Spatioselective Deposition of Passivating and Electrocatalytic Layers on Silicon Nanowire Arrays

Abstract

Metal–silicon nanowire array photoelectrodes provide a promising architecture for water-splitting because they can afford high catalyst loading and decouple charge separation from the light absorption process. To further improve and understand these hybrid nanowire photoelectrodes, control of the catalyst amount and location within the wire array is required. Such a level of control is currently synthetically challenging to achieve. Here, we report the synthesis of cm²-sized hybrid silicon nanowire arrays with electrocatalytically active Ni–Mo and Pt patches placed at defined vertical locations within the individual nanowires. Our method is based on a modified three-dimensional electrochemical axial lithography (3DEAL), which combines metal-assisted chemical etching (MACE) to produce Si nanowires with spatially defined SiO₂ protection layers to selectively cover and uncover specific areas within the nanowire arrays. This spatioselective SiO₂ passivation yields nanowire arrays with well-defined exposed Si surfaces, with feature sizes down to 100 nm in the axial direction. Subsequent electrodeposition directs the growth of the metal catalysts at the exposed silicon surfaces. As a proof of concept, we report photoelectrocatalytic activity of the deposited catalysts for the hydrogen evolution reaction on p-type Si nanowire photocathodes. This demonstrates the functionality of these hybrid metal/Si nanowire arrays patterned via 3DEAL, which paves the way for investigations of the influence of three-dimensional geometrical parameters on the conversion efficiency of nanostructured photoelectrodes interfaced with metal catalysts.

1. Introduction

Silicon nano- and microwire arrays provide versatile architecture with outstanding and tunable optoelectronic properties that can combine light trapping, Mie resonances, waveguiding, and diffractive effects.¹⁻⁶ These properties have enabled their use in a range of applications such as sensing, solar conversion, photodetection, and synthesis of optically controlled biointerfaces.⁷⁻²⁵ In most cases, the semiconductor wires are interfaced with metal structures that act either as plasmonic materials,^10,11 electrical contacts,²⁶ or catalysts.¹²⁻²⁴ The ability to precisely locate the metal within the nanostructured substrates is necessary to optimize and control their biological, photocatalytic, and optoelectronic properties.^{10−18,20−26} This is especially true for photo(electro)chemical systems that take advantage of the superior photonic properties of nanowire arrays.¹²⁻¹⁹ In these systems, a number of factors can affect the local catalyst activity due to spatially inhomogeneous mass transport, charge recombination, light absorption/scattering, catalyst coverage, and defect distribution.^16,17,27,28 Recent studies have shown that the deleterious effects of such inhomogeneities can be mitigated by optimizing catalyst location at the semiconductor surface.^20,21,27,29 However, such investigations are complex and synthetically challenging.

In the context of solar water-splitting, single-crystalline nanostructured silicon is the substrate of choice because of its abundance, defined and tunable electronic properties, appropriate band gap for solar light absorption, and proper conduction band edge energy for the hydrogen evolution reaction. Because of the slow reduction kinetic of protons, silicon photoelectrodes require the addition of catalytically active materials, which are usually based on noble metals, such as Pt.³⁰ Earth-abundant materials such as Ni, MoS₂, and Ni–Mo are attractive alternatives to the traditional and expensive Pt catalyst.^20,22,30,31 Their relatively lower activity requires the addition of larger amounts of catalyst on the semiconductor surface. In the case of planar photocathodes, this can lead to large optical losses caused by light absorption in the metal catalyst layer and a decreased photocathode efficiency in a front-face illumination configuration.²⁰ The use of micro- and nanowire arrays can mitigate this issue by decoupling light absorption and charge separation to allow higher catalyst loading without compromising the light absorption process.^{12−17,20−24} For example, Huskens et al. demonstrated high conversion efficiencies for the hydrogen evolution reaction (i.e., up to 10.8% in a regenerative cell) by locating the Ni–Mo catalyst at the top of buried n⁺/p Si microwires.²⁰ This demonstrates that the catalyst position within the Si wire arrays critically influences the conversion efficiencies. However, this effect remains largely unexplored because of the difficulty to precisely control its location within the array.

