Maskless Spatioselective Functionalization of Silicon Nanowires.

Spatioselective functionalization of silicon nanowires was achieved without using a masking material. The designed process combines metal-assisted chemical etching (MACE) to fabricate silicon nanowires and hydrosilylation to form molecular monolayers. After MACE, a monolayer was formed on the exposed nanowire surfaces. A second MACE step was expected to elongate the nanowires, thus creating two different segments. When monolayers of 1-undecene or 1-tetradecyne were formed on the upper segment, however, the second MACE step did not extend the nanowires. In contrast, nanowires functionalized with 1,8-nonadiyne were elongated, but at an approximately 8 times slower etching rate. The elongation resulted in a contrast difference in high-resolution scanning electron microscopy (HR-SEM) images, which indicated the formation of nanowires that were covered with a monolayer only at the top parts. Click chemistry was successfully used for secondary functionalization of the monolayer with azide-functionalized nanoparticles. The spatioselective presence of 1,8-nonadiyne gave a threefold higher particle density on the upper segment functionalized with 1,8-nonadiyne than on the lower segment without monolayer. These results indicate the successful spatioselective functionalization of silicon nanowires fabricated by MACE.

Introduction

Silicon nanostructures and nano/microwire arrays are attractive because of their optical, electronic, electrochemical, mechanical, and thermal properties.1 To tune these nanostructures towards a specific application, surface functionalization can offer versatility and specificity, for example, to couple photocatalysts onto the surface for hydrogen production.2 Whereas surface functionalization is mostly applied onto the entire surface of a Si micro/nanostructure, spatioselective functionalization of 3D nanostructures, i. e., the functionalization of parts of the structure, is useful when several functionalities should be combined.3 For example, oxidation and reduction reactions can be performed on two different ends of a nanowire when these are site‐specifically functionalized with the respective catalysts.4

Spatioselective functionalization of 3D nanostructures can be achieved by prohibiting access of functionalization agents to specific areas of the structures, i. e., by protecting specific segments with a masking material, such as a polymer (resist),4b,5 a metal,6 nanoparticles (NPs),7 or a membrane.8 These techniques, however, always require extra steps, such as the addition of the masking material, the removal of the masking material from the sites that should not be blocked, and finally the mask removal after functionalization. Very few studies provide processes that avoid such multiple masking/unmasking steps. For example, Milbrat et al. have developed a method for the spatioselective functionalization of silicon microwires with axial p‐n junctions.9 The charges directed by the junction, either in the dark or under illumination, were used to electrochemically deposit silver or platinum NPs in a spatially selective manner. This method, however, is applicable to metals in general but not to organic molecules. The use of molecular monolayers would offer versatility, for example, by secondary functionalization enabled by the chosen end moieties.10

Here, we aim for the spatioselective functionalization of silicon nanowires (Si NWs) with molecular monolayers without the use of a masking material. We have designed a process (Scheme 1a) based on metal‐assisted chemical etching (MACE) as the NW fabrication method, in which a metal film (silver or gold) with patterned gaps is etched into the underlying silicon substrate upon immersion in a HF/H2O2 solution.11 The diameter of the resulting nanowires is determined by the pattern in the metal film, whereas the length is determined by the etching time. Spatioselective functionalization was tested by two MACE steps with monolayer formation after the first step and functionalization of the top or bottom section of the NW after the second MACE step. The first MACE step is used to etch nanowires until a certain length, after which a monolayer is formed on the exposed silicon surface. A second MACE step is expected to elongate the nanowires and thus expose a non‐functionalized part of the silicon nanowires, which could then be functionalized differently in the final functionalization step.

Scheme 1

Schematic illustration of (a) the general process designed for the spatioselective functionalization of MACE‐fabricated silicon nanowires and (b, c) the two different routes tested here (a: MACE|1‐undecene/1‐tetradecyne|MACE|1,8‐nonadiyne|click chemistry, b: MACE|1,8‐nonadiyne|MACE|click chemistry).

