PCCP View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
PAPER
Cite this: Phys. Chem. Chem. Phys., 2015, 17, 13895
View Journal | View Issue
Insights into the gas phase oxidation of Ru(0001) on the mesoscopic scale using molecular oxygen Jan C. Goritzka,†a Benjamin Herd,†a Philipp P. T. Krause,a Jens Falta,b J. Ingo Flegeb and Herbert Over*a We present an extensive mesoscale study of the initial gas phase oxidation of Ru(0001), employing in situ low-energy electron microscopy (LEEM), micro low-energy electron diffraction (m-LEED) and scanning tunneling microscopy (STM). The initial oxidation was investigated in a temperature range of 500–800 K at a constant oxygen pressure of p(O2) = 4 10
5
mbar. Depending to the preparation
temperature a dramatic change of the growth morphology of the RuO2 film was observed. At lower Received 22nd December 2014, Accepted 15th April 2015
temperature (580 K) the RuO2(110) film grows anisotropically oriented along the high symmetry directions of
DOI: 10.1039/c4cp06010e
which are slightly rotated by up to 201 with respect to the high symmetry direction. These rotated RuO2(110)
www.rsc.org/pccp
domains grow along slightly rotated step edges and reveal an isotropic growth morphology. Both the growth speed and the nucleation rate differ from that of the oxide growth at lower temperature (580 K).
the Ru(0001) substrate. At higher temperature (680 K), new rotational domains of RuO2(110) begin to appear,
1. Introduction Ruthenium is one of the best studied transition metals with remarkable catalytic properties, for both the oxidation and the reduction of molecules in heterogeneous and homogeneous catalysis.1–3 Under oxidizing reaction conditions ruthenium forms only a single stable surface oxide, namely RuO2 which revealed higher catalytic oxidation activity than the pure metallic surface, e.g. in oxidation of CO and HCl.4 During the last decade, the gas phase oxidation of Ru(0001) using molecular oxygen was investigated by a variety of typical surface science techniques including photoelectron emission microscopy (PEEM),5,6 scanning photoemission microscopy (SPEM),7,8 scanning tunneling microscopy (STM),9–11 surface X-ray diffraction (SXRD)12 and density functional theory (DFT).13,14 Despite this extensive work a comprehensive understanding of the oxidation process has not be accomplished. Different oxygen rich structures have been discussed to play a role in the initial oxidation process of Ru(0001), including a so called transient surface oxide (TSO),8 an O–Ru–O trilayer7,8,13 and subsurface oxygen.5,6,15 Recent STM studies indicated that the oxidation proceeds via a modified nucleation and growth mechanism above temperatures of 550 K.16 In this process first three-dimensional oxide clusters are formed at double and multiple steps of the Ru(0001) surface which act as nucleation a
Department of Physical Chemistry, Justus-Liebig-University, 35390 Giessen, Germany. E-mail: [email protected] b Institute of Solid State Physics, University of Bremen, 28359 Bremen, Germany † Both authors contributed equally to this study.
This journal is © the Owner Societies 2015
sites for the subsequent growth of flat RuO2 domains with its (110) orientation along the high symmetry direction of the ruthenium substrate. Combined STM and SPEM experiments on Ru(0001) had demonstrated that the oxide film grows faster in the step bunching regions than on flat terraces. The formation of the oxide clusters is accompanied by corrosion of the step edges pointing to the existence of a mobile RuOx-precursor species which is formed by the detachment of undercoordinated ruthenium atoms from defects (i.e. adatoms or atoms from steps or kink sites). Subsequently the RuOx-precursors species diffuse across the Ru-surface and agglomerate into the oxide clusters at the nucleation sites. Induced by the formation of the metal–oxygen bonds, ruthenium atoms are released from the metallic network by breaking the metal–metal bonds, a process which is thermally activated. Also the dissociative adsorption of oxygen from the gas phase beyond a coverage of 1 monolayer (1 ML) can become the rate determining step in the initial oxidation of Ru(0001).10 In situ surface X-ray diffraction measurements point to a Avrami-like kinetic for the oxidation of Ru(0001) surface which confirms a nucleation and growth type mechanism.12 Despite these new insights into the initial oxidation process at the atomic scale, the film morphology of the resulting oxide film is still only poorly understood at larger scales. Extensive PEEM and ¨ttcher et al. exhibited the formation of three SPEM studies by Bo distinct surface phases depending on the applied oxygen pressure and oxidation temperature.5–8 These three phases were denoted as the low temperature phase (o650 K, p(O2) = 10 5 mbar), the medium temperature phase (650–700 K, p(O2) = 10 5 mbar) and the high temperature phase (4700 K, p(O2) = 10 5 mbar).
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13895
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
Paper
The medium temperature phase was characterized by anisotropic needle-like patches attributed to RuO2 that is consistent with the typical growth mode of RuO2(110) on Ru(0001).7,10 Both, the low and high temperature phase, show a remarkably different morphology. The low temperature phase grows as small round patches which are randomly distributed across the surface and have been ascribed to the formation of an unidentified oxide precursor species.5,7 The high temperature phase is characterized by large disc-like patches that are assigned to the formation of RuO2.5 The origin of the temperature dependent formation of these three surface phases and their distinct morphology are less understood and cannot be explained by a simple faceting of a rough RuO2(110) film at higher temperature.18 We should recall that PEEM and SPEM experiments5–8 are not straightforward to interpret because PEEM (in the mode ¨ttcher et al.) is only sensitive to the change of the used by Bo work function of the sample surface, giving no further information of the chemical or structural nature of the formed surface structures. SPEM can only discriminate between the chemisorbed oxygen phase on Ru(0001) and RuO2 but is not able to explain the observed contrast variations observed in PEEM. To obtain a full picture of the oxidation process on the mesoscale we investigated the oxidation of Ru(0001) in a temperature range of 500 K to 800 K at a constant oxygen pressure. We employed low energy electron microscopy (LEEM), a technique that combines the advantages of real-time mapping on the mesoscale with atomic-scale structural information. We show that the morphological change from needle-like oxide patches to round disc-like islands (medium temperature phase to high temperature phase) can be explained by the appearance of additional, slightly rotated domains of RuO2(110). At 550 K after an induction time of several minutes only distinct needlelike domains of RuO2(110) grow along the high symmetry direction of the Ru(0001) surface, starting from double and multi-step edges. At 680 K new slightly rotated RuO2(110) domains arise which can be distinguished by a different phase contrast in LEEM and additional m-LEED measurements. The slightly rotated oxide domains of RuO2(110) differ in their morphology to the needle-like domains formed at 580 K. At 780 K only isotropic randomly distributed round-shaped oxide islands are observed in LEEM.
2. Experimental details The low energy electron microscopy (LEEM) and m-LEED experiments were performed in a commercial LEEM III system (Elmitec) installed at the University of Bremen, Germany, while the scanning tunneling microscopy (STM) experiments were conducted in a home-built three-chamber system equipped with a commercial STM (VT-STM Omicron) at the University of Giessen. All experiments were carried out on Ru single crystals in (0001) orientation (Mateck) and under UHV-conditions. Due to differences in the UHV-chamber setups the procedures for cleaning the Ru single crystals varied slightly. In both chambers
13896 | Phys. Chem. Chem. Phys., 2015, 17, 13895--13903
PCCP
the Ru(0001) crystals were cleaned by Ar-Ion sputtering for 10 min ( p(Ar) = 10 5 mbar) at room temperature followed by heating in an atmosphere of molecular oxygen p(O2) = 10 7 mbar to remove carbon contamination. To anneal the sample surfaces after the sputter treatment the crystal temperature was raised to 1700 K for several seconds in the LEEM-chamber system. In the STMchamber system the crystal was held at only 1050 K for 45 minutes, which was the highest temperature achievable in this setup. For this reason the sample surface was on average rougher in the STM experiments than in the LEEM experiments. In the LEEM-chamber the temperature was measured with a type K thermocouple fixed to the sample support of the sample holder. The sample temperature in the STM-chamber system was measured by an IR-Pyrometer which was previously calibrated against a type K thermocouple spot welded to the Ru crystal. For all in situ oxidation experiments a fixed oxygen pressure of p(O2) = 4 10 5 mbar was applied, while the sample temperature was varied from 500 K to 800 K. The STM images were taken ex situ at room temperature (B300 K). The LEED pattern shown in Fig. 9a was taken in a dedicated LEED chamber described elsewhere.19,20 In the LEEM apparatus, electrons are generated at a LaB6 cathode and are accelerated to 20 keV towards the sample. To separate the incoming and reflected electron beams a magnetic sector field deflects the electron beams onto and off the optical axis of the objective lens by 601. The objective lens then focuses the electron beam from the crossover in its back focal plane to infinity, and simultaneously the retarding potential between objective and sample decelerates it to the desired low energy so that low energy electron diffraction conditions apply. After reaccelerating to the operating voltage, the resulting electron diffraction pattern is found in the back focal plane of the objective lens, and it is passed on through the magnetic sector field into the imaging column. Depending on the mode of operation, either the electron diffraction pattern (LEED mode) or the real-space image obtained after an additional Fourier transform (LEEM mode) is magnified and projected onto the detector. This detector consists of image grade multi channel plates and a fluorescent screen. The images from the screen are recorded by a high-resolution CCD camera. In LEEM mode, bright field images are obtained by selecting the specular (00) beam of the LEED pattern for imaging with a suitable aperture. However, if a non-specular reflection is selected for image formation, this is referred to as dark field imaging. Here only the parts of the sample exhibiting the crystal structure associated with this diffraction spot show high intensity while other regions, in which the selected crystal structure is not present, stay dark. This method was employed to distinguish between different surface structures that appear similar in bright field mode, e.g. rotational domains of one crystallographic phase. The intensity of the selected reflection at each point on the surface is represented by a grayscale. Two contrast mechanisms are important for the interpretation of bright-field LEEM images. The first is amplitude contrast stemming from a difference in diffraction conditions between structures with different crystallographic or chemical
This journal is © the Owner Societies 2015
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
PCCP
properties. The second is phase contrast, which is produced by a path length difference of the incident and the reflected electron wave that leads to interference effects. The two major examples for this phenomenon are a path length difference due to steps between different terraces on the surface and a quantum size effect. The quantum size effect can occur with ultra thin films when the incident electron wave is reflected from both the surface of the thin film and the interface between the thin film and the substrate and both interfere. To correlate local real space structure observed in bright-field LEEM with its local reciprocal space structure we employed m-LEED to observe the local LEED pattern of morphological features bigger than 250 nm in diameter. In doing so we limited the size of the incident electron beam by introducing an illumination aperture and positioning it on a surface feature of interest from which the local diffraction pattern is then acquired.21
3. Experimental results 3.1
Oxidation at low temperature (500 K and 580 K)
In the following we focus on the initial oxidation of Ru(0001) at low temperatures. Fig. 1 displays a low energy electron microscopy (LEEM) image (a) and a low energy electron diffraction (LEED) pattern (b) acquired after exposure of p(O2) = 4 10 5 mbar molecular oxygen at 500 K for 58 min. Neither LEEM nor LEED indicate the growth of an oxide. The LEED pattern as well as additional (I/V)-LEED curves are consistent with a formation of a hexagonal, chemisorbed (1 1)O-adlayer on Ru(0001).22 A survey of the surface reveals that this saturated chemisorption phase covers the entire surface and no other surface phases were found at this temperature. This finding is consistent with a previous SXRD-study by He et al. who found a temperature threshold for oxidation of 550 K.12 Elevating the temperature to 580 K and dosing oxygen onto the Ru(0001) surface for 63 min leads to the results shown in Fig. 2a. Roughly more than half of the imaged surface is covered by dark stripes consistent with the growth morphology previously
Fig. 1 (a) Low energy electron microscopy image (LEEM, bright field) of a Ru(0001) surface after exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 500 K for 58 min (LEEM 8.4 eV, 20 mm). Three step bunching regions can clearly be observed at this surface position (left, top and right). The dark spot in the upper right corner of the image is an artifact of the detector. (b) Corresponding low energy electron diffraction (LEED) image after 58 min.
