Nucleation Enhancement and Area-Selective Atomic Layer Deposition of Ruthenium using RuO₄ and H₂-gasMatthias M. Minjauw,a Hannes Rijckaert,b Isabel Van Driessche,b Christophe Detav-ernier,a Jolien Dendoovena

a Ghent University, Conformal Coating of Nanostructures (CoCooN), Department of Solid State Sci-ences, Krijgslaan 281 (S1), 9000 Ghent, Belgium.b Ghent University, Sol-gel Centre for Research on Inorganic Powders and Thin films Synthesis (SCRIPTS), Department of Chemistry, Krijgslaan 281 (S3), 9000 Ghent, Belgium. ABSTRACT: Inherent substrate selectivity is reported for the thermal RuO₄ (ToRuSTM)/H₂-gas atomic layer deposition process (ALD) on H-terminated Si (Si-H) versus SiO₂. In situ spectroscopic ellipsometry (SE) on blanket substrates shows that Ru growth occurs from the first cycle on Si-H, while on SiO₂ the growth is delayed, resulting in a substrate selectivity window of ~60 cycles, during which up to 15 nm of Ru can be grown on Si-H with negligible deposition onto SiO2. Area-selective growth was evaluated on a pat-terned substrate with 1-10 µm wide Si-H lines separated by 10 µm wide SiO₂ regions. Ex situ planar scan-ning electron microscopy and cross section high resolution transmission electron microscopy measure-ments showed that a 4.5 nm Ru film could be deposited on the Si-H, with no Ru detected on the SiO₂. The proposed mechanism behind the inherent substrate selectivity is the oxidation of the Si-H surface by RuO₄, which was confirmed by in vacuo X-ray photoelectron spectroscopy (XPS) experiments. A method-ology to enhance the nucleation of the RuO₄/H₂-gas process on oxide substrates is also reported. In situ SE and in vacuo XPS experiments show that the nucleation delay on SiO2 can be completely removed by exposing the surface to trimethylaluminum (TMA) just before the start of the ALD process. We found evi -dence that the TMA pulse makes the oxide surface reactive towards RuO₄, by introduction of surface methyl groups which can be combusted by RuO₄. As TMA is known to be reactive towards many oxide substrates, this methodology presents a way to achieve Ru metallization of virtually any surface. There-fore one can either (i) use the RuO₄/H₂-gas process to coat non-oxidized surfaces selectively with Ru, or (ii) by using a TMA-priming one can bypass the selectivity and coat a wide variety of surfaces non-selec-tively with Ru.

INTRODUCTIONRuthenium is a candidate to replace copper in

future sub-10 nm interconnects. Although Ru has a lower bulk resistivity, it shows a better electro-migration behavior, and most likely Ru intercon-nects will not need a barrier and liner.1-10 Atomic layer deposition (ALD) is a thin film deposition method in which the growing film is alternately exposed to a series of chemical precursors, each reacting with the surface in a self-limited way. This enables the deposition of thin films with pre-cise thickness control and excellent conformal-ity.11-13 By now, ALD of compounds has proven to be a key enabling technology for semiconductor device manufacturing. However, ALD blanket lay-ers still need to be patterned by lithography, and for future technology nodes it will be increasingly difficult to align subsequent lithography steps, such that bottom-up fabrication of patterned structures becomes desirable. Therefore area-se-

lective ALD, in which ALD film growth is achieved only on selected areas of the substrate, is of high interest. 14-16

For ALD of metals, and in particular for ALD of Ru, most processes show significant nucleation delays on a wide range of surfaces.13 When ALD of blanket Ru films is required, the long nucle-ation delays reported for most Ru ALD processes lead to increased process times, waste of expen-sive precursor, and layer closure when the thick-ness of the film exceeds 5-10nm, which limits the applicability of such processes. If the length of the nucleation delay depends on the nature of the surface, this can be exploited to achieve area-se-lective deposition, which has been reported for several Ru ALD processes.17-23 Only in one of these reports however, selective deposition of Ru is acquired directly onto the underlying patterned structure, which is called inherent substrate se-lectivity.22 The other approaches are using addi-

tional process steps to either block or activate Ru growth onto specific areas of the underlying structure. As these blocking or activation layers need to be deposited and aligned onto the pat-terned structure, this complicates fabrication and therefore makes these approaches less desirable. For the inherently selective Ru ALD process re-ported by Zyulkov et al, the observed selectivity window corresponds to a layer of 3.2 nm of Ru on SiCN with respect to amorphous C, and it was not reported if the resulting layer was continuous.22

