1 Introduction

Chemisorbed aldehyde has been proposed as key intermediate, an important surface and gas-phase product in the decomposition of primary alcohols (e.g.: catalytic steam reforming of ethanol) on metal surfaces [1,2,3,4,5,6]. Besides the environmental protection aspects [7, 8] this is the motivation which places the acetaldehyde chemistry in the focus of catalytic community. Noble metals, including Rh, proved to be excellent catalysts for the dehydrogenation of ethanol [9,10,11,12,13].

It is generally accepted that acetaldehyde bound in an η2(C,O) configuration on unsupported well-defined Pt metal surfaces [14,15,16,17]. η2-CH3CHOa decomposes to adsorbed products; CH2COa, CH3, CH2, CH, Ha, COa and carbon. CO and H2 are the final products in the gas phase. Gold-based classical high surface area catalysts have widely investigated for oxidation and hydrogenation reactions, but fundamental studies on model catalysts are lacking. Acetaldehyde partial oxidation on the Au(111) model catalyst surface via C–C bond activation occurs and formation of methyl acetate as an oxidative coupling product forms. Hydrogenation of aldehyde to ethanol was attempted by temperature programmed desorption (TPD) and density functional theory calculation (DFT) on Au(111) [18,19,20].

In the present work we aimed to investigate the surface chemistry of acetaldehyde on well-defined gold nanoparticles. It is well established that Au clusters supported on various metal oxides show strong catalytic activity for a wide variety of reactions. The most dramatic and well-known factor known to influence the catalytic activity of such systems is the size and shape of the Au nanoparticles [21,22,23,24]. The other important factor is the metal-support interaction which could basically influence the efficiency of the catalyst and so Au catalysts can almost never be considered independently from the support.

At present, two non-oxide supports are studied where the metal support interaction could be minimized. Graphene is in the focus of research and it is suitable to control the growth of metal including gold [25,26,27]. The other inert support is the boron-nitride prepared on single crystal surfaces. Hexagonal boron nitride (h-BN) on transition metal surfaces is free of oxygen and can help avoid ambiguities in the interpretation of the nature of active sites on Au nanocatalysts. Particularly interesting is the h-BN/Rh(111) system, especially since the discovery of a self-organized boron nitride superstructure on a Rh(111) surface by Corso et al. [28]. BN grows in a nanomesh form because of the stronger interfacial chemical bonding as a consequence of better orbital overlapping. The growth of Au and the surface properties of the formed gold nanoparticles were investigated by different surface science tools [28,29,30,31,32,33,34]. Scanning tunneling microscopy (STM) and low energy ion scattering (LEIS) measurements confirmed that at small coverages up to ΘAu = 0.1–0.2 ML, nearly two dimensional clusters form, while a clearly 3D growth was found at higher gold doses. Au nanoparticles nucleate preferentially at the pore-wire interface. Low temperature metal deposition yielded smaller clusters. X-ray photoelectron spectroscopy (XPS) analysis of core-level electronic states in the deposited Au shows strong final-state effects induced by restricted size dominating for low coverage [32, 35]. DFT calculation supports that all Au atoms in these one monolayer high clusters are negatively charged [31]. At higher coverages, the Au nanoparticles are less negatively charged; in particularly, topmost layer gold atoms are almost neutral [32]. Annealing the Au/h-BN/Rh(111) surface to higher temperatures resulted in the intercalation of Au below the h-BN layer according to LEIS, which is complete at 1050 K [35]. Desorption and agglomeration of Au also takes place.

In this work we studied the adsorption of acetaldehyde on Au covered Rh(111) and h-BN/Rh(111) surfaces. The stability and reactivity of adsorbed acetaldehyde is monitored by TPD and high-resolution electron energy loss spectroscopy (HREELS). This study is in connection with establishing of the reaction mechanism of ethanol transformation on the above mentioned, quasi inert surfaces. When different oxides are the support for gold, the ethanol decomposition runs differently further forming methane, ethane, hydrogen and CO [36]. The knowledge of the reactivity of acetaldehyde on Au covered h-BN/Rh(111) surface may give an answer to these differences and it is possible to investigate the reactivity of these nanoparticles with a lowest support impact.

