Comprehensive study of plasmonic materials in the ... - ACS Publications

properties for eight different plasmonic materials, namely Ag, Al, Au, Cu , Mg, Ni,. Pd, and ... nanostructures made from these materials up to 1100â—...
0 downloads 0 Views 4MB Size
Article Cite This: ACS Photonics XXXX, XXX, XXX−XXX

Comprehensive Study of Plasmonic Materials in the Visible and NearInfrared: Linear, Refractory, and Nonlinear Optical Properties Gelon Albrecht,*,†,‡ Monika Ubl,‡ Stefan Kaiser,†,‡ Harald Giessen,‡ and Mario Hentschel‡ †

Max Planck Institute for Solid State Research, Heisenbergstraße 1, 70569 Stuttgart, Germany 4 Physics Institute and Research Center SCoPE, University of Stuttgart, Pfaffenwaldring 57, 70569 Stuttgart, Germany

‡ th

S Supporting Information *

ABSTRACT: Plasmonic nanostructures are used today for a variety of applications. Choosing the best suited plasmonic material for a specific application depends on several criteria, such as chemical and thermal stability, bulk plasma frequency, nonlinear response, and fabrication constraints. To provide a comprehensive summary, we compare these properties for eight different plasmonic materials, namely, Ag, Al, Au, Cu, Mg, Ni, Pd, and Pt. All these materials can be fabricated with electron beam lithography and subsequent evaporation of the desired material. First, we heated rod-antenna-type nanostructures made from these materials up to 1100 °C in air and investigated their linear optical response. Most structures lose their plasmonic properties at temperatures far below the melting point of the respective material. Gold, silver, and platinum structurally deform, whereas the other materials appear to chemically degrade. Second, to improve the thermal stability, structures with a 4 nm thin Al2O3 capping layer are fabricated. The thermal stability is significantly increased with the capping layer for all materials except for copper and magnesium. Lastly, the laser damage threshold is investigated for silver, aluminum, gold, and copper, which exhibit high nonlinear optical susceptibilities and are therefore particularly interesting for nonlinear optical applications. KEYWORDS: plasmonics, thermal stability, third-harmonic generation, linear and nonlinear properties, material comparison

I

For real world applications further aspects are relevant. First, the chemical stability is of major importance. Most materials are affected by oxidization or hydration and lose their plasmonic properties. Basically only gold and platinum are unaffected by these processes and are long-term stable under ambient conditions. Second, the thermal stability may play a major role. Many applications such as nonlinear optics or photovoltaics use the increased absorption at the plasmon resonance. Therefore, the plasmonic material can be heated locally to high temperatures. Some materials such as gold are known to reshape at temperatures as low as 100 °C,28 which can easily be reached locally. Owing to the direct connection of geometrical shape to the properties of the plasmon resonance, this deformation leads to a low thermal stability. Besides the dimensional stability, the chemical stability also plays an important role at elevated temperatures, because higher temperatures lead to an increased chemical reaction speed. Third, for nonlinear applications the nonlinear optical susceptibility should be high. However, simultaneously the linear optical properties are also of significant importance owing to the local field enhancement. Last, the complexity to realize plasmonic nanostructures of high material quality varies strongly for different materials due to fabrication constraints. It is not straightforward to achieve long-term stability of plasmonic structures in a harsh environment. All bare

n recent years nanostructured plasmonic materials went from pure academic interest to applications. Localized surface plasmon resonances can be easily tailored to exhibit a specific optical response. Therefore, they are used in versatile applications such as plasmonically enhanced solar cells,1 cancer therapy,2 and optical sensing applications.3 Most of these applications, in fact, use gold as plasmonic material, as it exhibits excellent optical plasmonic properties, is chemically inert, and can easily be nanostructured. However, gold is not the only material with plasmonic properties. Alongside the development of the applications, many new plasmonic materials were presented. For good optical plasmonic properties, a large free charge carrier concentration is necessary.4 Therefore, most plasmonic materials are either metals or (doped) semiconductors. More exotic materials are Dirac systems such as topological insulators or graphene that can support edge state plasmons.5,6 Metallic systems exhibiting plasmonic resonances include silver,7−9 aluminum,10,11 copper,12 gallium,13 magnesium,14−16 molybdenum,17 nickel,18 palladium,19 platinum,20 and tungsten.17,21 Also nitrides such as TiN and ZrN22 and some hydrides such as YH223 provide good plasmonic resonances. Additionally, metal oxides such as WO3 and MoO3 as well as transition metal oxides such as ReO3 or VO2 have been demonstrated to display plasmonic properties.24 Furthermore, semiconductors such as transparent conducting oxides (ITO, AZO),25 germanium,26 and InSb27 are used as plasmonic materials. From the group of topological insulators mainly bismuth telluride selenide compounds are investigated.6 © XXXX American Chemical Society

Received: November 8, 2017 Published: January 3, 2018 A

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX

Article

ACS Photonics

Figure 1. Overview of the investigated materials. Top: Materials ordered according to their melting points.34 Bottom: Transmittance spectra of the different materials for the same physical dimensions of the rod structures (260 nm length, 60 nm width, 50 nm height, for gold 40 nm height) with and without a 4 nm alumina cover. The periodicity is 500 nm for all materials for both array dimensions.

