Formation of nitrile species on Ag nanostructures supported on a-Al2O3: a new corrosion route for silver exposed to the atmosphere

1 Adsorption of nitrile species on Ag nanostructures supported on a-Al2O3: a new corrosion route for silver exposed to the atmosphere R.J. Peláez,1 J. P. Espinós,2 C.N. Afonso1 1Laser Processing Group, Instituto de Optica, CSIC, Serrano 121, 28006, Madrid, Spain 2Surface Analysis Lab, Instituto de Ciencia de Materiales de Sevilla, CSIC-US, Américo Vespucio 49, 41092, Sevilla, Spain ABSTRACT The aging of supported Ag nanostructures upon storage in ambient conditions (air and room temperature) for 20 months has been studied. The samples are produced on glass substrates by pulsed laser deposition (PLD) by depositing first a 15 nm thick buffer layer of amorphous aluminum oxide (a-Al2O3) followed by PLD of Ag. The amount of deposited Ag ranges from that leading to a discontinuous layer up to an almost percolated layer with thicknesses < 6 nm. Some regions of the as grown silver layer are converted into round, isolated and with up to ~25 nm diameter nanoparticles (NPs) by laser irradiationinduced dewetting. The plasmonic, structural and chemical properties of both as grown and laser exposed regions upon aging have been followed through extinction spectroscopy, scanning electron microscopy (SEM) and X-ray photoelectron spectroscopy (XPS) respectively. The results show that the discontinuous as grown regions are optically and chemically unstable and that the metal becomes oxidized at a velocity that is higherfaster the smaller the amount of Ag. The corrosion leads to the formation of nitrile species due to the reaction between NOx species from the atmosphere adsorbed at the surface of Ag and hydrocarbons that were adsorbed in defects at the surface of the a-Al2O3 layer during the deposition of the Ag nanostructures and by PLD that migrated to the surface of the metal with time. The nitrile formation thus results in the main oxidation mechanism and inhibits almost completely the formation of sulphate/sulphide. Finally, the optical changes observed upon aging offer an easy to use tool to follow the aging process. They are dominated by an enhanced absorption in the UV side of the spectrum and a blue-shift of the surface plasmon resonance and an enhanced absorption in the UV side of the spectrumthat are respectively related . While the latter is mainly related to the formation of a dielectric overlayer of on the Ag nanostructure due to the oxidation process and , the former is Comentario [CNA1]: Es adsorption or fomation? RP: formation me parece mejor, aunque no soy experto. No se si cambiar el titulo a estas alturas puede complicar el proceso de envio del paper. Aunque en la web del autor TODO esta editable. J.P.: Tienes razón, formation es mas adecuado, dado que no partimos de nitrilo como gas adsorbible. Comentario [A2]: Respecto al comentario anterior aqui utilizamos ¡formation’ de forma creo que más adecuada. 2 caused by the dominant effect of changes in the dimensions/features of the nanostructures upon aging over the effect of the dielectric overlayer. 3 INTRODUCTION Nanostructures and nanoparticles (NPs) are characterized by their high surface to volume ratio that is responsible for their special properties when compared to those of the bulk material [1]. One of the most widely studied phenomenon is the resonant optical absorption of noble metal NPs. It is due to the excitation of surface plasmon resonances (SPRs) that are exploited for sensing, photovoltaic or metamaterials [2–4]. Silver nanostructures are the most attractive ones since their SPR is stronger than that of other noble metals like gold because it is well separated from interband transitions. SPRs of supported Ag NPs have been demonstrated to improve the efficiency of solar cells, and to lead to metamaterials for silver subwavelength slit gratings or plasmonactive tips-enhanced for Raman spectroscopy. However, surface tarnishing and/or chemical degradation of silver nanostructures upon exposure to the atmosphere decrease the quantum efficiency of the solar cells, reduce the extraordinary optical transmission of the metamaterials and limits the field enhancement of tips used for sensing [5–9]. Surface tarnishing of Ag is known since very long [10,11], and the main constituent reported upon exposure of silver to the atmosphere has been silver sulfide [12]. Although only a very thin layer is formed, it induces a dielectric overlayer that influences the optical response. The optical changes include the damping and shift of the SPR wavelength. However, different senses have been reported for the shift that have been suggested to be related to the NPs features: while Ag-nano objects with dimensions of tens up to hundreds of nm exhibit typically a red-shift [13,14], Ag NPs with small diameters exhibit a blue-shift [15]. Furthermore, the chemical activity of the latter type of Ag NPs is known since very long to depend also on the supporting substrate [16]. 