2 Experimental

Three UHV systems were applied in these studies. The operation of the first system was described previously [33]. In this two-level UHV chamber the routine base pressure of 5 × 10−10 mbar produced by a turbomolecular pump. It was equipped with facilities for auger electron spectroscopy (AES), HREELS and TPD. The HREEL spectrometer (LK, ELS 3000) is situated in the lower level of the chamber, and has a resolution of 20–40 cm–1 (FWHM). The count rates in the elastic peak were typically in the range of 1 × 104–1 × 105 counts per second (cps). All spectra reported were recorded with primary beam of 6.5 eV and at an incident angle of 60° with respect to the surface normal in the specular direction. A Balzers QMS was used for TPD situated close to the surface normal. 4 K sec–1 heating rate was used in these measurements.

The second one was equipped with a commercial room temperature (RT) scanning tunneling microscope (WA-Technology), a cylindrical mirror analyzer with a central electron gun (Staib-DESA-100), and a quadrupole mass spectrometer (Balzers–Prisma). Note that, in this work, AES and mass spectrometry (MS) techniques were mainly applied for checking the surface/ad-metal cleanliness and the gas phase composition. STM images of 256 × 256 pixels were generally recorded in constant current mode at a bias of − 1.0 V on the sample and a tunneling current of 0.1 nA. W tips were applied in the course of the experiments. The conditioning of the tip was a relatively simple procedure: several seconds at + 3 V and 10 nA. The constant current (cc) morphology images are shown in top-view representation, where brighter areas correspond to higher Z values. Images of enhanced lateral resolution were recorded in constant height (ch) mode, where the local-dependent variation of tunneling current (no feedback) served for construction of the image. In the case of some high resolution images, a gentle fast Fourier transformation (FFT) treatment was also applied.

In the third system a hemispherical electron energy analyzer (Leybold Heraeus) was used for XPS, AES and LEIS. An Al Kα X-ray source was applied for XPS, performed with constant pass energy. The binding energy scale was calibrated against the 4f7/2 peak of a thick Au layer, fixed at 84.0 eV and the 3d5/2 peak of the Rh single crystal, fixed at 307.2 eV. The detection angle was 16° with respect to the surface normal. For LEIS He+ ions of 800 eV kinetic energy were used, and the scattering angle was 95°. The probe was radiatively heated by a W filament placed behind the sample. Its temperature was measured with a chromel–alumel (K-type) thermocouple spot welded to the side of the sample.

The Rh(111) crystal used in these works was cut from single crystal boule and was a product of Materials Research Corporation (purity 99,99%). Initially, the samples were cleaned by repeated Ar+ sputtering (typically 1 kV, 1 × 10−7 mbar and 2 µA for 10–30 min) and annealing to 900–1100 K until no contamination were detected by AES. The sample was heated resistively from 140 to 1200 K in HREEL chamber. Acetaldehyde (CH3CHO) of 99% purity was the product of Sigma Aldrich. It was purified by several freeze-pump-thaw cycles and for adsorption it was dosed through a capillary which was ca. 1.5 cm from the sample.

Au was deposited by a commercial physical vapor deposition (PVD) source (Oxford Applied Research) using high-purity (99.95%) Au. The coverage of gold is expressed in monolayer equivalent (MLE), defined as the surface concentration of Rh(111): 1.60 × 1015 cm−2 [37]. In all cases the evaporation rate was ~ 0.2 MLE min–1.

Monolayer of h-BN was prepared at 1000–1050 K via thermal decomposition of > 99.8% purity borazine, a product of Katchem Ltd.