commonly used materials exists, beyond linear spectroscopy data for selected materials.14,20,33 Furthermore, we applied an alumina protection layer to all of these materials to increase their stability. Data for the more uncommon plasmonic materials germanium and YH2 can be found in the Supporting Information. The melting point of the investigated materials ranges from 649 °C for magnesium to 1769 °C for platinum.34 In Figure 1 all investigated materials as well as their melting points are depicted. Titanium and chromium were only used as adhesion layers because they do not exhibit a plasmon resonance in the investigated wavelength range. Additionally, the linear transmittance spectra for a given rod shape nanostructure are shown. All rods are designed to be 260 nm long, 60 nm wide, and 50 nm high. An exception is gold, with a height of 40 nm. Scanning electron microscope (SEM) images can be found in Figures 2, 3, and 4. Transmittance spectra for different rod lengths of all materials can be found in the Supporting Information. We prepared all structures using electron lithography and subsequent evaporation of the different materials. As substrate UV transparent Suprasil is used. Additionally, Suprasil exhibits a rather high softening temperature of 1600 °C. We investigated sapphire substrates before.31 However, we could not observe significant differences in the thermal stability. We choose Suprasil because of its lower refractive index (1.45 compared to 1.75). Nanostructures resonant at a certain wavelength have to be smaller on sapphire compared to Suprasil and accordingly exhibit lower dipole strength and a less modulated plasmon resonance. Based on previous experience and the results found during this study, the materials are prepared with different evaporation techniques and different adhesion and wetting layers in order to obtain the best possible material quality. The evaporation is done via electron beam evaporation for the material combinations Cr/Al, Cr/Cu, Cr/Ni, Cr/Pd, and Cr/ Pt. The thicknesses for all these combinations are 2 nm of chromium and 50 nm of the respective plasmonic material. For silver we use germanium as wetting layer, due to its reported superior growth on germanium.35 The evaporation is

nanostructured materials that we are aware of start to either deform or chemically deteriorate in air at temperatures above 500 °C. To increase the thermal stability, refractory materials such as TiN, ZrN, or tungsten were proposed,17,29 which are materials exhibiting a melting point higher than 2000 °C. However, especially the nitride-based materials are very prone to chemical degradation.30 Even the plasmonic resonances of nanostructured materials with extremely high melting points such as tungsten were reported to degrade rapidly at 600 °C in the presence of air.17 The most common way to improve the stability is using a protective coating. In a previous publication,31 we demonstrated the significantly increased thermal stability of gold nanostructures covered with a 4 nm thin alumina (Al203) protection layer. Alumina is a very durable dielectric material with a melting point of 2072 °C32 and additionally has a high chemical stability. Furthermore, alumina films deposited via atomic layer deposition (ALD) with thicknesses as thin as 4 nm form pore-free films. The advantage of ALD over other deposition methods such as evaporation or sputtering is the high homogeneity of the film. This film thickness is thin enough that the plasmonically enhanced near field reaches the outside of the protective layer and can be used for applications such as refractive index sensing. The oxide in the alumina layer is tightly bound and does not react with most materials. An exception is magnesium, where magnesium oxide is thermodynamically favorable to alumina. The alumina may be replaced with other materials such as HfO2 that can also be deposited with ALD.17 In addition to ALD also methods such as metalorganic vapor phase epitaxy (MOVPE) or self-assembled layers should provide highly homogeneous layers. For materials that react strongly with oxygen such as magnesium, oxygen-free cover materials such as Si3Ni4 may be advantageous. However, fabrication is more complicated. We compare in this article rod-shaped nanostructures made of silver, aluminum, gold, copper, magnesium, nickel, palladium, and platinum with respect to their linear plasmon resonance, their nonlinear optical properties, and their thermal and photostability. Until now, no comprehensive review of these B

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX

Article

ACS Photonics

compared to air. The largest shifts are visible for magnesium and copper. The covering process involves heating of the samples to 250 °C. Therefore, the larger shift may indicate some thermal degradation. Moreover, a possible explanation is related to the increased effective surface area of grainy rods, which might lead to a larger spectral shift when changing the effective refractive index. Additionally, it should be mentioned that the material dispersion of the different materials is different, which might also partially explain the different spectral shifts. For copper samples prepared under poor vacuum conditions (>10−5 mbar), we observed strongly deteriorated plasmon resonances after the ALD process. In the following paragraphs the thermal stability of the materials is investigated. The rod dimensions are as described above. The transmittance spectra of the covered and uncovered structures are measured after the preparation and after subsequent heat treatments at 500, 800, and 1100 °C in a furnace in air. The samples are kept at each temperature for 1 h. The shape of the sample is observed with a scanning electron microscope before the heat treatment and after the loss of the plasmonic resonance. We ordered the materials in three categories based on their melting point: low melting point for magnesium and aluminum; intermediate melting point for silver, gold, and copper; high melting point for nickel, palladium, and platinum. As first materials, magnesium and aluminum are investigated. Both exhibit a low melting point of 649 °C for magnesium and 660 °C for aluminum. Magnesium is a highly reactive material that quickly reacts with oxygen and water and loses its plasmonic properties. The added benefit of this high reactivity is the possibility to use magnesium for sensing applications, most prominent optical hydrogen sensing.15 The optical transmittance measurements as well as SEM images for both materials are depicted in Figure 2. Freshly prepared magnesium shows a reasonably well modulated plasmon resonance that appears to be stable under ambient conditions for several days, however exhibits deterioration of the plasmon resonance after about a month. Heating the sample to 500 °C strongly deteriorates the plasmon resonance immediately. Accordingly, the SEM micrograph shows a strong degradation of the structures. The material appears to be partly detached, and only fractions of the rods are still intact. For the SEM images it has to be stressed that it is not possible to use the usual spin-on conductive layer (Espacer (Showa Denko) or Electra (Allresist)) on top of magnesium. Both dissolve the magnesium nanostructures. For the uncovered structures the structure is instead metalized with a 2 nm chromium layer for SEM imaging. Accordingly, the SEM image before heat treatment is from a different sample with the same fabrication parameters. Covering the magnesium structures with alumina leads to a red-shift of the plasmon resonance, as expected. The modulation depth and the quality factor of the resonance are maintained. Heating the covered structures to 500 °C also results in a complete loss of the plasmonic resonance. The shape of the nanostructure appears in the SEM images unchanged compared to the initial shape. The standard electron potential of magnesium is far lower than that of aluminum, making it thermodynamical more favorable to form magnesium oxide and aluminum out of the magnesium and alumina layer. Therefore, we expect that the degradation is due to the chemical reaction of the magnesium with the alumina. From this observation no improvement in thermal