4 Metal NPs on oxide supports constitute an important class of heterogeneous catalysts that have been studied for a longtime [refs.5 y 6 de nuestro nanotechnology 2013]. There are support interactions that lead to a modification of the reaction and adsorption properties of the supported NPs. Most of these studies have been performed on crystalline substrates that were either single crystals or epitaxially grown films that require special conditions to be produced. Very little hasve been reported on the use of their amorphous counterparts that generally require no specific conditions or high temperatures in their production. For this work, we have selected Aamorphous alumina oxide (a-Al2O3) is used as a substratesupport for the metal nanostructures because on the hand, we have recently reported that Ag NPs on a-Al2O3 support become strongly oxidized under atmospheric conditions, sulfurization being negligible [15]. On the other hand, the adsorption of Ag on Al2O3 is known to be highly sensitive to its surface termination, being weak when it is Al or hidroxilated or water terminated surfaces and strong when it is O-terminated. When compared to the crystalline material, a-Al2O3 has a lower surface energy and more surface defects that favor the absorption of species.. Last but not least, Al2O3 has a high refractive index, high dielectric constant with low leakage current and an exceptional wear and scratch resistance that make it very attractive for a wide range of applications in optical communication networks, microelectronics and coatings. In order to provide a deeper understanding of the different corrosion mechanisms of Ag nanostructures on a-Al2O3 supports so far reported and clarify whether there is any dependence on the NPs dimension, we have monitored the plasmonic reponse, the morphological features and the chemical composition of Ag nanostructures supported on a-Al2O3 exposed to the atmosphere along 20 months. These nanostructures are are fabricated by pulsed laser deposition (PLD) by proddepositucing Comentario [A3]: Las referencias las pondré en formato final cuando todos estemos de acuerdo con el texto. Con formato: Subíndice Con formato: Subíndice Comentario [A4]: Phys.Chem.Chem .Phys., 2014, 16, 26645 A near ambient pressure XPS study of subnanometer silver clusters on Al2O3 and TiO2 ultrathin film supports† K. M. Bulanin, J. C. Lavalley and A. A. Tsyganenko, J. Phys. Chem., 1995, 99, 10294–10298. Comentario [A5]: Katiyar, P.; Jin, C.; Narayan, R. J. Electrical Properties of Amorphous Aluminum Oxide Thin Films. Acta Mater. 2005, 53, 2617−2622. Mardare, A. I.; Kaltenbrunner, M.; Sariciftci, N. S.; Bauer, S.; Hassel, A. W. Ultra-Thin Anodic Alumina Capacitor Films for Plastic Electronics. Phys. Status Solidi A 2012, 209, 813−818 Comentario [A6]: 10.1002/adfm.2016 00213 Con formato 5 a supporting a-Al2O3 layer first that is followed by the deposition of different amounts of Ag. that lead to nanostructures ranging from a discontinuous layer up to an almost percolated layer. Part of the nanostructured layers are converted into isolated and almost round NPs having up to ~25 nm diameter by laser irradiationinduced dewetting. A widespread range of dimensions and shapes are thus studied under the same aging conditions: flat (thickness < 6 nm) nanostructures ranging from a discontinuous layer up to an almost percolated layer and almost round and isolated NPs having diameters ~≤ 25 nm . The results obtained agree well with those reported earlier for flat nanostructuressmall NPs [15] and support a new corrosion route based on the formation of Ag-CN species from reduction of NOx at room temperature in which the a-Al2O3 support plays an essential role. . The formation of these nitrile species is found to depend strongly on the silver coverage and the nature of the substrate. EXPERIMENTAL The samples are produced by pulsed laser deposition (PLD) using an ArF excimer laser (=193 nm, =20 ns full width half maximum (FWHM)), focused at an angle of incidence of 45° on the target at a fluence of ~2.7 J cm-2. The substrates are standard cover glass slides held at room temperature that are rotated along an axis parallel to the plasma expansion axis and shifted a few mm in order to produce a homogenous deposit over an area >1 cm2. The ablation is performed in vacuum (P10-6 mbar) by first focusing the laser beam to the surface of a high purity ceramic Al2O3 target to produce a 15 nm thick buffer layer of amorphous aluminum oxide (a-Al2O3). Without breaking the vacuum, the laser is then focused to a 99.99% Ag target during 3 different times, in order to produce layers with effective thicknesses of 1.6 nm, 3.7 nm and 5.9 nm. Finally, the vacuum chamber is generally filled with 99.999% N2 to bring it Comentario [CNA7]: Contrastar con Fg.4 Son aprox 25 nm, pero no menor de 25 nm Con formato: Subíndice 6 to atmospheric pressure. A second series of three samples was are produced using the same conditions of the thinnest and thickest Ag layers but with no buffer layer, i.e. the Ag was is directly deposited on glass. These samples will from now on referred to as reference samples. Unless otherwise specified, all results presented will refer to samples having the a-Al2O3 buffer layer. In order to modify the initial nanostructure of the samples, we have used the same laser combined with a beam homogenizer that leads to a constant intensity (within 5%) over ~4x4 mm2 square regions, to irradiate regions of the thickest and thinnest samples in air immediately after deposition. We have used single pulses of fluences in the range 106-197 mJ cm-2. The samples are stored in air within single wafer shippers (47 mm diameter, ~3 cm3) made of natural polypropylene (Entegris). They are stored in Madrid, where the optical characterization starts immediately after laser processing and is repeated several times along 15 months. The samples travel 3 times to Sevilla where they stayed for chemical and structural characterization during the periods of days 14-91, 218-309 and from 441 on. These periods will be referred to from now on as 0.5 m, 7 m and 15 m where m stands for months. In addition, the optical results assigned to these periods correspond to measurements performed just before the period starts. Further details on the production and laser irradiation procedures as well as on the evolution of optical and structural properties of similar samples upon aging can be found elsewhere [15,17]. The optical response of both as grown and laser exposed regions has been analyzed through