This journal is © the Owner Societies 2015
Paper
reported for thin films of RuO2(110) growing incommensurately with three rotational domains on the Ru(0001) substrate in this temperature range.12 Fig. 2b shows a LEED pattern obtained from the same surface area as was shown in Fig. 2a. Different diffraction spots belonging to the hexagonal Ru(0001) substrate (orange circles) and three rotational domains of RuO2(110) are visible. The obtained unit cells of the three rotational domains are marked yellow, red and blue respectively. Fig. 2c shows a m-LEED image of one oxide patch observed in Fig. 2a revealing a diffraction pattern of only one rotational domain of RuO2(110) (unit cell and corresponding diffraction spots are marked blue). From the m-LEED pattern we determine the size of the unit cell as (6.38 0.03) Å (3.12 0.03) Å. These values correspond closely to the bulk-truncated unit cell of RuO2(110), namely 6.38 Å 3.11 Å.23 The diffraction spots of the Ru(0001) substrate are still visible despite the covering RuO2(110) domain. Therefore, the presence of diffraction spots of the Ru(0001) substrate indicates a very thin oxide layer thickness (o3 layers) that was corroborated by recent STM experiments performed in our group revealing the formation of small ultra thin RuO2(110) islands with an estimated thickness of 2–3 layers for pressures in the low 10 5 mbar region.16 Fig. 3a–f show a sequence of LEEM images for various elapsed times of the oxidation process. Fig. 3a displays the Ru(0001) surface after an oxygen exposure time of 18 min. No stripe-like oxide domains are visible yet. However, dark spots that point to the onset of oxidation begin to form (marked by red circles), preferentially in the step bunching regions. With progressing exposure time, (27 min, Fig. 3b) small dark stripes become visible that can be assigned to RuO2(110) domains. It should be emphasized that the oxide growth exclusively starts at step edges of the surface, rendering these specific heterogeneities important for the oxidation process. From this observation it can be deduced that the dark spots seen in Fig. 3a are nucleation centers for the growth of RuO2(110) patches. No such dark spots are found to form on flat terraces, while regions with high defect concentrations such as step bunching regions (denoted as A and B in Fig. 3) reveal an accelerated onset of oxidation. This corroborates previous evidence for the importance of step edges in the initial oxidation process.16,17,24–27 Fig. 3c displays the Ru surface after 35 min of oxygen exposure. The step bunching areas marked as A and B are already completely covered by oxide domains. Yet in defect poor surface areas, with less step edges, oxide islands only now start to grow (indicated by red circles in Fig. 3c). In the later stages of oxidation (as shown in Fig. 3d and f) a more homogeneous distribution of nucleation centers and oxide patches can be observed over the entire surface area. The kinetics of nucleation and growth type mechanisms can often be described by Avrami-kinetics.28–32 Fig. 4 shows a schematic Avrami growth curve of RuO2(110) on Ru(0001) as a function of the exposure time of molecular oxygen as it was obtained by in situ SXRD measurements.12 In these SXRD measurements the area under a diffraction signal typical for RuO2(110), is proportional to the total amount of RuO2(110) present. The Avrami curve starts with an induction period where
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13897
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
Paper
PCCP
Fig. 2 (a) Low energy electron microscopy image (bright field) of the Ru(0001) surface after exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 580 K for 63 min (LEEM 8.4 eV, 10 mm). RuO2(110)-domains are visible as dark stripes. The substrate (grey) is covered by a chemisorbed (1 1)O-phase. (b) Obtained ‘‘global’’ LEED-image after preparation. Diffraction spots of three rotational domains (colored yellow, blue and red) and the hexagonal Ru(0001) substrate (orange) are visible. (c) m-LEED-image of one rotational domain of RuO2(110). Only diffraction spots of one rotational domain (blue) and underlying substrate (orange) are visible. Unit cell: (6.38 0.03) Å (3.12 0.03) Å. The bright spot in the upper left part of b and c is due to inelastically scattered electrons that are displaced from the central beam by the sector field.
Fig. 3 Observation of the initial oxidation during oxygen exposure by in situ LEEM (bright field): (a–e) low energy electron microscopy images obtained during the exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 580 K after (a) 18 min, (b) 27 min, (c) 35 min, (d) 47 min and (f) 63 min. (a–c, f) 10 mm, (d) 20 mm. All images were obtained at 8.4 eV. Step bunching regions are marked with A and B. (e) Zoom into region marked in (d) revealing primary orientation of steps.
no diffraction signal of RuO2(110) is visible. This regime of oxidation is characterized by diffusion of mobile RuOx-species that agglomerate at step edges forming stable oxidic clusters that act as nucleation centers for the RuO2(110) domain growth.
13898 | Phys. Chem. Chem. Phys., 2015, 17, 13895--13903
Both the mobile precursors and the clusters are invisible to SXRD and LEEM. After a few minutes of oxygen exposure the signal intensity increases exponentially because the oxide film starts growing autocatalytically34 and approaches an apparent
This journal is © the Owner Societies 2015
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
PCCP
Paper
saturation, which can be traced to a full lateral coverage of the surface with a RuO2(110). At this point the oxidation can only continue by increasing the film thickness and the oxidation rate slows down remarkably. Our in situ LEEM results obtained at 580 K (Fig. 3a–f) are in good agreement with this kind of Avrami-type growth model. For comparison all oxidation times of the obtained LEEM images shown in Fig. 3 are marked on the schematic Avrami curve in Fig. 4. One should remember that the basic Avrami theory only describes phase transformation in a solid volume with completely random nucleation and isotropic growth. It should be noted that for systems like ours, where these assumptions do not hold true, the theory has to be modified if one wants to do quantitative data analysis. After 18 min (Fig. 3a), first small, noticeable oxide patches appear exclusively in step-bunching areas, leading to oxide growth that is limited to a small area of the surface only, which is comparable to the induction time. After 35 min (Fig. 3c), oxidation also starts in flatter, defect-poor regions of the surface where many oxide patches appear in short succession, which leads to a strong increase of the oxide amount as characterized by the increase of the Avrami curve. In this case, the end of the induction time is marked by the simultaneous formation of oxide nuclei in defect-poor regions and not by the first oxide formation in step bunching areas. After 63 min (Fig. 3f), only roughly more than
Fig. 4 Schematic representation of the development of an SXRD-diffraction spot of RuO2(110) during expose of molecular oxygen. The increasing spot intensity indicates an Avrami-like growth behavior of RuO2.
Fig. 5
one half of the substrate is covered by an oxide layer, which is consistent with a still growing Avrami curve. In previous STM studies we found that oxide clusters are formed predominantly at double steps and in step bunching regions.17 Experimental evidence gathered during these studies points to the formation of a mobile RuOx-species at step edges, which then agglomerates into the oxidic clusters. This is consistent with the findings presented in the previous section although the oxidic clusters themselves are not directly accessible via LEEM due to their size being below the instruments resolution. This reliance on RuOx agglomeration at step edges has remarkable influence on the resulting film morphology. First the final oxide film consists of many individual oxide domains that have grown independently of one another. The location of the oxide domains is furthermore dependent on the underlying substrate morphology. Regions that exhibit a higher concentration of defects, such as step bunching regions are oxidized far sooner than other regions of the surface that have a lower concentration of defects. To better estimate the relative abundance of the three rotational domains of RuO2(110) dark field images were acquired of the surface region shown in Fig. 2a (Fig. 5a–c). Here a specific diffraction spot of one domain of RuO2(110) is chosen. This way all RuO2(110) patches orientated along this distinct crystallographic direction exhibit the same bright contrast, while the rest of the surface appears dark. The three rotational domains are energetically degenerate and should therefore exhibit the same abundance on the surface on the macroscale. As a comparison between Fig. 5a–c indicates, this assumption breaks down on the mesoscale and one rotational domain is clearly preferred. The favored direction of oxide patch orientation coincides with the predominant orientation of steps on this particular area of the surface as seen in Fig. 3e. In their study of the oxidation of Ru(0001) via PEEM, ¨ttcher et al. observed at temperatures below 550 K what they Bo called the low temperature phase of oxidation.5,7 With SPEM this phase was identified as a structure that was named the transient surface oxide consisting of a O–Ru–O trilayer with the oxygen between the first two Ru layers occupying octahedral and tetrahedral sites.8
Dark-field images of the three rotational domains as seen in Fig. 2a. Oxide domains which share a rotational orientation appear bright.