performed via electron beam evaporation as well. The thickness of the germanium wetting layer is 1 nm, and the silver layer is 50 nm thick. We also fabricated silver on chromium as wetting layer. However, the growth of silver on chromium is strongly hindered, and we could not fabricate plasmonic rod structures with our standard fabrication parameters. For detailed information please refer to the Supporting Information. For magnesium we use titanium as adhesion layer. Titanium does not alloy with magnesium,36 which inhibits deterioration of the plasmonic resonance. The Ti/Mg is evaporated in a thermal evaporator with a thickness of 5 nm for titanium and 50 nm for magnesium. The Cr/Au system is evaporated in a thermal evaporator with a thickness of 2 nm for chromium and 40 nm for gold. The necessary vacuum conditions to fabricate nanostructures with a good plasmonic resonance vary strongly. Chromium and gold can be evaporated at pressures around 10−5 mbar with no effect on the plasmon resonance, whereas magnesium needs about 2 orders of magnitude lower pressures. Furthermore, aluminum and copper showed inferior thermal stability and plasmonic resonances for pressures above 5 × 10−6 mbar. A summary of the evaporation parameters can be found in the Supporting Information. More detailed studies on the growth of selected metallic films can be found in McPeak et al.37 For all materials each sample contains several arrays of rods varying in length. The array periodicity is 500 nm with a total size of 80 μm × 80 μm. All samples are prepared in pairs. From a sample pair, one sample is covered with a 4 nm thick alumina (Al2O3) layer. The cover is applied at a temperature of 250 °C with atomic layer deposition (Cambridge NanoTech Savannah 100) using water and trimethylaluminum (TMA) as precursor gases. The resulting growth rate is around 0.1 nm per cycle. Based on the transmittance spectra presented in Figure 1 the linear optical properties of the different materials can be compared. The linear optical spectroscopy is performed with a commercial FTIR spectrometer (Bruker Vertex 80) using a SiC Globar as light source. All linear measurements are performed with light polarized along the long axis of the nanostructures. The plasmon resonance along the short axis of the rod is located at around 500 nm. A good plasmonic material exhibits a plasmon resonance with a narrow full width at half-maximum (fwhm) and a high modulation depth. This indicates a strong near-field enhancement due to the high quality factor and a large interaction cross section. This is the case for silver, gold, and copper. Magnesium and aluminum exhibit intermediate optical properties, whereas nickel, palladium, and platinum exhibit the spectrally broadest resonances with the smallest modulation depth. The reduced oscillator strength is partially related to a reduced quasi-free electron density in the different materials. The higher the number of quasi-free conduction electrons, the stronger the plasmon mode. Good conductors therefore make the best plasmonic materials. The increased fwhm is related to larger ohmic losses, which reduce the resonance lifetime. The transmittance spectra of the nanostructures are additionally simulated based on literature values for the different materials. The simulations are shown in the Supporting Information and are in good agreement with the measurements. Additionally, the change from the bare structure to an alumina-covered structure is depicted in Figure 1. The plotted spectra are from the same structures before and after alumina covering. For all materials a small red-shift is visible, caused by the higher refractive index of alumina (n = 1.75 at 1000 nm) C

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX

Article

ACS Photonics

Figure 2. Structural and optical properties of magnesium and aluminum structures after different heat treatments. The melting point of bulk material is given with the chemical symbol. From the top: Bare Mg, Mg with 4 nm alumina cover, bare Al, and Al with 4 nm alumina cover. The left column displays the transmittance spectra before and after thermal treatment. The two right columns depict SEM micrographs after the fabrication of the nanostructures and after the loss of the plasmon resonance. The scale bar is 200 nm.

Figure 3. Structural and optical properties of silver, gold, and copper after different heat treatments. The melting point of bulk material is given with the chemical symbol. From the top: Bare Ag, Ag with 4 nm alumina cover, bare Au, Au with 4 nm alumina, bare Cu, and Cu with 4 nm alumina cover. The left column shows the transmittance spectra before and after thermal treatment. The two right columns depict SEM micrographs after the fabrication of the nanostructures and after the loss of the plasmon resonance. The scale bar is 200 nm.

stability by the alumina cover can be deduced for magnesium structures. The second low-melting material is aluminum. The structured material exhibits a good plasmonic resonance. Aluminum is oxidized in air; however, it forms a selfterminating surface oxide layer of 2.5−3 nm.38 The plasmonic resonance of the uncovered aluminum strongly degrades during the 500 °C temperature step. The SEM images suggest that the geometrical shape is intact, indicating a chemical change in the aluminum. For the covered structures, the thermal stability is clearly enhanced. After the 500 °C treatment the plasmon resonance has shifted its center wavelength and decreased in amplitude, but the plasmon resonance is still clearly observable, in stark contrast to the uncovered case. Since we assume a chemical reaction as cause for the deterioration, most likely the additional oxide layer reduces the diffusion rate and increases the thermal stability. After the 800 °C temperature step also the plasmon resonance for the covered aluminum vanishes. Comparing the SEM micrographs before and after the thermal treatment shows an expansion of the structure, whereas the shape is maintained. For both low melting point materials we observe a loss of the plasmonic resonance before the destruction of the geometrical shape. This behavior indicates that the materials chemically degrade, even for the covered cases. The second material class comprises silver, gold, and copper. Their melting points are around 1000 °C. The transmittance spectra and SEM micrographs are displayed in Figure 3.

Silver has the lowest melting point of this group. Freshly prepared silver nanostructures exhibit very narrow plasmon resonances with excellent modulation depth. We observed the nanostructured silver to be stable under ambient conditions for at least 2 weeks. Minor changes in the transmittance spectrum of the bare silver structures can be observed. The center wavelength shifts for various structures around 5 nm in 2 weeks. The covered structures exhibit a shift of around 2 nm in 2 weeks. Based on our experience the reproducibility of the measurements due to alignment imperfection is around 1 nm, which might explain the shift for the covered structures; however the small shift for the uncovered structures is very likely due to changes in the silver. The respective data can be found in the Supporting Information. The plasmon resonance of the uncovered silver structures vanishes after the first temperature step to 500 °C. SEM images show that the structures are strongly deformed and the shape has changed from the initial rods to discs. This indicates an Ostwald ripening process.39 Whether the silver has additionally chemically altered remains unclear. The germanium layer may D