the extinction spectra calculated as ln(1/T(  )), where T(  ) is the transmittance measured at normal incidence, using a Cary 5000 dual beam spectrophotometer. The structural properties are characterized by scanning electron microscopy (SEM) in a Zeiss CrossBeam 1540XB microscope. The chemical 7 composition of the surface is determined by X-ray photoelectron spectroscopy (XPS) using a SPECS Phoibos 100 DLD spectrometer and working in the pass energy constant mode at a value of 35 eV. The XPS spectra were excited with unmonochromatic Mg Kα (hυ = 1253.6 eV) radiation. The intensity of the main photoemission peaks detected, i.e. Al2p, O1s, Ag3d, C1s, N1s and S2p, are used to calculate the chemical composition of the surface. We have used Shirley type backgrounds to determine the area under the peaks [18], that are finally corrected for the escape depth of electrons [19], the photoionization cross sections of each photoelectron signal [20] and the transmission function of the analyzer [21]. The modified Auger parameter (α’) that is a key factor for determining the chemical state of silver is calculated from the binding energy of the Ag3d5/2 peak and the kinetic energy of the AgM4VV Auger peak. The percentage of metal Ag(0) and oxidized Ag(I) is derived from the AgMNN Auger signals rather than from the more intense Ag3d signals because the dependence of the shape and energetic features of the former on the chemical state of Ag is stronger [22]. As AgMNN reference spectra for pure Ag(0), we have used the one recorded from a ~6 nm thick Ag layer deposited by thermal evaporation in the XPS chamber, under ultrahigh vacuum conditions (P<10-9 mbar), and measured without breaking the vacuum. This reference spectrum was then subtracted from the experimental ones obtained in the samples studied in this work, in order to fully remove the contribution of the metallic component. In order to calculate the degree of silver oxidation, the resulting AgMNN spectra were then compared to the spectrum reported elsewhere [15] for a 1.6 nm thick Ag sample on a-Al2O3 that . The sample was prepared following the same procedure than in the present work. This sample and it was completely oxidized after 25 months in similar aging conditions and . This sample will from now on referred to as oxidized reference sample. 8 Further information on the chemical composition of a selection of regions and samples is obtained by Raman spectroscopy with a LabRam spectrophotometer (Horiba Jobin Yvon), by exciting with the non polarized beam from an Ar laser (  = 514.5 nm) at their front face, and collecting the backscattered light. We have used the minimum power required to obtain signals in order to prevent any possible transformation. RESULTS Figure 1a shows the extinction spectrum recorded at 0.5 m from the as grown region of the thickest sample. It exhibits a broad band peaking around 613 nm related to the SPR of the as grown region of the metal nanostructure that consists of irregular and almost percolated silver NPs, as seen in Figure 1c. Upon laser irradiation, the SPR narrows and shifts blue (see also Figure 1a), the amount of the shift depending on fluence as shown in the inset of Figure 1a, where the wavelength position of the SPR as a function of fluence is plot. The largest shift occurs for fluences in the range 120-152 mJcm-2 and is related to the complete break of the as grown layer first, into isolated irregular NPs for low fluences (Figure 1d) and into almost round and well isolated NPs for higher fluences (Figures 1e and 1f). Unless otherwise specified, the results presented from now on for laser exposed regions refer to regions irradiated with a fluence of 152 mJcm-2. Figure 1a also shows the spectra of both the as grown and laser exposed regions of the same sample at 15 m, where it is seen that they are similar to those recorded at 0.5 m. The inset in Figure 1a also shows that the SPR wavelength of the as grown regions at 15 m as a function of fluence at 15 m. It is seen that it follows the same dependence on fluence than that at 0.5 m, although the values are slightly blueshifted as it becomes evident in the spectrum. 9 Figure 1b includes de same type of spectra than Figure 1a but for the thinnest sample. The spectrum of the as grown region at 0.5 m exhibits a narrower band than that of the same region of the thickest sample, peaking around 505 nm. This result is consistent with the discontinuous character of the sample or its smaller effective thickness. Upon laser irradiation, the SPR also narrows and shifts blue. At 15 m, significant changes are observed in the as grown region: a blue-shift of the SPR in the as grown region and a strong increase of the extinction at the UV edge of the spectrum, the latter also occurring the laser exposed region. It is worth pointing out here that results for wavelengths shorter than 300 nm are meaningless due to the absorption of the glass substrate. The detailed values of the SPR wavelength of all studied as grown and laser exposed regions as a function of aging time are shown in Figure 2. It is seen that the as grown regions show a continuous blue-shift as time progresses, the lower the metal content the higher the shift. The laser exposed regions show much smaller shift that becomes almost negligible in for the thickest sample. Figure 3 shows SEM images of all studied as grown regions at 0.5 m, 7 m and 15 m. It is seen that the nanostructure evolves from almost isolated small NPs (thinnest sample, Figure 3a), to bigger and coalesced NPs (intermediate thickness sample, Figure 3d) to an almost percolated structure (thickest sample, Figure 3g). Elongated structures appear to develop on top of this initial nanostructure at 7 m in the thinnest sample (Figure 3b), while similar features