Skip to content

Numerical Simulation and Mechanism of Line Uniformity for Aerosol Jet-Printed Diamond Coatings

FieldValue
JournalCoatings
Published2026-08-10
AuthorsHao Chang, Qingyu Yao, Xiaofei Xie, Mohammad Uddin
DOI10.3390/coatings16080948

The transient CFD model predicted a 141.20 µm line spacing for maximum spreading at 65 mL/min. Printing tests found the best coalescence at 140 µm. The paper reports a 1.20 µm difference between the two values. This agreement supports CFD as a tool for setting line spacing on aerosol jet-printed diamond coatings.

The result applies to the paper’s validation setup. It used a 65 mL/min main airflow, a 1:2 main-to-sheath ratio, and a 2 mm print-head standoff. The substrate was tungsten carbide. It is not a universal gap for other inks, nozzles, or machine settings. The authors target high-aspect-ratio micro end mills, but the experiment used a flat tungsten carbide coupon rather than a finished cutter.

The aerosol generator atomizes the ink into micron-sized droplets. A carrier gas carries them through a micro-nozzle and along a fixed trajectory. The model treats the droplet stream as one aggregated cluster because uniform atomization was assumed to remove meaningful droplet-to-droplet variation. Two droplets represent adjacent lines. The first reaches equilibrium before the second arrives.

The VOF method tracks the transient liquid-gas interface. The DPM tracks diamond particles under drag, gravity, pressure gradient, and virtual mass. Their coupling resolves droplet impact, spreading, and particle transport. Surface tension, inertia, adhesion, and viscosity govern the calculated motion.

ANSYS Fluent 2023 ran the transient calculations. A 1 µm hexahedral mesh contained 2,499,860 elements. Refining it to 0.5 µm changed the key outputs by less than 3%, so the authors selected the coarser mesh. The time step was 0.02 µs. Low-speed calculations used a cylindrical domain 55 µm high and 240 µm in diameter. High-speed calculations used a 110 µm high and 480 µm diameter domain.

The model treats the substrate as ideally smooth and neglects substrate temperature. It also ignores the droplet velocity change over the working distance. These assumptions limit transfer to rough tools, heated substrates, and long deposition distances.

The modeled ink combined deionized water, 95% ethanol, and 60,000 mesh diamond particles. The diamond particles had an average diameter of 0.25 µm. The model used an ink density of 1140 kg/m³ and viscosity of 4.7 × 10⁻³ kg/(m·s). It used a 61.2° contact angle and a 44 mN/m surface tension coefficient. Contact angle was measured with an SR-180S tester, and viscosity with an LDWND instrument.

The experimental formulation in Table 4 also included SiO2, sodium dodecyl sulfate, polyethylene glycol 400, acetone, and adhesive. It contained 1 mL of polycrystalline diamond. The 20 µm simulation droplet was estimated to contain 3300 diamond particles.

The low-speed case used a 10 mL/min main airflow, a 100 µm nozzle, and a 21.22 m/s calculated droplet speed. A 20 µm droplet began 30 µm above the substrate, and adjacent droplets were 50 µm apart.

Impact separated into three stages. Initial contact occurred from 0 to 1.0 µs. Rapid spreading ran from 1.0 to 3.0 µs. Contraction and stabilization continued to 100.0 µs. Inertia drove the early spread, while surface tension and viscosity controlled contraction.

The diameter increased from about 21.0 µm to 51.2 µm during the first 2.5 µs. That 158.1% increase marked maximum spreading. The diameter then converged to 33.1 µm by 100 µs. Thickness fell rapidly during impact, held near 4.4 µm from 2.5 to 30 µs, and then rebounded to 11.8 µm by 70 µs. It remained at 11.8 µm through 100 µs.

The authors calculated a 42.15 µm critical spacing from a 25.6 µm maximum spreading radius and a 16.55 µm equilibrium radius. At 42.15 µm, the droplet edges met and formed a transient bridge. The bridge later receded, leaving separate droplets and poor line uniformity.

At 40 µm, the bridge expanded until the droplets became one continuous liquid film. At 36 µm, they aggregated immediately, reached equilibrium sooner, and showed higher coalescence efficiency. The paper therefore concludes that spacing below 42.15 µm enables effective fusion and continuous lines without gaps.

