Large-Scale Synthesis of Highly Uniform Silicon Nanowire Arrays Using Metal-Assisted Chemical Etching

The combination of metal-assisted chemical etching (MACE) with colloidal lithography has emerged as a simple and cost-effective approach to nanostructure silicon. It is especially efficient at synthesizing Si micro- and nanowire arrays using a catalytic metal mesh, which sinks into the silicon substrate during the etching process. The approach provides a precise control over the array geometry, without requiring expensive nanopatterning techniques. Although MACE is a high-throughput solution-based approach, achieving large-scale homogeneity can be challenging because of the instability of the metal catalyst when the experimental parameters are not set appropriately. Such instabilities can lead to metal film fracture, significantly damaging the substrate and thus compromising the nanowire array quality. Here, we report on the critical parameters that influence the stability of the metal catalyst layer for achieving large-scale homogeneous MACE: etchant composition, metal film thickness, adhesion layer thickness, nanowire diameter and pitch, metal film coverage, Si/Au/etchant interface length, and crystalline quality of the colloidal template (grain size and defects). Our results investigate the origin of the catalyst film fracture and reveal that MACE experiments should be optimized for each Si wire array geometry by keeping the etch rate below a certain threshold. We show that the Si/Au/etchant interface length also affects the etch rate and should thus be considered when optimizing the MACE experimental parameters. Finally, our results demonstrate that colloidal templates with small grain sizes (i.e., <100 μm2) can yield significant problems during the pattern transfer because of a high density of defects at the grain boundaries that negatively affects the metal film stability. As such, this work provides guidelines for the large-scale synthesis of Si micro- and nanowire arrays via MACE, relevant for both new and experienced researchers working with MACE.


Synthetic protocols
Colloidal sphere synthesis PS colloidal particles were synthesized, self-assembled at the air-water interface and used as mask as described elsewhere. 1 SiO2-PNiPAm core-shell particles with an inner diameter of 160 nm (±10 nm) and a shell of 480 nm were synthesized as described elsewhere. 4

Preparation of substrates
Colloidal sphere monolayer were obtained by self-assembly at the water-air interface and by spin coating. The substrates were either used directly (2" wafers) or cut to the appropriate size using a glass cutter. Prior to self-assembly, the substrates were cleaned via sonication in acetone, ethanol and MilliQ water for 4 min each, followed by an oxygen plasma treatment (50 W, 10 mL/min, 4 min) to render the surface clean and hydrophilic.

Self-Assembly at the water/air interface
Prior to self-assembly at the water/air interface, the PS colloidal particles (d = 590 nm, d = 1100 nm) were cleaned and purified by three centrifugation-redispersion steps. 5 Typically, 2 mL of the PS solution was mixed with 20 mL MilliQ water and 20 mL ethanol (absolute) and centrifuged at 4000 rpm for 15 minutes. The supernatant was removed, and the PS colloidal particles were redispersed in 20 mL MilliQ water and 20 mL ethanol (absolute). The procedur was repeated three times. The final redispersion was done with 1 mL MilliQ water and 1 mL ethanol (absolute).
For the spreading onto the water/air interface a syringe pump (Landgraf HLL LA120) was used ( Figure S1). A 1 mL syringe was filled with 500 µL PS spheres and 500 µL ethanol (absolute) for the 1100 nm PS colloidal particles and with 750 µL PS sphere solution and 250 µL ethanol (absolute) for the 590 nm PS colloidal particles. A big beaker (Ø = 140 mm; height: 70 mm) was filled with MilliQ water and a syringe was bent with a 90 ° angle and placed in such way that half of the opening at the syringe tip was immersed in the water. During self-assembly, the dispersions were added slowly (1-30µl/min) via the syringe to the air/water interface to minimize disturbance.
When the water surface was fully covered by a monolayer, the syringe pump was turned off.
Silicon substrates were immersed in the water phase and elevated under a shallow angle to transfer the monolayer. The substrates were left to dry in air on top of a tissue and at an angle of ~ 45 °.
After removal of each sample, the syringe pump was turned on again, and additional PS colloidal particles were added to the interface via the syringe pump to replenish the interfacial assembly.
Several substrates (~ 10 -15) can be pulled out until the process has to be started again, by filling up the solution in the syringe and renewing the water in the large beaker. The procedure delivered large-area colloidal crystals with grains in the mm² range ( Figure S2).
The SiO2-PNiPAm core-shell particles were self-assembled at the water/air interface using a Langmuir-Blodgett trough (KSVNIMA) (area = 243 cm², width = 7.5 cm) with Delrin barriers and the surface pressure was measured by a Wilhelmy plate. Four substrates were mounted vertically onto the dipper and the trough was filled with Milli-Q water. The core-shell particle suspension was diluted to 0.5 wt.%, mixed with 30 wt.% ethanol as the spreading agent, and spread at the water/air interface of the trough using a regular 100 µL pipette. After 10 min of equilibration, the barriers were compressed to a constant surface pressure of 30 mN/m while the dipper was lifted by 0.8 mm/min. After deposition, the SiO2-PNiPAm core-shell particles were exposed to oxygen plasma for 10 min to remove the organic PNiPAm shell, leaving a non-close packed array of silica particles at the surface. 4 Self-Assembly using spin coating