3 Results and Discussion

3.1 Adsorption and Decomposition of Acetaldehyde on Clean and Gold Decorated Rh(111)

In our previous study we investigated the adsorption properties of acetaldehyde on clean Rh(111) surface [17]. Adsorption of acetaldehyde at ~ 120 K produced various types of adsorption forms: η1-(O)-CH3CHO and η2-(O,C)-CH3CHO were developed and characterized by HREELS. η1-(O)-CH3CHO desorbed at 150 K, another part of these species are incorporated in trimer, linear 2D polymer and three-dimensional surface species. Above 225 K acetaldehyde either desorbed (in monomer and trimer forms) or bonded as a stable surface intermediate (η2-(O,C)-CH3CHO) on the rhodium surface. It was the first observation of the polymerization process of acetaldehyde on Rh(111) [17]. The surface intermediates at higher temperature decomposed to CO, Ha, CH4 and surface carbon. In our first experiment in this study we adsorbed acetaldehyde on the clean rhodium surface (Fig. 1.) to have base comparison spectra before we investigate the gold decorated substrate. We adsorbed ~ 10 L of acetaldehyde at 170 K on the clean Rh(111) single crystal. At ~ 170 K, we observed peaks at 430, 610, 920, 1120, 1460 and at ~ 3000 cm−1 characteristic to adsorbed acetaldehyde molecule. Increasing the temperature to 200 K, we recognize small shifts in peak positions and new peaks appear at 650, 750, 1180 and 1560 cm−1. Small downward shifts also occurred for the loss at ~ 1100 cm−1. Similarly to our former observation the features in HREELS at 200 K, originates from the formation of surface oligomers and the peaks mentioned above belong to O–C–O vibrations (see Table 1.). The losses connected to η2(C,O)-acetaldehyde at low exposure (790, 1270, 1460 and ~ 2800 cm−1) are missing. Upon heating the layer above 230 K, the peaks at 650 and 1180 cm−1 disappeared from the spectra. Nevertheless, all of the peaks were attenuated at 300 K, although they are still discernible at 610, 950, 1120, 1400 and 2940 cm−1. These features can be attributed to different hydrocarbon fragments. At the same time, CO surface vibrations as new peaks appeared at 460, ~ 1800 and at 2040 cm−1; however, the hydrocarbon species still dominated the spectra at 300 K. The CO losses are the clear sign that acetaldehyde started to dissociate to CO, CHx and Ha on the surface. At 400 K only the peaks characteristic to CO adsorption were observed; at the next temperature step, above 500 K, all peak disappeared from the spectra (not shown).

Fig. 1
figure 1

HREEL spectra of adsorbed CH3CHO at 170 K on the clean Rh(111) surface followed stepwise annealing at different temperatures

Full size image
Table 1 Characteristic vibrations of acetaldehyde on single crystal surfaces
Full size table

Following the measurements on clean Rh(111) surface, similar experiments were also performed on ~ 0.5 MLE gold modified surface. In this case the Au was deposited at 500 K in order to eliminate carbon monoxide adsorption from the background and to prevent the surface mixing of Au and Rh [37]. The morphology STM image of 45 × 45 nm2 is shown in Fig. 2a, where the Au islands marked on the image form well separated 2D plates on the Rh(111) terraces or they can be condensed to Rh atomic steps (step flow growth mode) [37]. This type of growth mode suggests that in this state relatively large clean rhodium terraces of the Rh(111) surface are also available. Using this sample, the acetaldehyde was exposed at Ta = 170 K. The HREEL spectra of acetaldehyde recorded after preadsorption of gold shows similar features as on the clean Rh surface at low temperature (Fig. 3a). Increasing the temperature of the adsorbed layer we observed several characteristic vibrations of η2-(O,C)-CH3CHO at 640, 750, 930, 1080, 1125, 1195, 1340, 1420, 2890 and 2970 cm−1 in harmony with the work by Houtman and Barteau [16] and similarly to the results on Ru(0001) at 300 K [14]. However, the peaks at 640 and the shoulder at 1590 cm−1 at RT could be also the sign of surface oligomer formation, they belong very likely to the δ(OCO) and νa(OCO) modes, respectively [17]. Based on these results, we cannot rule out the interaction between neighboring η1-(O)-CH3CHO molecules, the production of oligomers or polymers. The very intensive peak at 2030 cm−1 at RT can be assigned to the formation of CO in on top position and there is no doubt that these molecules originated from the dissociation of η2-acetaldehyde connected to rhodium atoms. Above 300 K CO is not stable on gold. This statement is based not only on literature data but also on our own test measurements (not shown). Another remarkable finding is that on gold decorated surface the CO peak is the dominant loss already at 300 K and no contribution observed at around 1860 cm−1.

