Nanosphere lithography (NSL) (also known as colloidal lithography) is a simple, bottom-up fabrication technique that enables pattern generation at the nanoscale via self-assembly of nanospheres on a substrate. [1] Compared to top-down nanoscale patterning techniques such as deep UV or electron-beam lithography, NSL has advantages of high throughput, low cost, and versatility. Specifically, lift-off patterning using NSL can be used to create patterns of metals, insulators, and organic semiconductors on a variety of substrate materials with sub-100 nm feature sizes. [2] A variety of nanoscale patterns have been demonstrated with NSL, including: triangles, rings, dots, rods, bimetallic "cup-like structures", and wires. [3,4] The success of NSL is due in large part to the ability to manufacture nanospheres with tight size tolerances (e.g., <5% size variation [2]), and the ability to form singleand multilayer nanosphere patterns on a wide variety of substrates using facile methods such as spin-coating, [2] solvent evaporation, [5] drop-casting, [6] dip coating, [7] and electrophoretic deposition, [8,9] to name a few. As a result, researchers have used NSL as a patterning approach for numerous applications including: surfaceenhanced Raman scattering, [10,11] lightemitting diode micro-lenses, [12] nanogap electrodes for single-molecule electronics, [13] semiconductor nanowires for photovoltaic cells, [14,15] and nanohole arrays for transparent conducting electrodes, [16-18] as examples. One of the main factors that limits current practical application of NSL is the ability to reproducibly pattern large areas of defect-free nanosphere monolayers or multi-layers. In a perfectly assembled monolayer NSL pattern, the nanospheres are arranged in hexagonal close-packed (HCP) formation. In practice, self-assembled monolayers are subject to crystalline defects, incomplete surface coverage, and multi-layer stacking that limit the useable surface area for NSL applications. [1] The origin of these defects depends on the self-assembly method, and may include grain boundary formation originating from "impurity atoms" (i.e., nanospheres that fall outside strict size and/or shape tolerances), [9] an imbalance between rates of capillary-induced self-assembly versus convective transport of nanospheres to the air-liquid contact line, [19,20] a hydrophilic substrate, [19] and/or post-growth crack formation during drying. [20,21] Although nanosphere monolayers can be formed using simple methods such as those listed previously, the resulting patterns-including defects and areas of continuous coverage-are highly sensitive to assembly parameters. For example, nanosphere spin-coating is sensitive to spin speeds, ramp rates, sphere concentration, solvent volatility, and substrate surface chemistry; [1,22] more generally, convective self-assembly mechanisms depend on wetting angles, temperature and relative

Nanosphere lithography employs single- or multilayer self-assembled nanospheres as a template for bottom-up nanoscale patterning. The ability to produce self-assembled nanospheres with minimal packing defects over large areas is critical to advancing applications of nanosphere lithography. Spin coating is a simple-to-execute, high-throughput method of nanosphere self-assembly. The wide range of possible process parameters for nanosphere spin coating, howeverand the sensitivity of nanosphere self-assembly to these parameterscan lead to highly variable outcomes in nanosphere configuration by this method. Finding the optimum process parameters for nanosphere spin coating remains challenging. This work adopts a design-of-experiments approach to investigate the effects of seven factorsnanosphere wt%, methanol/water ratio, solution volume, wetting time, spin time, maximum revolutions per minute, and ramp rateon two response variablespercentage hexagonal close packing and macroscale coverage of nanospheres. Single-response and multiple-response linear regression models identify main and two-way interaction effects of statistical significance to the outcomes of both response variables and enable prediction of optimized settings. The results indicate a tradeoff between the high ramp rates required for large macroscale coverage and the need to minimize high shear forces and evaporation rates to ensure that nanospheres properly self-assemble into hexagonally packed arrays.