In a previous publication, we reported a thermal Ru ALD process using the inorganic RuO4-precur-sor (ToRuSTM, Air Liquide) in combination with H2-gas.24 The ALD process was shown to have a nar-row temperature window near 100°C, where high quality low-resistivity Ru films can be deposited with a high growth rate of 0.1 nm/cycle. The fol-lowing half reactions were proposed for the process in steady growth conditions:Ru (s) + RuO4 (g) 2RuO2 (s) (1)

RuO2 (s) + 2H2 (g) Ru (s) + 2H2O (g) (2)

As soon as the Ru surface is fully oxidized, half reaction (1) stops, while half reaction (2) stops as soon as the RuO2 surface is fully reduced to Ru. This explains the self-limiting nature of the process. At higher temperatures, thermal decom-position of RuO4 occurs:RuO4 (g) RuO2 (s) + O2 (g) (3)

Such that half reaction (1) is not self-limiting, and this determines the upper limit of the tempera-ture window. Half reaction (2) is inefficient at tem-peratures below 100°C, which determines the lower limit of the temperature window. Based on this last observation, a plasma-enhanced Ru ALD process using RuO4 and H2-plasma at 50°C was developed by the authors.25 As half reaction (1) stops as soon as the Ru surface is fully oxidized, one could expect that an oxidizable surface is needed to nucleate Ru.24,26 In this paper we report strong inherent selectivity of the RuO4/H2-gas process on H-terminated Si (Si-H) vs. SiO2, as a result of the delayed Ru nucleation on SiO2 com-pared to Si-H. The selectivity window was found to be ~ 60 cycles corresponding to ~ 15 nm Ru on Si-H.

Although the delayed nucleation on oxides is beneficial for achieving area-selective Ru growth, some applications might require Ru metallization of an oxide surface. In this case, using the RuO4/H2-process would be problematic, as the large nucleation delay leads to a loss of precursor and Ru films with a high roughness. Therefore we developed a method to enhance the nucleation on oxides. In this paper we show that the Ru nu-cleation delay on SiO2 can be completely re-moved by pretreatment of the surface with trimethylaluminum (TMA). Therefore one can ei-

ther (i) use the RuO4/H2-gas process to coat non-oxidized surfaces selectively with Ru, or (ii) by using a TMA-priming one can bypass the selectiv-ity and coat a wide variety of surfaces non-selec-tively with Ru and without potential nucleation difficulties. EXPERIMENTAL SECTION

Deposition System. The ALD depositions were performed in an experimental high vacuum ALD reactor with a base pressure of 10-7 mbar. This pressure is achieved by using a turbomolecu-lar pump in combination with a rotary vane back-ing pump. Samples were resistively heated inside the reactor chamber to a temperature of 125°C, and it was verified that RuO4 did not thermally decompose at this temperature. The walls of the reactor were heated to 80°C. Precursors are evap-orated in stainless steel containers, and delivered to the reactor through stainless steel tubing and pneumatically controlled inlets. Both the RuO4 (ToRuSTM, Air Liquide) and TMA (STREM CHEMICALS, min. 98%) precursors were at room temperature, with stainless steel tubing heated to 35°C. The reactor is equipped with a remote in-ductively coupled RF plasma source (13.56 MHz).

Deposition Process. Ru ALD was achieved using the RuO4/H2-gas process.10 RuO4 was sup-plied to the reactor by using the ToRuSTM-precur-sor, which is a solution of RuO4 in a methyl-ethyl fluorinated solvent developed and produced by Air Liquide. H2-gas was supplied by using a 20% mixture of H2 in Ar. Both ToRuS and H2 pulses were of a static nature. This means that the valve between the chamber and the turbo pump was closed to allow the pressure to build up by inject-ing the gas over a time ti, after which the pneu-matic inlet was closed and the chamber was held at a constant pressure Ps for a time ts. After this, the chamber was evacuated again by first using the backing pump, which is bypassed to the reac-tor to allow pumping down to roughing vacuum, and afterwards using the turbomolecular pump to pump down to base pressure. Ps , ti , ts for ToRuS and H2-gas pulses were 1.8 mbar , 8s , 10s and 5 mbar , 8s , 20s respectively.

Substrate Preparation. Planar SiO2 sub-strates were 100 nm SiO2 films grown on Si by plasma-enhanced chemical vapour deposition. Air storage lead to carbonaceous contamination of the SiO2 surface, which was removed by in situ O2 plasma (O2*) cleaning before Ru deposition. The plasma cleaning involved 5 pulses of 10s O2-plasma using 99.999% pure O2-gas at 0.005 mbar pressure and with a power of 250W. Successful removal of carbon was verified by in vacuo XPS. For an SiO2 surface treated with TMA this stan-dard O2* cleaning was followed by a single 10s pulse of TMA at a pressure of 0.008 mbar. For TMA-treatment followed by an O2-plasma, the standard O2* cleaning was followed by a single 10s pulse of TMA at a pressure of 0.008 mbar and

a single 10s O2* pulse (250W) at 0.005 mbar pressure. Planar Si and patterned Si/SiO2 sub-strates were cleaned by a hydrofluoric acid (HF) dip using a 2% HF in H2O solution for 60s. After this the substrates were rinsed in DI water and blown dry using pure nitrogen. The transfer time to the ALD reactor after this procedure was less than 10 minutes.