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX

Article

ACS Photonics influence the adhesion properties in comparison to the chromium layers. As indicated before, germanium was chosen as wetting layer due to the poor growth properties of silver on chromium. For a comparison of germanium and chromium wetting layers, please refer to the Supporting Information. The alumina cover improves the thermal stability of the silver structures significantly. There is a rather small change in the center wavelength and amplitude of the plasmon resonance after the first temperature step to 500 °C. However, the structures completely vanish after the next temperature step to 800 °C. Neither in the optical microscope images nor in the SEM micrographs can any remaining structures be found; even large structures of 100 μm size vanished completely. We assume that the protective layer is cracked and the silver is redistributed across the sample and accumulates at larger dirt particles or millimeter scale marker structures, where we cannot differentiate it from the original material. The main results for gold were previously presented.31 The uncovered gold structures are nearly unaffected by heating to 500 °C. After the 800 °C step the structures have deformed and exhibit a plamonic resonance at lower wavelengths. The covered gold structures exhibit even superior temperature stability, remaining unaffected up to 800 °C. Heating them to 1100 °C, however, results in a complete loss of the plasmonic response. At higher temperatures the protective layer is most likely destroyed and the structures vanish. SEM inspection indeed shows only strongly deteriorated structures. We demonstrated31 that a 40 nm thick protective layer can confine the structures even at temperatures up to 1100 °C; however they strongly deform inside the protective layer. The last material of this category is copper. Similarly to silver and gold, copper exhibits good plasmon resonances. The material also seems to be stable under ambient condition for at least 1 week without any immediate degradation despite its larger chemical reactivity. Heating the uncovered copper structures to 500 °C leads to a complete loss of the plasmonic response. SEM inspection shows that the shape of the antennas is maintained. We thus assume that the material has chemically degraded. Covering the copper structures with an alumina layer does not improve thermal stability. After heating the structures to 500 °C the plasmon resonance has as well vanished. When inspecting the structures in an SEM, we observe a fragmentation of the structures despite the alumina layer. We can thus conclude that copper is a good plasmonic material under ambient conditions; yet it is not suitable for even slightly elevated temperatures. The category of the high melting point materials is composed of nickel, palladium, and platinum. All these materials exhibit a melting point far above the last heating step of our used temperature cycle at 1100 °C. The transmittance spectra and SEM micrographs are shown in Figure 4. Freshly prepared nickel structures exhibit a rather unpronounced plasmon resonance. The oscillator strength as well as the quality factor are significantly lower than for the previously discussed materials. This reduced performance is most likely linked to the larger intrinsic material losses of nickel when compared to, for example, gold. Heating the bare nickel structures to 500 °C results in a complete loss of the plasmonic response. Because the melting point is significantly higher, we expect that the structures oxidized. SEM inspection indeed reveals that the shape of the structures is nearly retained, yet grains have formed on the nickel surface. A chemical degradation is the most likely explanation. After covering

Figure 4. Structural and optical properties of nickel, palladium, and platinum after different heat treatments. The melting point of bulk material is given with the chemical symbol. From the top: Bare Ni, Ni with 4 nm alumina cover, bare Pd, Pd with 4 nm alumina, bare Pt, and Pt with 4 nm alumina cover. The left column shows the transmittance spectra before and after thermal treatment. The two right columns depict SEM micrographs after the fabrication of the nanostructures and after the loss of the plasmon resonance. The scale bar is 200 nm.

with alumina, the transmittance spectrum remains nearly unchanged after the 500 °C step. Yet, the structure deteriorates after the second temperature step to 800 °C. Similarly to the uncoverd case, it appears that grains have grown on the surface of the nickel structure. Palladium exhibits the next higher melting point. Similarly to nickel, palladium exhibits broad plasmonic resonances of low modulation and quality factor. Again, we assume that larger intrinsic losses are the main reason. The uncovered palladium structures show a loss of the plasmon resonance at 500 °C, whereas they appear dimensionally stable in the SEM micrographs. Therefore, palladium appears to degrade chemically at rather low temperatures. The alumina cover helps to reduce the chemical degradation. The loss of the plasmon resonance is only observed at the temperature step to 800 °C. As in the uncovered case, the structures have retained their geometrical shape, making a chemical degradation likely. Finally, platinum has the highest melting point of all investigated materials and is chemically as inert as gold. The E

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX

Article

ACS Photonics plasmonic resonance is of higher quality as compared to nickel and palladium, yet shows lower modulation and quality factor when compared to gold. Heating the uncovered structures to 500 °C results in a slight increase in the quality of the plasmonic resonance, most likely due to a sintering effect, which has also been observed previously for gold structures.40 Heating the structures further to 800 °C leads to a structural deformation and thus to a shift of the plamon resonance. However, the plasmonic response is retained and the shift can be fully ascribed to the shape change of the nanostructure. For the covered structures we observe once more an increase in resonance quality after heating to 800 °C. Structural degradation only sets in at the last temperature step to 1100 °C. Compared to the case of gold, the platinum structures have not changed their shape and single rods appear to be still intact. However, based on the strong contrast differences between different rods most of them seem to have lost a substantial amount of the platinum. In addition, we fabricated platinum rods with a 40 nm alumina cover layer. The plasmon resonance persists even after heating at 1100 °C. However, the center wavelength shifts significantly between all heating steps. The spectra can be found in the Supporting Information. To sum up, platinum and gold form the most stable structures. Silver, aluminum, nickel, and palladium can be stabilized with the alumina cover up to 500 °C. Magnesium and copper do not benefit from an alumina cover layer and still degrade at 500 °C. The high stability of platinum and gold meets the expectation41 and so does the poor stability of magnesium. Especially the poor stability of copper may be surprising. Copper belongs to the same subgroup as silver and gold and due to its filled d-band.41 Basically, no significant influence of the melting point on the thermal stability is observed. Both gold with a melting point of 1063 °C and platinum with a melting point of 1769 °C start to deform at temperatures of 800 °C. Silver might be stabilized with different wetting layers. For all other materials the chemical degradation appears to be the key issue for the thermal stability, which might be improved with different protection layers. An important factor for the dimensional stability is the adhesion of the material to the substrate. Thin adhesion layers are frequently used to improve the mechanical stability of layers on different substrates. Gold, for example, is known to adhere only weakly on bare glass substrate. Wetting layers, on the other hand, are used to improve the growth of the subsequent layer. One material can also act for both purposes at the same time. Both types of layers may have an influence on the thermal stability. Two samples of gold rods are prepared to investigate the influence of the chromium adhesion layer. One is fabricated with a 2 nm chromium adhesion layer, and the other one is prepared without this adhesion layer. The comparison is shown in Figure 5. The transmittance spectra are measured after preparation and after several heating steps on a hot plate ranging from 50 to 500 °C. Panel (a) depicts the linear spectra for both cases directly after preparation (black) and after the 500 °C heating step (magenta). As discussed previously, in the case of the 2 nm chromium adhesion layer, the temperature treatment leads to an increased quality of the plasmon resonance, regarding both modulation depth and quality factor increase. As expected, the stability of the gold structures without the adhesion layer is severely reduced. The heating leads to a strong resonance blueshift and a reduction in modulation that is caused by

