is preliminary and is based solely on our QMSmeasurements combined with literature reports from catalysisand surface science. Figure 4 aims at simplifying the complex

Ru ALD chemistry using the concept of catalytic combustionreactions, which we envision to proceed as follows. During theRu precursor pulse, the precursor’s hydrocarbon ligands arelargely dehydrogenated on the O-covered Ru surface,producing chemisorbed C-rich species, while most of the His released into the gas phase as H2. The precursor’s carbonylligands may be directly released as gas-phase C16O ormolecularly chemisorb onto the Ru surface as C16O ordissociatively chemisorb (see Results section) as C and 16O.

Figure 4. Schematic of the catalytic combustion reactions driving thesurface processes during CpRu(CO)2Et and O2 pulses of the Ru ALDprocess. At the start of the CpRu(CO)2Et pulse, the initial surfaceconsists of a Ru layer with an O coverage, θO ∼ 4,5 ML, determinedfrom our experiments. As CpRu(CO)2Et adsorbs onto the surface, theorganic ligands are dehydrogenated, and the dehydrogenationproducts react with each other or with surface O, releasing H2, CO2,H2O, and CO into the gas phase. As a result of the dehydrogenationreactions, a layer of carbon-rich species remains on the surface. As theO2 dissociatively chemisorbs onto this surface during the O2 pulse, itcombusts the carbon-rich species, releasing CO2 and CO into the gasphase. It also restores the O coverage on the Ru surface before thenext Ru precursor pulse.



Subsequently, a fraction of the surface dehydrogenationproducts and the carbonyl ligands can react with the surface18O or 16O to form CO2 and H2O, and thereby the surface Ruatoms are reduced to their pure metallic state during thecombustion process. When the 18/16O surface reservoir isdepleted, predominantly partial combustion reactions occur,producing mainly gaseous C16O or C18O. At the end of the Rupulse, carbon-rich species remain on the Ru surface thatcorrespond to the ∼30% of the total carbon eventually releasedfrom the Ru surface in one complete ALD cycle. In the 18O2pulse, the 18O2 chemisorbs dissociatively onto the Ru surface,and the reactive 18O atoms (and some residual 16O from the Ruhalf-cycle) react with the remaining surface species, liberatingCO and CO2. In both half-cycles, combustion productscontaining both oxygen isotopes were detected, which confirmsthe dissociative chemisorption of C16O ligands in CpRu-(C16O)2Et.Additional Considerations on the Basis of Surface

Science Investigations. Ru vs RuO2. From additionalinformation reported in the surface science literature, moresubtle speculations can be made concerning the formation ofCO2 and H2O during the Ru ALD process. In a simplistic view,the origin of CO2 and H2O is a result of complete combustionreactions of the Ru precursor’s ligands on O-rich Ru surfaces.From surface science studies it is possible to refine this view byproviding evidence that, in fact, these reactions are mostefficient on RuO2 regions. Depending on the O2 pressure anddose, and the surface temperature, RuO2 domains can exist on aRu surface, which can have a very different catalytic activity. Onthe basis of the reaction products measured, we speculate thatcombustion-like reactions that lead to the release of CO2 andH2O occur primarily on RuO2 domains, whereas thedehydrogenation reactions that lead to the release of H2occur primarily on Ru surfaces.One pathway for the formation of CO2 is via the oxidation of

CO,56,87 which is either released thermally from the Ruprecursor or generated due to partial combustion of theprecursor’s hydrocarbon ligands. Below we speculate on thepossible pathways for CO oxidation. As mentioned in theIntroduction, the CO oxidation reaction on Ru(0001) surfaceshas been well explored, and it has been reported that O-chemisorbed Ru(0001) surfaces with 0 < θO < 3 ML do not playa significant role in the CO oxidation reaction under UHVconditions56,88,89 because of the strong Ru−O bond and thelow desorption temperature for adsorbed CO.67 Consequently,under UHV conditions the chemisorbed CO desorbs withoutreacting with the chemisorbed O on the Ru surface. Otherexperiments suggest that the RuO2 microregions formed at acoverage value of ≥3 ML are significantly more active incombusting C-containing species than Ru(0001) because thecoordinatively unsaturated Ru atoms in RuO2 act as “danglingbonds” and can provide the adsorption sites for CO.90−92 Inprevious studies it was shown that the CO oxidation reactionproceeds via the Mars−Van Krevelen (reduction−reoxidation)mechanism93 in which an O atom is abstracted from the RuO2lattice to form gaseous CO2 leaving behind an O vacancy. Ifpresent, gaseous O2 dissociates on the surface, repopulating theO vacancies.92 In the absence of gas-phase O2, the surface Ruatoms are reduced to metallic Ru as a result of CO oxidation onthe RuO2 surface.