This journal is © the Owner Societies 2015
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13899
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
Paper
PCCP
In our low temperature oxidation experiment we did not observe the formation of any other species besides the (1 1)O–Ru(0001) chemisorbed phase. An incorporation of oxygen between the first and second layer of Ru and the formation of a transient surface oxide would have certainly led to a widening of the crystal lattice. This widening should be readily discernible by contrast change in LEEM, which however has not been observed. The low temperature oxidation experiments that we performed with STM however yielded the formation of a large number of oxidic clusters (o10 nm) that might be connected to the additional species found in the low temperature SPEM measurements.7,8 SPEM cannot resolve the oxide clusters so that the SPEM image averages oxide cluster with (1 1)O domains. With m-LEED we could clearly establish that the growing film already solely consists of RuO2(110) which is consistent with SPEM observations under similar oxidation conditions.7 3.2
Oxidation at high temperature (680 K and above)
Exposure of molecular oxygen at 680 K results in a pronounced alteration of the oxide morphology and the occurrence of an additional phase on the Ru(0001)-surface. Fig. 6a reveals three distinct surface phases (I, II, III) in the LEEM image after an oxygen exposure of 43 min at a constant oxygen pressure of p(O2) = 4 10 5 mbar. These surface phases are characterized by their micro-LEED patterns shown in Fig. 6b–d. The microLEED pattern of phase I (Fig. 6b) shows a sixfold symmetric diffraction pattern consistent with a (1 1)O-phase on hexagonal Ru(0001). The diffraction pattern of phase II (Fig. 6c) can easily be compared to Fig. 2c and thereby be identified as one of the known rotational domains of RuO2(110) that are oriented along the high symmetry directions of the substrate. Phase III displays the same diffraction pattern of a rotational domain of RuO2(110), but here the observed diffraction spots are rotated clockwise by 151 to those in phase II making it a previously unknown rotational domain of RuO2(110). Both RuO2(110) diffraction pattern of phase II and phase III reveal virtually identical unit cell dimensions of (6.41 0.03) Å (3.10 0.03) Å for phase II and (6.42 0.03) Å (3.11 0.03) Å for phase III that are close to the bulk-truncated unit cell, reported as (6.38 Å 3.11 Å).23 In both m-LEED images (Fig. 6c and d) it is noteworthy that again substrate reflections are discernible which help estimate the film thickness of the oxide patch to be below 3 ML. Fig. 6e displays a schematic representation of both observed phase II & III of RuO2(110). Panel (A) indicates the three known epitaxial and incommensurate rotational domains of RuO2(110) on Ru(0001). Panel (B) indicates the schematic real space unit cells of the 151 rotated RuO2(110) rotational domains on Ru(0001). These rotated RuO2(110) domains are also incommensurate to the substrate but fit in one direction to the hexagonal arrangement of Ru atoms of Ru(0001). A change in the interface between RuO2(110) and Ru(0001) could additionally explain the contrast difference between phase II and phase III RuO2(110) islands although the films have nearly the same structure. LEEM intensity, unlike traditional LEED, is very sensitive to the full electronic structure of the probed material. The electronic structure should in turn depend on the interface
13900 | Phys. Chem. Chem. Phys., 2015, 17, 13895--13903
Fig. 6 Low energy electron microscopy (bright field) and diffraction images of the Ru(0001) surface after exposure of molecular oxygen at p(O2) = 4 105 mbar and 680 K for 43 min (LEEM 8.4 eV, 10 mm). (a) The acquired LEEM pattern displays three different phases (I, II, III) grown on the Ru(0001) substrate. (b) m-LEED pattern of phase I. Hexagonal diffraction spots of a (1 1)O-phase are visible. (c) m-LEED pattern of phase II. Diffraction spots of one rotational domain of RuO2(110) are visible. (d) m-LEED image of phase III. The obtained LEED image also reveal diffraction spots of RuO2(110). The diffraction pattern is rotated 151 to the diffraction pattern of phase II. Both oxide phases show similar unit cells (e) schematic real space unit cells of the observed RuO2(110) phases II (A) and III (B) on Ru(0001).
between substrate and oxide, especially for the thin films observed here. The growth of slightly rotated RuO2(110) oxide domains has not been observed at lower temperature. Fig. 7 displays a set of LEEM images acquired during exposure 4 10 5 mbar of molecular oxygen at 680 K. The first image Fig. 7a shows the surface after 25 min of exposure. The three different phases I, II, III are visible in which phase I can directly be assigned to the chemisorbed (1 1)O-phase, covering the free Ru(0001) substrate. At first, isolated, small, round-shaped islands of phase III (rotated RuO2(110)) are formed predominantly on the surface. Additionally a large grey line of phase III is visible which coincides with an extended step bunching region of the Ru(0001) crystal. The preferred growth of the rotated RuO2(110) phase atop this line defect suggests a high nucleation rate in this region favored by a high concentration of
This journal is © the Owner Societies 2015
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
PCCP
Paper
Fig. 7 Low energy electron microscopy images obtained during the exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 680 K after (a) 25 min (b) 37 min and (c) 43 min. During the initial oxidation process two different phases of RuO2(110) arise on the surface. Phase III grows as round islands with no preferential growth direction while phase II still shows preference for certain growth directions (red arrows).
available defects. Also small islands of phase II are found to grow starting at the outer boundary of the small phase III islands. No isolated islands of phase II were found in this surface region. Fig. 7b displays the surface after oxygen exposure of 37 min. The rotated RuO2(110) domains appear to nucleate more rapidly, leading to a higher number of separate oxide patches. The not rotated RuO2(110) oxide domains (phase II) appear to nucleate slower but each oxide patch grows on average faster than those of phase III. While at lower temperatures the oxide grows exclusively along high symmetry directions of Ru(0001), at higher temperatures this preference is partially lost. Although phase II retains some resemblance of anisotropic film morphology, the rotated RuO2(110) oxide patches grow on average in all directions equally fast, leading to disc-like shape oxide patches. Fig. 7c displays the surface after an O2 exposure time of 43 min. Almost the entire surface is now covered by the oxide film. The significant differences in growth speed and nucleation rate between the rotated RuO2(110) and the not rotated RuO2(110) phases happen to balance each other out at the experimental conditions chosen so that they both cover nearly the same surface area at the end of the in situ oxidation. The formation of the rotated RuO2(110) phases only at higher temperatures suggests that for lower temperature the
growth mode is controlled by reaction kinetics. The reason for the observed preferred growth along the crystallographic axis at lower temperatures (580 K) could be that one or multiple elementary reaction steps (i.e. oxygen penetration into the substrate or the removal of Ru atoms) face lower activation barriers along the high symmetry directions of the Ru substrate. In Fig. 8a comparison of the shape of the resulting oxide domains of phase II grown at 580 K and 680 K is shown. While at 580 K the growth occurs mostly one-dimensionally along the main crystallographic direction of the Ru substrate (needle-like domains in Fig. 8a and b), additional wide oxide domains become visible at 680 K which grow equally fast perpendicular to the needle-like domains (Fig. 8c and d). Similar behavior was previously observed with NO2 as the oxidizing agent.27 To gain further insight into the formation of the slightly rotated oxide domains, a sample oxidized under high temperature conditions 740 K, p(O2) = 3 10 5 mbar was investigated with LEED and STM. The global LEED pattern (Fig. 9a) taken after 60 min of oxidation shows a rotational smearing of the (110) oxide related reflections that is consistent with a rotation of the RuO2(110) domain. Fig. 9b displays a STM image taken after 8 min of oxygen exposure. The colored RuO2 island is formed in an step bunching region of the Ru substrate. In addition to the
Fig. 8 Low energy electron microscopy images obtained during the exposure of molecular oxygen at p(O2) = 4 10 5 mbar at 580 K after a exposure time of 25 min (a) and 37 (b) and at 680 K after exposure time of 25 min (c) and 37 min (d). While at 580 K (a, b) the growth of RuO2 domains occur only one-dimensional along the main symmetry direction of the Ru(0001) (white arrows), at 680 K broad oxide islands appear which are characterized by broad growth perpendicular to the main growth directions.