Figure 5. Effect of the chromium adhesion layer on the thermal stability of gold rods. The left column depicts the measured data for arrays of gold rods with a chromium adhesion layer on a Suprasil substrate. The right column depicts the data for bare gold rods on a Suprasil substrate without adhesion layer. (a) Transmittance spectra after fabrication at T = 20 °C and after a heating cycle up to 500 °C for a rod length of 320 nm. fwhm and the center wavelength of the plasmon resonance are subsequently extracted from the spectrum for all temperatures of the cycle and for three different antenna lengths. These are shown as a function of temperature in panel (b).

deformation of the nanostructures. The quality factor for the freshly prepared structures is slightly larger in the case of no adhesion layer. This behavior can be explained by an increased damping due to the lossy chromium layer, which is in accordance with previous reports.42,43 To gain further insight and to determine whether or not the effect is size dependent, we show the results for all temperature steps and three selected rod lengths of 140, 240, and 320 nm in Figure 5(b). For the sake of clarity, the fwhm and the center wavelength are extracted from the transmittance spectra as relevant quantities. For the structures with a chromium adhesion layer, a small and uniform change in the center wavelength is observed for all lengths. Moreover, the fwhm is decreasing for all three lengths. For the sample without an adhesion layer, the decrease in center wavelength is substantially stronger. Furthermore, an increase in the fwhm is visible, which indicates an inhomogeneous structural deformation and broadening of the size distribution of the nanostructures. As the trend is very similar for all rod lengths, a major influence of the structure size is unlikely. Most plasmonic applications use optical methods to interact with the plasmon. Therefore, besides the thermal and chemical F

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX

Article

ACS Photonics

Figure 6. (a) Experimental setup for third-harmonic spectroscopy. (b) Construction of THG over time. The left panel shows the linear spectrum of silver rods before (black) and after (magenta) laser exposure. The laser spectrum is depicted in gray. The middle panel shows the initial TH spectrum (black) and after 15 min of laser exposure (magenta). The right panel shows the evolution of the integrated TH intensity over time. (c) Linear and nonlinear response of silver, aluminum, gold, and copper. The time evolution of the THG and the transmittance spectrum before and after laser exposure are depicted for each material in the covered and uncovered version.

the intensity can be attenuated to 0.3 GW/cm2. The TH is detected with a Peltier-cooled charge-coupled device. In front of the spectrometer the fundamental beam is suppressed by a 6 mm thick Schott KG5 and a 3 mm thick Schott KG4 filter. The linear properties are measured with a commercial FTIR spectrometer before and after laser exposure. All measurements are performed with light polarized along the long axis of the nanostructures. The 1 mm Suprasil substrate generates a weak TH signal. It is significantly weaker than the response from, for example, gold nanostructures. No influence of the 4 nm thin alumina coating on the THG is observed. Both observations are explained by the 3 orders of magnitude smaller nonlinear susceptibilities of fused silica and alumina compared to gold.48 In Figure 6(b) we plot detailed results of silver structures. The left column depicts the linear spectra of the antenna array before (black) and after (magenta) the laser exposure as well as the spectrum of the impinging laser source. The middle column displays the TH spectra at the beginning (black) of the time series and at the end (magenta). The right column shows the integrated radiated TH signal over time, which is obtained by integrating the TH spectra. The two spectra in the middle column thus correspond to t = 0 min and t = 15 min. The time evolution of the TH response shows a clear and steady decrease

stability, also the photostability is of concern. Especially under intense laser illumination the structures are exposed to strong electric fields that can lead to a reshaping of the structures.44 Plasmonic nanostructures have at resonance a larger absorption cross section than their physical cross section. Accordingly, the nanostructures are locally heated with high efficiency,45 which can lead to a deterioration of the plasmon resonance as discussed above. As a measure of photostability, third-harmonic generation (THG) is used. Nonlinear generation from plasmonic nanostructures is a widely used application of plasmonic materials, and it is a very sensitive tool to investigate changes in the plasmon resonance.46 The nonlinear spectroscopy is performed with an in-house-built laser system.47 The scheme of the laser setup is shown in Figure 6(a). The laser source is an Yb:KGW solid-state laser with a repetition rate of 44 MHz that is subsequently spectrally broadened to a spectral range from 900 to 1150 nm and temporally recompressed to 16 fs.47 Before the sample all wavelength components below 700 nm are filtered out with a 3 mm thick Schott RG715 filter. At the sample, up to 180 mW average power is available and focused down to a spot size of 30 μm fwhm. The maximum intensity on the sample is 10.3 GW/cm2. With a neutral density filter wheel G