92

Most surface science studies are based on single-crystalRu(0001) surfaces, and this is an important differencecompared to ALD grown films, which are polycrystalline. In

terms of O coverage, the few reports available show thatRu(101 0) surfaces can also take up chemisorbed O beyond 1ML.94,95 The polycrystalline nature of the films seems tobecome more important at high O coverages because thecrystal orientation of the developing RuO2 patches was claimedto be determined by that of the underlying Ru domains.92,96,97

The RuO2 domain orientation, on the other hand, was reportedto be decisive in the catalytic activity of RuO2 in the specificsurface reactions.96 Concerning dehydrogenation reactions,little is known about Ru crystal orientations other thanRu(0001). These dehydrogenation reactions have beenreported to occur for CH4 on Ru(112 0) surfaces,44 and theyare known to occur on polycrystalline Ru catalysts.Both Ru and RuO2 are known to be good dehydrogenation

catalysts, but the proposed microscopic dehydrogenationmechanisms are different for each of these surfaces.98 Thedehydrogenation of hydrocarbon species on RuO2 surfacesleads to the preferential formation and liberation of CO2 andH2O,

98,99 whereas on Ru surfaces, H2 is the main gas-phasedehydrogenation product due to the lack of surface O.49 In thecatalytic oxidation of hydrocarbons on supported Ru catalystsat low temperatures (300−600 K), the formation of CO2 andH2O was also attributed to the Mars−Van Krevelen mechanismoccurring on thin RuO2 layers or domains.

100−102

In Ru ALD, the substrate temperature is ∼300 °C (∼573 K),and O2 exposure is ∼105 L, which coincides with the typicalexperimental conditions at which these RuO2 patchesdevelop.103 In addition, we observed that ∼70% of the C16Oligands were oxidized on the surface to CO2 during the Ruprecursor pulse. From these facts, it is plausible that RuO2patches are present on the surface during ALD. It is unlikelythat the entire surface consists of RuO2 due to the observationthat high amounts of H2 are liberated, which is a fingerprint forRu surfaces. The transition from oxygen adsorption to oxideformation on the Ru(0001) surface is quite complex and leadsto the coexistence of (1 × 1)O−Ru(0001) and RuO2(110)patches.55 It is therefore possible that our Ru surface consists ofa heterogeneous system of RuO2 and O-rich Ru patches at thestart of the ALD cycle. Upon the interaction of the organicprecursor ligands, the oxide surface would be entirely reducedto metallic Ru by the end of the Ru pulse. However, it will benecessary to study the Ru surface during the ALD cycle bymeans other than QMS to confirm this view and draw finalconclusions.

Saturation Mechanism during Ru ALD. In general, thesaturation of surface half-reactions in ALD occurs whenadsorption of new molecules is inhibited. This can occur eitherdue to steric hindrance provided by the chemisorbed surfacespecies or due to the occupation of surface reactive sites.Previous literature46−48 and the QMS data during Ru ALD ofthis study suggest that during the metal pulse the surfacereactions are terminated due to the C-rich species remaining onthe surface, which inhibit further precursor adsorption. On theother hand, during the O2 pulse, the surface reactions areterminated when all of the remaining C-rich species areeliminated: the relevant parameters include surface temperatureand O2 pressure and exposure time. Specifically, we postulatethat during the Ru precursor pulse hydrocarbon dehydrogen-ation reactions play an important role. On the basis of thesurface science literature, it is known that on Ru(0001) surfacesdehydrogenation reactions can occur at low temperatures (105K)104 but are complete only at high temperatures (700−800K), leaving a layer of C atoms on the Ru surface.44,49,84 The



deposition temperature (600 K) for the Ru ALD processdiscussed herein is somewhat lower than the temperature forcomplete dehydrogenation, and therefore, we expect that thespecies remaining on the surface at the end of the Ru pulse areprimarily C atoms with only a small fraction of H atoms. This isin agreement with the fact that we observed almost no H-containing species during the following 18O2 pulse. In fact, wewere able to estimate the coverage of these C-rich surfacespecies after the Ru precursor cycle to be ∼1 ML: this coveragewas calculated using the number density of CO2 and COreleased into the gas phase during the 18O2 cycle (see Table 1)and relating this number density to the amount of carbon onthe Ru surface using the same procedure described earlier fordetermining the surface O coverage (see Results section).Furthermore, a study on ethylene dehydrogenation onRu(0001) surfaces (for the purpose of depositing graphene)showed that a monolayer of C-atoms was sufficient todeactivate the dehydrogenation of incoming ethylene moleculeson the Ru surface.105 As a consequence, we speculate thateventually most of the available Ru sites are occupied by Cspecies, thereby blocking the adsorption of incoming CpRu-(CO)2Et molecules and leading to the saturation of the ALDsurface reactions during the Ru precursor pulse.Temperature Dependence of the Ru ALD Process.

Figure 5 shows that over the range of process conditions

studied for Ru ALD the GPC was low at temperatures below200 °C. Several other Ru ALD processes using organometallicprecursors and O2 gas show a very similar trend,26,28,37,75−77,106

and in most cases, film deposition was reported at ∼300 °C. Apossible surface reaction that would require these elevatedtemperatures is the dehydrogenation of the Ru precursor’shydrocarbon ligands. However, these dehydrogenation reac-tions can be excluded as the rate-limiting step as they have beenreported to occur on pristine Ru(0001) surfaces at temper-atures as low as 105 K104 and at even lower temperatures on O-covered Ru(0001).41 Therefore, we postulate that the rate-limiting step for Ru ALD is during the O2 cycle where O2 gashas to dissociate on the carbon-covered Ru surface obtainedafter the Ru precursor pulse. These carbon atoms have to beremoved via combustion-like reactions by the dissociated O2gas. Previous surface science literature on Ru shows that thetransition from Ru to RuO2 during O2 exposure occurs over thetemperature range of 500−550 K.107 Interestingly, combustion

reactions of C-containing species on fully oxidized RuO2surfaces have also been shown to require a thresholdtemperature of 500−600 K.98,103,108 Since Ru ALD occursprecisely over the temperature range of 500−550 K, wespeculate that RuO2 domains may contribute to the eliminationof surface carbon.