Fig. 2
figure 2

STM cc-images detected a on 0.5 MLE gold deposited (Td = 500 K) and b subsequently annealed (1000 K, 5 min) Rh (111) surfaces. c High resolution STM ch-image of 5 × 5 nm2 taken on atomic terraces shown in b. d Schematic model concerning the ordered (1 × 2) Au–Rh alloy phase

Full size image
Fig. 3
figure 3

a HREEL spectra of adsorbed CH3CHO at 170 K and the subsequent annealing at higher temperatures a on a ~ 0.5 ML gold decorated Rh(111) surface shown in Fig. 2a. b On the Au–Rh alloy surface shown in Fig. 2b

Full size image

In the subsequent part of this work we present similar acetaldehyde adsorption measurements as above on a Rh–Au alloy surface containing also approximately 0.5 MLE gold. The recipe is taken from our former work, in which the gold growth and surface alloying were studied in detail on Rh(111) surface [37]. Figure 2b shows an STM image of 50 × 50 nm2 recorded on gold (~ 0.5 MLE at 500 K) deposited Rh(111) surface after annealing it at 1000 K for 5 min. As it was demonstrated in [37], following the higher temperature (T > 700 K) treatment, these bulk immiscible metals form Au–Rh surface alloy, which can form a random or an (1 × 2) ordered phase as a function of the added amount of Au. Figure 2c image shows the latter unique ordered surface structure appearing on an STM ch-image of 5 × 5 nm2 and illustrated by a ball model in Fig. 2d. The substrate surface layer prepared in this way consists of gold and rhodium atoms mixed nearly atomically in the topmost layer. More details on the structure of this bimetallic surface can be found in [37]. To learn about the adsorption properties of this bimetallic layer, it was exposed to 5 L acetaldehyde at 170 K. Following the adsorption HREELS peaks appeared at 500, 840, 920, 1120, 1200, 1430, 1650, 1730 and at 3000 cm−1 (Fig. 3b.). These features show that acetaldehyde adsorbed molecularly and very possibly as multilayer, evidenced by the 1730 cm−1 loss belonging to ν(CO) mode in acetaldehyde molecules in the multilayer phase. Increasing the temperature to 230 K caused a remarkable attenuation of all peaks and small shifts in peak positions. Interestingly the peaks at around 610 and 750 cm−1 miss also at this temperature, however, the peak at ~ 1650 cm−1 still exists. These differences on the HREEL spectra suggest a slight discrepancy in the adsorption characteristics, which might be caused not only by the absence of η2-(O,C)-CH3CHO but possibly also by the adsorption of acetaldehyde on gold atoms. The stability of these species on the mixed layer is very low compared to the clean Rh(111) surface considering that at 300 K only a small indication of adsorbed molecules remains on the HREEL spectra. Although the intensity of these peaks at RT very low, we can recognize the similarity between the nature of this layer and those, which was pictured on the clean surface (e.g.: the peak at 610 cm−1 appeared). Dissociation of these surface intermediates resulted in the formation of adsorbed CO indicated by the small peaks at 400 K. HREELS results showed that on the mixed surface layer acetaldehyde adsorbed possibly weakly also on gold atoms. Another part of molecules connected to rhodium atoms but mainly in η1-(O) form. HREELS identification, including the lack of characteristic vibrations of ν(CO) at 230 K also suggests that the suppressed number of accessible adsorption sites, on which η1-(O) form can adsorb caused a reduced tendency toward polymerization.

Summarizing the above results, the behavior of acetaldehyde is obviously controlled by the modified adsorption sites. Namely, upon the formation of Au–Rh surface alloy there are only a few rhodium atoms on the surface, which neighbored exclusively by Rh atoms and form adsorption centers characteristic for a clean Rh(111) surface. This morphology causes probably that the low coordinated (η1-(O)) molecules desorb from the surface without dissociation below 300 K and only a small amount of molecules remained bonded with two legs to the rhodium, resulting in adsorbed CO at and above 400 K. Although we have to take into consideration for this scheme that the annealed layer can contain surface carbon remained on the surface as a contamination after dissociation of acetaldehyde, it cannot be the reason for this dramatically different behavior. Whilst on the carbon contaminated surface the carbon decreased the uptake of adsorbed acetaldehyde, nevertheless it facilitated also the C–O bond scission and CO formation with 40–50 K lower temperature [17].

3.2 Preparation and Characterization of Boron Nitride Layer on Rh(111) by HREELS

In the next step we prepared a so called h-BN “nanomesh” layer on Rh(111) by a CVD method decomposing borazine at 1000 K. The h-BN/Rh(111) system is interesting also because it allows preparation of metal clusters with a narrow size distribution [28]. This periodically corrugated super lattice can serve as a template, on which molecules and nanoparticles can be readily adsorbed in a tailored arrangement on atomic length scale [32, 33]. The preparation method of this monolayer 2D film in our system was described in a previous work in detail [33]. In this study, first we prepared the boron nitride layer on Rh(111) and titrated with CO at 300 K. (Fig. 4a, b).