Monolayer formation has been achieved by hydrosilylation of alkenes or alkynes, i. e., by linking terminally unsaturated carbon‐carbon bonds onto H‐terminated Si surfaces.12 This technique results in direct Si−C bond formation and avoids the presence of silicon oxide in between the substrate and the monolayer. If silicon oxide would be present, the second MACE step would etch away the silicon oxide on the first part including the monolayer. Specifically, 1‐undecene or 1‐tetradecyne were chosen as (methyl‐terminated) dummy molecules and the dialkyne 1,8‐nonadiyne, which can potentially be reacted further by click chemistry with azide‐functionalized model catalysts, as a functional adsorbate. We here compare the different process parameters that influence the success rate of the selectively functionalized NW fabrication process.

Results and Discussion

Two routes were envisaged towards the spatioselective functionalization of silicon nanowires (Scheme 1b, c). Both routes are based on the same order of steps: i) MACE, ii) monolayer formation, iii) MACE, iv) functionalization. The first route (Scheme 1b) was designed using monolayer formation with a dummy molecule on the upper part, i. e., an alkene/alkyne (1‐undecene or 1‐tetradecyne, Scheme S1) with a non‐reactive methyl end moiety. As a requirement, these monolayers should survive the second MACE step. After the second MACE step, monolayer formation with a dialkyne (1,8‐nonadiyne, Scheme S1) could be employed, which then selectively forms on the lower part since the upper segment is already covered by a monolayer of 1‐undecene or 1‐tetradecyne. To visualize the difference between the two segments, the 1,8‐nonadiyne monolayer could be functionalized further by copper‐catalyzed click chemistry10a (Scheme S1) with an azide‐functionalized dye (characterized by fluorescence microscopy) or azide‐functionalized gold NPs (characterized by HR‐SEM). The second route (Scheme 1c) involves the formation of a 1,8‐nonadiyne monolayer after the first MACE step. After the second MACE step to elongate the nanowires, the lower segment remains non‐functionalized, and click chemistry onto the exposed alkyne moieties attached to the upper NW segments could be used to visualize the spatial selectivity.

First MACE Step and Monolayer Formation

Silicon nanowires were created by combining nanosphere lithography and MACE.13 Nanosphere lithography (Scheme S2) was used to form a close‐packed array of polystyrene spheres on a cleaned p‐type silicon substrate. The diameter of the spheres was decreased from 447 nm (dried state) to about 280 nm by oxygen reactive ion etching. After sputtering a metal film (silver or gold) on top, lift‐off of the polystyrene spheres resulted in a metal film with patterned openings. Exposure of these metal‐coated samples to an aqueous solution of HF/H2O2 ensured wet etching of the silicon substrate underneath the metal layer. Immersion for 7 min (Ag) or 15 min (Au) resulted in nanowires with a length of about 2.1 and 1.0‐1.1 μm, and a diameter of 390 and 260 nm, respectively, as shown by HR‐SEM imaging (Figure 1a, c, e).

Figure 1

Cross‐sectional HR‐SEM images of Si NWs using (a‐b) a Ag or (c‐f) a Au film, imaged (a, c, e) after the first MACE step and (b, d, f) after the second MACE step on samples with photochemically formed monolayers of (a–d) 1‐undecene or (e, f) 1,8‐nonadiyne.

Monolayer formation was achieved by thermal or photochemical hydrosilylation (Scheme S1).12 This oxide‐free monolayer formation technique significantly increases the chance that the monolayer stays intact during the second MACE step, since HF, present in the MACE solution, quickly removes silicon oxide. Monolayer formation was first tested by the thermal hydrosilylation method, since this generally results in higher quality monolayers than the photochemical route. The temperature used for this reaction (180 °C for 1‐undecene in mesitylene), however, resulted in agglomeration of the gold film (Figure S1). The photochemical route was therefore used by reaction under illumination of 420 nm light, which did not affect the metal films.