■ CONCLUSIONSIn this work, QMS was used to obtain quantitative informationon the surface reactions involved in Ru ALD from CpRu-(CO)2Et and oxygen isotope 18O2. The gas-phase reactionproducts were CO2, CO, H2, and H2O during the Ru precursorpulse and mainly CO2 and CO during the O2 pulse. Most of thereaction products were formed during the Ru pulse of the ALDprocess: Approximately 70% of the carbon contained in theprecursor was released during this pulse of which ∼90% was inthe form of CO2 and the rest as CO. During the same metalprecursor pulse, ∼70% of the carbonyl ligands reacting with thesurface were oxidized, and the remaining ∼30% were liberatedduring the O2 pulse in the form of C16O and C16O18O.Furthermore, during the Ru pulse, ∼100% of the hydrogencontained in the precursor was released as H2 (85%) and H2O(15%). From these observations and on the basis of surfacescience and catalysis literature, we concluded that the complexchemistry taking place on the Ru surface during ALD can bedescribed by catalytic combustion reactions, which wereoriginally introduced in the field of catalysis. These reactionsconsist of the dissociative chemisorption of specific species on acatalytic surface (organic ligands during the Ru pulse andoxygen in the O2 pulse) accompanied with the oxidation of theC- and H-rich species with the adsorbed oxygen. On the basisof the surface science and catalysis literature on Ru surfaces, wehave speculated on specific properties of the Ru ALD process:the role of RuO2 and Ru due to the different catalytic chemistryof both materials, the deposition temperature dependence, andthe saturation mechanism. These speculations provide oppor-tunities to test and refine the surface reaction mechanism suchthat Ru ALD processes can be tuned further.

■ ASSOCIATED CONTENT*S Supporting InformationAdditional information on the identification of the reactionproducts, the procedure to correct the QMS signal intensityused in the quantification of the reaction products, and thecalibration of the QMS signal intensity are provided in aseparate document. This material is available free of charge viathe Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONNotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe thank C.A.A. van Helvoirt for his technical supportthroughout this study and Dr. H.C.M. Knoops both from theEindhoven University of Technology for very fruitfuldiscussions. This work was financially supported by theDutch Technology Foundation STW, which is part of TheNetherlands Organization for Scientific Research (NWO), andspecifically the work of one of the authors (W.M.M. Kessels) issupported through the VICI program on “Nanomanufacturing”.S. A. gratefully acknowledges financial support from the NSF

Figure 5. Growth-per-cycle during Ru ALD as a function of thesubstrate temperature as measured by spectroscopic ellipsometry.



http://pubs.acs.org

CAREER program (Grant No. CBET-0846923) and from theFoundation for Fundamental Research on Matter (FOM) forhis sabbatical leave.

■ REFERENCES(1) Liu, C.; Wang, C.-C.; Kei, C.-C.; Hsueh, Y.-C.; Perng, T.-P.Atomic Layer Deposition of Platinum Nanoparticles on CarbonNanotubes for Application in Proton-Exchange Membrane Fuel Cells.Small 2009, 5, 1535−1538.(2) ITRS www.itrs.net - Process integration, devices, and structures,2010.(3) Kim, S. K.; Lee, S. W.; Han, J. H.; Lee, B.; Han, S.; Hwang, C. S.Capacitors with an Equivalent Oxide Thickness of <0.5 nm forNanoscale Electronic Semiconductor Memory. Adv. Funct. Mater.2010, 20, 2989−3003.(4) Kittl, J. A.; Opsomer, K.; Popovici, M.; Menou, N.; Kaczer, B.;Wang, X. P.; Adelmann, C.; Pawlak, M. A.; Tomida, K.; Rothschild, A.;et al. High-k dielectrics for Future Generation Memory Devices.Microelectron. Eng. 2009, 86, 1789−1795.(5) Kohl, D. The Role of Noble Metals in the Chemistry of Solid-State Gas Sensors. Sens. Actuators, B: Chem. 1990, 1, 158−165.(6) Patil, L. A.; Bari, A. R.; Shinde, M. D.; Deo, V.; Kaushik, M. P.Detection of Dimethylmethylphosphonate − a Simulant of Sarin: TheHighly Toxic Chemical Warfare − Using Platinum ActivatedNanocrystalline ZnO Thick Films. Sens. Actuators, B: Chem. 2012,161, 372−380.(7) Bartlett, P. N.; Guerin, S. A Micromachined Calorimetric GasSensor: an Application of Electrodeposited Nanostructured Palladiumfor the Detection of Combustible Gases. Anal. Chem. 2002, 75, 126−132.(8) Ukisu, Y.; Miyadera, T. Dechlorination of PolychlorinatedDibenzo-p-dioxins Catalyzed by Noble Metal Catalysts under MildConditions. Chemosphere 2002, 46, 507−510.(9) Wang, J.; Zhong, J.; Gong, M.; Liu, Z.; Zhao, M.; Chen, Y.Remove Cooking Fume Using Catalytic Combustion over Pt/La-Al2O3. J. Environ. Sci. 2007, 19, 644−646.(10) Bhattacharyya, S.; Das, R. K. Catalytic Control of AutomotiveNOx: A Review. Int. J. Energy Res. 1999, 23, 351−369.(11) George, S. M. Atomic Layer Deposition: An Overview. Chem.Rev. 2010, 110, 111−131.(12) Comstock, D. J.; Christensen, S. T.; Elam, J. W.; Pellin, M. J.;Hersam, M. C. Tuning the Composition and Nanostructure of Pt/IrFilms via Anodized Aluminum Oxide Templated Atomic LayerDeposition. Adv. Funct. Mater. 2010, 20, 3099−3105.(13) Knoops, H. C. M.; Mackus, A. J. M.; Donders, M. E.; Van deSanden, M. C. M.; Notten, P. H. L.; Kessels, W. M. M. Remote PlasmaALD of Platinum and Platinum Oxide Films. Electrochem. Solid-StateLett. 2009, 12, G34−G36.(14) Christensen, S. T.; Feng, H.; Libera, J. L.; Guo, N.; Miller, J. T.;Stair, P. C.; Elam, J. W. Supported Ru−Pt Bimetallic NanoparticleCatalysts Prepared by Atomic Layer Deposition. Nano Lett. 2010, 10,3047−3051.(15) Biener, J.; Baumann, T. F.; Wang, Y.; Nelson, E. J.; Kucheyev, S.O.; Hamza, A. V.; Kemel, M.; Ritala, M.; Leskela, M. Ruthenium/Aerogel Nanocomposites via Atomic Layer Deposition. Nanotechnol-ogy 2007, 18, 55303.(16) Joo, J. H.; Seon, J. M.; Jeon, Y. C.; Oh, K. Y.; Roh, J. S.; Kim, J.J.; Choi, J. T. Investigation of Ruthenium Electrodes for (Ba,Sr)TiO3