Fig. 4
figure 4

a HREEL spectra of the (i) clean and (ii) CO dosed (10 L) BN/Rh(111) compared to (iii) CO saturated rhodium. b HREEL spectra taken on after adsorption of 5 L acetaldehyde on h-BN/Rh(111) surface at 170 K and followed by annealing at higher temperatures

Full size image

The observed HREEL spectrum (i) in Fig. 4a corresponds well with the literature data, where hexagonal boron nitride layer (h-BN) was prepared and investigated on different single crystal surfaces by HREELS [33, 42]. The peak at 790 cm−1 corresponds to transverse optical (TO) phonon with out-of-plane polarization. The peaks at 1350 and 1510 cm−1 are assigned to phonons with in-plane polarization; the former one is originated from the TO phonon, while the latter one from the longitudinal optical (LO) phonon. Titrating the 2D film by carbon monoxide (10 L) at 300 K caused no change in the spectra (ii) in Fig. 4a. The lack of the substrate metal-CO (ν(M-CO)) and ν(CO) vibrations (see (iii) curve in Fig. 4a recorded on Rh(111) surface saturated by CO at 300 K) is the clear evidence that the h-BN layer covers completely the surface of Rh(111) and the sticking probability of CO on h-BN layer is very low at room temperature. Figure 4b displays HREEL spectra of h-BN/Rh(111) after its exposure to acetaldehyde at 170 K and annealing at different temperatures. Similarly to carbon monoxide, acetaldehyde does not adsorb on the clear nanomesh layer at his temperature or it was under the detection limit of HREELS. It suggests that we didn’t find any sign of substrate metal effect through boron nitride layer toward the adsorption of acetaldehyde.

3.3 Adsorption of Acetaldehyde on Gold Nanoparticles Prepared on h-BN Covered Rh(111)

As we have seen above, hexagonal boron nitride proved to be completely inert toward acetaldehyde at 170 K. In this part of our work the effect of deposited gold will be investigated on the adsorption behavior of CH3CHO molecule. It seemed to be interesting, partially because the h-BN nanomesh template offers the possibility to produce Au nanoparticles in quasi uniform size and ordering determined by the pore regions of the h-BN mesh [29, 32, 35]. Note, the amplitude of the corrugation of the BN layer is about 0.55 Å, while the periodicity of the hexagonal mesh is 3.2 ± 0.2 nm. Another interesting aspect of the Au/h-BN/Rh(111) system is that regarding h-BN as an essentially inert support for gold, it is possible to investigate the reactivity of these nanoparticles with a lowest support impact.

LEIS showed that the Au LEIS intensity enhanced sharply at low gold coverage and with the increase of the gold quantity it moderated and did not reach saturation even at 27 MLE of gold. Since LEIS is an extremely surface sensitive method it suggests that at low coverage (up to ~ 0.2 MLE) 2D, while at higher coverages 3D Au clusters formed on top of h-BN.

First gold was evaporated onto the h-BN covered Rh(111) surface at 300 K, followed by exposure of acetaldehyde (5 L) at 170 K. After the HREEL spectra were recorded at low temperature, we monitored the changes during a stepwise annealing of the sample. At small gold coverages (ΘAu < 0.3 MLE) the adsorption of acetaldehyde was barely observable by HREELS (not shown); nevertheless, for increasing amounts of gold the adsorption ability obviously enhanced. At ΘAu ~ 1 MLE coverage the adsorbed acetaldehyde was clearly detectable; however, the stability of the adsorbed molecules was very low: at 300 K almost all signs of surface fragments disappeared, only the phonon losses belonging to boron nitride showed up. In Fig. 5a, b we demonstrate the characteristic vibrations of acetaldehyde on gold nanoparticles for 2 MLE (a) and 5 MLE (b) gold coverages.