This journal is © the Owner Societies 2015
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13901
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
Paper
PCCP
Fig. 9 (a) Obtained LEED image taken after 60 min at 740 K and p(O2) = 3 10 5 mbar. The LEED image shows the rotational smearing of the (110) typical oxide reflections that is consistent with a rotation of the RuO2(110) domain. (b) STM-images of RuO2(110) islands formed in a step bunching area at 740 K and p(O2) = 3 10 5 mbar after 8 min (taken in separate chamber after the identical oxidation treatment). The RuO2(110) islands reveal two different kinds of rotation domains (marked brown and pink). The step edges are covered by oxidic clusters that serve as nuclei for oxidation (marked with a green circle) (470 nm 470 nm, U = 1, 1 V, I = 1 A). The images (c) and (d) displays the atomically resolved surface of two different rotation domain (Obr rows are visible) (20 nm 20 nm, U = 1, 1 V, I = 1 A), while the domains A and B are rotated 1201 against each other, the domain C is rotated 201 against domain A.
oxide islands, oxidic clusters are formed predominantly at step edges (marked by a green circle in Fig. 9b). We were able to identify three distinct rotational domains in one single STM image. Two of them (colored pink, A and B) are rotated 1201 from one another and can be identified as the known incommensurate RuO2(110) phase. The two RuO2(110) domains marked with A are orientated along the step edges of the underlying Ru substrate, as it was found for the needle-like oxide domains grown at 580 K. The third oxide domain (C, colored brown) grows slightly rotated to A and B. This observation is corroborated by comparing the orientation of bridging oxygen (Obr) rows in atomically resolved STM images as shown in Fig. 9c and d revealing a rotation angle of about 201. Just like the domain A is aligned along the direction of a step edge of the substrate the same can be observed for the rotated oxide domain C which is orientated along (slightly rotated: 201) multi step edges of the Ru substrate (marked by an orange arrow in Fig. 9b). These observations inclined us to presume that the orientation of the rotated oxide domain is caused by rotated step edges of the Ru substrate. In this case the oxide would grow first along the slightly rotated step edges where highly undercoordinated Ru-atoms are present. Further growth into the Ru-terraces occurs without any noticeable direction dependent kinetic growth barrier. This would in turn explain the observed higher nucleation rate for the slightly rotated oxide domains, seeing as these kinked step edges show a lower coordination than normal step edges making them more active for agglomerating mobile Ru–O species. Fig. 10 displays the progress of initial oxidation at 780 K and with a constant oxygen pressure of p(O2) = 4 10 5 mbar. Here only round shaped oxide islands grow, exhibiting a complex multicomponent contrast. Phase I is again identified as Ru(0001)–(1 1)O. The assignment of the oxide phases via
13902 | Phys. Chem. Chem. Phys., 2015, 17, 13895--13903
Fig. 10 Low energy electron microscopy images obtained during the exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 780 K after (a) 20 min, (b) 60 min, all LEEM images 20 mm, 8.4 eV.
m-LEED however was rendered unfeasible due to the small size of individual sub-patches. Instead we relied on the phase assignment via (I/V)-LEEM data,33 which will be presented in a separate publication. These (I/V)-LEEM measurements identify the oxide patches to consist of a mixture of RuO2(110), RuO2(100) and RuO2(101). Again a distinct change in oxidation behavior of Ru(0001) can be observed. The formation of surface oxide patches is far slower at 780 K than it was observed for lower temperatures. Previous thermal desorption spectroscopy (TDS) experiments indicate an oxygen desorption signal at 750 K that has been attributed to the decomposition of oxide cluster.24 The facile decomposition of these nuclei at high temperatures explains the observed lower oxide nucleation rate. Furthermore, the anisotropy in oxide growth is lost under these oxidation conditions leading to round-shaped islands. This behavior is consistent with the change from the medium temperature phase to the high temperature phase as observed ¨ttcher et al.5,7 with PEEM and reported by Bo
This journal is © the Owner Societies 2015
View Article Online
PCCP
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
4. Summary and conclusion Ru(0001) was oxidized at different temperatures from 500 K to 780 K at a constant oxygen pressure p(O2) = 4 10 5 mbar for various exposure times. The oxidation process on the mesoscale was followed by in situ LEEM, LEED and STM. While at 500 K no oxide formation was encountered, small needle-like domains of RuO2(110) nucleate exclusively at double and multisteps of the Ru(0001) substrate at 580 K, pointing to an important role of step edges in the initial oxidation process. The three rotational domains of RuO2(110) observed at low oxidation temperature reveal a characteristic stripe-like morphology along the high symmetry directions of the underlying Ru(0001) substrate. The formation of RuO2(110) domains directly at the step edges is reconciled with a heterogeneous nucleation and growth oxidation mechanism that was established in a recent STM study.16 No other oxidic species besides rutile RuO2(110) (such as a subsurface oxygen species or the proposed O–Ru–O-trilayer phase) could be identified. Increasing the temperature to 680 K leads to a significant change in the morphology of the resulting RuO2(110) film. At this temperature additional domains of RuO2(110) appear which are slightly rotated to the previously observed three rotation domain at 580 K. The appearance of rotated RuO2(110) domains is closely related to the development of step edges which are slightly twisted against the high symmetry directions. While the characteristic RuO2(110) domains orientated along the high symmetry directions still show a preferential growth direction thus leading to a stripe-like morphology, this anisotropy is lost for the slightly rotated domains which grow isotropically into almost disk-like shape. The appearance of rotated RuO2 phase explains the ¨ttcher previously introduced high temperature phase of Bo observed in PEEM and SPEM.5,7 At 780 K the oxide film consists of big round-shaped oxide islands.
Acknowledgements Jan Goritzka thanks the Justus Liebig University for financial support in form of the JLU Graduiertenstipendium. We thank Daniel Langsdorf for fruitful discussions.
Notes and references 1 E. A. Seddon and K. R. Seddon, The Chemistry of Ruthenium, Elsevier, Amsterdam, The Netherlands, 1984. 2 J. F. Weaver, Chem. Rev., 2013, 113, 4164–4215. 3 H. Over, Chem. Rev., 2012, 112, 3356. 4 (a) H. Over, Y. D. Kim, A. P. Seitsonen, S. Wendt, E. Lundgren, M. Schmid, P. Varga, A. Morgante and G. Ertl, Science, 2000, 287, 1474–1476; (b) S. Zweidinger, J. P. Hofmann, O. Balmes, E. Lundgren and H. Over, J. Catal., 2010, 272, 169–175. ¨ttcher, B. Kreuzer, H. Conrad and H. Niehus, Surf. Sci., 5 A. Bo 2002, 504, 42–58. ¨ttcher, B. Kreuzer, H. Conrad and H. Niehus, Surf. Sci., 6 A. Bo 2000, 466, 811–820.
This journal is © the Owner Societies 2015
Paper
¨ttcher, U. Starke, H. Conrad, R. Blume, H. Niehus, 7 A. Bo J. Gregoratti, B. Kaulich, A. Barinov and M. Kiskinova, J. Chem. Phys., 2002, 117, 17. ¨ttcher, L. Aballe, 8 R. Blume, H. Niehus, H. Conrad, A. Bo L. Gregoratti, A. Barinov and M. Kiskinova, J. Phys. Chem. B, 2005, 109, 14052. 9 S. H. Kim and J. Wintterlin, J. Chem. Phys., 2009, 131, 064705. 10 H. Over, A. P. Seitsonen, E. Lundgren, M. Schmid and P. Varga, Surf. Sci., 2002, 515, 143. 11 H. Over, A. P. Seitsonen, M. Knapp, E. Lundgren, M. Schmid and P. Varga, ChemPhysChem, 2004, 5, 167. 12 Y. B. He, M. Knapp, E. Lundgren and H. Over, J. Phys. Chem. B, 2005, 109, 21825. 13 K. Reuter, C. Stampfl, M. V. Ganduglia-Pirovano and M. Scheffler, Chem. Phys. Lett., 2002, 352, 311–317. 14 K. Reuter and M. Scheffler, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 65, 035406. ¨ttcher and H. Niehus, J. Chem. Phys., 1999, 110, 15 A. Bo 3186–3195. 16 B. Herd, M. Knapp and H. Over, J. Phys. Chem. C, 2012, 116, 24649–24660. 17 P. Dudin, A. Barinov, L. Gregoratti, M. Kiskinova, A. Goriachko and H. Over, J. Electron Spectrosc. Relat. Phenom., 2008, 89, 166. ¨ffler, 18 J. Aßmann, D. Crihan, M. Knapp, E. Lundgren, E. Lo M. Muhler, V. Narkhede, H. Over, M. Schmid and P. Varga, Angew. Chem., Int. Ed., 2005, 116, 939–942. 19 H. Over, H. Bludau, M. Skottke-Klein, W. Moritz, G. Ertl and C. T. Campbell, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 45, 8638. 20 Y. D. Kim, Y. J. Zhu, S. Wendt, A. P. Seitsonen, S. Schwegmann, H. Bludau, H. Over, A. Morgante and K. Christmann, Phys. Rev. B: Condens. Matter Mater. Phys., 2000, 61(12), 8455–8461. 21 J. I. Flege, W.-X. Tanq and M. Altman, in Low-Energy Electron Microscopy, Characterization of Materials, ed. E. N. Kaufmann, John Wiley & Sons, Inc., 2nd edn, 2012, pp. 1808–1827. 22 C. Stampfl, S. Schwegmann, H. Over, M. Scheffler and G. Ertl, Phys. Rev. Lett., 1996, 77(16), 3371–3374. 23 Y. D. Kim, A. P. Seitsonen, S. Wendt, J. Wang, C. Fan, K. Jacobi, H. Over and G. Ertl, J. Phys. Chem. B, 2001, 105, 3752–3758. 24 B. Herd, J. C. Goritzka and H. Over, J. Phys. Chem. C, 2013, 117, 15148–15154. 25 B. Herd and H. Over, Surf. Sci., 2014, 622, 24–34. 26 J. I. Flege, J. Hrbek and P. Sutter, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 78, 165407. 27 J. I. Flege and P. Sutter, J. Phys.: Condens. Matter, 2009, 21, 314018. 28 A. N. Komogorov, Izv. Akad. Nauk. URSS, Ser. Mater., 1937, 3, 355. 29 W. A. Johnson and R. Mehl, Trans. AIME, 1939, 185, 416. 30 M. Avrami, J. Chem. Phys., 1939, 7(12), 1103–1112. 31 M. Avrami, J. Chem. Phys., 1940, 8(2), 212–224. 32 M. Avrami, J. Chem. Phys., 1941, 9(2), 177–184. 33 J. I. Flege and E. E. Krasovskii, Phys. Status Solidi RRL, 2014, 8, 463–477. 34 H. Over and A. P. Seitsonen, Science, 2002, 297, 2003–2004.