Thin Films. Jpn. J. Appl. Phys. 1998, 37, 3396−3401.(17) Swerts, J.; Armini, S.; Carbonell, L.; Delabie, A.; Franquet, A.;Mertens, S.; Popovici, M.; Schaekers, M.; Witters, T.; Tokei, Z.; Beyer,G.; Van Elshocht, S.; Gravey, V.; Cockburn, A.; Shah, K.; Aubuchon, J.Scalability of Plasma Enhanced Atomic Layer Deposited RutheniumFilms for Interconnect Applications. J. Vac. Sci. Technol. A 2012, 30,01A103−6.(18) Kim, J.-Y.; Kil, D.-S.; Kim, J.-H.; Kwon, S.-H.; Ahn, J.-H.; Roh,J.-S.; Park, S.-K. Ru Films from Bis(ethylcyclopentadienyl)ruthenium

Using Ozone as a Reactant by Atomic Layer Deposition for CapacitorElectrodes. J. Electrochem. Soc. 2012, 159, H560−H564.(19) Park, S. J.; Kim, W. H.; Lee, H. B. R.; Maeng, W. J.; Kim, H.Thermal and Plasma Enhanced Atomic Layer Deposition Rutheniumand Electrical Characterization as a Metal Electrode.Microelectron. Eng.2008, 85, 39−44.(20) Kwon, O. K.; Kwon, S. H.; Park, H. S.; Kang, S. W. PEALD of aRuthenium Adhesion Layer for Copper Interconnects. J. Electrochem.Soc. 2004, 151, C753−C756.(21) Aaltonen, T.; Alen, P.; Ritala, M.; Leskela, M. Ruthenium ThinFilms Grown by Atomic Layer Deposition. Chem. Vap. Deposition2003, 9, 45−49.(22) Kim, S. K.; Lee, S. Y.; Lee, S. W.; Hwang, G. W.; Hwang, C. S.;Lee, J. W.; Jeong, J. Atomic Layer Deposition of Ru Thin Films using2,4-(Dimethylpentadienyl)(ethylcyclopentadienyl)Ru by a LiquidInjection System. J. Electrochem. Soc. 2007, 154, D95−D101.(23) Kim, S. K.; Hoffmann-Eifert, S.; Waser, R. Growth of NobleMetal Ru Thin Films by Liquid Injection Atomic Layer Deposition. J.Phys. Chem. C 2009, 113, 11329−11335.(24) Kukli, K.; Kemell, M.; Puukilainen, E.; Aarik, J.; Aidla, A.;Sajavaara, T.; Laitinen, M.; Tallarida, M.; Sundqvist, J.; Ritala, M.;Leskela, M. Atomic Layer Deposition of Ruthenium Films from(Ethylcyclopentadienyl)(pyrrolyl)-ruthenium and Oxygen. J. Electro-chem. Soc. 2011, 158, D158−D165.(25) Kukli, K.; Ritala, M.; Kemell, M.; Leskela, M. High TemperatureAtomic Layer Deposition of Ruthenium from N,N-Dimethyl-1-ruthenocenylethylamine. J. Electrochem. Soc. 2010, 157, D35−D40.(26) Kukli, K.; Aarik, J.; Aidla, A.; Uustare, T.; Jogi, I.; Lu, J.;Tallarida, M.; Kemell, M.; Kiisler, A. A.; Ritala, M.; Leskela, M.Structure and morphology of Ru films grown by atomic layerdeposition from 1-ethyl-1′-methyl-ruthenocene. J. Cryst. Growth 2010,312, 2025−2032.(27) Gregorczyk, K.; Henn-Lecordier, L.; Gatineau, J.; Dussarrat, C.;Rubloff, G. Atomic Layer Deposition of Ruthenium Using the NovelPrecursor bis(2,6,6-trimethyl-cyclohexadienyl)ruthenium. Chem.Mater. 2011, 23, 2650−2656.(28) Choi, S.-H.; Cheon, T.; Kim, S.-H.; Kang, D.-H.; Park, G.-S.;Kim, S. Thermal Atomic Layer Deposition (ALD) of Ru Films for CuDirect Plating. J. Electrochem. Soc. 2011, 158, D351−D356.(29) Geidel, M.; Junige, M.; Albert, M.; Bartha, J. W. In-situ Analysison the Initial Growth of Ultra-Thin Ruthenium Films with AtomicLayer Deposition. Microelectron. Eng. 2013, 107, 151−155.(30) Aaltonen, T.; Ritala, M.; Tung, Y. L.; Chi, Y.; Arstila, K.;Meinander, K.; Leskela, M. Atomic Layer Deposition of Noble Metals:Exploration of the Low Limit of the Deposition Temperature. J. Mater.Res. 2004, 19, 3353−3358.(31) Choi, J.; Choi, Y.; Hong, J.; Tian, H.; Roh, J. S.; Kim, Y.; Chung,T. M.; Oh, Y. W.; Kim, C. G.; No, K. Composition and ElectricalProperties of Metallic Ru Thin Films Deposited using Ru(C6H6)-(C6H8) Precursor. Jpn. J. Appl. Phys. 2002, 41, 6852−6856.(32) Kwon, O. K.; Kwon, S. H.; Park, H. S.; Kang, S. W. Plasma-Enhanced Atomic Layer Deposition of Ruthenium Thin Films.Electrochem. Solid-State Lett. 2004, 7, C46−C48.(33) Yim, S. S.; Lee, M. S.; Kim, K. S.; Kim, K. B. Formation of RuNanocrystals by Plasma Enhanced Atomic Layer Deposition forNonvolatile Memory Applications. Appl. Phys. Lett. 2006, 89, 93115.(34) Aaltonen, T.; Rahtu, A.; Ritala, M.; Leskela, M. ReactionMechanism Studies on Atomic Layer Deposition of Ruthenium andPlatinum. Electrochem. Solid-State Lett. 2003, 6, C130−C133.(35) Christensen, S. T.; Elam, J. W. Atomic Layer Deposition of Ir-PtAlloy Films. Chem. Mater. 2010, 22, 2517−2525.(36) Kessels, W. M. M.; Knoops, H. C. M.; Dielissen, S. A. F.;Mackus, A. J. M.; Van de Sanden, M. C. M. Surface Reactions DuringAtomic Layer Deposition of Pt Derived from Gas Phase InfraredSpectroscopy. Appl. Phys. Lett. 2009, 95, 013114.(37) Aaltonen, T.; Ritala, M.; Arstila, K.; Keinonen, J.; Leskela, M.Atomic Layer Deposition of Ruthenium Thin Films from Ru(thd)3and Oxygen. Chem. Vap. Deposition 2004, 10, 215−219.