Metal film deposition
After converting the close-packed sphere monolayer into a non-close packed architecture, a thin metal film is deposited by sputter coating or thermal evaporation. The spheres protect the surface from the metal and a metal hole array is created. Gold was used for all substrates in this work, due to its excellent chemical stability.
Before gold sputtering a thin adhesion layer of Al-doped ZnO was sputtered directly on the nonclose packed colloidal monolayer, using a Clustex 100M sputtering system from Leybold Optics.
The substrates were fixed on the sample holder and inserted into the reaction chamber. The chamber was evacuated to around 1*10 -6 mbar. Sputtering was performed for 1 second using Argon gas at a pressure at 3*10 -3 mbar and a power in the 75-200 W range depending on the PS particle size. After sputtering of the adhesion layer, gold was immediately sputtered using a

Removal of the colloidal template
After metal deposition, the PS particles were removed using adhesive tape (Scotch Magic Tape).
The adhesive tape was carefully placed on top of the substrate and then brought into contact with the backside of a tweezer or by using a finger ( Figure S3). A gold film without patterning, i.e. corresponding to an area where no PS spheres were present before the gold sputtering, usually delaminated during the removal of adhesive tape. To avoid gold film delamination the tape removal was always initiated at a patterned side of the substrate. Finally, the substrates were cleaned from residual tape with oxygen plasma using the Emitech K1050x at 50 W for 5 min with an oxygen flow of 10 mL/min.

Synthesis of Si nanowire arrays via metal-assisted chemical etching (MACE)
MACE was used to prepare Si nanowire (SiNW) arrays by immersing the Si/Au hole array substrates in an aqueous HF/H2O2 mixture. 2, 3, 6 Caution: Appropriate safety precautions have to be observed when working with hydrofluoric acid (HF): HF is a contact poison! The experimental build up is shown in Figure S4. All HF steps were performed inside a HF specific fume hood, using HF resistant plastic or Teflon tools, beakers and butyl gloves (Butoject 897+ from Honeywell). For safety precautions a 2 L beaker was filled with 1 L liquid calcium gluconate, as quick washing of the exposed skin in the event of an accident can significantly limit the damages caused by HF. In addition, a 2.5 % calcium gluconate gel (C-gel) was commercially bought for the same purpose and kept close to the hood where the experiment was performed. A second 2 L beaker was filled with 1 -1.5 L of calcium chloride (CaCl2) aqueous solution. CaCl2 scavenges the toxic fluoride ions by quickly forming an inert precipitate of CaF2, thus significantly reducing the toxicity of the HF waste and reducing the chances of HF poisoning and contaminations. The various HF etchings and subsequent rinsing were carried out in plastic beakers in a plastic box inside the fume hood. Different home-made 3D printed polymer (polylactic acid, PLA) sample holders were prepared to avoid using tweezers, allowing reproducible etching of several samples at the same time ( Figure S4). At the end of the experiment, all pieces of equipment used were fully immersed into the calcium chloride solution, thoroughly rinsed with deionized water, dried and stored inside a fume hood until further use.
The MACE solution was freshly prepared before etching and was composed of 10 mL of

Scanning Electron Microscopy (SEM)
Secondary electron (SE) and back-scattered electron (BSE) images of the arrays were acquired using a Zeiss Ultra Plus 55 at a working distance of ~ 2.5 -4 mm, equipped with an InLens SE detector, an InLens BSE detector (EsB) and an angle selective BSE detector (AsB) located below the pole piece. The accelerating voltage was adjusted between 1 and 10 kV depending on the sample to limit charging.

Calculation of the gold coverage and the normalized Au/Si/etchant interface length
The gold coverage and wire density were calculated assuming 100 % defect-free monolayer. The interface length was calculated by taking the circumference of a wire multiplied by the number of wires for a specific substrate area. The interface length shown in Figure 5b (main text) was normalized to the largest calculated value, obtained for a hexagonal close-packed wire array with 590 diameter and a 590 nm pitch, corresponding to a 9.31 % Au coverage. (a) Growth of a PS sphere monolayer. First Image: The PS sphere solution is added to the air/water interface using a syringe pump. Inset shows the immersion of the syringe tip from side and front at the water-air interface. Second Image: Structural colors (diffraction) occurring from the closepacked, enable a simple, visual quality control of the assembly process. Third image: Nearly finished monolayer on the water surface. The Inset shows a PS colloidal monolayer on a large Si substrate. (b) Transfer of a PS sphere monolayer onto a Si substrate by manual immersion and withdrawal of the substrate. A light source under a shallow angle illuminates the air/water interface to increase the visibility of the interfacially assembled monolayer.