Materials and Process Characterization. In situ spectroscopic ellipsometry (SE) was per-formed in between ALD cycles using a Woollam M-2000 spectrometer fitted directly onto the ALD reactor. For metallic films, the thickness and opti-cal model are correlated. In addition it is typical for metal ALD that the metal nucleates as parti-cles rather than a continuous film. Therefore, no optical model was built to fit the data and extract the Ru thickness after every cycle. As SE is very surface sensitive, the deposited material on the surface will be negligible when the raw SE spectra do not change. Therefore the change of the raw data as a function of the number of ALD cycles can be used to give us an accurate indication of the nucleation period. As the amplitude ratio Ψ at a wavelength of 515 nm was found to be most sensitive to changes at the surface, we used this value as an indicator for change of SE data.

In vacuo X-ray Photoelectron Spectroscopy (XPS) measurements were performed using a Thermo ScientificTM Theta Probe X-ray Photoelec-tron Spectrometer System, which is directly at-tached to the ALD reactor through an UHV con-nection. This setup allows through-vacuum sam-ple transfer times below 60s from the ALD cham-ber (10-7 mbar) to the XPS analysis chamber (10-10

mbar). For most types of surfaces, this is suffi-ciently fast to avoid carbon contamination or des-orption of surface species. As the Ru3d region overlaps with C1s, we used Ru3p as a means to detect ruthenium on the surface. The energy axis of the spectra was calibrated using either the Si2p position for SiO2 substrates (103.5 eV) or the Si2p3/2 (99.8eV) position for Si-H substrates.27

Area-selective Ru deposition on patterned sub-strates was evaluated by Scanning Electron Mi-croscopy (SEM) and Scanning Transmission Elec-tron Microscopy (STEM). SEM was performed us-ing a FEI Quanta 200F instrument, combined with an EDAX silicon drift detector to perform Energy-dispersive X-ray spectroscopy (EDX). (S)TEM was performed using a JEOL JEM 2200-FS instrument. Samples for (S)TEM were prepared by cutting a cross-sectional lamella via the Focused Ion Beam (FIB) technique in a FEI Nova 600 Nanolab Dual Beam FIB-SEM. The lamella were extracted using the in situ lift out procedure with an Omniprobe extraction needle.28

The thickness of blanket Ru films was deter-mined by X-ray Reflectivity (XRR) measurements, using a Bruker D8 diffractometer (CuKα source). For some of these samples, X-ray Fluorescence

(XRF) was used to determine the thickness, as the films were too rough for XRR. This was done by first constructing a calibration curve of the integrated Ru L peak intensity versus the XRR thickness of smooth Ru films, and afterwards cali-brating the XRF data of rough films using this curve. The root mean square (RMS) roughness was determined by Atomic Force Microscopy, us-ing a Bruker Dimension Edge system. The electri-cal resistivity of blanket Ru films was determined by using the 4-point probes method.RESULTS AND DISCUSSION

Area-Selective Ru ALD. The RuO4/H2-gas process was monitored by in situ SE on blanket SiO2 and Si-H substrates (Figure 1). At cycle 0, the amplitude ratio Ψ(515 nm) corresponds to the substrate, and is therefore different for both cases. The real differ-ence between both cases is that the value of Ψ(515 nm) changes from the first cycle for Si-H, while for SiO2 the change starts at ~ 60 cycles. Hence, growth occurs from the first cycle on Si-H, and a nucleation delay of about 60 cycles is present on SiO2. The Ru film thickness acquired after 75 cycles on Si-H was 18.5 nm, which after interpolation leads to a thickness of roughly 15 nm after 60 cycles. Therefore we can conclude that the RuO4/H2-gas process has an inherent sub-strate selectivity window of about 60 cycles, dur-ing which up to ~ 15 nm Ru can be grown onto Si-H while avoiding growth on SiO2. This is in agree-ment with our previous results obtained using in situ synchrotron X-ray Fluorescence spectroscopy measurements.26

Area-selective ALD was evaluated on a pat-terned Si-H/SiO2 substrate. The RuO4/H2-gas process was run for 20 cycles, and the sample was analyzed by ex situ planar SEM-EDX and cross-section TEM. In Figure 2, the ex situ planar SEM-EDX analysis results are shown. No sign of Ru was detected by EDX in the SiO2-regions, while a clear RuL peak was present on the Si-H lines.