www.itrs.net

(38) Knapas, K.; Ritala, M. In situ Reaction Mechanism Studies onAtomic Layer Deposition of Ir and IrO2 from Ir(acac)3. Chem. Mater.2011, 23, 2766−2771.(39) Rauscher, H.; Jakob, P.; Menzel, D.; Lloyd, D. R. TheAdsorption and Reaction of Toluene on Ru(001), and Coadsorptionwith CO. Surf. Sci. 1991, 256, 27−41.(40) Weiss, M. J.; Hagedorn, C. J.; Weinberg, W. H. Trapping-Mediated and Multilayer-Induced Dissociative Chemisorption ofCyclobutane on Ru(001). J. Am. Chem. Soc. 1999, 121, 5047−5055.(41) Garcia, A. R.; Barros, R. B.; Ilharco, L. M. Reactivity of 3-hexyneon Oxygen Modified Ru(001) Surfaces: Observation of Oxametalla-cycles by RAIRS. Surf. Sci. 2009, 603, 380−386.(42) Ilharco, L. M.; Garcia, A. R.; Hargreaves, E. C.; Chesters, M. A.A Comparative Reflection-Absorption Infrared Spectroscopy Study ofthe Thermal Decomposition of 1-Hexene on Ru(0001) and onPt(111). Surf. Sci. 2000, 459, 115−123.(43) Hrbek, J. Weakly Bound Hydrogen on Oxygen ModifiedRu(001) Surface. J. Catal. 1986, 100, 523−525.(44) Wu, M. C.; Xu, Q.; Goodman, D. W. Investigations of GraphiticOverlayers Formed from Methane Decomposition on Ru(0001) andRu(112 0) Catalysts with Scanning-Tunneling-Microscopy and High-Resolution Electron-Energy-Loss Spectroscopy. J. Phys. Chem. 1994,98, 5104−5110.(45) Elliott, S. D. Mechanism, Products, and Growth Rate of AtomicLayer Deposition of Noble Metals. Langmuir 2010, 26, 9179−9182.(46) Mackus, A. J. M.; Leick, N.; Baker, L.; Kessels, W. M. M.Catalytic Combustion and Dehydrogenation Reactions during AtomicLayer Deposition of Platinum. Chem. Mater. 2012, 24, 1752−1761.(47) Erkens, I. J. M.; Mackus, A. J. M.; Knoops, H. C. M.; Smits, P.;Van de Ven, T. H. M.; Roozeboom, F.; Kessels, W. M. M. MassSpectrometry Study of the Temperature Dependence of Pt FilmGrowth by Atomic Layer Deposition. ECS J. Solid State Sci. Technol.2012, 1, P255−P262.(48) Leick, N.; Agarwal, S.; Mackus, A. J. M.; Kessels, W. M. M.Dehydrogenation Reactions during Atomic Layer Deposition of Ruusing O2. Chem. Mater. 2012, 24, 3696−3700.(49) George, P. M.; Avery, N. R.; Weinberg, W. H.; Tebbe, F. N.Formation and Identification of Methylene on Ru(001). J. Am. Chem.Soc. 1983, 105, 1393−1394.(50) Eom, T. K.; Sari, W.; Choi, K. J.; Shin, W. C.; Kim, J. H.; Lee, D.J.; Kim, K. B.; Sohn, H.; Kim, S. H. Low Temperature Atomic LayerDeposition of Ruthenium Thin Films Using Isopropylmethylbenzene-Cyclohexadiene-Ruthenium and O2. Electrochem. Solid-State Lett. 2009,12, D85−D88.(51) Bradford, M. C. J. Two-step Methane Conversion to HigherHydrocarbons: Comment on the Relevance of Metal-Carbon BondStrength. J. Catal. 2000, 189, 238−243.(52) Choudhary, T. V; Goodman, D. W. Methane Activation onRuthenium: The Nature of the Surface Intermediates. Top. Catal.2002, 20, 35−42.(53) Ditlevsen, P. D.; Vanhove, M. A.; Somorjai, G. A. MolecularModeling of Ethylene Decomposition on Platinum Surfaces. Surf. Sci.1993, 292, 267−275.(54) Ertl, G. Reactions at Surfaces: From Atoms to Complexity(Nobel lecture). Angew. Chem., Int. Ed. 2008, 47, 3524−3535.(55) Over, H.; Seitsonen, A. P. Surface Chemistry: Oxidation ofMetal Surfaces. Science 2002, 297, 2003−2005.(56) Bottcher, A.; Niehus, H.; Schwegmann, S.; Over, H.; Ertl, G.CO Oxidation Reaction over Oxygen-rich Ru(0001) Surfaces. J. Phys.Chem. B 1997, 101, 11185−11191.(57) Blume, R.; Niehus, H.; Conrad, H.; Bottcher, A.; Aballe, L.;Gregoratti, L.; Barinov, A.; Kiskinova, M. Identification of SubsurfaceOxygen Species Created during Oxidation of Ru(0001). J. Phys. Chem.B 2005, 109, 14052−14058.(58) Blume, R.; Havecker, M.; Zafeiratos, S.; Teschner, D.;Kleimenov, E.; Knop-Gericke, A.; Schlogl, R.; Barinov, A.; Dudin,P.; Kiskinova, M. Catalytically Active States of Ru(0001) Catalyst inCO Oxidation Reaction. J. Catal. 2006, 239, 354−361.