The photochemical hydrosilylation technique (Scheme S1) was used to form monolayers onto the exposed silicon nanowire surfaces after the first MACE step. On one half of the sample with nanowires, an H‐terminated silicon surface was created by short immersion in an aqueous 1% HF solution. Subsequently, a monolayer of 1‐undecene, 1‐tetradecyne, or 1,8‐nonadiyne with Si−C bonds was formed under illumination with a 420 nm LED. Whereas the hydrosilylation of 1‐undecene and 1‐tetradecyne was performed in a 5% v/v solution in mesitylene, 1,8‐nonadiyne was used in its pure form to suppress back bending of the second alkyne group towards the hydrogen‐terminated silicon surface. Cleaning was performed without ultrasonication in order to avoid damage to the metal film. The other half of the sample was used as a control, on which the second MACE step was tested without preceding monolayer formation. For the first route (Scheme 1b), presence of the monolayer was indicated by a contact angle of 130.7°±2.1 (1‐undecene) or 132.6°±2.2 (1‐tetradecyne) on top of the nanowires, confirming the combined effect of nanostructuring and the presence of the hydrophobic monolayer. Nanowires with a monolayer of 1,8‐nonadiyne (Scheme 1c) were less hydrophobic due to their alkyne end moieties, as shown by a contact angle of 112.1°±3.6. The nanowires without monolayer showed a contact angle of 40.4°±3.5, 70.7°±6.5, or 89.0°±7.2, indicative of (varying degrees of) regrowth of native silicon oxide.

Images after the first and second MACE steps indicated a preference for using a gold film over a silver film, since i) a gold film gave more homogeneously etched samples (Figure 1) and ii) the silver film was sometimes dissolved and redeposited on the lower segment of the nanowires, after which the bottom of the nanowires got etched until they broke off (data not shown). The etch rate, however, was lower for samples with a gold film compared to a silver film. This is a disadvantage since the second MACE step, and thus the exposure of the monolayer on the upper segment to the MACE solution, has to be longer.

Second MACE Step and Monolayer Formation

Upon the second MACE step, the nanowires with a monolayer of 1‐undecene (Scheme 1b) were unexpectedly not elongated after 15 min (Ag) or 45 min (Au) of MACE (Figure 1b, d). HR‐SEM images showed that the non‐functionalized nanowires increased in length to 3.7 (Ag) and 4.8 μm (Au) in the same reaction times (Figure S2a, b), while the nanowires functionalized with 1‐undecene remained 2.1 and 1.0 μm, respectively (Figure 1b, d). The same effect was observed for silicon nanowires functionalized with 1‐tetradecyne (data not shown).

Although these data already indicated a clear effect of the presence of a monolayer on the etching behavior, contact angle measurements could not be used to verify whether the monolayers on the nanowires survived the second MACE step, since the top of the nanowires got roughened and resulted in contact angles of 130.8°±2.0 and 138.8°±0.5 for the functionalized (1‐tetradecyne) and non‐functionalized samples, respectively. As a control, therefore, a planar substrate with a monolayer of 1‐tetradecyne was immersed in the MACE solution for 10, 20, and 40 min, after which the contact angle was measured (Figure 2, left axis). The contact angle decreased due to exposure to the etching solution, thus indicating that the monolayer had deteriorated. These values were converted into the fraction of monolayer that was still intact (assuming a full monolayer at t=0) and the fraction of Si−H formed (contact angle 78° for a fully H‐terminated surface) using Cassie's law (Figure 2, right axis). This indicates that about 44% of the monolayer was still intact after 40 min exposure to the MACE solution, whereas a second MACE step of 45 min had not lengthened the nanowires yet. A longer second MACE step would induce even more damage to the monolayer, if it would be possible to elongate the nanowires at all. More importantly, when viewing the process envisaged in Scheme 1b, the incomplete monolayer at the upper segment would be back‐filled with 1,8‐nonadiyne molecules in the second functionalization step. Spatioselective functionalization can thus not be achieved by this route, since 1,8‐nonadiyne would be present on both segments.