Figure 1. Amplitude ratio Ψ at 515 nm as a function of the number of cycles for the RuO4/H2-gas process on blanket Si-H and SiO2 substrates.

Figure 2. Planar view SEM analysis of patterned Si-H lines (3 µm width, and 10 µm separation) on SiO2 which were exposed to 20 cycles of the RuO4/H2-gas process. The SEM micrograph was acquired after the EDX line scan perpendicular to the Si lines. The car-bon track left by the electron beam can be seen on the micrograph, and is overlaid with the correspond-ing integrated EDX data for the Ru, O and Si-peaks.

In Figure 3, ex situ cross-section (S)TEM images are shown. From the overview TEM image it is clear that Ru was only deposited on the Si-H lines, and not on the SiO2. Also, the thickness unifor-mity of the Ru film on Si-H is good. The thick Pt overlayer is present due to the FIB preparation. From Figure 3 (b), a thickness of 4.5 nm was ex-tracted for the Ru film. In addition, an interface is visible between the Ru film and the Si, for which a thickness of 2.7 nm was found. From Figure 3 (c), it is clear that the edge of the Ru film near the transition of Si to SiO2 is well defined. The Ru film is slightly overhanging the SiO2 slope due to the conformality of the ALD-process. This result is similar to what Kalanyan et al found for area-se-lective ALD of tungsten using the SiH4/(WF6+H2)-process.6 The interface between Ru and Si has a similar contrast as the SiO2 region, and therefore we expect that the interface is a layer of oxidized Si.

Figure 3. Cross section (S)TEM analysis of patterned Si-H lines (7.2 µm width) on SiO2 which were ex-posed to 20 cycles of the RuO4/H2-gas process. (a) Overview TEM image. The Pt on top is deriving from the FIB preparation. (b) HRTEM image of the Ru film deposited on the Si-H. (c) HRTEM image near the edge of the Si-H with the SiO2.

Figure 4. The proposed surface reactions occurring during the very first half cycle of the RuO4/H2-gas process, explaining the observed selectivity between SiO2 and Si-H. (a) The fully oxidized SiO2-surface is unlikely to react with the RuO4-precursor. (b) The Si-H surface can be oxidized by RuO4 resulting in the deposition of a thin RuO2 layer.

Figure 5. In vacuo XPS data showing the effect of a single RuO4-pulse on (a) SiO2 and (b) Si-H. The pass energy setting of the XPS-spectrometer used to acquire each spectrum is shown in the upper left corner. In the case of Si-H we are able to resolve the spin-orbit split components Si2p3/2 and Si2p1/2 of the Si2p peak, by using a low pass energy (high resolution).

We propose that the mechanism behind the instant nucleation of Ru on Si-H is the oxidation of Si-H to SiO2 by RuO4. During this oxidation, RuO2 is deposited on the surface. In the case of SiO2 the surface is already fully oxidized, such that a large nucleation delay is present (Figure 4). To support this hypothesis, in vacuo XPS experi-ments were conducted. From Figure 5 (a) it is clear that a single precursor pulse on SiO2 does not lead to a detectable amount of Ru on the sur-face by XPS, due to the absence of a Ru3p peak. For Si-H on the other hand (Figure 5 (b)), a single RuO4 pulse leads to deposition of Ru on the sur-face. The Si2p peak (Si2p3/2 at 99.2 eV and Si2p1/2 at 99.8eV) for the Si-H substrate is at a lower binding energy compared to the one for SiO2 (103.5 eV). After exposing Si-H to RuO4 an additional broad peak at higher binding energy (102.3 eV) appears in the Si2p spectrum. This proves that the Si surface is oxidized by RuO4. As the Si2p peak for bulk SiO2 is expected at a higher binding energy (103.5 eV), most likely a sub-oxide is formed at the surface.29 In the O1s spectrum, two peaks can be distinguished. The one at higher binding energy (531.6 eV) can be assigned to oxygen atoms bound to Si,27 while the one at lower energy (529.7 eV) can be assigned to oxygen bound to Ru.30

Nucleation Enhancement on Oxides. The large nucleation delay observed on SiO2 was ex-plained by the inability of RuO4 to react with the fully oxidized SiO2 surface. RuO4 is known as a

strong oxidizing agent in organic chemistry.31,32

Therefore we developed the idea to enhance the Ru nucleation by introducing hydrocarbon func-tional groups on the SiO2 surface. It is established that TMA functionalizes the SiO2-surface with methyl groups.11 In Figure 6 it can be seen that by giving a single pulse of TMA before the RuO4/H2-gas process, the nucleation delay on SiO2 is com-pletely removed. When an O2*-pulse is given after the TMA-pulse, a nucleation delay is introduced again.