are seen in the intermediate thickness sample at 15 m (Figure 3f) and are not seeing at allnever observed in the thickest sample one (Figure 3g-i). The average Feret length and aspect ratio of these elongated structures are shown in Figure 4a as a function of aging time. It is seen that once they appear their length changes little with time and they are ~5 times longer in the intermediate thickness 16 Overall, Table 1 evidences that the main element at the surface is C having concentrations ranging from 33.5% to 55.7%. The C1s related region in the XPS spectra is shown in Figure 9b. They show a single asymmetric peak with a longer tail at the high binding energy side with little changes in the binding energy position upon aging. In spite of its apparent simplicity, four components are needed to fit these spectra properly for all the spectra acquired at different aging stages, as illustrated in Figure 11 for the as-grown region of the thinnest sample at 0.5 m. In order to have chemical consistency, the four components have the same shape. The peak energy position and FWHM are included in Table 2 for all aging stages. The first component, CI at 284.65 ±0.05 eV, corresponds to C-H and C-C species in both aliphatic and aromatic organic molecules. They are very common adsorbates found on the surface of any kind of solid exposed to the atmosphere for long and are usually referred to as “adventitius carbon”. In spite of the great care taken in the storage, the surface concentration of these species is very high irrespective of the thickness or the treatment. The components at higher binding energies are usually assigned to C atoms bonded to heteroatoms (O, N, S, Cl, F, …), the higher the electronegativity of the heteroatom or the higher the bond order, the higher the binding energy. The next two components, CII at 285.35 ± 0.05 eV and CIII at 286.7 ± 0.1 eV, should be assigned to C atoms singled or multiple bonded to N (amines, amides, nitriles, …) and single bonded to O (alcohols, ethers, …), respectively. The last component CIV, at 288.4 ± 0.3 eV, should be ascribed to C=O species (cetones, aldehides, esters, …). Note that CII is the only component whose relative intensity changes significantly over aging. Finally, the most unexpected result with respect to surface composition is the appearance of nitrogen upon aging with surface concentrations that can be up to 6.2% (see Table 1). No chemicals containing nitrogen have been used for the preparation of Comentario [CNA9]: Se podría suprimir la figura 11. Es un detalle muy técnico y para especialistas y no para el “common reader” que dice el referee 1. Escribire en la respuesta al refereee una nota al respecto por si os gustara la idea, pero no hago nada en el texto. RP: contestado en la respuesta al referee. J.P. Si no la piden los referees, nos la podemos evitar. Pero para un referee serio, si no la hubieramos puesto, lo lógico es que la pidiera. 17 the samples or during their storage. In addition, supplementary analyses have shown that no N1s signal is ever detected either on the surface of the a-Al2O3 support or on the surface of the bare glass substrates. Furthermore, no nitrogen was detected either in a sample in which the silver NPs were fully buried in a-Al2O3 or were supported directly on the glass substrate [15]. Figure 12 shows N/Ag as a function of aging time for the different samples and regions studied. It is seen that the N signal appears at 7 m and becomes higher at 15 m, this increase being higher the smaller the amount of silver (see also Table 21). In order to analyze if N keeps increasing with time, the as grown regions of the samples where measured again at 20 m and Figure 12 shows that the N/Ag ratio where it is seen that this ratio tends to saturate after 15 m. The evolution of the N1s region upon aging is illustrated in Figure 9c for the case of the as grown region of the thinnest sample. The photemission signal is quite symmetric and narrow (FWHM=1.55 ±0.05eV), which suggests that it is caused by a single chemical species. The binding energy position, 399.1 ±0.1 eV, is compatible with nitrogen bonded to silver, aluminum [30] or carbon [31–34] and excludes the possibility of N bonded to oxygen. The N1s binding energy in binary aluminum nitrides films has been reported to range between 399.7 eV and 397.8 eV, depending on its purity and microstructure [35,36], while for nitrogen coordinated to other metals (i.e. silver nitride) this energy range could be even higher depending on the nature of the metal and the bonding structure (amides, imides or nitrides) [30]. However, silver nitride is an extremely explosive compound that slowly decomposes at room temperature [37]. It has never been synthesized by dry methods and, to our knowledge, it has never been studied by XPS or any other ultra-high-vacuum analysis technique, very likely due to its instability under vacuum and X-ray illumination. Finally, the bonding configurations of Comentario [A10]: Esto debe ser tabla 1 ¿¿verdad?? J.P: Si, es la Tabla I 18 carbon-nitrogen systems can be very complex (amines, nitriles, amides, …) and the binding energies can range between 398 and 401 eV[38]. In order to investigate further the chemical nature of the detected nitrogen species, a selection of samples / regions has been characterized by Raman spectroscopy. The results are shown in Figure 13 shows where the spectra for of as-grown regions of the thinnest and thickest samples at 7 m are plotted together with the spectra of the corresponding reference samples on glass, i.e. same silver content but no a-Al2O3 buffer layer. Although the samples have undergone the same aging process, nitrogen species were only sharply detected by XPS on the surface of the samples in which Ag was on a- Al2O3 