Direct Gravure Printing of Silicon Nanowires Using Entropic Attraction Forces

Background

Solution-based NW assembly methods, including capillary and convective assembly, require a solvent evaporation step, so the motion of suspended NWs during evaporation determines where they end up. In earlier contact-line methods the NWs could only be aligned at the contact line between the solvent and the substrate and only parallel to it, which limited the possible array patterns.

Design and approach

Si NWs made by aqueous electroless etching (AEE) are dispersed by brief sonication in polar or nonpolar solvents and dropped onto a Si substrate with micro-engraved structures. The structures are etched in tetramethylammonium hydroxide (TMAH), giving Si(111) facets with an inward slope of 54.7°. During evaporation, internal flow carries the NWs to the structures, and entropic depletion attraction holds them at the corners. For printing, a rigid, low-adhesion polydimethylsiloxane (PDMS) stamp removes NWs outside the grooves, a soft, high-adhesion PDMS stamp lifts the aligned arrays, and the arrays are bonded with an SU-8 epoxy layer to a polyethylene terephthalate (PET) substrate carrying an indium tin oxide (ITO) gate and an Al2O3 dielectric deposited by atomic layer deposition.

How it was evaluated

Assembly was compared across polar solvents (DI water, isopropyl alcohol, methyl alcohol, acetone) and nonpolar solvents (chloroform, toluene, methylene chloride) using optical microscopy and scanning electron microscopy (SEM), and evaporation was followed with an in-situ optical microscope. Alignment angles were measured on line patterns 5 to 30 μm wide, sequential deposition was examined by SEM in a 150 μm × 150 μm truncated pyramid, and an assembled array in a micro-engraved structure was cross-sectioned with a focused ion beam system, and NW diameters and lengths were measured to model the depletion interaction. Back-gated FETs with a channel length of approximately 8 μm were characterized by transfer curves at a drain-source voltage of 5 V and by output curves.

Key results

The assembly selectivity, defined as the percentage of NWs located in the micro-engraved regions, was about 95% for DI water and about 32% for chloroform; DI water gave the highest selectivity among the polar solvents, and NWs in the nonpolar solvents assembled randomly. In polar solutions more than about 73% of the NWs lay within 10° of the line direction. NWs stayed aligned when the line width was smaller than the NW length (10 μm or less), while selectivity did not change with line width.

In-situ observation showed that NWs moved actively in polar solutions and were driven into the grooves, whereas their movement was much more restricted in nonpolar solutions. The authors relate this to the evaporation mode: in polar solutions assembly occurred mostly during the pinned contact mode, which drives outward flow toward the contact line, whereas the depinned contact mode dominates for nonpolar solvents.

Because flow alone cannot account for a selectivity of about 95%, the authors examined deposition inside a 150 μm × 150 μm truncated pyramid, where NWs with larger diameters nucleated first along the edges and smaller NWs followed. The NWs ranged from 25 to 940 nm in diameter and 2.5 to 15 μm in length and fell into two groups averaging 190 nm/5.5 μm and 600 nm/9 μm. Their depletion-attraction model gave a free energy decrease of 8.6 kBT when a large NW nucleated at a notch, much larger than the thermal energy.

The rigid PDMS step removed more than 80% of the redundant NWs, and the printed patterns consisted of a single or bi-layer of Si NWs. The back-gated FETs on PET showed p-type behavior with an on/off ratio of about 2.6 × 10⁵, a subthreshold slope of 500 mV/decade, a transconductance of 92 nS, a threshold voltage of 2.1 V and a hole mobility of 17.1 cm²·V⁻¹·s⁻¹.

Open questions

The output curves did not show a clear saturation regime. The authors suggest that carrier scattering in the complex NW network, the large surface roughness of the NWs and poor contacts caused by residual polymer between the NWs and electrodes are possible causes.

How it connects to other research

The related paper on a monolayer resist transfers aligned nanowires by patterning adhesion differences in a light-patterned monolayer, whereas this study places the nanowires with evaporation-driven flow and engraved grooves and then lifts them with stamps of different adhesion.