(59) Blume, R.; Niehus, H.; Conrad, H.; Bottcher, A. Oxide-freeOxygen Incorporation into Ru(0001). J. Chem. Phys. 2004, 120, 3871−3879.(60) Bottcher, A.; Niehus, H. Formation of Subsurface Oxygen atRu(0001). J. Chem. Phys. 1999, 110, 3186−3195.(61) Stampfl, C.; Schwegmann, S.; Over, H.; Scheffler, M.; Ertl, G.Structure and Stability of a High-coverage (1 × 1) Oxygen Phase onRu(0001). Phys. Rev. Lett. 1996, 77, 3371−3374.(62) Kim, Y. D.; Seitsonen, A. P.; Wendt, S.; Wang, J.; Fan, C.;Jacobi, K.; Over, H.; Ertl, G. Characterization of Various OxygenSpecies on an Oxide Surface: RuO2(110). J. Phys. Chem. B 2001, 105,3752−3758.(63) Meinel, K.; Wolter, H.; Ammer, C.; Beckmann, A.;Neddermeyer, H. Adsorption Stages of O on Ru(0001) Studied byMeans of Scanning Tunnelling Microscopy. J. Phys.: Condens. Matter.1997, 9, 4611−4619.(64) Reuter, K.; Ganduglia-Pirovano, M. V.; Stampfl, C.; Scheffler, M.Metastable Precursors during the Oxidation of the Ru(0001) Surface.Phys. Rev. B 2002, 65, 165403.(65) Todorova, M.; Reuter, K.; Scheffler, M. Density-functionalTheory Study of the Initial Oxygen Incorporation in Pd(111). Phys.Rev. B 2005, 71, 195403.(66) Peden, C. H. F.; Goodman, D. W.; Weisel, M. D.; Hoffmann, F.M. In-situ FT-IRAS Study of the CO Oxidation Reaction overRu(001): I. Evidence for an Eley-Rideal Mechanism at High Pressures?Surf. Sci. 1991, 253, 44−58.(67) Bonn, M.; Funk, S.; Hess, C.; Denzler, D. N.; Stampfl, C.;Scheffler, M.; Wolf, M.; Ertl, G. Phonon- versus Electron-mediatedDesorption and Oxidation of CO on Ru(0001). Science 1999, 285,1042−1045.(68) Kim, Y. D.; Seitsonen, A. P.; Over, H.; Kim, D.; Seitsonen, P.The Atomic Geometry of Oxygen-rich Ru(0001) Surfaces: Coex-istence of (1 × 1)O and RuO2(110) Domains. Surf. Sci. 2000, 465, 1−8.(69) Somorjai, G. A.; Li, Y. Selective Nanocatalysis of OrganicTransformation by Metals: Concepts, Model Systems, and Instru-ments. Top. Catal. 2010, 53, 832−847.(70) Huff, M.; Schmidt, L. D. Ethylene Formation by OxidativeDehydrogenation of Ethane over Monoliths at Very Short ContactTimes. J. Phys. Chem. 1993, 97, 11815−11822.(71) Mantzaras, J. In Handbook of Combustion; Wiley-VCH VerlagGmbH & Co. KGaA: New York, 2010.(72) Lee, J. H.; Trimm, D. L. Catalytic Combustion of Methane. FuelProcess. Technol. 1995, 42, 339−359.(73) Reinke, M.; Mantzaras, J.; Schaeren, R.; Bombach, R.; Inauen,A.; Schenker, S. High-Pressure Catalytic Combustion of Methane overPlatinum: In situ Experiments and Detailed Numerical Predictions.Combust. Flame 2004, 136, 217−240.(74) Van Hemmen, J. L.; Heil, S. B. S.; Klootwijk, J. H.; Roozeboom,F.; Hodson, C. J.; Van de Sanden, M. C. M.; Kessels, W. M. M. Plasmaand Thermal ALD of Al2O3 in a Commercial 200 mm ALD Reactor. J.Electrochem. Soc. 2007, 154, G165−G169.(75) Leick, N.; Verkuijlen, R. O. F.; Lamagna, L.; Langereis, E.;Rushworth, S.; Roozeboom, F.; Van de Sanden, M. C. M.; Kessels, W.M. M. Atomic Layer Deposition of Ru from CpRu(CO)2Et Using O2