supports. The spectra in Figure 13 evidence several broad signals in the 500-3200 cm-1 energy range. The two more intense bands, at 1565±5 and 1355±5 cm-1, are ascribed to C=C bond vibrations caused by in-plane vibrational modes of sp2 C=C bonds in amorphous carbon [39]. They are usually referred to as “G band”, where G means graphitic, and “D band”, where D means disordered. The bands at 2905±5 cm-1 and 670-800 cm-1, also found in amorphous carbon, are ascribed to the stretching and bending vibrational modes of C-H bonds, respectively. In addition, a new band peaking at 2138±2 cm-1 is found in the samples having Ag on a-Al2O3 whose intensity is much higher for the case of the thinnest sample. This band should be assigned to the stretching vibrational mode of nitrile (also known as “cyanide”) species, -C≡N and its energy is known to be slightly dependent on the nature of the metallic species they are linked to and the adsorption mode [40]. More specifically this band has been reported at 2142 cm- 1 for nitrile species bonded to Ag+ through a carbon atom (Ag-C≡N), at 2165 cm-1 in bulk AgCN for Ag(CN)2- species in aqueous solutions [41], and at 2130 cm-1 when (CN)- is adsorbed on Ag NPs supported on Al2O3 [42,43]. In addition, for nitrile species 19 bonded to Al3+, two vibrational bands (CN) have been reported at 2165 cm-1 and 2155 cm-1 [44]. DISCUSSION The main and unexpected result of this work is the appearance of nitrogen on the surface of the samples upon aging. To our knowledge, and before our earlier work [15], this phenomenon has never been reported upon aging at room temperature and atmospherice conditions either in this system or in other common metal/oxide (M/BOx) systems widely used as supported heterogeneous catalysts where M=Pt, Pd, Au, Ni,… and BOx=Al2O3, SiO2, TiO2, CeO2, ZrO2,… [1] Furthermore, we have earlier reported [15] that the incorporation of nitrogen was almost negligible when the Ag nanostructures were prepared directly on a bare glass substrate, i.e. with no a-Al2O3 buffer layer. In this work, we have confirmed further this result through the test Ag nanostructures produced directly on glass and undergoing the same aging conditions than the ones on a-Al2O3 studied in this work. The main findings of this work with respect to the incorporation of nitrogen together with those achieved in our earlier work [15] can be summarized as follows: - Initially (at 0.5 m), the surface of all studied samples and regions only contains metallic Ag and is free of any trace of nitrogen. - Upon aging, the surface of silver becomes oxidized to Ag(I), the smaller the amount of silver the faster the oxidation and the higher the percentage of oxidized Ag atoms. - Ag(I) is bonded to non oxygenated anions rather than to oxygen forming silver oxides. - Nitrogen appears appearance occurs sooner and its content is higherupon, the smaller the amount of silver, the higher the N content and the sooner it appears (at 7 m in the thinnest studied sample). In addition, it nitrogen tends to saturate upon aging after 15 m. 20 - Nitrogen is not observed either in bare a-Al2O3 supports or glass substrates upon aging as well as in samples having the metal NPs buried in a-Al2O3. - Nitrogen is only clearly detected on the surface of samples in which Ag is on a-Al2O3 supports. - The nitrogen that appears at the surface is forming nitrile species. The above list of evidences suggests that the observed nitrile species are bonded to Ag(I) atoms located at the surface of the silver nanostructures and are produced during their storage in atmospheric conditions rather than during the fabrication of the samples. A first hypothesis to explain the origin of these species is surface pollution of samples. To analyze this hypothesis, we have measured the composition of both the bare and a-Al2O3 coated glass substrates by XPS as well as the inner and outer surfaces of the plastic containers used for the storage of the samples. No traces of nitrogen were ever detected. We have even revised thoroughly the experimental set up and the preparation and storage protocols. No sources of nitriles (C2N2, HCN, inorganic cyanides, or organic nitrile solvents as acrylonitrile, acetonitrile, etc.) could be found in the immediate vicinity of the samples. Finally, we have repeated the preparation several times after our first observation [15], with the same qualitative result. The nitrogen species have always been found although their amount varies slightly from one set of samples to another following the slight changes of the Ag amount and/or nanostructure features. These results allow us ruling out the direct adsorption of any nitrile reagent as possible source for the observed nitrile species. Consequently, the nitrile species must result from a chemical reaction of atmospheric reactants at the Ag nanostructures / a-Al2O3 interface. Three compounds can provide the required nitrogen: N2, NO and NO2. Among them, the N2 molecule is the main component of the atmosphere (78%). However, it is chemically inert and is 21 only known to react with some metals (Cr, Si, Ti, Al, B, Be, Mg, Ba Sr, Ca andy Li) to produce nitrides at very high temperature, or with O2, C and H2 to produce NO, NO2, HCN and NH3 at high temperatures and high pressure. To our knowledge, it has never been reported the reaction of N2 at room temperature and atmospheric pressure with Ag, C or Al2O3. At present, the reaction at room temperature of the N2 molecule with transition metals has only been partially reached within the coordination sphere of highly sophisticated organometallic complexes of metals in the Groups 6 to 9 of the Periodic Table and by means of strong reducing agents [45]. Exposure of Ag to dry nitrogen atmosphere was reported to lead to a red-shift of the