Figure 6. In situ spectroscopic ellipsometry data for the RuO4/H2-gas process performed on SiO2, with different in situ surface pretreatments. For the curve indicated by “SiO2”, the surface received the stan-dard in situ O2* pretreatment discussed in the exper-imental section. For the curve indicated by “TMA”,

this O2* pretreatment was followed by a single TMA-pulse, while for “TMA + O2*” it was followed by one TMA pulse and one final O2* pulse.

We propose that the mechanism behind the prompt Ru nucleation on TMA-treated SiO2 is due to the reaction of RuO4 with the surface -CH3 groups introduced by TMA (Figure 7). In vacuo XPS experiments were performed to support this hypothesis. From Figure 8 (a) and (b) it can be seen that a single TMA pulse leads to a de-tectable amount of Al and C atoms on the SiO2-surface by XPS. This illustrates that XPS is suffi-ciently surface-sensitive to detect a monolayer of adsorbed TMA molecules, which has been shown previously by Geidel et al.33 From Figure 8 (c) it can be seen that O2-plasma removes all the car-bon from the surface. The Al2p signal is still visi-ble, so most likely the O2-plasma oxidizes the surface, removing the methyl groups and leaving behind oxidized Al atoms (Figure 7). If we now expose each of these three surfaces to RuO4 (Fig-ure 8), this leads to a detectable amount of Ru only in the case where methyl groups are initially present, which supports our hypothesis

Figure 7. The proposed surface reactions during the very first half cycle of the RuO4/H2-gas process on SiO2, after different surface pretreatments. (a) The fully oxidized SiO2-surface is unlikely to react with the RuO4-precursor. (b) Exposure of the SiO2-surface to TMA leads to adsorbed methyl groups, which can be oxidized by RuO4. (c) When the surface methyl groups introduced by TMA are removed by O2-plasma, the surface is again unlikely to react with the RuO4-precursor.

Ru Thin Film Properties. The thickness, RMS roughness, resistivity and impurity content of four thick Ru films deposited on planar Si-H and SiO2 substrates with different pretreatments are pre-sented in Table 1. Comparing the thickness of these films, it can be seen that 200 cycles of RuO4/H2 are needed on SiO2 to achieve the same

Ru film thickness as is obtained on Si-H after 75 cycles. We know that TMA treatment of SiO2 leads to an enhancement of the nucleation. However, if we compare the thickness of the Ru films de-posited on SiO2 + TMA and Si-H for the same number of cycles, we can conclude that the nu-cleation must still be better on Si-H. If TMA treat-ment of SiO2 is followed by O2-plasma, the thick-ness of the Ru film is lower compared to that on plain SiO2 for the same number of cycles. Next, if we compare the RMS roughness of these four Ru films, the films that nucleate most easily, Si-H and SiO2 + TMA, have the lowest roughness. The resistivities of Ru films grown on Si-H and SiO2 + TMA are similar, and have a sufficiently low value.4 Ex situ XPS depth profiling was used to determine the impurity content of the Ru films. In the bulk of the films, the only detected impurity was oxygen. From Table 1 it is evident that the bulk oxygen impurity content was low and does not depend on the nature of the starting surface. At the surface of all Ru films, fluorine was de-tected. Fluorine was also detected during the in vacuo XPS experiments which studied the nucle-ation regime. As no fluorine was detected in the bulk of the Ru films, this suggests that the fluo-rine remains at the top of the Ru film during growth. The fluorine is most likely deriving from the ToRuS solvent, and therefore the solvent might be involved in some way in the surface reactions.34 However, this needs further investiga-tion.

We can conclude that for thick Ru films de-posited by the RuO4 / H2 process, the material properties are more or less the same for all start-ing surfaces, except for the roughness. The low-est roughness is found for films that nucleate most easily: Si-H and TMA-treated SiO2, which can be expected.Table 1. Thickness Δ, RMS roughness R, electri-cal resistivity ρ and bulk atomic concentration of oxygen relative to ruthenium O of Ru thin films deposited on different surfaces.a

Substrate cycles Δ (nm)

R (nm)

ρ (µΩ.cm)

O (%)

Si-H 75 18.5 0.3 36 4SiO2 200 18.4 2.2 37 5SiO2 + TMA

75 13.3 0.4 37 4

SiO2 + TMA + O2*

200 12.5 2.2 22 5

aThese Ru films were obtained during the in situ SE experiments of which the data are shown in Fig-ure 1 and Figure 6.