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13903
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
PAPER
Cite this: Phys. Chem. Chem. Phys., 2015, 17, 13895
View Journal | View Issue
Insights into the gas phase oxidation of Ru(0001) on the mesoscopic scale using molecular oxygen Jan C. Goritzka,†a Benjamin Herd,†a Philipp P. T. Krause,a Jens Falta,b J. Ingo Flegeb and Herbert Over*a We present an extensive mesoscale study of the initial gas phase oxidation of Ru(0001), employing in situ low-energy electron microscopy (LEEM), micro low-energy electron diffraction (m-LEED) and scanning tunneling microscopy (STM). The initial oxidation was investigated in a temperature range of 500–800 K at a constant oxygen pressure of p(O2) = 4 10
5
mbar. Depending to the preparation
temperature a dramatic change of the growth morphology of the RuO2 film was observed. At lower Received 22nd December 2014, Accepted 15th April 2015
temperature (580 K) the RuO2(110) film grows anisotropically oriented along the high symmetry directions of
DOI: 10.1039/c4cp06010e
which are slightly rotated by up to 201 with respect to the high symmetry direction. These rotated RuO2(110)
www.rsc.org/pccp
domains grow along slightly rotated step edges and reveal an isotropic growth morphology. Both the growth speed and the nucleation rate differ from that of the oxide growth at lower temperature (580 K).
the Ru(0001) substrate. At higher temperature (680 K), new rotational domains of RuO2(110) begin to appear,
1. Introduction Ruthenium is one of the best studied transition metals with remarkable catalytic properties, for both the oxidation and the reduction of molecules in heterogeneous and homogeneous catalysis.1–3 Under oxidizing reaction conditions ruthenium forms only a single stable surface oxide, namely RuO2 which revealed higher catalytic oxidation activity than the pure metallic surface, e.g. in oxidation of CO and HCl.4 During the last decade, the gas phase oxidation of Ru(0001) using molecular oxygen was investigated by a variety of typical surface science techniques including photoelectron emission microscopy (PEEM),5,6 scanning photoemission microscopy (SPEM),7,8 scanning tunneling microscopy (STM),9–11 surface X-ray diffraction (SXRD)12 and density functional theory (DFT).13,14 Despite this extensive work a comprehensive understanding of the oxidation process has not be accomplished. Different oxygen rich structures have been discussed to play a role in the initial oxidation process of Ru(0001), including a so called transient surface oxide (TSO),8 an O–Ru–O trilayer7,8,13 and subsurface oxygen.5,6,15 Recent STM studies indicated that the oxidation proceeds via a modified nucleation and growth mechanism above temperatures of 550 K.16 In this process first three-dimensional oxide clusters are formed at double and multiple steps of the Ru(0001) surface which act as nucleation a
Department of Physical Chemistry, Justus-Liebig-University, 35390 Giessen, Germany. E-mail: [email protected] b Institute of Solid State Physics, University of Bremen, 28359 Bremen, Germany † Both authors contributed equally to this study.
This journal is © the Owner Societies 2015
sites for the subsequent growth of flat RuO2 domains with its (110) orientation along the high symmetry direction of the ruthenium substrate. Combined STM and SPEM experiments on Ru(0001) had demonstrated that the oxide film grows faster in the step bunching regions than on flat terraces. The formation of the oxide clusters is accompanied by corrosion of the step edges pointing to the existence of a mobile RuOx-precursor species which is formed by the detachment of undercoordinated ruthenium atoms from defects (i.e. adatoms or atoms from steps or kink sites). Subsequently the RuOx-precursors species diffuse across the Ru-surface and agglomerate into the oxide clusters at the nucleation sites. Induced by the formation of the metal–oxygen bonds, ruthenium atoms are released from the metallic network by breaking the metal–metal bonds, a process which is thermally activated. Also the dissociative adsorption of oxygen from the gas phase beyond a coverage of 1 monolayer (1 ML) can become the rate determining step in the initial oxidation of Ru(0001).10 In situ surface X-ray diffraction measurements point to a Avrami-like kinetic for the oxidation of Ru(0001) surface which confirms a nucleation and growth type mechanism.12 Despite these new insights into the initial oxidation process at the atomic scale, the film morphology of the resulting oxide film is still only poorly understood at larger scales. Extensive PEEM and ¨ttcher et al. exhibited the formation of three SPEM studies by Bo distinct surface phases depending on the applied oxygen pressure and oxidation temperature.5–8 These three phases were denoted as the low temperature phase (o650 K, p(O2) = 10 5 mbar), the medium temperature phase (650–700 K, p(O2) = 10 5 mbar) and the high temperature phase (4700 K, p(O2) = 10 5 mbar).
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13895
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
Paper
The medium temperature phase was characterized by anisotropic needle-like patches attributed to RuO2 that is consistent with the typical growth mode of RuO2(110) on Ru(0001).7,10 Both, the low and high temperature phase, show a remarkably different morphology. The low temperature phase grows as small round patches which are randomly distributed across the surface and have been ascribed to the formation of an unidentified oxide precursor species.5,7 The high temperature phase is characterized by large disc-like patches that are assigned to the formation of RuO2.5 The origin of the temperature dependent formation of these three surface phases and their distinct morphology are less understood and cannot be explained by a simple faceting of a rough RuO2(110) film at higher temperature.18 We should recall that PEEM and SPEM experiments5–8 are not straightforward to interpret because PEEM (in the mode ¨ttcher et al.) is only sensitive to the change of the used by Bo work function of the sample surface, giving no further information of the chemical or structural nature of the formed surface structures. SPEM can only discriminate between the chemisorbed oxygen phase on Ru(0001) and RuO2 but is not able to explain the observed contrast variations observed in PEEM. To obtain a full picture of the oxidation process on the mesoscale we investigated the oxidation of Ru(0001) in a temperature range of 500 K to 800 K at a constant oxygen pressure. We employed low energy electron microscopy (LEEM), a technique that combines the advantages of real-time mapping on the mesoscale with atomic-scale structural information. We show that the morphological change from needle-like oxide patches to round disc-like islands (medium temperature phase to high temperature phase) can be explained by the appearance of additional, slightly rotated domains of RuO2(110). At 550 K after an induction time of several minutes only distinct needlelike domains of RuO2(110) grow along the high symmetry direction of the Ru(0001) surface, starting from double and multi-step edges. At 680 K new slightly rotated RuO2(110) domains arise which can be distinguished by a different phase contrast in LEEM and additional m-LEED measurements. The slightly rotated oxide domains of RuO2(110) differ in their morphology to the needle-like domains formed at 580 K. At 780 K only isotropic randomly distributed round-shaped oxide islands are observed in LEEM.
2. Experimental details The low energy electron microscopy (LEEM) and m-LEED experiments were performed in a commercial LEEM III system (Elmitec) installed at the University of Bremen, Germany, while the scanning tunneling microscopy (STM) experiments were conducted in a home-built three-chamber system equipped with a commercial STM (VT-STM Omicron) at the University of Giessen. All experiments were carried out on Ru single crystals in (0001) orientation (Mateck) and under UHV-conditions. Due to differences in the UHV-chamber setups the procedures for cleaning the Ru single crystals varied slightly. In both chambers
13896 | Phys. Chem. Chem. Phys., 2015, 17, 13895--13903
PCCP
the Ru(0001) crystals were cleaned by Ar-Ion sputtering for 10 min ( p(Ar) = 10 5 mbar) at room temperature followed by heating in an atmosphere of molecular oxygen p(O2) = 10 7 mbar to remove carbon contamination. To anneal the sample surfaces after the sputter treatment the crystal temperature was raised to 1700 K for several seconds in the LEEM-chamber system. In the STMchamber system the crystal was held at only 1050 K for 45 minutes, which was the highest temperature achievable in this setup. For this reason the sample surface was on average rougher in the STM experiments than in the LEEM experiments. In the LEEM-chamber the temperature was measured with a type K thermocouple fixed to the sample support of the sample holder. The sample temperature in the STM-chamber system was measured by an IR-Pyrometer which was previously calibrated against a type K thermocouple spot welded to the Ru crystal. For all in situ oxidation experiments a fixed oxygen pressure of p(O2) = 4 10 5 mbar was applied, while the sample temperature was varied from 500 K to 800 K. The STM images were taken ex situ at room temperature (B300 K). The LEED pattern shown in Fig. 9a was taken in a dedicated LEED chamber described elsewhere.19,20 In the LEEM apparatus, electrons are generated at a LaB6 cathode and are accelerated to 20 keV towards the sample. To separate the incoming and reflected electron beams a magnetic sector field deflects the electron beams onto and off the optical axis of the objective lens by 601. The objective lens then focuses the electron beam from the crossover in its back focal plane to infinity, and simultaneously the retarding potential between objective and sample decelerates it to the desired low energy so that low energy electron diffraction conditions apply. After reaccelerating to the operating voltage, the resulting electron diffraction pattern is found in the back focal plane of the objective lens, and it is passed on through the magnetic sector field into the imaging column. Depending on the mode of operation, either the electron diffraction pattern (LEED mode) or the real-space image obtained after an additional Fourier transform (LEEM mode) is magnified and projected onto the detector. This detector consists of image grade multi channel plates and a fluorescent screen. The images from the screen are recorded by a high-resolution CCD camera. In LEEM mode, bright field images are obtained by selecting the specular (00) beam of the LEED pattern for imaging with a suitable aperture. However, if a non-specular reflection is selected for image formation, this is referred to as dark field imaging. Here only the parts of the sample exhibiting the crystal structure associated with this diffraction spot show high intensity while other regions, in which the selected crystal structure is not present, stay dark. This method was employed to distinguish between different surface structures that appear similar in bright field mode, e.g. rotational domains of one crystallographic phase. The intensity of the selected reflection at each point on the surface is represented by a grayscale. Two contrast mechanisms are important for the interpretation of bright-field LEEM images. The first is amplitude contrast stemming from a difference in diffraction conditions between structures with different crystallographic or chemical
This journal is © the Owner Societies 2015
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
PCCP
properties. The second is phase contrast, which is produced by a path length difference of the incident and the reflected electron wave that leads to interference effects. The two major examples for this phenomenon are a path length difference due to steps between different terraces on the surface and a quantum size effect. The quantum size effect can occur with ultra thin films when the incident electron wave is reflected from both the surface of the thin film and the interface between the thin film and the substrate and both interfere. To correlate local real space structure observed in bright-field LEEM with its local reciprocal space structure we employed m-LEED to observe the local LEED pattern of morphological features bigger than 250 nm in diameter. In doing so we limited the size of the incident electron beam by introducing an illumination aperture and positioning it on a surface feature of interest from which the local diffraction pattern is then acquired.21
3. Experimental results 3.1
Oxidation at low temperature (500 K and 580 K)
In the following we focus on the initial oxidation of Ru(0001) at low temperatures. Fig. 1 displays a low energy electron microscopy (LEEM) image (a) and a low energy electron diffraction (LEED) pattern (b) acquired after exposure of p(O2) = 4 10 5 mbar molecular oxygen at 500 K for 58 min. Neither LEEM nor LEED indicate the growth of an oxide. The LEED pattern as well as additional (I/V)-LEED curves are consistent with a formation of a hexagonal, chemisorbed (1 1)O-adlayer on Ru(0001).22 A survey of the surface reveals that this saturated chemisorption phase covers the entire surface and no other surface phases were found at this temperature. This finding is consistent with a previous SXRD-study by He et al. who found a temperature threshold for oxidation of 550 K.12 Elevating the temperature to 580 K and dosing oxygen onto the Ru(0001) surface for 63 min leads to the results shown in Fig. 2a. Roughly more than half of the imaged surface is covered by dark stripes consistent with the growth morphology previously
Fig. 1 (a) Low energy electron microscopy image (LEEM, bright field) of a Ru(0001) surface after exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 500 K for 58 min (LEEM 8.4 eV, 20 mm). Three step bunching regions can clearly be observed at this surface position (left, top and right). The dark spot in the upper right corner of the image is an artifact of the detector. (b) Corresponding low energy electron diffraction (LEED) image after 58 min.