Figure 2

Contact angle measurements (black squares, left axis) of a 1‐tetradecyne monolayer on a p(100) planar sample measured after the reaction and after exposure to the MACE solution, and the monolayer fraction (blue circles, right axis) that is still intact as determined by Cassie's law.

For the other route, with 1,8‐nonadiyne on the upper segment and a non‐functionalized lower segment (Scheme 1c), the second MACE step was tested as well. Only samples with a gold film were tested for this adsorbate for the reasons mentioned above. Additionally, it was observed that a silver film could be destroyed by contact with 1,8‐nonadiyne (Figure S3). Contrary to the results with 1‐undecene/1‐tetradecyne, a second MACE step of 30 min after 1,8‐nonadiyne attachment resulted in an increase of the length of the nanowires of 200–300 nm, from 1.1 μm to 1.3–1.4 μm (Figure 1f). The non‐functionalized control samples were elongated up to 3.0 μm (Figure S2c), which shows that the presence of the monolayer has a retarding effect (of approximately a factor 8) on the second MACE step. The effect of 30 min exposure of the 1,8‐nonadiyne monolayer to the MACE solution could not be tested by contact angle measurements as described above for 1‐tetradecyne (Figure 2), since the contact angle of a 1,8‐nonadiyne monolayer on a planar sample was too close (82°)14 to that of an absent or fully deteriorated monolayer (Si−H, 78°). This made a proper determination of the fraction of intact monolayer impossible.

The HR‐SEM data do not clarify what causes the retardation or complete inhibition of the second MACE step. The difference in etching between the 1,8‐nonadiyne and 1‐undecene/1‐tetradecyne samples seems to originate from the monolayer formation step, since the non‐functionalized control samples were etched as expected. The main differences between the two routes (Scheme 1b, c) include the adsorbates (1,8‐nonadiyne versus 1‐undecene/1‐tetradecyne) and the solvents (none versus mesitylene). Two main explanations are hypothesized: i) the MACE solution is not able to enter the voids when a hydrophobic monolayer is present and ii) a contamination at the metal/silicon interface is preventing the second MACE step. To the first point, the more hydrophilic the nanowires are, the higher the chance that the aqueous MACE solution enters the voids to continue etching. As shown by contact angle measurements, the monolayers of 1‐undecene and 1‐tetradecyne were more hydrophobic than those of 1,8‐nonadiyne and could thus prevent the MACE solution from reaching the metal film at the bottom of the voids. To the second point, if something would be present on the metal/silicon interface, the second MACE step could be delayed or prohibited. A layer on top of the metal film, instead, does not prohibit MACE,11c i. e., only the interface between the metal film and the underlying silicon substrate is of importance. After the first MACE step, porous silicon is formed underneath the metal film,11c which could theoretically be filled up with reagents or solvents (from the reaction or cleaning steps). This would, however, not explain why the samples with 1,8‐nonadiyne could be etched further in the second MACE step. Control experiments were performed by overnight immersion of silicon samples with a gold‐patterned film after lift‐off, i. e., before the first MACE step, in mesitylene, 1‐undecene, or 1,8‐nonadiyne. The first MACE step was successful for all of these samples, although the etch rate was lower than for samples without immersion. Preliminary tests on immersion of silicon nanowires in mesitylene and a subsequent second MACE step, did not conclusively show whether the nanowires could be elongated or not. Further research is thus required to verify whether these hypotheses are correct.