Fig. 5
figure 5

HREEL spectra showing the effect of annealing after adsorption of acetaldehyde (5 L) at 170 K on h-BN/Rh(111) surface decorated by a 2 MLE, and b 5 MLE of gold

Full size image

In the latter case (5 MLE), the h-BN HREELS losses were screened completely by the admetal. Consequently, there were no signs of BN phonon vibrations before the adsorption of acetaldehyde at 170 K (not shown). The effects of surface metal on the substrate spectra and the possible screening by inducing image dipoles were previously investigated and described [43, 44]. Two important observations were made. We experienced that the BN phonon vibrations were hardly observable in HREELS for 5 MLE gold deposition, on the other hand in our LEIS studies the gold signal did not saturate even at coverages above ~ 20–30 MLE. This comparison suggests that the above-mentioned screening effect is rather strong in HREELS, and does not require a complete and continuous gold layer. The reason of this strong effect could be that the first phonon loss at ~ 790 cm−1 (the TO phonon with out-of-plane polarization) has a dynamic dipole nature (evidenced by off specular measurements) which affected by the gold on the surface.

In both cases (2 and 5 MLE) the observed vibrations are characteristic of acetaldehyde; however, the peak positions are slightly different compared to those observed on Rh(111). Losses appeared at ~ 850, 1120, 1180, 1350, 1425 and at 2980 cm−1after 170 K adsorption of acetaldehyde on 2 MLE gold covered Rh(111). These features are in good agreement with the loss spectra of acetaldehyde observed on annealed gold layer on clean Rh(111) (Fig. 3b). There we experienced peaks at 850, 920, 1100, 1160, 1380, 1450, a small shoulder at 1650 and broad peak at 3000 cm−1 at 230 K after multilayer desorption. In harmony with the TPD results of the mentioned analogical experiment from gold covered BN/Rh(111) layer, only molecular acetaldehyde desorption (at ~ 220 K) was detected without any sign of dissociation products.

Figure 6a shows the temperature programmed desorption spectra (m/e = 44) after the adsorption of 5 L acetaldehyde on h-BN/Rh(111) at increasing gold coverages (iii, iv, v) The first two spectra (i) and (ii) were taken on the clean rhodium and the h-BN covered Rh surface, respectively. It was verified that only the molecular acetaldehyde (the parent molecule) desorption contributed to the intensity of the appeared peaks. On the clean surface we can observe possibly the stabilizing effect of the polymerized form of acetaldehyde (Tp around 300 K). These polymers strengthen the thermic stability of acetaldehyde molecules in contrast to the lower coverage case where acetaldehyde starts to decompose already at 300 K to CO and H2 [17]. On the modified single crystal surfaces we made efforts to detect CO, H2, C2H5OH or several other possible reaction products (oligomers too), but it did not succeed. At the same time the adsorbed amount of acetaldehyde gradually increased from close to zero (on h-BN/Rh(111)) up to saturation value at around 2 MLE of gold coverage (Fig. 6b). After saturation it did not change during our measurement series even when BN phonon vibrations quasi disappeared from HREEL spectra at around 5 MLE.

Fig. 6
figure 6

a TPD spectra recorded after the adsorption of acetaldehyde (5 L) at 170 K for (i) clean Rh(111), (ii) h-BN covered Rh(111) and Au/h-BN/Rh(111) surfaces of different gold coverages: (iii) 0.3 MLE, (iv) 1 MLE and (v) 2 MLE. b Integrated peak areas of amu 44 (acetaldehyde) as a function of gold coverage on BN/Rh(111). STM cc-images recorded on c h-BN/Rh(111)surface and Au covered, d 1 MLE and e 2 MLE surfaces surface(Tdep = 300 K)

Full size image

There is no sign of polymerization reaction on the HREEL spectra. TPD results at amu 44 shows also only one type of acetaldehyde desorption which does not confirm the production of oligomers on the gold covered h-BN/Rh(111) system. Our HREELS and TPD results on different surface layers propose that the possibility of polymerization and the production of oligomers is high only if the interaction between the substrate metal and acetaldehyde is strong enough. Only in that case metal can affect the charge distribution and this way facilitate the connection of the neighboring molecules.

Figure 6c shows a typical STM image of the as-synthesized h-BN layer on Rh(111) surface. The nanomesh indicates a typical 3.3 (± 0.2) nm lateral periodicity, which coincides with the superlattice’s unit cell’s corners. These data are in a good agreement with earlier work [28, 45]. Depositing 1 MLE Au at 300 K results in the formation of three-dimensional (3D) nanoparticles on the h-BN film showing wide size distribution, with 1.9 × 1012 cm−2 particle concentration. The highest particles typically consists of 4–5 atomic layers (~ 5% of the NPs). STM pictures of the investigated adsorption layers (0, 1 and 2 ML of Au on h-BN/Rh(111)) are exhibited on Fig. 6c–e.