The spatioselective decoration of vertically aligned silicon nanowire (VA-SiNW) arrays with metals is conventionally performed using a polymeric membrane that selectively protects the bottom of the wires and directs the functionalization at the uncovered top parts of the Si wires.^12,20,32 This strategy provides no real control over the catalyst location in the axial direction (i.e., along the nanowire length). Templated electrochemical techniques have emerged as promising alternatives to synthesize and control heterostructures at the nanoscale.³³⁻⁴⁶ For example, the on-wire lithography³⁸⁻⁴¹ and the coaxial lithography⁴²⁻⁴⁴ have allowed the structuring of electrochemically grown nanowires in both axial and radial directions. More recently, the three-dimensional electrochemical axial lithography (3DEAL) has presented a flexible platform to interface materials around silicon nanowires with a controlled axial deposition.⁴⁶ The original 3DEAL yields silicon nanowires decorated with well-defined metal rings that are electrically insulated from each other by a thin SiO₂ layer, which impedes the charge transport. This complicates the direct use of 3DEAL to prepare Si–metal hybrid photoelectrodes.

Here, we present an electrochemical method of controlling catalyst position along the Si nanowires using a modified version of the 3DEAL. Our approach affords localized passivation of the Si surface, with feature lengths down to 100 nm in the axial direction, which we subsequently use to electrochemically grow Pt and Ni–Mo catalysts at specific locations along the nanowires (Figure 1).

Figure 1.

Schematic illustration of the axially controlled deposition of catalytically active materials within the Si nanowire arrays via modified 3DEAL. A gold nanohole array (yellow) is formed via colloidal lithograhy (see Supporting Information for more details). (1) Metal-assisted chemical etching (MACE): The gold film etches into the substrate to fabricate silicon nanowire arrays (gray). (2) A protective electrically insulating sub-5 nm SiO₂ shell (dark blue) is deposited on the Si nanowires via sol–gel chemistry. (3) Planar films of gold (yellow) and nickel (green) are sequentially electrochemically grown on the gold base layer. (4) After chemical removal of the gold film, the array is then coated with a conformal SiO₂ film via sol–gel chemistry (white, 5). Subsequent selective etching of the nickel layer (6) leads to silicon nanowire arrays that are patterned with a thick nanostructured SiO₂ layer. Thermal annealing and KOH etching (7) dissolve both the residual horizontal SiO₂ layer deposited on top of the nickel layer and the sub-5 nm SiO₂ shell, yielding SiO₂ passivated silicon nanowire arrays with exposed areas at defined axial positions. Spatioselective electrodeposition of metal particles (green, 8) can now be realized at these defined locations.

2. Results and Discussion

2.1. SiO₂ Patterning via 3DEAL

The p-type VA-SiNW arrays were synthesized via metal-assisted chemical etching (MACE; Figure 1).⁴⁷⁻⁵⁰ In this process, a gold nanohole array produced via colloidal lithography is used to anisotropically etch silicon using a HF/H₂O₂ solution (details in the Supporting Information). After the MACE process, this gold base layer is present at the bottom of the SiNW array (Figures 1 and 2) and can be electrically connected to the working electrode of a potentiostat to electrochemically grow additional metal films on top (Figures 2b and S1). To ensure a homogeneous layered growth of the metal films at the bottom of the array, a sub-5 nm thin SiO₂ protective shell is chemically deposited onto the Si NWs before the electrodeposition step. This protection prevents metal nucleation at the top of the wires (Figure S2).⁴⁶ The first metal film, made of gold, grows uniformly around the silicon nanowires from the bottom to the top (Figure 2b). Nickel is subsequently grown on top of the gold layer (Figures S2 and S3) to form a bilayered Au/Ni film embedded within the nanowire array. Selective etching of the gold film in the KI/I₂ solution yields a VA-SiNW array with a continuous nickel film of constant thickness and a defined distance from the bottom of the wires, as controlled by the thickness of the sacrificial gold film (Figures 2c and S4). The sol–gel deposition of SiO₂ generates a conformal SiO₂ film around the silicon substrate and the nickel film, as can be seen in Figure 2d,e.^46,51,52 Electron microscopy shows that the SiO₂ shell is slightly thicker above the nickel film, most likely due to mass-transport limitation caused by the continuous nickel film (Figure S5). However, the wires are homogeneously passivated with SiO₂ above and below the nickel film as detailed below. The complete chemical dissolution of the nickel film with HNO₃ leads to a nanostructured SiO₂ layer coating the VA-SiNWs (Figures 2e and S6). This is similar to the reported acid-induced etching of gold and nickel nanowires through SiO₂ shells grown under the same conditions, suggesting that the SiO₂ layer exhibits some level of porosity.^51,52 The areas initially protected by the nickel film are now exposed but remain covered with a sub-5 nm thick SiO₂ shell, which was initially deposited prior to the electrodeposition of the metal films. The rest of the Si NWs remain passivated with a thick SiO₂ film (i.e., >80 nm). Energy-dispersive X-ray spectroscopy (EDX) shows that the dissolution of both Au and Ni films is complete across the whole cm²-substrates (Figures S3, S4, and S6). Further thermal annealing (10 h at 500 °C) was found to improve the homogeneity of the SiO₂ shells (Figure S7),⁵³ yielding VA-SiNW arrays homogeneously patterned with insulating thick conformal SiO₂ shells and defined Si areas coated with a sub-5 nm SiO₂ layer (Figure 2e). The location and height of these unprotected areas are controlled by the thicknesses of the metal films deposited by electrochemical deposition after the MACE fabrication (Figure S8), down to feature sizes of around 100 nm (Table S1 and Figure S8c).