Gas and O2 Plasma. J. Vac. Sci. Technol. A 2011, 29, 021016−0210167.(76) Wang, H. T.; Gordon, R. G.; Alvis, R.; Ulfig, R. M. Atomic LayerDeposition of Ruthenium Thin Films from an Amidinate Precursor.Chem. Vap. Deposition 2009, 15, 312−319.(77) Park, S. K.; Kanjolia, R.; Anthis, J.; Odedra, R.; Boag, N.;Wielunski, L.; Chabal, Y. J. Atomic Layer Deposition of Ru/RuO2

Thin Films Studied by In situ Infrared Spectroscopy. Chem. Mater.2010, 22, 4867−4878.(78) Knoops, H. C. M.; Langereis, E.; Van de Sanden, M. C. M.;Kessels, W. M. M. Reaction Mechanisms of Atomic Layer Depositionof TaNx from Ta(NMe2)5 Precursor and H2-based Plasmas. J. Vac. Sci.Technol. A 2012, 30, 01A1011−01A10110.(79) NIST Chemistry WebBook; NIST Standard Reference DatabaseNumber 69, NIST: Maryland.



(80) Austin, W. E.; Holme, A. E.; Leck, J. H. In Quadrupole MassSpectrometry and Its Applications; Dawson, P. H., Ed.; AmericanInstitute of Physics: Amsterdam, 1997.(81) Singh, H.; Coburn, J. W.; Graves, D. B. Appearance of PotentialMass Spectrometry: Discrimination of Dissociative IonizationProducts. J. Vac. Sci. Technol. A 2000, 18, 299−305.(82) Hwang, W.; Kim, Y. K.; Rudd, M. E. New Model for Electron-impact Ionization Cross Sections of Molecules. J. Chem. Phys. 1996,104, 2956−2966.(83) Morrow, B. A.; Ramamurthy, P. Infrared Studies of theFormation of Hydroxyl Groups during Hydrogen-Oxygen Reactionson Noble Metal Catalysts. J. Phys. Chem. 1973, 77, 3052−3058.(84) Hills, M. M.; Parmeter, J. E.; Mullins, C. B.; Weinberg, W. H.Interaction of Ethylene with the Ruthenium(001) Surface. J. Am.Chem. Soc. 1986, 108, 3554−3562.(85) Getman, R. B.; Xu, Y.; Schneider, W. F. Thermodynamics ofEnvironment-Dependent Oxygen Chemisorption on Pt(111). J. Phys.Chem. B 2008, 112, 9559−9572.(86) Taylor, J. L.; Ibbotson, D. E.; Weinberg, W. H. TheChemisorption of Oxygen on the (110) Surface of Iridium. Surf. Sci.1979, 79, 349−384.(87) Gao, F.; Wang, Y.; Cai, Y.; Goodman, D. W. CO Oxidation overRu(0001) at Near-Atmospheric Pressures: From ChemisorbedOxygen to RuO2. Surf. Sci. 2009, 603, 1126−1134.(88) Kim, Y. D.; Seitsonen, A. P.; Over, H. The Atomic Geometry ofOxygen-rich Ru(0001) Surfaces: Coexistence of (1 × 1)O andRuO2(110) Domains. Surf. Sci. 2000, 465, 1−8.(89) Lee, H.-I.; White, J. M. Carbon Monoxide Oxidation overRu(001). J. Catal. 1980, 63, 261−264.(90) Over, H.; Seitsonen, A. P. Surface Chemistry: Oxidation ofMetal Surfaces. Science 2002, 297, 2003−2005.(91) Over, H.; Kim, Y. D.; Seitsonen, A. P.; Wendt, S.; Lundgren, E.;Schmid, M.; Varga, P.; Morgante, A.; Ertl, G. Atomic-Scale Structureand Catalytic Reactivity of the RuO2(110) Surface. Science 2000, 287,1474−1476.(92) Over, H. Ruthenium Dioxide, a Fascinating Material for AtomicScale Surface Chemistry. Appl. Phys. A: Mater. Sci. Process. 2002, 75,37−44.(93) Mars, P.; Van Krevelen, D. W. Oxidations Carried out by Meansof Vanadium Oxide Catalysts. Chem. Eng. Sci. 1954, 3 (Supplem), 41−59.