In view of the potential application of Ru films in sub-10 nm interconnects, the resistivity of Ru films deposited on Si-H and TMA-treated SiO2 was also determined as a function of thickness (Figure

9). On Si-H no real trend can be observed with decreasing Ru film thickness, and a resistivity of ~40 µΩ.cm is obtained for a film with a thickness of only 3.6 nm. This result suggests that the Ru films are continuous early on during deposition, which is promising for applications and in agree-ment with our previous results.26 When depositing on TMA-treated SiO2, the resistivity increases with decreasing Ru film thickness below 5 nm. Most likely this is caused by the more difficult nucle-ation on SiO2 + TMA compared to Si-H. Although this larger resistivity might limit the applicability of this approach for sub-10 m interconnects, the Ru films are still continuous and one could try to improve the resistivity by optimizing the deposi-tion conditions or performing post deposition an-nealing.4

Figure 9. Resistivity as a function of thickness for Ru films deposited on Si-H and TMA-treated SiO2 sub-strates.

CONCLUSIONSThe RuO4/H2 process shows an inherent sub-

strate selectivity window of ~60 cycles, during which up to 15 nm of Ru can be grown on Si-H while avoiding growth on SiO2. In vacuo XPS ex-periments suggest that this results from the strong oxidative nature of the RuO4-precursor, which allows reaction with the Si-H surface during the first half-cycle, but not with SiO2. Using this inherent substrate selectivity, area-selective ALD was demonstrated on a Si/SiO2 patterned sub-strate. Based on these results and the established strong oxidative nature of RuO4, we expect that area-selective ALD with the RuO4/H2 process can be achieved more generally with other material systems, e.g. to achieve selective metal-on-metal growth on metal/dielectric patterned structures.24

Fully oxidized regions can be used to inhibit Ru growth, while Ru growth is achieved on regions which can be oxidized by RuO4.

Although the substrate selectivity is crucial for achieving area-selective ALD, some applications might require ALD deposition of a good quality, continuous Ru film on an oxidized surface. We showed by in situ SE experiments that the Ru nucleation delay on SiO2 can be eliminated by first exposing the surface to TMA. In vacuo XPS experiments suggest that the methyl groups on the TMA-exposed SiO2 surface by RuO4 are re-sponsible for this enhanced nucleation, most likely through combustion by RuO4. As most oxide surfaces react with TMA in a similar fashion, it is likely that this methodology presents a way to perform Ru metallization of virtually any oxide surface.

Figure 8. In vacuo XPS data showing the effect of a single RuO4-pulse on (a) SiO2, (b) SiO2 + TMA and (c) SiO2 + TMA + O2*. The pass energy setting of the XPS-spectrometer used to acquire each spectrum was 200 eV.

AUTHOR INFORMATIONCorresponding Author* E-mail: Jolien.dendooven@ugent.beAuthor ContributionsAll authors have given approval to the final version of the manuscript. Funding SourcesThis research was funded by Fonds Wetenschap-pelijk Onderzoek Vlaanderen (FWO Vlaanderen), the special research fund BOF at Ghent University (GOA 01G01513) and the Flemish Government (Medium-scale research infrastructure funding, Hercules fund-ing). M.M.M and J.D. received financial support through a personal FWO research grant. J.D. also received funding through the FWO “Krediet aan Na-vorsers” (1527916N).

ACKNOWLEDGMENT

The authors acknowledge: Lode Tassignon, Davy Deduytsche and Stefaan Broekaert for technical assistance with the construction of the in vacuo ALD-XPS setup; Lin-Lin Wang and Marc Schaekers for providing the patterned Si/SiO2 substrates; and Olivier Janssens for performing SEM measurements.

REFERENCES(1) Adelmann, C.; Wen, L. G.; Peter, A. P.; Siew, Y. K.;

Croes, K.; Swerts, J.; Popovici, M.; Sankaran, K.; Pour-tois, G.; Van Elshocht, S.; et al. Alternative Metals for Advanced Interconnects. 2014 IEEE Int. Interconnect Technol. Conf. / Adv. Met. Conf. IITC/AMC 2014 2014, 173–176.

(2) Wen, L. G.; Adelmann, C.; Pedreira, O. V.; Dutta, S.; Popovici, M.; Briggs, B.; Heylen, N.; Vanstreels, K.; Wilson, C. J.; Van Elshocht, S.; et al. Ruthenium Metal-lization for Advanced Interconnects. 2016 IEEE Int. Interconnect Technol. Conf. / Adv. Met. Conf. IITC/AMC 2016 2016, 34–36.

(3) Wen, L. G.; Roussel, P.; Pedreira, O. V.; Briggs, B.; Groven, B.; Dutta, S.; Popovici, M. I.; Heylen, N.; Ciofi, I.;

Vanstreels, K.; et al. Atomic Layer Deposition of Ruthe-nium with TiN Interface for Sub-10 Nm Advanced Inter-connects beyond Copper. ACS Appl. Mater. Interfaces 2016, 8, 26119–26125.