Particle tracking showed why the merged deposit need not have uniform particle distribution. The later droplet spread into the earlier one and formed a crescent interface. At 36 µm, radial flow reached 12 m/s at 2 µs. By 10 µs, the maximum velocity had fallen to 0.5 m/s, and contact-interface flow had reversed.

The high-speed model used a 65 mL/min main airflow and a calculated droplet speed of 137.06 m/s. Its fixed line gap was 150 µm. The droplets stabilized by 20.0 µs, compared with 100.0 µs in the low-speed case. They spread into dispersed ring-shaped patterns, with less uniform films but larger coverage areas.

At 20.0 µs, the two calculated spreading diameters were 138.91 µm and 143.49 µm. The model identified 141.20 µm as the spacing for maximum spreading efficiency.

The validation used a 10 mm × 10 mm × 0.3 mm tungsten carbide substrate. The print head was 2 mm above it. Main airflow was 65 mL/min, and the main-to-sheath ratio was 1:2. Tested gaps were 130, 140, 150, 180, 200, 220, and 250 µm. Low-power microscopy and FEI Quanta 250 FEG scanning electron microscopy assessed the lines.

At 150 µm, the lines remained separated. At 130 µm, local overlap appeared. The 140 µm condition gave coalescence without significant overlap. The paper reports a 1.20 µm difference from the 141.20 µm prediction. Its conclusion gives the high-speed relative error as +0.85%.

Accuracy, prior work, and engineering limits

Section titled “Accuracy, prior work, and engineering limits”

At low speed, a spherical-cap calculation gave a 31.849 µm equilibrium diameter. The simulation gave 33.1 µm, a 1.251 µm difference and a +3.9% relative error. The authors place both speed cases within an absolute simulation error of ±10%. This makes the model reliable for the tested conditions, but not a general validation across inks or machines.

Earlier work used related methods for different questions. Jin et al. used EWF-VOF to convert liquid film into the fluid phase. Their DPM-VOF structure converted impacting particles into fluid-phase volume. Liu et al. coupled CFD with DPM, Taylor analogy breakup, and Saffman lift, then checked line widths. Elkaseer et al. modeled droplet volume, contact angle, and fusion in drop-on-demand printing. Shah et al. used VOF to examine impact, spreading, and rebound. The present model adds sequential adjacent droplets and line-spacing analysis under low and high jet speeds.

Several limits remain. The low-speed behavior was simulated because the authors could not directly capture the complete process at that speed. The high-speed check used a flat coupon, not the complex cutting edges and chip-clearance grooves of a micro end mill. The study also stops before solvent removal, binder burnout, and sintering. It does not report coating microstructure, adhesion, wear resistance, or cutting performance. No alternative diamond grades were compared, so the study cannot rank diamond material for the final coating.

For the tested 65 mL/min configuration, 140 µm is the experimental reference spacing. For the low-speed model, spacing below 42.15 µm defines the predicted fusion regime. Applying either result to a real cutter requires checking roughness, temperature, ink composition, and geometry, because the model excludes or simplifies those effects.

View Original Abstract

The large aspect ratio micro end mill is a critical tool for microstructure machining, and its performance directly determines processing quality and efficiency. Diamond coatings are commonly applied to cutting edges to enhance wear resistance and extend tool life. However, existing coating techniques often suffer from poor uniformity and inadequate consistency, limiting batch production and process stability. Aerosol jet printing (AJP) offers a cost-effective and highly controllable alternative for the efficient, large-scale deposition of diamond coatings on micro end mills, where precise control of line spacing is essential to achieving coating uniformity. In this study, a transient numerical model of droplet deposition in AJP is developed using computational fluid dynamics (CFD). The volume of fluid (VOF) method and the discrete phase model (DPM) are coupled to track liquid-gas interface deformation and diamond particle motion, enabling the dynamic evolution of droplet deposition to be captured. The effects of inter-droplet distance on deposition, spreading, coalescence, and line uniformity are systematically investigated. Droplet deposition mechanisms are analyzed under low-speed jetting conditions, while high-speed jetting simulations are conducted to reflect industrial processing scenarios. The results show that under low-speed jetting, droplets undergo spreading, contraction, and rebound, eventually forming a uniform cap-like structure. Under high-speed jetting, droplets exhibit a dispersed ring-shaped spreading pattern; although uniformity is slightly reduced, the spreading area and deposition efficiency are significantly increased. These findings provide a theoretical basis for optimizing AJP process parameters to achieve high-quality diamond coatings on micro end mills.