(94) Baraldi, A.; Lizzit, S.; Paolucci, G. Identification of AtomicAdsorption Site by Means of High-resolution Photoemission SurfaceCore-Level Shift: Oxygen on Ru(1010). Surf. Sci. 2000, 457, L354−L360.(95) Schwegmann, S.; Seitsonen, A. P.; De Renzi, V.; Dietrich, H.;Bludau, H.; Gierer, M.; Over, H.; Jacobi, K.; Scheffler, M.; Ertl, G.Oxygen Adsorption on the Ru(101 0) Surface: Anomalous CoverageDependence. Phys. Rev. B 1998, 57, 15487−15495.(96) Assmann, J.; Narkhede, V.; Breuer, N. A.; Muhler, M.;Seitsonen, A. P.; Knapp, M.; Crihan, D.; Farkas, A.; Mellau, G.;Over, H. Heterogeneous Oxidation Catalysis on Ruthenium: Bridgingthe Pressure and Materials Gaps and Beyond. J. Phys.: Condens. Matter2008, 20, 184017.(97) Kim, Y. D.; Schwegmann, S.; Seitsonen, A. P.; Over, H. EpitaxialGrowth of RuO2(100) on Ru(101 0): Surface Structure and OtherProperties. J. Phys. Chem B 2001, 105, 2205−2211.(98) Over, H.; He, Y. B.; Farkas, A.; Mellau, G.; Korte, C.; Knapp,M.; Chandhok, M.; Fang, M. Long-term Stability of Ru-basedProtection Layers in Extreme Ultraviolet Lithography: A SurfaceScience Approach. J. Vac. Sci. Technol. B 2007, 25, 1123−1138.(99) Knapp, M.; Crihan, D.; Seitsonen, A. P.; Resta, A.; Lundgren, E.;Andersen, J. N.; Schmid, M.; Varga, P.; Over, H. Unusual Process ofWater Formation on RuO2(110) by Hydrogen Exposure at RoomTemperature. J. Phys. Chem. B 2006, 110, 14007−14010.(100) Liu, H.; Iglesia, E. Selective Oxidation of Methanol andEthanol on Supported Ruthenium Oxide Clusters at Low Temper-atures. J. Phys. Chem. B 2005, 109, 2155−63.

(101) Okal, J.; Zawadzki, M. Catalytic Combustion of Butane on Ru/γ-Al2O3 Catalysts. Appl. Catal. B: Environ. 2009, 89, 22−32.(102) Knapp, M.; Crihan, D.; Seitsonen, A. P.; Lundgren, E.; Resta,A.; Andersen, J. N. Complex Interaction of Hydrogen with theRuO2(110) Surface. J. Phys. Chem. C 2007, 111, 5363−5373.(103) Villani, K.; Kirschhock, C. E. A.; Liang, D.; Van Tendeloo, G.;Martens, J. A. Catalytic Carbon Oxidation Over Ruthenium-BasedCatalysts. Angew. Chem., Int. Ed. 2006, 45, 3106−3109.(104) Henderson, M. A.; Radloff, P. L.; Greenlief, C. M.; White, J.M.; Mims, C. A. Surface Chemistry of Ketene on Ruthenium(001). 2.Surface Processes. J. Phys. Chem. 1988, 92, 4120−4127.(105) Loginova, E.; Bartelt, N. C.; Feibelman, P. J.; McCarty, K. F.Factors Influencing Graphene Growth on Metal Surfaces. New J. Phys.2009, 11, 63046.(106) Kim, S. K.; Han, J. H.; Kim, G. H.; Hwang, C. S. Investigationon the Growth Initiation of Ru Thin Films by Atomic LayerDeposition. Chem. Mater. 2010, 22, 2850−2856.(107) He, Y. B.; Knapp, M.; Lundgren, E.; Over, H. Ru(0001) ModelCatalyst under Oxidizing and Reducing Reaction Conditions: In-SituHigh-Pressure Surface X-ray Diffraction Study. J. Phys. Chem. B 2005,109, 21825−21830.(108) Paulus, U. A.; Wang, Y.; Bonzel, H. P.; Jacobi, K.; Ertl, G.Adsorption and Interaction of Ethylene on RuO2(110) Surfaces. J.Phys. Chem. B 2004, 109, 2139−2148.



catalytic combustion reactions during atomic layer deposition of ru studied using 18 ...

Documents