(4) Dutta, S.; Sankaran, K.; Moors, K.; Pourtois, G.; Van Elshocht, S.; Bömmels, J.; Vandervorst, W.; Tokei, Z.; Adelmann, C. Thickness Dependence of the Resistiv-ity of Platinum-Group Metal Thin Films. J. Appl. Phys. 2017, 122, 025107.

(5) Varela Pedreira, O.; Croes, K.; Lesniewska, A.; Wu, C.; Van Der Veen, M. H.; De Messemaeker, J.; Vandersmissen, K.; Jourdan, N.; Wen, L. G.; Adelmann, C.; et al. Reliability Study on Cobalt and Ruthenium as Alternative Metals for Advanced Interconnects. IEEE Int. Reliab. Phys. Symp. Proc. 2017, 6B2.1-6B2.8.

(6) Dutta, S.; Kundu, S.; Wen, L.; Jamieson, G.; Croes, K.; Gupta, A.; Bommels, J.; Wilson, C. J.; Adelmann, C.; Tokei, Z. Ruthenium Interconnects with 58 nm2 Cross-Section Area Using a Metal-Spacer Process. 2017 IEEE International Interconnect Technology Conference (IITC) 2017, 1–3.

(7) Dutta, S.; Kundu, S.; Gupta, A.; Jamieson, G.; Gomez Granados, J. F.; Boemmels, J.; Wilson, C. J.; Tokei, Z.; Adelmann, C. Highly Scaled Ruthenium Intercon-nects. IEEE Electron Device Lett. 2017, 38, 949-951.

(8) Dutta, S.; Moors, K.; Vandemaele, M.; Adelmann, C. Finite Size Effects in Highly Scaled Ruthenium Inter-connects. IEEE Electron Device Lett. 2018, 39, 268–271.

(9) Beyne, S.; Dutta, S.; Pedreira, O. V.; Bosman, N.; Adelmann, C.; De Wolf, I.; Tokei, Z.; Croes, K. The First Observation of P-Type Electromigration Failure in Full Ruthenium Interconnects. IEEE Int. Reliab. Phys. Symp. Proc. 2018, 2018–March, 6D.71-6D.79.

(10) Hu, C. K.; Kelly, J.; Huang, H.; Motoyama, K.; Shobha, H.; Ostrovski, Y.; Chen, J. H. C.; Patlolla, R.; Peethala, B.; Adusumilli, P.; et al. Future On-Chip Inter-connect Metallization and Electromigration. IEEE Int. Reliab. Phys. Symp. Proc. 2018, 2018–March, 4F.11-4F.16.

(11) Puurunen, R. L. Surface Chemistry of Atomic Layer Deposition: A Case Study for the Trimethylalu-minum/Water Process. J. Appl. Phys. 2005, 97, 121301.

(12) Miikkulainen, V.; Leskela, M.; Ritala, M.; Puu-runen, R. L. Crystallinity of Inorganic Films Grown by Atomic Layer Deposition: Overview and General Trends. J. Appl. Phys. 2013, 113 (2), 021301.

(13) Hamalainen, J.; Ritala, M.; Leskela, M. Atomic Layer Deposition of Noble Metals and Their Oxides. Chem. Mater. 2014, 26, 786–801.

(14) Mackus, A. J. M.; Bol, A. A.; Kessels, W. M. M. The Use of Atomic Layer Deposition in Advanced Nanopat-terning. Nanoscale 2014, 6, 10941-10960.

(15) Kalanyan, B.; Lemaire, P. C.; Atanasov, S. E.; Ritz, M. J.; Parsons, G. N. Using Hydrogen To Expand the Inherent Substrate Selectivity Window During Tungsten Atomic Layer Deposition. Chem. Mater. 2016, 28, 117–126.

(16) Lemaire, P. C.; King, M.; Parsons, G. N. Under-standing Inherent Substrate Selectivity during Atomic Layer Deposition: Effect of Surface Preparation, Hy-droxyl Density, and Metal Oxide Composition on Nucle-ation Mechanisms during Tungsten ALD. J. Chem. Phys. 2017, 146, 052811.

(17) Yang-Kyu Choi; Tsu-Jae King; Chenming Hu. A Spacer Patterning Technology for Nanoscale CMOS. IEEE Trans. Electron Devices 2002, 49, 436–441.

(18) Park, K. J.; Doub, J. M.; Gougousi, T.; Parsons, G. N. Microcontact Patterning of Ruthenium Gate Elec-

trodes by Selective Area Atomic Layer Deposition. Appl. Phys. Lett. 2005, 86 , 051903.