Secondary Functionalization

In the second route (Scheme 1c), the copper‐catalyzed click reaction10a,15 was used to investigate whether the 1,8‐nonadiyne monolayer was still present after the second MACE step and spatioselective functionalization could be achieved. In this reaction, an azide‐functionalized molecule or particle is coupled onto the alkyne end moiety of the 1,8‐nonadiyne monolayer using Cu(I) as the catalyst (Scheme S1). Either an azide‐functionalized dye was chosen for detection by fluorescence microscopy or azide‐functionalized gold NPs to allow detection by high‐resolution scanning electron microscopy. A Cu(I) stabilizing ligand (TBTA) was used to increase the rate of the click chemistry within the voids between the nanowires.16

After functionalization with azide‐fluor 488, the nanowires were scraped off the surface and immobilized onto a microscope glass slide. As a disadvantage, this induced a high background fluorescence and showed irregularly shaped fluorescent objects, probably due to residuals of the gold film used for MACE. Nonetheless, fluorescence microscopy showed needle‐shaped objects that are most likely the functionalized nanowires based on their size (Figure S4). Comparison of the length of the wires in the fluorescence and brightfield images indicates, however, that the resolution of these images is too low to conclude whether the 200–300 nm bottom segments had remained non‐functionalized. Confocal microscopy did not solve this resolution issue. A longer second MACE step would be needed to increase the contrast between the two segments, but could also damage the 1,8‐nonadiyne monolayer on the upper segment, and thus was not pursued further here.

To circumvent the practical issues associated with fluorescent dye functionalization and imaging, click chemistry with azide‐functionalized gold NPs (10 nm diameter) was tested. In this way, the nanowires did not have to be transferred to a microscope slide, since HR‐SEM imaging of sample cross sections was sufficient, and a higher resolution could be obtained. After click chemistry, HR‐SEM images showed a clear contrast between the upper functionalized segment and the lower non‐functionalized segment (Figure 3a). This contrast was also visible when imaging before click chemistry (Figure S5a), so in the absence of nanoparticles but in the presence of a 1,8‐nonadiyne monolayer, but was more clearly visible in the higher resolution image of Figure 3a. Since the contrast was not visible on the control samples in the absence of a monolayer (Figure S5b), it is assumed that this indicates the presence of a 1,8‐nonadiyne monolayer on the upper segment. A carbon‐containing layer is known to give a darker contrast in SEM images compared to the, most likely, silicon oxide‐containing lower segment which is much brighter. The height of the lower segment in the HR‐SEM image (Figure 3a) corresponds well with the nanowire length increase observed for the second MACE step, as discussed above.

Figure 3

Cross‐sectional HR‐SEM images of Si NWs (Au film) after click chemistry with azide‐functionalized gold NPs on (a) nanowires functionalized with a 1,8‐nonadiyne monolayer on the upper segment, with (b) the corresponding ESB image, (c) the particle count along the nanowire length (bin size 50 nm) and (d) the number of bins from (c) sorted on particle density for the two segments, and cross‐sectional HR‐SEM images after click chemistry with azide‐functionalized gold NPs on nanowires fully functionalized with a monolayer of (e) 1‐undecene and (f) 1,8‐nonadiyne.

After click chemistry, the azide‐functionalized gold NPs should be present on the upper segment only. An energy‐selective backscatter (ESB) detector was used to reflect compositional variations on the sample based on atomic number, which showed NPs over the entire nanowire length (Figure 3b). Nonetheless, particle counting along the nanowire length (Figure 3c) clearly showed more NPs on the segment above the contrasting line. Quantitatively, the average particle densities, as calculated from the SEM images, were 136±70 NPs/μm2 on the 1,8‐nonadiyne functionalized upper segment (0‐0.8 μm) and 45±27 NPs/μm2 on the non‐functionalized lower segment (1.0–1.3 μm), where the gradual transition in contrast between 0.8 and 1.0 μm was not taken into account. The threefold difference between the two segments (p<0.01, Student's t‐test) is also reflected in Figure 3d, where the upper segment shows higher particle densities than the lower segment. Nonspecific adsorption of the gold NPs was also observed on control samples with a monolayer of 1‐undecene on the full nanowires (Figure 3e, 81±15 NPs/μm2), but the average particle density was much lower than on the same nanowires with 1,8‐nonadiyne (Figure 3f, 212±17 NPs/μm2). Although the NPs were inhomogeneously distributed along the nanowires that were fully functionalized with 1,8‐nonadiyne, particles were clearly observed at the lower part of these nanowires. The much lower amount of particles present at the lower part of the nanowires shown in Figure 3b should thus be due to the absence of a 1,8‐nonadiyne monolayer. Overall, on nanowires fabricated by the 1,8‐nonadiyne route (Scheme 1c) the differences in contrast and in the amount of NPs present on the two segments clearly indicate successful spatioselective functionalization.