This journal is © the Owner Societies 2015
Paper
reported for thin films of RuO2(110) growing incommensurately with three rotational domains on the Ru(0001) substrate in this temperature range.12 Fig. 2b shows a LEED pattern obtained from the same surface area as was shown in Fig. 2a. Different diffraction spots belonging to the hexagonal Ru(0001) substrate (orange circles) and three rotational domains of RuO2(110) are visible. The obtained unit cells of the three rotational domains are marked yellow, red and blue respectively. Fig. 2c shows a m-LEED image of one oxide patch observed in Fig. 2a revealing a diffraction pattern of only one rotational domain of RuO2(110) (unit cell and corresponding diffraction spots are marked blue). From the m-LEED pattern we determine the size of the unit cell as (6.38 0.03) Å (3.12 0.03) Å. These values correspond closely to the bulk-truncated unit cell of RuO2(110), namely 6.38 Å 3.11 Å.23 The diffraction spots of the Ru(0001) substrate are still visible despite the covering RuO2(110) domain. Therefore, the presence of diffraction spots of the Ru(0001) substrate indicates a very thin oxide layer thickness (o3 layers) that was corroborated by recent STM experiments performed in our group revealing the formation of small ultra thin RuO2(110) islands with an estimated thickness of 2–3 layers for pressures in the low 10 5 mbar region.16 Fig. 3a–f show a sequence of LEEM images for various elapsed times of the oxidation process. Fig. 3a displays the Ru(0001) surface after an oxygen exposure time of 18 min. No stripe-like oxide domains are visible yet. However, dark spots that point to the onset of oxidation begin to form (marked by red circles), preferentially in the step bunching regions. With progressing exposure time, (27 min, Fig. 3b) small dark stripes become visible that can be assigned to RuO2(110) domains. It should be emphasized that the oxide growth exclusively starts at step edges of the surface, rendering these specific heterogeneities important for the oxidation process. From this observation it can be deduced that the dark spots seen in Fig. 3a are nucleation centers for the growth of RuO2(110) patches. No such dark spots are found to form on flat terraces, while regions with high defect concentrations such as step bunching regions (denoted as A and B in Fig. 3) reveal an accelerated onset of oxidation. This corroborates previous evidence for the importance of step edges in the initial oxidation process.16,17,24–27 Fig. 3c displays the Ru surface after 35 min of oxygen exposure. The step bunching areas marked as A and B are already completely covered by oxide domains. Yet in defect poor surface areas, with less step edges, oxide islands only now start to grow (indicated by red circles in Fig. 3c). In the later stages of oxidation (as shown in Fig. 3d and f) a more homogeneous distribution of nucleation centers and oxide patches can be observed over the entire surface area. The kinetics of nucleation and growth type mechanisms can often be described by Avrami-kinetics.28–32 Fig. 4 shows a schematic Avrami growth curve of RuO2(110) on Ru(0001) as a function of the exposure time of molecular oxygen as it was obtained by in situ SXRD measurements.12 In these SXRD measurements the area under a diffraction signal typical for RuO2(110), is proportional to the total amount of RuO2(110) present. The Avrami curve starts with an induction period where
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13897
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
Paper
PCCP
Fig. 2 (a) Low energy electron microscopy image (bright field) of the Ru(0001) surface after exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 580 K for 63 min (LEEM 8.4 eV, 10 mm). RuO2(110)-domains are visible as dark stripes. The substrate (grey) is covered by a chemisorbed (1 1)O-phase. (b) Obtained ‘‘global’’ LEED-image after preparation. Diffraction spots of three rotational domains (colored yellow, blue and red) and the hexagonal Ru(0001) substrate (orange) are visible. (c) m-LEED-image of one rotational domain of RuO2(110). Only diffraction spots of one rotational domain (blue) and underlying substrate (orange) are visible. Unit cell: (6.38 0.03) Å (3.12 0.03) Å. The bright spot in the upper left part of b and c is due to inelastically scattered electrons that are displaced from the central beam by the sector field.
Fig. 3 Observation of the initial oxidation during oxygen exposure by in situ LEEM (bright field): (a–e) low energy electron microscopy images obtained during the exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 580 K after (a) 18 min, (b) 27 min, (c) 35 min, (d) 47 min and (f) 63 min. (a–c, f) 10 mm, (d) 20 mm. All images were obtained at 8.4 eV. Step bunching regions are marked with A and B. (e) Zoom into region marked in (d) revealing primary orientation of steps.
no diffraction signal of RuO2(110) is visible. This regime of oxidation is characterized by diffusion of mobile RuOx-species that agglomerate at step edges forming stable oxidic clusters that act as nucleation centers for the RuO2(110) domain growth.
13898 | Phys. Chem. Chem. Phys., 2015, 17, 13895--13903
Both the mobile precursors and the clusters are invisible to SXRD and LEEM. After a few minutes of oxygen exposure the signal intensity increases exponentially because the oxide film starts growing autocatalytically34 and approaches an apparent
This journal is © the Owner Societies 2015
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
PCCP
Paper
saturation, which can be traced to a full lateral coverage of the surface with a RuO2(110). At this point the oxidation can only continue by increasing the film thickness and the oxidation rate slows down remarkably. Our in situ LEEM results obtained at 580 K (Fig. 3a–f) are in good agreement with this kind of Avrami-type growth model. For comparison all oxidation times of the obtained LEEM images shown in Fig. 3 are marked on the schematic Avrami curve in Fig. 4. One should remember that the basic Avrami theory only describes phase transformation in a solid volume with completely random nucleation and isotropic growth. It should be noted that for systems like ours, where these assumptions do not hold true, the theory has to be modified if one wants to do quantitative data analysis. After 18 min (Fig. 3a), first small, noticeable oxide patches appear exclusively in step-bunching areas, leading to oxide growth that is limited to a small area of the surface only, which is comparable to the induction time. After 35 min (Fig. 3c), oxidation also starts in flatter, defect-poor regions of the surface where many oxide patches appear in short succession, which leads to a strong increase of the oxide amount as characterized by the increase of the Avrami curve. In this case, the end of the induction time is marked by the simultaneous formation of oxide nuclei in defect-poor regions and not by the first oxide formation in step bunching areas. After 63 min (Fig. 3f), only roughly more than
Fig. 4 Schematic representation of the development of an SXRD-diffraction spot of RuO2(110) during expose of molecular oxygen. The increasing spot intensity indicates an Avrami-like growth behavior of RuO2.
Fig. 5
one half of the substrate is covered by an oxide layer, which is consistent with a still growing Avrami curve. In previous STM studies we found that oxide clusters are formed predominantly at double steps and in step bunching regions.17 Experimental evidence gathered during these studies points to the formation of a mobile RuOx-species at step edges, which then agglomerates into the oxidic clusters. This is consistent with the findings presented in the previous section although the oxidic clusters themselves are not directly accessible via LEEM due to their size being below the instruments resolution. This reliance on RuOx agglomeration at step edges has remarkable influence on the resulting film morphology. First the final oxide film consists of many individual oxide domains that have grown independently of one another. The location of the oxide domains is furthermore dependent on the underlying substrate morphology. Regions that exhibit a higher concentration of defects, such as step bunching regions are oxidized far sooner than other regions of the surface that have a lower concentration of defects. To better estimate the relative abundance of the three rotational domains of RuO2(110) dark field images were acquired of the surface region shown in Fig. 2a (Fig. 5a–c). Here a specific diffraction spot of one domain of RuO2(110) is chosen. This way all RuO2(110) patches orientated along this distinct crystallographic direction exhibit the same bright contrast, while the rest of the surface appears dark. The three rotational domains are energetically degenerate and should therefore exhibit the same abundance on the surface on the macroscale. As a comparison between Fig. 5a–c indicates, this assumption breaks down on the mesoscale and one rotational domain is clearly preferred. The favored direction of oxide patch orientation coincides with the predominant orientation of steps on this particular area of the surface as seen in Fig. 3e. In their study of the oxidation of Ru(0001) via PEEM, ¨ttcher et al. observed at temperatures below 550 K what they Bo called the low temperature phase of oxidation.5,7 With SPEM this phase was identified as a structure that was named the transient surface oxide consisting of a O–Ru–O trilayer with the oxygen between the first two Ru layers occupying octahedral and tetrahedral sites.8
Dark-field images of the three rotational domains as seen in Fig. 2a. Oxide domains which share a rotational orientation appear bright.