(19) Farm, E.; Kemell, M.; Ritala, M.; Leskela, M. Se-lective-Area Atomic Layer Deposition Using Poly ( Methyl Methacrylate ) Films as Mask Layers. J. Phys. Chem. 2008, 112, 15791–15795.

(20) Farm, E.; Kemell, M.; Santala, E.; Ritala, M.; Leskela, M. Selective-Area Atomic Layer Deposition Using Poly(Vinyl Pyrrolidone) as a Passivation Layer. J. Electrochem. Soc. 2010, 157, K10.

(21) Farm, E.; Lindroos, S.; Ritala, M.; Leskela, M. Microcontact Printed RuO x Film as an Activation Layer for Selective-Area Atomic Layer Deposition of Ruthe-nium. Chem. Mater. 2012, 24, 275–278.

(22) Zyulkov, I.; Krishtab, M.; De Gendt, S.; Armini, S. Selective Ru ALD as a Catalyst for Sub-Seven-Nanome-ter Bottom-Up Metal Interconnects. ACS Appl. Mater. Interfaces 2017, 9, 31031–31041.

(23) Junige, M.; Löffler, M.; Geidel, M.; Albert, M.; Bartha, J. W.; Zschech, E.; Rellinghaus, B.; Dorp, W. F. van. Area-Selective Atomic Layer Deposition of Ru on Electron-Beam-Written Pt(C) Patterns versus SiO2 Sub-stratum. Nanotechnology 2017, 28, 395301.

(24) Minjauw, M. M.; Dendooven, J.; Capon, B.; Schaekers, M.; Detavernier, C. Atomic Layer Deposition of Ruthenium at 100 °C Using the RuO4-Precursor and H2. J. Mater. Chem. C 2015, 3, 132–137.

(25) Minjauw, M. M.; Dendooven, J.; Capon, B.; Schaekers, M.; Detavernier, C. Near Room Temperature Plasma-Enhanced Atomic Layer Deposition of Ruthe-nium Using the RuO4-Precursor and H2-Plasma. J. Mater. Chem. C 2015, 3, 4848–4851.

(26) Dendooven, J.; Solano, E.; Minjauw, M. M.; Van de Kerckhove, K.; Coati, A.; Fonda, E.; Portale, G.; Gar-reau, Y.; Detavernier, C. Mobile Setup for Synchrotron Based in Situ Characterization during Thermal and Plasma-Enhanced Atomic Layer Deposition. Rev. Sci. Instrum. 2016, 87, 113905.

(27) NIST X-ray Photoelectron Spectroscopy Data-base, NIST Standard Reference Database Number 20, National Institute of Standards and Technology, Gaithersburg MD, 20899 (2000), doi: 10.18434/T4T88K, (retrieved 26th june 2018).

(28) Rijckaert, H.; Pollefeyt, G.; Sieger, M.; Hanisch, J.; Bennewitz, J.; De Keukeleere, K.; De Roo, J.; Hühne, R.; Backer, M.; Paturi, P.; et al. Optimizing Nanocomposites through Nanocrystal Surface Chemistry: Superconduct-ing YBa2Cu3O7 Thin Films via Low-Fluorine Metal Or-ganic Deposition and Preformed Metal Oxide Nanocrys-tals. Chem. Mater. 2017, 29, 6104–6113.

(29) Hattori, T. High Resolution X-Ray Photoemission Spectroscopy Studies of Thin SiO2 and Si/SiO2 Inter-faces. J. Vac. Sci. Technol. B Microelectron. Nanom. Struct. 1993, 11, 1528.

(30) Morgan, D. J. Resolving Ruthenium: XPS Studies of Common Ruthenium Materials. Surf. Interface Anal. 2015, 47, 1072–1079.

(31) Plietker, B. Selectivity versus Reactivity — Re-cent Advances in RuO4-Catalyzed Oxidations. ChemIn-form 2006, 37, 2453–2472.

(32) Piccialli, V. Ruthenium Tetroxide and Perruthen-ate Chemistry. Recent Advances and Related Transfor-mations Mediated by Other Transition Metal Oxo-Species. Molecules 2014, 19, 6534–6582.

(33) Geidel, M.; Knaut, M.; Albert, M.; Bartha, J. W. In Situ XPS Investigation of the Chemical Surface Compo-sition during the ALD of Ultra-Thin Aluminum Oxide Films. IEEE 2011 Semiconductor Conference Dresden, 2011, 1–4.

(34) Gatineau, J.; Yanagita, K.; Dussarrat, C. A New RuO4 Solvent Solution for Pure Ruthenium Film Deposi-tions. Microelectron. Eng. 2006, 83, 2248–2252.