Conclusions

In summary, the spatioselective functionalization of silicon nanowires was tested using consecutively: i) MACE, ii) monolayer formation, iii) MACE, iv) functionalization by click chemistry. Nanowires were successfully fabricated and subsequently functionalized with 1‐alkenes/alkynes by hydrosilylation. When the top segment, exposed after the first MACE step, was covered with a monolayer of 1‐undecene or 1‐tetradecyne, the second MACE step inexplicably did not elongate the nanowires. In contrast, nanowires functionalized with 1,8‐nonadiyne could be lengthened from 1.1 μm to 1.3–1.4 μm. A contrast difference observed in HR‐SEM images of the 1,8‐nonadiyne‐functionalized nanowires indicated the formation of the 1,8‐nonadiyne monolayer only on the top section of the nanowires. Click chemistry with an azide‐functionalized dye was possible, but presence of the dye only on the top section could not be verified. The reaction with azide‐functionalized nanoparticles, however, confirmed the successful secondary functionalization and showed a higher particle density on the upper segment. These results indicate that spatioselective functionalization of silicon nanowires by MACE is possible.

Further research could elucidate why/when the second MACE step does (not) work. To conclude whether the hydrophobicity of the monolayer is the origin, testing the second MACE step on nanowires functionalized with a hydrophilic alkyne, for example with a carboxylic acid end group,17 could be useful. Alternatively, it might help to add a surfactant, e. g., ethanol18 or sodium dodecyl sulfate, to lower the interfacial tension. Next, the influence of the metal/silicon interface could be studied further, i. e., whether something present at this interface is prohibiting the second MACE step. As an example, the role of mesitylene as a solvent can be eliminated by monolayer formation in pure 1‐undecene/1‐tetradecyne or in a solution of 1,8‐nonadiyne in mesitylene. Considering the specific click chemistry reaction used, the nonspecific adsorption of the azide‐functionalized gold NPs on the segment without 1,8‐nonadiyne should be lowered, for example using a poly(ethylene glycol) antifouling monolayer.

The spatioselective formation of a 1,8‐nonadiyne monolayer was shown here by a contrast difference in the HR‐SEM images, and azide‐functionalized nanoparticles were successfully used to functionalize the nanowires further. The compounds used here, however, are just an example for the applicability of this process. End group of the alkyne‐containing molecules used in the monolayer formation can be chosen as desired, and the subsequent secondary functionalization can be tuned accordingly. The fabrication method also offers versatility, since i) the height of each segment could be tuned, depending on the application, for instance different amounts of catalysts based on their respective catalytic activities, ii) more than two segments could be made by adding more MACE/functionalization cycles. Thus, this fabrication method offers a relatively simple solution for the creation of spatioselectively functionalized nanowires without the use of a masking material.