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX

Article

ACS Photonics

the poor signal-to-noise ratio. After doubling the exposure intensity the TH shows a continuous decay and is therefore considered to be not stable any more. The covered material appears to be stable at 1.1 GW/cm2. At the doubled intensity a continuous change is observed. In contrast to the former cases an increase in THG is observed. A change in the linear properties may lead to an increased THG owing to the fact that THG depends on the overlap of the linear resonance with the laser spectrum. If the laser-induced changes increase the overlap of the fundamental resonance with the laser spectrum, also an increased THG is possible. Silver and gold exhibit the strongest THG; copper and aluminum show far less THG. The highest photostability is exhibited by gold. Alumina-covered gold withstands the maximum available intensity of the laser of 10.3 GW/cm2. Also aluminum and copper withstand rather high intensities. Silver has the poorest photostability, as expected based on the poor melting point and good absorption properties. Comparing these results to the thermal measurements indicates different mechanisms for the deterioration. Copper for example is thermally far less stable than silver, however significantly more stable under laser exposure. Laser exposure could also lead to an ablation-driven deterioration of the materials.51 This mechanism would be driven more efficiently for silver due to the excellent near-field enhancement compared to all other studied materials. For a thermal destruction, the conversion of the incident light into heat has to be studied in more detail. This mechanism may be influenced by parameters such as the heat conductivity to the materials and the heat distribution to the substrate as well as the absorption cross section of the different materials. In conclusion, the thermal properties of the materials can be ordered from stable to unstable from platinum, gold, palladium, nickel, silver, and aluminum to copper and magnesium. For all materials except the most unstable ones, copper and magnesium, the 4 nm alumina covered structures show a substantially increased thermal stability. The beneficial influence of a chromium adhesion layer on the thermal stability of gold nanostructures is demonstrated, and a major influence of the adhesion layer on the dimensional stability can be deduced. Furthermore, for silver, aluminum, gold, and copper the photostability and nonlinear response are investigated with THG spectroscopy. Here a significant increase in the photostability is observed for alumina-covered gold, silver, and copper structures. Especially for silver and copper the alumina cover is highly advantageous because the uncovered materials can only withstand intensities far below the commonly used ones. All in all, the thin alumina cover improves the thermal and photostability of most tested materials significantly and enables the use of these materials at higher temperatures and under more intense illumination conditions.

in intensity. This indicates that the plasmon resonance of the structures is deteriorating, most likely due to local heating and structure deformation or chemical deterioration. The left column underpins this interpretation, as one can see a dramatic change in the linear response. From this measurement we thus conclude that silver is a source of strong TH; yet, it is not photostable. To compare the results of silver, aluminum, gold, and copper, we are reducing the information and are only showing the extracted values in panel (c). We have studied all materials with and without an alumina cover layer. For each case the time evolution of the TH signal as well as the linear response before and after the exposure series is shown. We have not found THG exceeding the THG generated from the Suprasil substrate for the other materials. The data for gold were presented in a previous publication.31 The most susceptible material to laser exposure appears to be silver. This is expected due to fact that silver has the lowest melting point of the studied materials and the highest absorption efficiency and highest electric field enhancement.14 The uncovered silver structures show a stable response at 0.3 GW/cm2, however, with a strong fluctuation due to the poor signal-to-noise ratio. At higher intensities a fast degradation is observed. Furthermore, a clear change in the linear spectrum is visible. The covered sample is more stable, yet at 1.1 GW/cm2 the THG decays rapidly as well. Moreover, for this case the linear spectrum changes as well but far less when compared to the uncovered structures. The covered and uncovered aluminum rods generate a stable TH signal for an incident laser intensity of 2.3 GW/cm2. Doubling the fundamental power results in a slight decrease of the radiated TH intensity over time. This behavior indicates that the structures are no longer stable under the mentioned laser irradiation. For both cases basically no changes are visible in the linear spectra. The observed photostability qualifies aluminum for the usage in nonlinear spectroscopy. However, the overall THG is far less than the THG of silver for the same exposure intensity, indicating a smaller nonlinear susceptibility. Gold has a comparable nonlinear susceptibility to silver,48 which is also visible in a comparable nonlinear signal. The TH of the uncovered gold rods is slowly changing for the maximal available exposure intensity of 10.3 GW/cm2. In comparison, the TH generated from the covered structures does not change after the initial time step. The used intensity of 10.3 GW/cm2 is around 1 order of magnitude larger than the usually used intensity for nonlinear spectroscopy of gold nanostructures,46 which is around 1 GW/cm2. For both cases only minor changes are visible in the linear spectra. The fwhm of the transmittance decreases for the covered structures, whereas the center wavelength remains constant. Therefore, the quality factor of the resonance is increased, which we relate to a sintering effect.40 This is in accordance with an increased THG that substantially depends on the quality factor of the resonance.49 For the uncovered sample, mainly a small shift to shorter wavelengths is visible. The THG is highly sensitive to the overlap of the laser spectrum with the transmittance spectrum of the structure. Therefore, small changes in the linear properties can cause large changes in the nonlinear response.50 Copper exhibits a weak nonlinear response that is comparable to aluminum. Therefore, at low exposure powers the signal-to-noise is rather low. For the uncovered structure an exposure intensity of 0.6 GW/cm2 does not show any continuous changes, however, with a high fluctuation due to



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsphotonics.7b01346. Linear spectroscopy data and SEM micrographs for Ge and YH2; linear spectra for different rod sizes of all materials; simulations of transmittance spectra of bare nanostructures; transmittance spectra of silver nanoH