SPR of Ag NPs, an effect that was explained in terms of activation of N2 molecules by plasma species (radicals, ions, …) produced prior to the deposition process [46]. In our work, the first exposure to dry N2 occurs just after sample preparation by PLD to bring the samples to atmospheric pressure. In order to explore if this exposure could have play any role on the observed formation of nitrile species upon aging, we have produced one sample under the same conditions than the intermediate thickness sample and brought it to atmospheric pressure by filling the vacuum chamber with high purity Ar. No significant differences could be appreciated between the two samples upon aging including the nitrile content [15]. These results and the aforementioned chemical inertness of N2 lead us to discard reactions of this molecule as responsible for the formation of nitrile adsorbates. The next two nitrogen containing gas molecules in the atmosphere, i.e. NO an NO2, thus become the only realistic potential candidates. These nitrogen oxides, which are generically named NOx, are pollutants produced in small quantities during high temperature combustion and their concentration in the air of highly populated cities can reach levels higher than ~5x10-6 % (or 50 ppb). In order to reduce this concentration, 22 heterogeneous catalysts that promote the reaction of NOx with reductant reagents in the atmosphere, like CO, hydrocarbons or alcohols have successfully been developed [47]. Among them, several formulations based on Ag/Al2O3 catalysts have proved to be particularly active [44,48–50] and although there is still some speculation about the mechanism for NOx reduction, -CN and –NCO species have been proposed as key intermediates [42]. According to these authors, when CO is used as reductant, the process starts with the adsorption of CO and NOx on the same Ag NP where they are both dissociated. Later on, activated C* and N* atoms react and Ag-CN species become formed. Like nitrogen oxides, CO is also a common pollutant in the air of large urban areas where it reaches concentrations of 3x10-3 % or 30 ppm that are nearly three orders of magnitude higher than those typically found of NOx. Consequently, urban air would could provide simultaneously the required CO and NOX to produce the Ag-CN species. However, the great amount of carbon detected on the surface of our samples lead us to consider in addition the existence of adsorbed -CHx species (hydrocarbons) in addition to CO. They are most likely related to the method used for producing the samples. It is well known that PLD involves high energetic ions that lead to surface sputtering as well as sub-surface metal implantation [51]. For the case of Ag and the laser fluence used in this work to produce the samples, ~25% of arriving silver species have kinetics energies >200 eV [52]. This bombardment is responsible for the very dense, flat and free from nanoscopic voids a-Al2O3 layers produced by PLD as reported elsewhere. However, it can also produce oxygen vacancies, coordinately unsaturated Al+3 sites and microstructural defects on the a-Al2O3 surface that can act as sites for the chemisorption of hydrocarbons at the vicinity of the Ag nanostructures. Once hydrocarbons become Comentario [CNA11]: Termino correcto??? RP: yo creo que si. J.P.: Ver en respuesta a los referees la posibilidad de emplear los términos micropores o mesopores. Comentario [CNA12]: Referir a: “ The importance of the energetic species in pulsed laser deposition for nanostructuring”, A Castelo, C N Afonso, E Pesce and E Piscopiello, Nanotechnology 23 (2012) 105603 Ok, cuando realice la version definitive. 23 adsorbed, they can migrate to the surface of the Ag NPs where the formation of nitriles will be catalyzed upon exposure to atmosphere. The hypothesis of hydrocarbons adsorbed on the surface of a-Al2O3 instead of CO directly adsorbed on Ag as a reductant of the NOx species allows us explaining the different concentration of nitriles (catalytic activity) found in the several samples. For both reactants to adsorb it is thus necessary that the amount of Ag is below the percolation limit in order to lead to regions of uncovered a-Al2O3 from where the chemisorbed hydrocarbons can later migrate to the surface of the metal. In other words, the reactivity of the system will tend to saturate as the amount of deposited Ag increases and will be blocked once the thickness of the Ag is above a critical value (above the percolation limit), in agreement with our experimental observations. Further support to this interpretation is provided by the fact that the nitrile surface concentration (N/Ag in Figure 12 and Table 1) is approximately the same in as grown and laser exposed regions in spite of the strong morphological changes induced by the laser treatment on the Ag nanostructure. The results presented in this work in addition evidence that the optical response of the Ag nanostructures exhibits significant changes upon aging that are stronger the smaller the amount of silver. The main findings can be summarized as follows: - The SPR wavelength of as grown regions (discontinuous metal layers) blueshifts upon aging, the smaller the effective thickness of the nanostructures the higher the blue-shift. - The SPR wavelength of laser exposed regions (isolated and almost round nanoparticles) changes much less upon aging than that as grown regions. - The extinction at the UV side of the spectrum of both as grown and laser exposed regions increases significantly upon aging in the thinnest sample. 