Figure 2.

SiO₂ patterning via 3DEAL. Top row: synthesis scheme. Bottom row: secondary electron scanning electron microscopy (SEM) cross-sectional images. (a) VA-SiNW array after MACE and deposition of a sub-5 nm conformal SiO₂ shell. (b) Planar electrodeposited gold and nickel films that are embedded within the VA-SiNWs. (c) Selective etching of gold with an aqueous KI/I₂ solution. (d) Deposition of a conformal SiO₂ thick film. The inset shows higher magnification. (e) Selective etching of nickel yields VA-SiNW arrays patterned with continuous thick SiO₂ shells. The area in between is passivated with a sub-5 nm SiO₂ layer (shown in blue) that can be etched at a later stage via KOH. The inset shows higher magnification.

2.2. Spatioselective Deposition of the Metal Catalysts

Next, the silicon substrate is used as a working electrode for the direct electrodeposition of the metal catalysts (Figure 3). A KOH etching step is first used to remove the flat SiO₂ horizontal layers and the sub-5 nm SiO₂ shell in the patterned regions (Figure 1). Remarkably, the horizontal SiO₂ layers dissolve much faster than the SiO₂ shells around the Si NWs. We expect that repeated pitting of these horizontal SiO₂ layers promotes the KOH solution infiltration within the interlayer region (i.e., where the nickel film was originally), dissolving SiO₂ from top and bottom, which accelerates the etching process. This is not possible for the SiO₂ shells covering the Si NWs, which can only be etched from one side, and thus dissolve slower. This KOH treatment also dissolves the sub-5 nm thick silicon oxide layer on the nonpassivated area (shown in dark blue in Figure 2e), which allows for the subsequent spatioselective electrodeposition of metals on the Si NWs (Figure 3e–l). As seen with SEM, scanning transmission electron microscopy (STEM), and EDX analysis (Figures 3a–d and S9), the silicon surface of the unprotected areas is a bit rough, which is, in our experience, typical for Si wires prepared via MACE using a sputtered gold nanomesh as a mask, while the protected areas have a relatively smooth surface that originates from the successive sol–gel deposition steps. Longer KOH treatments etch into the exposed areas of the VA-SiNWs and can lead to ill-defined patterning (Figure S10). As an example of catalytically active material, we electrochemically deposit platinum at the exposed areas of the Si NW arrays from a homemade hexachloroplatinic acid aqueous solution at a constant potential. Acidic plating solutions (5 mM H₂PtCl₆ and 0.5 M Na₂SO₄, initial pH = 3) yield considerable parasitic deposition over the entire wire surface (i.e., even on the passivated SiO₂ shells; Figure S11). We tentatively attribute this failure of spatial localization to the generation of H₂ bubbles, which can electrochemically reduce the Pt ions and therefore induce heterogeneous nucleation of Pt nanoparticles at both Si and SiO₂ nanowire surfaces. The use of a buffered solution (15 mM H₂PtCl₆, 200 mM Na₂HPO₄, initial pH = 7), circumvents these problems and leads to the exclusive deposition of Pt on the predetermined unprotected regions of the nanowires (Figure 3i–l). Besides Pt, nanostructured earth-abundant Ni–Mo catalyst was deposited as previously described using a homemade aqueous Ni–Mo solution (1.3 M NiSO₄·6H₂O, 25 mM sodium molybdate, 0.5 M H₃BO₃) at a constant potential (Figure 3e–h).^20,22,31 As shown by EDX analysis (Figure S12), the electrodeposited Ni–Mo catalysts have around 8% of Mo atoms incorporated, corresponding to a Ni_0.92–Mo_0.08 alloy, close to the expected value for this electrochemical bath.^20,22,31 The SEM/STEM/EDX analysis in Figure 3 clearly demonstrates the controlled placement of different, catalytically active materials at controlled axial locations within the VA-SiNW arrays: The metal depositions occur selectively on the unprotected regions. In addition, the defined position of the catalyst in each single nanowire extends uniformly throughout the macroscopic substrate, as seen in the low-magnification SEM images (Figure S13). We also report the functionalization of patterns at different locations along the nanowires (bottom, center, and top, Figures 3 and S14). This validates our experimental approach to site-specifically pattern VA-SiNW with insulating and electrocatalytic nanostructured layers.