Experimental Section

Materials

Silicon wafers (<100>‐oriented, 100 mm diameter, single side polished) were obtained from Okmetic (Finland) as p‐type (boron, resistivity 5–10 Ω cm, thickness 525 μm). Polystyrene spheres of 500 nm diameter functionalized with carboxyl groups were obtained from Polysciences. Azide‐functionalized gold NPs of 10 nm diameter were obtained from NanoCS, with a particle concentration of 0.5 mg/mL in water (based on gold salt, 2.8×1013 particles/mL), a size distribution <15% and a poly(ethylene glycol) linker between the NPs and the azide groups. Mesitylene (>98%, Sigma‐Aldrich) and 1,8‐nonadiyne (98%, Sigma‐Aldrich) were dried over molecular sieves (0.3 nm). Dichloromethane (99.7%, Actu‐All) was dried over anhydrous magnesium sulfate (Merck). Acetone (pure, VWR), L‐ascorbic acid (>99%, Sigma‐Aldrich), azide‐fluor 488 (>90%, Sigma‐Aldrich), copper(II) sulfate pentahydrate (99.995% metals basis, Cu(II)SO4.5H2O, Sigma‐Aldrich), ethanol (absolute, VWR), hydrofluoric acid 1% (aqueous, VLSI, 1% HF, Technic France), hydrofluoric acid 50% (aqueous, VLSI, 50% HF, BASF), hydrogen peroxide (VLSIn, H2O2, 31%, BASF), sodium dodecyl sulfate (SDS, >99%, Sigma‐Aldrich), 1‐tetradecyne (>97%, Sigma‐Aldrich) and 1‐undecene (97%, Sigma‐Aldrich) were used as received. Tris‐(benzyltriazolylmethyl)amine (TBTA) was synthesized according to a procedure from the literature.19 Hexane was obtained from a solvent purification system (MB SPS‐800). Milli‐Q water with a resistivity >18 MΩ⋅cm was obtained from a Milli‐Q Integral water purification system (Merck Millipore). Glassware used for the hydrosilylation reactions was dried overnight at 120 °C.

Fabrication of Silicon Nanowires

Silicon nanowires were fabricated by combining nanosphere lithography (Scheme S2) and MACE. A close‐packed array of 500 nm polystyrene (PS) spheres was formed at the water/air interface and transferred onto a silicon substrate according to a procedure adapted from the literature.20 Specifically, small 2x2 cm2 samples were diced from silicon wafers (p‐type), which were rinsed with ethanol and sonicated in acetone for 10 min to remove particles generated during dicing. A 0.65% w/v solution of 500 nm PS spheres in 1 : 1 water:ethanol was sonicated for several hours to make a uniform suspension. The silicon samples, a glass Petri dish, and a small microscope glass slide were turned hydrophilic by oxygen plasma (SPI Plasma Prep II, 40 mA, 10 min). The glass slide was put in the middle of the Petri dish, after which water was added until the level was just above the glass slide without wetting the top of the glass slide. The PS solution (120 μL) was dispersed onto the glass slide in order to diffuse the spheres from the glass slide to the water/air interface. After slowly increasing the water level, 50 μL of an aqueous 2% w/v SDS solution was added as surfactant. Opposite to the side where the PS spheres started to cluster, the glass slide was carefully removed from the solution and a silicon substrate was submerged. The silicon substrate was moved underneath the PS spheres and the water level was carefully lowered by removing all water. The rest of the solution evaporated over time, resulting in well‐ordered hexagonally packed PS arrays reflecting blue light when seen under an angle.

The diameter of the PS spheres was reduced by reactive ion etching using a custom‐built etching machine (10 mTorr, 20 sccm O2, 20 W, 2.5 min). Then, a silver layer of 40 nm (100 W, 1.0 min) or a gold layer of 30 nm (200 W, 45 s) was sputtered on top using a custom‐built sputtering system. Lift‐off of the PS spheres was achieved by 90 min immersion and 5 min sonication in dichloromethane. For the MACE step,11c,13 the substrates were immersed in an aqueous solution of 5.8 M HF/0.1 M H2O2 (50% HF, 31% H2O2 and water mixed at 20/0.9/79.1 v/v%) for 7 min (Ag) or 15 min (Au) to obtain silicon nanowire arrays. The samples were rinsed with water and dried under nitrogen. The samples were cut into two halves to use one half as a control without monolayer formation. After monolayer formation on the other half, the second MACE step was performed on both halves in the same MACE solution for 15 min (Ag) and 30 or 45 min (Au).