24 Comparing these findings to the chemical changes and taking into account the fact that the nitrogen incorporation follows the same trend evolution with time in as grown and laser exposed regions, one can conclude that the changes observed in the wavelength of the SPR are not only linked to the nitrile formation. It is well known that the SPR wavelength and width are very sensitive to the dimensions and shape of the nanostructures and the generally SPR generally shifts blue and becomes narrower when the dimensions of the NPs are reduced and/or the NPs approach spheres [53]. This is what we observe in Figure 1 at 0.5 m when comparing laser exposed to as grown regions and this result is consistent with SEM observations (Figure 3 and Figure 5). Upon aging, the most significant structural change is the appearance of elongated features at 7 m in as grown regions of the thinnest sample and at 15 m in the intermediate thickness sample (Figure 3). Comparing their dependence on aging time included in Figure 4a to the dependence of SPR wavelength included in Figure 2, it becomes clear that the changes in the former are not mirrored in the latter. Notice for instance that the elongated features are not observed in the intermediate thickness sample at 7 m (Figure 3e) when a significant shift of the SPR is already observed (Figure 2). It can be thus concluded that the changes in the SPR wavelength are not linked to the appearance of the elongated nanostructures and this result agrees well with our earlier conclusion that the elongated structures have no significant impact on the optical spectra [15]. The origin of these elongated nanostructures and whether they relate to the nitrogen incorporation observed in these samples is at this stage unclear and requires further investigation. On the one hand, the results obtained from SEM analysis on the features of the underlying nanostructure of the as grown regions over which the elongated structures develop show that the metal coverage is reduced and metal agglomeration occurs upon 25 aging (Figure 4b). Agglomeration is a process that leads to the morphology of the nanostructures approaching that of spheres and this process should result in a blue-shift of the SPR wavelength as observed in as grown regions upon aging. On the other hand, the formation of a dielectric overlayer on the surface of the metal nanostructures by silver oxidation is expected to lead to a red-shift of the SPR wavelength as has been reported both upon oxidation and sulfurization [13] due to the higher refractive index of the shell material with respect to that of air. However, in our earlier work [15], we have also observed a blue-shift upon aging rather than a red-one. It was suggested that the blue-shift could relate to the much smaller dimensions of NPs with respect to those reported in literature that lead to a red-shift (few nm diameter compared to tens of nm). The small dimensions imply that the surface to bulk atoms ratio is very high and when oxidation occurs, an oxidized overlayer grows at the expense of the metal effectively reducing the dimension of the metal core. The effect of this reduction is much stronger than that of the refractive index of the overlayer being higher than that of air thus fully compensating the red-shift. The agglomeration of nanostructures in as grown regions together with the fact that the SPR wavelength of the almost round NPs produced by laser irradiation show a very reduced shift upon aging supports further this conclusion since the surface-to-volume atom ratio is minimum for the almost round NPs produced by laser irradiationinduced dewetting. The extinction at the UV side of the spectrum increases strongly upon aging in the thinnest sample that is the one that also becomes strongly oxidized. Furthermore, this increase is similarly high in both as grown and laser exposed regions (see Figure 1b) and both regions show similar amount of nitrogen incorporation (see Figure 12). Therefore, the extinction at the UV side of the spectrum must be linked to the chemical changes. Since this extinction has been reported to be a means to distinguish hetero 32 E 2014 Toward a Silver–Alumina Model System for NO x Reduction Catalysis J. Phys. Chem. C 118 24556–61 [50] Ralphs K, Chansai S, Hardacre C, Burch R, Taylor S F R and James S L 2015 Mechanochemical preparation of Ag catalysts for the n-octane-SCR de-NOx reaction: Structural and reactivity effects Catal. Today 246 198–206 [51] Perea A, Gonzalo J, Budtz-Jorgensen C, Epurescu G, Siegel J, Afonso C N and Garca-Lpez J 2008 Quantification of self-sputtering and implantation during pulsed laser deposition of gold J. Appl. Phys. 104 084912 [52] Baraldi G, Perea A and Afonso C N 2011 Dynamics of ions produced by laser ablation of several metals at 193 nm J. Appl. Phys. 109 043302 [53] Lee K-C, Lin S-J, Lin C-H, Tsai C-S and Lu Y-J 2008 Size effect of Ag nanoparticles on surface plasmon resonance Surf. Coatings Technol. 202 5339–42 [54] Xiao-Yong G, Hong-Liang F, Jiao-Min M and Zeng-Yuan Z 2010 Spectroscopic ellipsometric study of the optical properties of Ag 2 O film prepared by directcurrent magnetron reactive sputtering Chinese Phys. B 19 90701 [55] Encina E R and Coronado E A 2010 On the Far Field Optical Properties of Ag−Au Nanosphere Pairs J. Phys. Chem. C 114 16278–84 [56] Sheikholeslami S, Jun Y, Jain P K and Alivisatos A P 2010 Coupling of Optical Resonances in a Compositionally Asymmetric Plasmonic Nanoparticle Dimer Nano Lett. 10 2655–60 [57] Sazama P, Capek L, Drobna H, Sobalik Z, Dedecek J, Arve K and Wichterlova B 2005 Enhancement of decane-SCR-NO over Ag/alumina by hydrogen. Reaction kinetics and in situ FTIR and UV–vis study J. Catal. 