Figure 3.

Spatioselective metal deposition characterized via SEM, STEM, TEM, and EDX analysis. (a–d) Nanowires patterned with SiO₂. (e–h) Nanowire arrays patterned with SiO₂ and Ni–Mo. (i–l) Nanowire arrays patterned with SiO₂ and Pt. (a, e, i) Synthetic schemes. (b, f, j) Top: STEM secondary electron images. Bottom: EDX overlay maps. Red: Si; blue: O; green: metal (f: Ni and j: Pt). (c, g, k) High-magnification bright-field TEM (c) and secondary electron STEM (g, k) images at the boundary between the SiO₂-passivated segments and the unpassivated regions coated with the metal catalyst. (d, h, l) Low-magnification SEM cross-sectional images.

2.3. Photoelectrochemical Tests

As a proof-of-principle experiment, we prepared photocathodes from the arrays and showed photoelectrochemical activity of the patterned hybrid VA-SiNW arrays synthesized via 3DEAL. The arrays were electrically connected to fluorine-doped tin oxide (FTO) glass slides using a Ga–In eutectic mixture. Linear sweep voltammetry in 0.1 M H₂SO₄ (pH = 1) under light irradiation (a 300 W Xe lamp equipped with an AM1.5G filter calibrated to an intensity of ca. 120 mW cm^–2), shown in Figure S15, demonstrates the successful functionalization of the nanostructured silicon photocathodes. Two substrates were tested: an array composed of bare VA-SiNWs (no catalyst) and a VA-SiNW array patterned with SiO₂ and Ni–Mo (the STEM/EDX images of this sample are shown in Figures 3e–h and S13). Both samples were prepared from the same native Si wire array substrate, which was split into smaller pieces before electrochemical testing. Under light irradiation, the onset potential, defined here as the potential at which the current reaches 1 mA·cm^–2, goes from ca. −0.17 V for the bare wires to ca. +0.11 V (vs RHE, reversible hydrogen electrode) for the array partially coated with Ni–Mo. Additionally, the current density under light irradiation is larger for the array coated with the Ni–Mo catalysts. These observations are in line with previous studies based on metal-coated VA-SiNW array photoelectrodes synthesized via a combined MACE/colloidal lithography route.¹⁸ These results confirm the viability of 3DEAL to pattern VA-SiNW photoelectrodes with functional metal electrocatalysts.

3. Conclusions

To conclude, we report a simple electrochemical approach to pattern Si NW arrays with passivating and nanostructured electrocatalytic layers. The modified 3DEAL is based on the electrochemical deposition, sol–gel chemistry, and selective etching and is compatible with the patterning of densely packed VA-SiNW arrays with features down to 100 nm in the axial direction. The wire arrays selectively patterned with Ni–Mo nanoparticles show activity for the hydrogen evolution reaction at acidic pH. This demonstrates the potential of the technique to investigate the structure–property relationships linking the precise position of catalyst particles with the resulting (photo)electrochemical conversion performance of these complex structures. Such studies provide design rules to optimize and minimize catalyst loading within nanostructured electrodes. The process is flexible with respect to the active material decorating the silicon nanowires but could also be transferred to different semiconducting wire materials, therefore providing a flexible toolbox to design tunable and well-controlled hybrid nanowire arrays.

4. Experimental Section

Silicon nanowire arrays were synthesized via colloidal lithography and metal-assisted chemical etching (MACE) following the protocols we have previously developed.^6,46,49 Further details can be found in the Supporting Information.