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX

Article

ACS Photonics



localized surface plasmon dynamics, and interface charge. Langmuir 2008, 25, 924−930. (14) Sanz, J.; Ortiz, D.; Alcaraz De La Osa, R.; Saiz, J.; González, F.; Brown, A.; Losurdo, M.; Everitt, H.; Moreno, F. UV plasmonic behavior of various metal nanoparticles in the near-and far-field regimes: geometry and substrate effects. J. Phys. Chem. C 2013, 117, 19606−19615. (15) Sterl, F.; Strohfeldt, N.; Walter, R.; Griessen, R.; Tittl, A.; Giessen, H. Magnesium as novel material for active plasmonics in the visible wavelength range. Nano Lett. 2015, 15, 7949−7955. (16) Duan, X.; Kamin, S.; Liu, N. Dynamic plasmonic colour display. Nat. Commun. 2017, 8, 14606. (17) Park, J. H.; Han, S. E.; Nagpal, P.; Norris, D. J. Observation of thermal beaming from tungsten and molybdenum bull’s eyes. ACS Photonics 2016, 3, 494−500. (18) Chen, J.; Albella, P.; Pirzadeh, Z.; Alonso-González, P.; Huth, F.; Bonetti, S.; Bonanni, V.; Åkerman, J.; Nogués, J.; Vavassori, P.; Dmitriev, A.; Aizpurua, J.; Hillenbrand, R. Plasmonic nickel nanoantennas. Small 2011, 7, 2341−2347. (19) Tittl, A.; Mai, P.; Taubert, R.; Dregely, D.; Liu, N.; Giessen, H. Palladium-based plasmonic perfect absorber in the visible wavelength range and its application to hydrogen sensing. Nano Lett. 2011, 11, 4366−4369. (20) Zoric, I.; Zäch, M.; Kasemo, B.; Langhammer, C. Gold, platinum, and aluminum nanodisk plasmons: material independence, subradiance, and damping mechanisms. ACS Nano 2011, 5, 2535− 2546. (21) Arpin, K. A.; Losego, M. D.; Cloud, A. N.; Ning, H.; Mallek, J.; Sergeant, N. P.; Zhu, L.; Yu, Z.; Kalanyan, B.; Parsons, G. N.; Girolami, G. S.; Abelson, J. R.; Fan, S.; Braun, P. V. Three-dimensional self-assembled photonic crystals with high temperature stability for thermal emission modification. Nat. Commun. 2013, 4, 2630. (22) Guler, U.; Shalaev, V. M.; Boltasseva, A. Nanoparticle plasmonics: going practical with transition metal nitrides. Mater. Today 2015, 18, 227−237. (23) Strohfeldt, N.; Tittl, A.; Schäferling, M.; Neubrech, F.; Kreibig, U.; Griessen, R.; Giessen, H. Yttrium hydride nanoantennas for active plasmonics. Nano Lett. 2014, 14, 1140−1147. (24) Comin, A.; Manna, L. New materials for tunable plasmonic colloidal nanocrystals. Chem. Soc. Rev. 2014, 43, 3957−3975. (25) Naik, G. V.; Kim, J.; Boltasseva, A. Oxides and nitrides as alternative plasmonic materials in the optical range. Opt. Mater. Express 2011, 1, 1090−1099. (26) Grinblat, G.; Li, Y.; Nielsen, M. P.; Oulton, R. F.; Maier, S. A. Enhanced third harmonic generation in single germanium nanodisks excited at the anapole mode. Nano Lett. 2016, 16, 4635−4640. (27) Hanham, S.; Fernández-Domínguez, A.; Teng, J. H.; Ang, S.; Lim, K.; Yoon, S. F.; Ngo, C.; Klein, N.; Pendry, J.; Maier, S. A. Broadband terahertz plasmonic response of touching InSb disks. Adv. Mater. 2012, 24, OP22610.1002/adma.201202003 (28) Petrova, H.; Juste, J. P.; Pastoriza-Santos, I.; Hartland, G. V.; Liz-Marzán, L. M.; Mulvaney, P. On the temperature stability of gold nanorods: comparison between thermal and ultrafast laser-induced heating. Phys. Chem. Chem. Phys. 2006, 8, 814−821. (29) Guler, U.; Boltasseva, A.; Shalaev, V. M. Refractory plasmonics. Science 2014, 344, 263−264. (30) Bagheri, S.; Zgrabik, C. M.; Gissibl, T.; Tittl, A.; Sterl, F.; Walter, R.; de Zuani, S.; Berrier, A.; Stauden, T.; Richter, G.; Hu, E. L.; Giessen, H. Large-area fabrication of TiN nanoantenna arrays for refractory plasmonics in the mid-infrared by femtosecond direct laser writing and interference lithography. Opt. Mater. Express 2015, 5, 2625−2633. (31) Albrecht, G.; Giessen, H.; Kaiser, S.; Hentschel, M. Refractory Plasmonics without Refractory Materials. Nano Lett. 2017, 17, 6402− 6408. (32) Patnaik, P. Handbook of Inorganic Chemicals; McGraw-Hill: New York, 2003; Vol. 529.

particles in air after different storage times; used deposition parameters; transmittance spectra of platinum nanostructures covered with 40 nm alumina after heating; SEM and AFM images as well as transmittance measurements of silver grown on chromium and germanium (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Gelon Albrecht: 0000-0002-5456-2864 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We acknowledge support from ERC (Complexplas), BMBF, DFG (SPP1839), MWK BW under the Juniorprofessorenprogramm, Daimler und Benz Stiftung, and Baden-Württemberg Stiftung. G.A. acknowledges funding from Max Planck Institute for Solid State Research. We thank Julian Karst und Florian Sterl for magnesium evaporation and Bettina Frank for her help with the AFM measurements. Especially we want to sincerely thank Marion Hagel for fabricating the ALD layers.



REFERENCES

(1) Atwater, H. A.; Polman, A. Plasmonics for improved photovoltaic devices. Nat. Mater. 2010, 9, 205. (2) Lal, S.; Clare, S. E.; Halas, N. J. Nanoshell-enabled photothermal cancer therapy: impending clinical impact. Acc. Chem. Res. 2008, 41, 1842−1851. (3) Liu, N.; Mesch, M.; Weiss, T.; Hentschel, M.; Giessen, H. Infrared perfect absorber and its application as plasmonic sensor. Nano Lett. 2010, 10, 2342−2348. (4) Boltasseva, A.; Atwater, H. A. Low-loss plasmonic metamaterials. Science 2011, 331, 290−291. (5) Stauber, T. Plasmonics in Dirac systems: from graphene to topological insulators. J. Phys.: Condens. Matter 2014, 26, 123201. (6) Di Pietro, P.; Ortolani, M.; Limaj, O.; Di Gaspare, A.; Giliberti, V.; Giorgianni, F.; Brahlek, M.; Bansal, N.; Koirala, N.; Oh, S.; Calvani, P.; Lupi, S. Observation of Dirac plasmons in a topological insulator. Nat. Nanotechnol. 2013, 8, 556−560. (7) Mock, J.; Barbic, M.; Smith, D.; Schultz, D.; Schultz, S. Shape effects in plasmon resonance of individual colloidal silver nanoparticles. J. Chem. Phys. 2002, 116, 6755−6759. (8) Nagpal, P.; Lindquist, N. C.; Oh, S.-H.; Norris, D. J. Ultrasmooth patterned metals for plasmonics and metamaterials. Science 2009, 325, 594−597. (9) Wang, X.; Santschi, C.; Martin, O. J. Strong Improvement of Long-Term Chemical and Thermal Stability of Plasmonic Silver Nanoantennas and Films. Small 2017, 13, 1700044. (10) Knight, M. W.; Liu, L.; Wang, Y.; Brown, L.; Mukherjee, S.; King, N. S.; Everitt, H. O.; Nordlander, P.; Halas, N. J. Aluminum plasmonic nanoantennas. Nano Lett. 2012, 12, 6000−6004. (11) Thyagarajan, K.; Rivier, S.; Lovera, A.; Martin, O. J. Enhanced second-harmonic generation from double resonant plasmonic antennae. Opt. Express 2012, 20, 12860−12865. (12) Wang, D.-D.; Ge, C.-W.; Wu, G.-A.; Li, Z.-P.; Wang, J.-Z.; Zhang, T.-F.; Yu, Y.-Q.; Luo, L.-B. A sensitive red light nanophotodetector propelled by plasmonic copper nanoparticles. J. Mater. Chem. C 2017, 5, 1328−1335. (13) Wu, P. C.; Losurdo, M.; Kim, T.-H.; Giangregorio, M.; Bruno, G.; Everitt, H. O.; Brown, A. S. Plasmonic gallium nanoparticles on polar semiconductors: interplay between nanoparticle wetting, I