This journal is © the Owner Societies 2015
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13899
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
Paper
PCCP
In our low temperature oxidation experiment we did not observe the formation of any other species besides the (1 1)O–Ru(0001) chemisorbed phase. An incorporation of oxygen between the first and second layer of Ru and the formation of a transient surface oxide would have certainly led to a widening of the crystal lattice. This widening should be readily discernible by contrast change in LEEM, which however has not been observed. The low temperature oxidation experiments that we performed with STM however yielded the formation of a large number of oxidic clusters (o10 nm) that might be connected to the additional species found in the low temperature SPEM measurements.7,8 SPEM cannot resolve the oxide clusters so that the SPEM image averages oxide cluster with (1 1)O domains. With m-LEED we could clearly establish that the growing film already solely consists of RuO2(110) which is consistent with SPEM observations under similar oxidation conditions.7 3.2
Oxidation at high temperature (680 K and above)
Exposure of molecular oxygen at 680 K results in a pronounced alteration of the oxide morphology and the occurrence of an additional phase on the Ru(0001)-surface. Fig. 6a reveals three distinct surface phases (I, II, III) in the LEEM image after an oxygen exposure of 43 min at a constant oxygen pressure of p(O2) = 4 10 5 mbar. These surface phases are characterized by their micro-LEED patterns shown in Fig. 6b–d. The microLEED pattern of phase I (Fig. 6b) shows a sixfold symmetric diffraction pattern consistent with a (1 1)O-phase on hexagonal Ru(0001). The diffraction pattern of phase II (Fig. 6c) can easily be compared to Fig. 2c and thereby be identified as one of the known rotational domains of RuO2(110) that are oriented along the high symmetry directions of the substrate. Phase III displays the same diffraction pattern of a rotational domain of RuO2(110), but here the observed diffraction spots are rotated clockwise by 151 to those in phase II making it a previously unknown rotational domain of RuO2(110). Both RuO2(110) diffraction pattern of phase II and phase III reveal virtually identical unit cell dimensions of (6.41 0.03) Å (3.10 0.03) Å for phase II and (6.42 0.03) Å (3.11 0.03) Å for phase III that are close to the bulk-truncated unit cell, reported as (6.38 Å 3.11 Å).23 In both m-LEED images (Fig. 6c and d) it is noteworthy that again substrate reflections are discernible which help estimate the film thickness of the oxide patch to be below 3 ML. Fig. 6e displays a schematic representation of both observed phase II & III of RuO2(110). Panel (A) indicates the three known epitaxial and incommensurate rotational domains of RuO2(110) on Ru(0001). Panel (B) indicates the schematic real space unit cells of the 151 rotated RuO2(110) rotational domains on Ru(0001). These rotated RuO2(110) domains are also incommensurate to the substrate but fit in one direction to the hexagonal arrangement of Ru atoms of Ru(0001). A change in the interface between RuO2(110) and Ru(0001) could additionally explain the contrast difference between phase II and phase III RuO2(110) islands although the films have nearly the same structure. LEEM intensity, unlike traditional LEED, is very sensitive to the full electronic structure of the probed material. The electronic structure should in turn depend on the interface
13900 | Phys. Chem. Chem. Phys., 2015, 17, 13895--13903
Fig. 6 Low energy electron microscopy (bright field) and diffraction images of the Ru(0001) surface after exposure of molecular oxygen at p(O2) = 4 105 mbar and 680 K for 43 min (LEEM 8.4 eV, 10 mm). (a) The acquired LEEM pattern displays three different phases (I, II, III) grown on the Ru(0001) substrate. (b) m-LEED pattern of phase I. Hexagonal diffraction spots of a (1 1)O-phase are visible. (c) m-LEED pattern of phase II. Diffraction spots of one rotational domain of RuO2(110) are visible. (d) m-LEED image of phase III. The obtained LEED image also reveal diffraction spots of RuO2(110). The diffraction pattern is rotated 151 to the diffraction pattern of phase II. Both oxide phases show similar unit cells (e) schematic real space unit cells of the observed RuO2(110) phases II (A) and III (B) on Ru(0001).
between substrate and oxide, especially for the thin films observed here. The growth of slightly rotated RuO2(110) oxide domains has not been observed at lower temperature. Fig. 7 displays a set of LEEM images acquired during exposure 4 10 5 mbar of molecular oxygen at 680 K. The first image Fig. 7a shows the surface after 25 min of exposure. The three different phases I, II, III are visible in which phase I can directly be assigned to the chemisorbed (1 1)O-phase, covering the free Ru(0001) substrate. At first, isolated, small, round-shaped islands of phase III (rotated RuO2(110)) are formed predominantly on the surface. Additionally a large grey line of phase III is visible which coincides with an extended step bunching region of the Ru(0001) crystal. The preferred growth of the rotated RuO2(110) phase atop this line defect suggests a high nucleation rate in this region favored by a high concentration of
This journal is © the Owner Societies 2015
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
PCCP
Paper
Fig. 7 Low energy electron microscopy images obtained during the exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 680 K after (a) 25 min (b) 37 min and (c) 43 min. During the initial oxidation process two different phases of RuO2(110) arise on the surface. Phase III grows as round islands with no preferential growth direction while phase II still shows preference for certain growth directions (red arrows).
available defects. Also small islands of phase II are found to grow starting at the outer boundary of the small phase III islands. No isolated islands of phase II were found in this surface region. Fig. 7b displays the surface after oxygen exposure of 37 min. The rotated RuO2(110) domains appear to nucleate more rapidly, leading to a higher number of separate oxide patches. The not rotated RuO2(110) oxide domains (phase II) appear to nucleate slower but each oxide patch grows on average faster than those of phase III. While at lower temperatures the oxide grows exclusively along high symmetry directions of Ru(0001), at higher temperatures this preference is partially lost. Although phase II retains some resemblance of anisotropic film morphology, the rotated RuO2(110) oxide patches grow on average in all directions equally fast, leading to disc-like shape oxide patches. Fig. 7c displays the surface after an O2 exposure time of 43 min. Almost the entire surface is now covered by the oxide film. The significant differences in growth speed and nucleation rate between the rotated RuO2(110) and the not rotated RuO2(110) phases happen to balance each other out at the experimental conditions chosen so that they both cover nearly the same surface area at the end of the in situ oxidation. The formation of the rotated RuO2(110) phases only at higher temperatures suggests that for lower temperature the
growth mode is controlled by reaction kinetics. The reason for the observed preferred growth along the crystallographic axis at lower temperatures (580 K) could be that one or multiple elementary reaction steps (i.e. oxygen penetration into the substrate or the removal of Ru atoms) face lower activation barriers along the high symmetry directions of the Ru substrate. In Fig. 8a comparison of the shape of the resulting oxide domains of phase II grown at 580 K and 680 K is shown. While at 580 K the growth occurs mostly one-dimensionally along the main crystallographic direction of the Ru substrate (needle-like domains in Fig. 8a and b), additional wide oxide domains become visible at 680 K which grow equally fast perpendicular to the needle-like domains (Fig. 8c and d). Similar behavior was previously observed with NO2 as the oxidizing agent.27 To gain further insight into the formation of the slightly rotated oxide domains, a sample oxidized under high temperature conditions 740 K, p(O2) = 3 10 5 mbar was investigated with LEED and STM. The global LEED pattern (Fig. 9a) taken after 60 min of oxidation shows a rotational smearing of the (110) oxide related reflections that is consistent with a rotation of the RuO2(110) domain. Fig. 9b displays a STM image taken after 8 min of oxygen exposure. The colored RuO2 island is formed in an step bunching region of the Ru substrate. In addition to the
Fig. 8 Low energy electron microscopy images obtained during the exposure of molecular oxygen at p(O2) = 4 10 5 mbar at 580 K after a exposure time of 25 min (a) and 37 (b) and at 680 K after exposure time of 25 min (c) and 37 min (d). While at 580 K (a, b) the growth of RuO2 domains occur only one-dimensional along the main symmetry direction of the Ru(0001) (white arrows), at 680 K broad oxide islands appear which are characterized by broad growth perpendicular to the main growth directions.