In a parallel measurement we followed the changes of the Au 4f7/2 XPS peak position and the Au 4f integrated area during deposition of increasing amounts of gold on h-BN/Rh(111) at 300 K (Fig. 7a).

Fig. 7
figure 7

a Changes in the Au 4f7/2 XPS peak position and integrated area as a function of gold coverage on h-BN/Rh(111). The Rh 4s contribution was subtracted. The solid lines are to guide the eye. b Effect of gold coverage on the ν(CO) HREELS peak after adsorption of 5 L CO on clean and gold decorated Rh(111) at 170 K

Full size image

At ΘAu = 0.045 MLE the Au 4f7/2 peak was found at ~ 83.7 eV, while the bulk value is 84.0 eV. The rather low binding energy observed at small coverages is attributed to the presence of negative extra charge on Au clusters due to an electron transfer from h-BN (and Rh(111)) to Au nanoparticles, in accordance with previous DFT studies [30,31,32]. At higher coverages the peak position gradually approaches the bulk value, because the amount of extra charge per Au atom decreases, and the electronic structure of larger Au clusters is bulk-like. Interestingly, the amount of acetaldehyde adsorbed on Au/h-BN/Rh(111) saturates at the same coverage ΘAu ~ 2 MLE, where the Au 4f7/2 binding energy is already very near to the bulk value (Fig. 6b). In a hydrogen chemisorption experiment on Al2O3 supported gold nanoparticles, published by van Bokhoven et al. a lower hydrogen uptake is measured after increasing the gold particle size by high-temperature reduction [46]. The reason of this similar observation was that the dissociation and adsorption of hydrogen was limited to the gold atoms on corner and edge positions. In our measurements acetaldehyde adsorption shows very similar behavior—saturation at a certain gold coverage—which strengthens their statement about the active site of gold particles. Please note that edge and corner atoms of gold clusters were suggested to be the active sites also for oxygen and CO bonding [47]. The slope of the Au 4f area curve gets smaller above ΘAu ~ 0.2 MLE confirming the 3D growth of Au at larger coverages (Fig. 7, inset).

On Fig. 7b we presented the effect of gold coverage on the ν(CO) HREEL peak after adsorption of 5 L CO on clean and gold decorated Rh(111) at 170 K. These measurements were carried out to gain some information about the charge on Au nanoparticles with the help of ν(CO) HREELS peak position. The size of nanoparticles and possibly the negative charge on gold particles also influenced the position of the CO HREELS peak after adsorption of CO on Au/h-BN/Rh(111). The small shift in ν(CO) HREELS peak position at this temperature ~ 170 K suggests a charge transfer which is resulted -at small size clusters- a negatively charged particles. Peak appeared ~ 2070 cm−1 at 0.2 MLE and it is shifted to higher energies at ~ 1 MLE. Our explanation is, similarly as it was discussed in the work of McKee earlier [34], that the highly undercoordinated perimeter edge and corner sites bind CO more strongly than bulk Au surfaces and that the small size Au cluster also contains a sizable negative charge, which is localized on the perimeter atoms. Literature data also deal with the phenomenon of negatively charged gold clusters and with the shift from lower energy value to the higher one tend toward the gas phase value at 2143 cm−1 [48,49,50].

4 Conclusions

In this work we investigated the adsorption properties and reactions of acetaldehyde on clean and gold decorated rhodium and BN/Rh(111) surfaces by HREELS and TPD methods. STM and XPS were used to characterize the gold nanoparticles.

  • Our results on different surface layers propose that the possibility of polymerization is requires stronger interaction between the substrate metal and acetaldehyde. According to this we observed the sign of oligomers only when clean rhodium terraces are present on the gold modified surfaces.

  • The surface alloy formation of gold and rhodium on the clean substrate eliminates the formation of CO. Nevertheless, on an inert BN support increased coverage of gold enhanced the amount of adsorbed aldehyde at low temperature.

  • In our measurements acetaldehyde adsorption on gold covered BN/Rh(111) shows saturation at a certain gold coverage which suggests that the active sites of gold clusters on h-BN are possibly edge and corner atoms and very likely at small coverages we need to count with a charge transfer between the substrate and gold which is resulted a negatively charged particles and this way gives enhanced reactivity of Au/h-BN/Rh(111).