232 302–17 33 Table 1: Silver photoelectronic energy parameters (Ag3d5/2 binding energy, AgM4VV kinetic energy and silver modified Auger parameter (*)), percentage of silver in metallic state (Ag(0)), surface composition (atomic %) and some relevant atomic ratios obtained from X-ray photoemission spectral peaks for both as grown and laser exposed regions at different aging stages. Samples Ag3d5/2 AgM4VV (*) α* (eV) % Ag(0) Surface Composition (Atomic %) Atomic ratios Time (m) Thick (nm) Binding Energy (eV) FWHM (eV) Kinetic Energy (eV) Ag(0) Ag(I) O Al Ag C N S Ag/Al N/Ag S/Ag O/(1/2Ag +3/2Al) As grown 0.5 1.6 368.4 1.60 357.1 725.5 100 30.7 17.8 14.6 36.9 0.0 0.0 0.83 0.0 0.0 0.90 3.7 368.3 1.45 357.3 725.6 100 29.3 13.7 21.2 35.8 0.0 0.0 1.57 0.0 0.0 0.94 5.9 368.2 1.40 357.5 725.7 100 18.1 8.3 25.1 48.5 0.0 0.0 3.07 0.0 0.0 0.72 7 1.6 368.7 2.15 357.3 353.3 726.0 722.0 40 19.1 11.3 14.5 50.8 3.9 0.4 1.28 0.27 0.03 0.79 3.7 368.3 1.65 357.5 725.8 100 17.8 9.0 23.6 46.9 2.3 0.4 2.61 0.10 0.02 0.70 5.9 368.3 1.40 357.6 725.9 100 15.6 5.6 26.9 50.5 1.4 0.3 4.68 0.05 0.01 0.71 15 1.6 368.9 2.00 357.2 353.2 726.1 722.1 20 17.2 11.0 15.1 50.0 6.2 0.4 1.36 0.41 0.03 0.71 3.7 368.3 1.85 357.5 725.8 75 16.6 9.0 20.3 51.3 2.5 0.3 2.25 0.13 0.02 0.70 5.9 368.3 1.40 357.5 725.8 100 15.1 6.0 27.0 50.0 1.4 0.4 4.32 0.05 0.02 0.67 20 1.6 368.9 2.00 357.2 353.2 726.1 722.1 20 16.6 10.6 14 52.9 5.4 0.5 1.32 0.39 0.04 0.72 3.7 368.3 1.85 357.5 725.8 75 13.6 6.4 19.0 58.2 2.4 0.4 2.97 0.12 0.02 0.71 5.9 368.3 1.40 357.5 725.8 100 12.2 4.25 22.0 59.5 1.5 0.6 5.1 0.06 0.03 0.70 Laser exposed 0.5 1.6 368.3 1.45 357.1 725.4 100 34.0 21.8 10.7 33.5 0.0 0.0 0.49 0.0 0.0 0.89 5.9 368.4 1.30 357.2 725.6 100 30.9 20.6 12.6 35.9 0.0 0.0 0.62 0.0 0.0 0.83 7 1.6 368.1 1.95 357.4 725.5 90 24.8 15.9 13.3 43.5 2.1 0.4 0.84 0.16 0.03 0.84 5.9 368.3 1.30 357.4 725.7 100 24.2 15.5 13.9 45.7 0.4 0.3 0.90 0.03 0.02 0.80 15 1.6 368.6 2.30 357.1 353.1 725.6 721.7 40 17.0 10.5 10.3 57.6 4.1 0.4 0.98 0.41 0.04 0.81 5.9 368.2 1.30 357.5 725.7 100 21.6 14.9 16.5 46.3 0.5 0.3 1.11 0.03 0.02 0.70 34 Table 2: C1s fitting parameters of photoemission peak for as grown regions of the 1.6 nm thick sample at different aging stages. Time (m) CI CII CIII CIV ΔCII/Δ N Energy (eV) FWHM (eV) % Energy (eV) FWHM (eV) % Energy (eV) FWHM (eV) % Energy (eV) FWHM (eV) % 0.5 284.6 1.6 72.3 285.3 1.6 16.8 286.8 1.6 3.8 288.1 2.1 7.1 0 7 284.7 1.6 66.0 285.3 1.6 24.2 286.6 1.6 4.7 288.3 2.1 5.0 1.6 15 284.7 1.6 64.7 285.4 1.6 26.4 286.8 1.6 4.6 288.5 2.1 4.3 1.1 35 FIGURE CAPTIONS FIGURE 1. Extinction spectra of (full line) as grown and (dashed lines) laser exposed regions at F=152 mJcm-2 of the (a) 5.9 nm and (b) 1.6 nm thick samples and measured at (black) 0.5 m and (grey) 15 m. SEM images at same magnification of the nanostructure of the thickest sample at 0.5 m (c) as grown and laser exposed regions at (d) 145 mJcm-2, (e) 158 mJcm-2 and (f) 196 mJcm-2. The inset shows the evolution of the SPR wavelength as a function of laser exposure fluence in the thickest sample at (black) 0.5 m and (grey) 15 m; the SPR values at 0 fluence correspond to as grown regions. FIGURE 2. SPR wavelength as a function of aging time for (full lines and symbols) as grown and (dashed lines and open symbols) laser exposed regions at 152 mJcm-2 in the (,) 5.9 nm, () 3.7 nm and (,) 1.6 nm thick samples. Lines are guidelines. FIGURE 3. SEM images at same magnification of as grown regions of the (a-c) 1.6 nm, (d-f) 3.7 nm and (g-i) 5.9 nm thick samples after (a,d,g) 0.5 m, (b,e,h) 7 m and (c,f,i) 15 m. FIGURE 4. (a) (full symbols and lines) Mean Feret length and (open symbols and dashed lines) aspect ratio of the elongated structures seen in Figure 3b and 3c and f as a function of aging stage for as grown regions of the (,) 1.6 nm and (,) 3.7 nm thick samples. (b) Metal coverage and (c) mean Feret diameter of the nanostructures seen in SEM images as a function of aging stage for (full symbols and lines) as grown and (open symbols and dashed line) laser regions of the () 3.7 nm and (,) 5.9 nm thick samples. Lines are guidelines. 36 FIGURE 5. SEM images at same magnification of laser exposed regions of the (a,b) 1.6 nm and (c,d) 5.9 nm thick samples after (a,c) 0.5 m and (b,d) 15 m. FIGURE 6. X-ray photoemission survey spectra of as grown regions of the 1.6 nm thick sample at (black) 0.5 m and (grey) 7 m. FIGURE 7. High energy resolution X-ray photoemission spectral region of (a,c) Ag3d and (b,c) AgMNN at 0.5 m of the (a, b) as grown and (c, d) laser exposed regions of the (green) 1.6 nm, (red) 3.7 nm and (black) 5.9 nm thick samples. FIGURE 8. High energy resolution X-ray photoemission spectral regions of (a) Ag3d and (b) AgMNN, of as grown regions of the 1.6 nm thick sample at (black) 0.5 m, (red) 7 m and (green) 15 m and (blue) oxidized reference sample. FIGURE 9. High energy resolution X-ray photoemission spectral regions of (a) S2p, (b) C1s and (c) N1s of as grown regions of the 1.6 nm thick sample at (black) 0.5 m, (red) 7 m and (green) 15 m. FIGURE 10. Ag to Al atomic ratio as a function of thickness for (full symbols and lines) as grown and (open symbols and dashed lines) laser exposed regions at (,) 0.5 m and (,) 7 m. Lines are guidelines. FIGURE 11. Fitting analysis of the C1s photoemission spectral region of as grown regions of the thinnest sample at 0.5 m. See text for assignation of the carbon chemical states. Comentario [EMJP14]: Efectivamente, había un error en las posiciones relativas de las subfiguras b) y c). Ahora, el pie de figura coincide con el texto. Y he modificado la figura, incluyendo etiquetas de región espectral y de tiempo de exposición. 37 FIGURE 12. N to Ag atomic ratio as a function of aging stage for (full symbols and lines) as grown and (open symbols and dashed lines) laser exposed regions of the (,) 5.9 nm, () 3.7 nm and (,) 1.6 nm thick samples. Lines are guidelines. FIGURE 13. Raman spectra of as grown regions at 7 m of the (black) 5.9 nm and (green) 1.6 nm thick samples (full line) with and (dashed line) without a-Al2O3 buffer layer.