4.1. Patterning of the SiO₂ Insulating Layers

The process starts with the electrodeposition of Au and Ni at the bottom of the Si wire array (Figure S1). Before the electrode position, a thin protective sub-5 nm SiO₂ shell is deposited for 30 min around the wires via a sol–gel process⁵¹ (more details in the Supporting Information) to avoid the parasitic growth of metal particles on the silicon wires (Figure S2).⁴⁶ Electrochemical deposition was performed using Keithley Sourcemeter 2400 operated with a three-electrode configuration and a homemade electrochemical Teflon cell (Figure S1). A Pt mesh was used as the counter electrode, Ag/AgCl as the reference electrode, while the Au film at the bottom of the silicon wire arrays (remaining after MACE) was used as the working electrode, similar to the original 3DEAL.⁴⁶ After metal deposition, the gold film below the nickel is selectively etched in a KI/I₂ solution. The dissolution was monitored by the SEM EDX analysis (Figure S4). At this stage of the synthesis, the wires are protected with a continuous flat nickel film, located at a fixed distance from the bottom of the wires. A SiO₂ shell is then deposited around the wires for 8.5 h (with the nickel film) using the same sol–gel process. Homogeneous and conformal SiO₂ films are obtained via successive 1-h deposition steps with a freshly prepared sol–gel solution, which allowed us to avoid the unwanted nucleation of SiO₂ particles in solution that occurs during longer sol–gel deposition steps. The nickel film is etched in a 1:1 H₂O/HNO₃ solution (monitored by SEM EDX measurements, Figure S6), which reveals a spatioselectively patterned silicon wire array with SiO₂. The substrates are then annealed for 10 h at 500 °C, which increases the homogeneity of the SiO₂ shell (Figure S7). The location and height of the patterned area can be precisely controlled by the electrodeposition duration of Au and Ni (Figures S8, S14 and Table S1). More details can be found in the Supporting Information.

4.2. Spatioselective Deposition of the Metal Catalysts

Prior to the spatioselective deposition of metal catalysts, the protective sub-5 nm SiO₂ shell is removed via a KOH etching step (Figures S9 and S10). The metal catalyst electrodeposition was performed immediately after KOH etching to avoid the growth of the native SiO₂ layer. Ni–Mo and Pt were deposited using homemade aqueous solutions at a constant potential. Additional details can be found in the Supporting Information.

4.3. Preparation of the Photocathodes

The backside of the Si substrate was scratched with a glass cutter and a Ga–In eutectic mixture was added on top. The Si substrate was then placed on a glass slide coated with FTO (resistivity 6–9 Ω cm). The back, sides, and front of the photoelectrodes were covered with an electrically insulating epoxy. The photoelectrodes were photographed and the active surface area was measured with the open-source software ImageJ. The photoelectrode active surface areas were in the 0.3–0.8 cm² range.

4.4. Photoelectrochemical Measurements

Photoelectrochemical measurements were performed using a VersaSTAT 4 potentiostat in a three-electrode setup with a Pt wire counter electrode and a Ag/AgCl (3M NaCl) reference electrode. A solar simulator composed of a 300 W Xe lamp (LotOriel) equipped with an AM1.5G filter with an intensity of ca. 120 mW cm^–2 calibrated with a Si photodiode was used for light irradiation. The irradiated area was circular with a diameter of 8.5 mm. All measurements were performed in a front-face illumination configuration. The electrolyte, an aqueous 0.1 M H₂SO₄ solution (pH = 1), was purged with Ar for 30 min prior to the measurement. The electrode potential was initially cycled several times in the dark between 0 and −1.6 V (vs Ag/AgCl) until the current value stabilized. The electrode potential was then cycled between −0.1 and −0.8 V (vs Ag/AgCl) under light illumination until the current value stabilized. Linear sweep voltammograms were acquired at a scan rate of 100 mV s^–1. An automatic optical shutter was used to observe the rapid changes in current induced by the incident light (Figure S15). E_Ag/AgCl, the electrode potential vs Ag/AgCl was converted vs the potential of the reversible hydrogen electrode (RHE) and expressed as E_RHE, defined by the following equation

where E_{Ag/AgCl vs SHE}⁰ is the potential of the Ag/AgCl reference electrode vs SHE (standard hydrogen electrode), which is 0.209 V for an Ag/AgCl electrode stored in 3 M NaCl. Therefore, the overall shift in the potentials for the RHE was calculated using E(RHE) = E(Ag/AgCl) + 0.268 V.

Glossary

Abbreviations

3DEAL

three-dimensional electrochemical axial lithography

EDX

energy-dispersive X-ray spectroscopy

FTO

fluorine-doped tin oxide

MACE

metal-assisted chemical etching

RHE

reversible hydrogen electrode

SEM

scanning electron microscopy

SHE

standard hydrogen potential

STEM

scanning transmission electron microscopy

VA-SiNW

vertically aligned Si nanowires

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.0c14013.

• Materials and details of the experimental methods, tables listing the dimensions of the Si NW arrays, and additional SEM images are included in the Supporting Information (PDF)

Author Contributions

^# F.J.W. and M.A. contributed equally to this work. G.R.B. designed and supervised the project. F.J.W. and M.A. developed the synthesis. F.J.W., M.A., and H.B. synthesized the nanowire arrays. M.A. and V.W. performed the photoelectrochemical measurements. M.R., E.S.A.G., and N.V. performed the colloidal self-assembly and lithography. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.