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX

Article

ACS Photonics (33) Gao, H.; Henzie, J.; Lee, M. H.; Odom, T. W. Screening plasmonic materials using pyramidal gratings. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 20146−20151. (34) Smithells, C. Metals Reference Book, 5th ed.; Elsevier Science, 1976; pp 186−187. (35) Logeeswaran, V.; Kobayashi, N. P.; Islam, M. S.; Wu, W.; Chaturvedi, P.; Fang, N. X.; Wang, S. Y.; Williams, R. S. Ultrasmooth silver thin films deposited with a germanium nucleation layer. Nano Lett. 2008, 9, 178−182. (36) Baldi, A.; Palmisano, V.; Gonzalez-Silveira, M.; Pivak, Y.; Slaman, M.; Schreuders, H.; Dam, B.; Griessen, R. Quasifree Mg-H thin films. Appl. Phys. Lett. 2009, 95, 071903. (37) McPeak, K. M.; Jayanti, S. V.; Kress, S. J.; Meyer, S.; Iotti, S.; Rossinelli, A.; Norris, D. J. Plasmonic films can easily be better: Rules and recipes. ACS Photonics 2015, 2, 326−333. (38) Langhammer, C.; Schwind, M.; Kasemo, B.; Zoric, I. Localized surface plasmon resonances in aluminum nanodisks. Nano Lett. 2008, 8, 1461−1471. (39) Beszeda, I.; Gontier-Moya, E.; Imre, A. Surface Ostwaldripening and evaporation of gold beaded films on sapphire. Appl. Phys. A: Mater. Sci. Process. 2005, 81, 673−677. (40) Bosman, M.; Zhang, L.; Duan, H.; Tan, S. F.; Nijhuis, C. A.; Qiu, C.-W.; Yang, J. K. Encapsulated annealing: enhancing the plasmon quality factor in lithographically-defined nanostructures. Sci. Rep. 2015, 4, 5537. (41) Hammer, B.; Norskov, J. Why gold is the noblest of all the metals. Nature 1995, 376, 238−240. (42) Habteyes, T. G.; Dhuey, S.; Wood, E.; Gargas, D.; Cabrini, S.; Schuck, P. J.; Alivisatos, A. P.; Leone, S. R. Metallic adhesion layer induced plasmon damping and molecular linker as a nondamping alternative. ACS Nano 2012, 6, 5702−5709. (43) Ghorbanpour, M.; Falamaki, C. A novel method for the production of highly adherent Au layers on glass substrates used in surface plasmon resonance analysis: substitution of Cr or Ti intermediate layers with Ag layer followed by an optimal annealing treatment. Journal of Nanostructure in Chemistry 2013, 3, 66. (44) Shi, L.; et al. Self-optimization of plasmonic nanoantennas in strong femtosecond fields. Optica 2017, 4, 1038−1043. (45) Caldarola, M.; Albella, P.; Cortés, E.; Rahmani, M.; Roschuk, T.; Grinblat, G.; Oulton, R. F.; Bragas, A. V.; Maier, S. A. Non-plasmonic nanoantennas for surface enhanced spectroscopies with ultra-low heat conversion. Nat. Commun. 2015, 6, 791510.1038/ncomms8915. (46) Metzger, B.; Hentschel, M.; Lippitz, M.; Giessen, H. Thirdharmonic spectroscopy and modeling of the nonlinear response of plasmonic nanoantennas. Opt. Lett. 2012, 37, 4741−4743. (47) Metzger, B.; Steinmann, A.; Giessen, H. High-power widely tunable sub-20fs Gaussian laser pulses for ultrafast nonlinear spectroscopy. Opt. Express 2011, 19, 24354−24360. (48) Boyd, R.; Prato, D. Nonlinear Optics, 3rd ed.; Nonlinear Optics Series; Elsevier Science, 2008; p 212. (49) Hentschel, M.; Utikal, T.; Giessen, H.; Lippitz, M. Quantitative modeling of the third harmonic emission spectrum of plasmonic nanoantennas. Nano Lett. 2012, 12, 3778−3782. (50) Gui, L.; Bagheri, S.; Strohfeldt, N.; Hentschel, M.; Zgrabik, C. M.; Metzger, B.; Linnenbank, H.; Hu, E. L.; Giessen, H. Nonlinear refractory plasmonics with titanium nitride nanoantennas. Nano Lett. 2016, 16, 5708−5713. (51) Plech, A.; Kotaidis, V.; Lorenc, M.; Boneberg, J. Femtosecond laser near-field ablation from gold nanoparticles. Nat. Phys. 2006, 2, 44−47.

J

DOI: 10.1021/acsphotonics.7b01346 ACS Photonics XXXX, XXX, XXX−XXX