This journal is © the Owner Societies 2015
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13901
View Article Online
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
Paper
PCCP
Fig. 9 (a) Obtained LEED image taken after 60 min at 740 K and p(O2) = 3 10 5 mbar. The LEED image shows the rotational smearing of the (110) typical oxide reflections that is consistent with a rotation of the RuO2(110) domain. (b) STM-images of RuO2(110) islands formed in a step bunching area at 740 K and p(O2) = 3 10 5 mbar after 8 min (taken in separate chamber after the identical oxidation treatment). The RuO2(110) islands reveal two different kinds of rotation domains (marked brown and pink). The step edges are covered by oxidic clusters that serve as nuclei for oxidation (marked with a green circle) (470 nm 470 nm, U = 1, 1 V, I = 1 A). The images (c) and (d) displays the atomically resolved surface of two different rotation domain (Obr rows are visible) (20 nm 20 nm, U = 1, 1 V, I = 1 A), while the domains A and B are rotated 1201 against each other, the domain C is rotated 201 against domain A.
oxide islands, oxidic clusters are formed predominantly at step edges (marked by a green circle in Fig. 9b). We were able to identify three distinct rotational domains in one single STM image. Two of them (colored pink, A and B) are rotated 1201 from one another and can be identified as the known incommensurate RuO2(110) phase. The two RuO2(110) domains marked with A are orientated along the step edges of the underlying Ru substrate, as it was found for the needle-like oxide domains grown at 580 K. The third oxide domain (C, colored brown) grows slightly rotated to A and B. This observation is corroborated by comparing the orientation of bridging oxygen (Obr) rows in atomically resolved STM images as shown in Fig. 9c and d revealing a rotation angle of about 201. Just like the domain A is aligned along the direction of a step edge of the substrate the same can be observed for the rotated oxide domain C which is orientated along (slightly rotated: 201) multi step edges of the Ru substrate (marked by an orange arrow in Fig. 9b). These observations inclined us to presume that the orientation of the rotated oxide domain is caused by rotated step edges of the Ru substrate. In this case the oxide would grow first along the slightly rotated step edges where highly undercoordinated Ru-atoms are present. Further growth into the Ru-terraces occurs without any noticeable direction dependent kinetic growth barrier. This would in turn explain the observed higher nucleation rate for the slightly rotated oxide domains, seeing as these kinked step edges show a lower coordination than normal step edges making them more active for agglomerating mobile Ru–O species. Fig. 10 displays the progress of initial oxidation at 780 K and with a constant oxygen pressure of p(O2) = 4 10 5 mbar. Here only round shaped oxide islands grow, exhibiting a complex multicomponent contrast. Phase I is again identified as Ru(0001)–(1 1)O. The assignment of the oxide phases via
13902 | Phys. Chem. Chem. Phys., 2015, 17, 13895--13903
Fig. 10 Low energy electron microscopy images obtained during the exposure of molecular oxygen at p(O2) = 4 10 5 mbar and 780 K after (a) 20 min, (b) 60 min, all LEEM images 20 mm, 8.4 eV.
m-LEED however was rendered unfeasible due to the small size of individual sub-patches. Instead we relied on the phase assignment via (I/V)-LEEM data,33 which will be presented in a separate publication. These (I/V)-LEEM measurements identify the oxide patches to consist of a mixture of RuO2(110), RuO2(100) and RuO2(101). Again a distinct change in oxidation behavior of Ru(0001) can be observed. The formation of surface oxide patches is far slower at 780 K than it was observed for lower temperatures. Previous thermal desorption spectroscopy (TDS) experiments indicate an oxygen desorption signal at 750 K that has been attributed to the decomposition of oxide cluster.24 The facile decomposition of these nuclei at high temperatures explains the observed lower oxide nucleation rate. Furthermore, the anisotropy in oxide growth is lost under these oxidation conditions leading to round-shaped islands. This behavior is consistent with the change from the medium temperature phase to the high temperature phase as observed ¨ttcher et al.5,7 with PEEM and reported by Bo
This journal is © the Owner Societies 2015
View Article Online
PCCP
Published on 20 April 2015. Downloaded by East Carolina University on 14/09/2015 20:08:56.
4. Summary and conclusion Ru(0001) was oxidized at different temperatures from 500 K to 780 K at a constant oxygen pressure p(O2) = 4 10 5 mbar for various exposure times. The oxidation process on the mesoscale was followed by in situ LEEM, LEED and STM. While at 500 K no oxide formation was encountered, small needle-like domains of RuO2(110) nucleate exclusively at double and multisteps of the Ru(0001) substrate at 580 K, pointing to an important role of step edges in the initial oxidation process. The three rotational domains of RuO2(110) observed at low oxidation temperature reveal a characteristic stripe-like morphology along the high symmetry directions of the underlying Ru(0001) substrate. The formation of RuO2(110) domains directly at the step edges is reconciled with a heterogeneous nucleation and growth oxidation mechanism that was established in a recent STM study.16 No other oxidic species besides rutile RuO2(110) (such as a subsurface oxygen species or the proposed O–Ru–O-trilayer phase) could be identified. Increasing the temperature to 680 K leads to a significant change in the morphology of the resulting RuO2(110) film. At this temperature additional domains of RuO2(110) appear which are slightly rotated to the previously observed three rotation domain at 580 K. The appearance of rotated RuO2(110) domains is closely related to the development of step edges which are slightly twisted against the high symmetry directions. While the characteristic RuO2(110) domains orientated along the high symmetry directions still show a preferential growth direction thus leading to a stripe-like morphology, this anisotropy is lost for the slightly rotated domains which grow isotropically into almost disk-like shape. The appearance of rotated RuO2 phase explains the ¨ttcher previously introduced high temperature phase of Bo observed in PEEM and SPEM.5,7 At 780 K the oxide film consists of big round-shaped oxide islands.
Acknowledgements Jan Goritzka thanks the Justus Liebig University for financial support in form of the JLU Graduiertenstipendium. We thank Daniel Langsdorf for fruitful discussions.
Notes and references 1 E. A. Seddon and K. R. Seddon, The Chemistry of Ruthenium, Elsevier, Amsterdam, The Netherlands, 1984. 2 J. F. Weaver, Chem. Rev., 2013, 113, 4164–4215. 3 H. Over, Chem. Rev., 2012, 112, 3356. 4 (a) H. Over, Y. D. Kim, A. P. Seitsonen, S. Wendt, E. Lundgren, M. Schmid, P. Varga, A. Morgante and G. Ertl, Science, 2000, 287, 1474–1476; (b) S. Zweidinger, J. P. Hofmann, O. Balmes, E. Lundgren and H. Over, J. Catal., 2010, 272, 169–175. ¨ttcher, B. Kreuzer, H. Conrad and H. Niehus, Surf. Sci., 5 A. Bo 2002, 504, 42–58. ¨ttcher, B. Kreuzer, H. Conrad and H. Niehus, Surf. Sci., 6 A. Bo 2000, 466, 811–820.
This journal is © the Owner Societies 2015
Paper
¨ttcher, U. Starke, H. Conrad, R. Blume, H. Niehus, 7 A. Bo J. Gregoratti, B. Kaulich, A. Barinov and M. Kiskinova, J. Chem. Phys., 2002, 117, 17. ¨ttcher, L. Aballe, 8 R. Blume, H. Niehus, H. Conrad, A. Bo L. Gregoratti, A. Barinov and M. Kiskinova, J. Phys. Chem. B, 2005, 109, 14052. 9 S. H. Kim and J. Wintterlin, J. Chem. Phys., 2009, 131, 064705. 10 H. Over, A. P. Seitsonen, E. Lundgren, M. Schmid and P. Varga, Surf. Sci., 2002, 515, 143. 11 H. Over, A. P. Seitsonen, M. Knapp, E. Lundgren, M. Schmid and P. Varga, ChemPhysChem, 2004, 5, 167. 12 Y. B. He, M. Knapp, E. Lundgren and H. Over, J. Phys. Chem. B, 2005, 109, 21825. 13 K. Reuter, C. Stampfl, M. V. Ganduglia-Pirovano and M. Scheffler, Chem. Phys. Lett., 2002, 352, 311–317. 14 K. Reuter and M. Scheffler, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 65, 035406. ¨ttcher and H. Niehus, J. Chem. Phys., 1999, 110, 15 A. Bo 3186–3195. 16 B. Herd, M. Knapp and H. Over, J. Phys. Chem. C, 2012, 116, 24649–24660. 17 P. Dudin, A. Barinov, L. Gregoratti, M. Kiskinova, A. Goriachko and H. Over, J. Electron Spectrosc. Relat. Phenom., 2008, 89, 166. ¨ffler, 18 J. Aßmann, D. Crihan, M. Knapp, E. Lundgren, E. Lo M. Muhler, V. Narkhede, H. Over, M. Schmid and P. Varga, Angew. Chem., Int. Ed., 2005, 116, 939–942. 19 H. Over, H. Bludau, M. Skottke-Klein, W. Moritz, G. Ertl and C. T. Campbell, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 45, 8638. 20 Y. D. Kim, Y. J. Zhu, S. Wendt, A. P. Seitsonen, S. Schwegmann, H. Bludau, H. Over, A. Morgante and K. Christmann, Phys. Rev. B: Condens. Matter Mater. Phys., 2000, 61(12), 8455–8461. 21 J. I. Flege, W.-X. Tanq and M. Altman, in Low-Energy Electron Microscopy, Characterization of Materials, ed. E. N. Kaufmann, John Wiley & Sons, Inc., 2nd edn, 2012, pp. 1808–1827. 22 C. Stampfl, S. Schwegmann, H. Over, M. Scheffler and G. Ertl, Phys. Rev. Lett., 1996, 77(16), 3371–3374. 23 Y. D. Kim, A. P. Seitsonen, S. Wendt, J. Wang, C. Fan, K. Jacobi, H. Over and G. Ertl, J. Phys. Chem. B, 2001, 105, 3752–3758. 24 B. Herd, J. C. Goritzka and H. Over, J. Phys. Chem. C, 2013, 117, 15148–15154. 25 B. Herd and H. Over, Surf. Sci., 2014, 622, 24–34. 26 J. I. Flege, J. Hrbek and P. Sutter, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 78, 165407. 27 J. I. Flege and P. Sutter, J. Phys.: Condens. Matter, 2009, 21, 314018. 28 A. N. Komogorov, Izv. Akad. Nauk. URSS, Ser. Mater., 1937, 3, 355. 29 W. A. Johnson and R. Mehl, Trans. AIME, 1939, 185, 416. 30 M. Avrami, J. Chem. Phys., 1939, 7(12), 1103–1112. 31 M. Avrami, J. Chem. Phys., 1940, 8(2), 212–224. 32 M. Avrami, J. Chem. Phys., 1941, 9(2), 177–184. 33 J. I. Flege and E. E. Krasovskii, Phys. Status Solidi RRL, 2014, 8, 463–477. 34 H. Over and A. P. Seitsonen, Science, 2002, 297, 2003–2004.
Phys. Chem. Chem. Phys., 2015, 17, 13895--13903 | 13903