Monolayer Formation

Monolayers of 1‐undecene, 1‐tetradecyne or 1,8‐nonadiyne were formed by hydrosilylation (Scheme S1). The adsorbate solution (3–6 mL), consisting of a 5% v/v solution of 1‐undecene in mesitylene, a 5% v/v solution of 1‐tetradecyne in mesitylene or pure 1,8‐nonadiyne, was degassed by four freeze‐pump‐thaw cycles for the hydrosilylation reaction. After MACE and HR‐SEM imaging, a hydrogen‐terminated surface was reactivated on the p‐type silicon nanowires by 1 min immersion in an aqueous 1% HF solution, to remove the native oxide grown during HR‐SEM imaging. The substrates were then immediately immersed in the degassed adsorbate solution inside a nitrogen glovebox. The reaction flask was equipped with a capillary as a nitrogen inlet and a reflux condenser. The hydrosilylation reactions were performed overnight under continuous nitrogen flow at 180 °C (for 1‐undecene) for the thermal route and under illumination of a 420 nm LED (130 mW) at a 1 cm distance for the photochemical route (for 1‐undecene, 1‐tetradecyne or 1,8‐nonadiyne). The nanowires were cleaned by immersion in hexane, rinsed with ethanol and dichloromethane, and subsequently dried in a stream of nitrogen. The samples were not sonicated in order to keep the metal film intact.

Click Chemistry

Copper‐catalyzed azide‐alkyne cycloaddition (click chemistry, Scheme S1) was used to couple the fluorescent dye azide‐fluor 488 or azide‐functionalized gold NPs onto a 1,8‐nonadiyne monolayer. Silicon nanowires were overnight incubated with 60 μL of the azide solution (2 mM azide‐fluor 488 in water, or azide‐functionalized gold NPs as received) and 60 μL of the catalyst solution (1.5 mM Cu(II)SO4.5H2O, 1.5 mM TBTA, 60 mM L‐ascorbic acid in ethanol/water, ratio 5 : 2 v/v) in a silicone isolator (Electron Microscopy Sciences). A glass slide on top was used to avoid solvent evaporation. Afterwards, the samples were sequentially rinsed with water, ethanol, water, ethanol, immersed in acetone to remove the glue of the isolator, sonicated in ethanol for 1 min and dried in a stream of nitrogen.

Contact Angle Measurements

Static contact angles were measured with Milli‐Q water on a Krüss G10 Contact Angle Measuring Instrument equipped with a CCD camera and drop shape analysis software. Contact angles were measured directly after the hydrosilylation reaction and averaged over three drops.

High‐resolution Scanning Electron Microscopy

HR‐SEM images of cross sections of the nanowire arrays were obtained with a FEI Sirion HR‐SEM with a through‐the‐lens detector (TLD) or a Zeiss Merlin HR‐SEM system with an InLens or ESB detector, operated at typical acceleration voltages of 10–15 kV and 1.6 kV, respectively. Fresh cuts were made to assess the cross sections after each step.

Fluorescence Microscopy

Silicon nanowires functionalized with azide‐functionalized dye were removed from the surface using a scalpel and immersed in a small volume of ethanol. The solution was pipetted onto a microscope glass slide, after which the ethanol was evaporated to confine the nanowires onto the glass slide. Fluorescence microscopy images were acquired in air on an Olympus inverted research microscope IX71 equipped with a mercury burner U‐RFL−T as light source and a digital Olympus DP70 camera. Blue excitation (λex=490–510 nm) and green emission (λem=520–550 nm) were filtered using a Chroma filter cube.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.