Influence of Different Surface Morphologies on the Performance of High Voltage, Low Resistance Diamond Schottky Diodes

TL;DR

This study investigates how different surface morphologies of diamond influence the performance of high-voltage, low-resistance Schottky diodes, revealing that surface quality impacts breakdown voltage, current density, and device efficiency.

Contribution

It provides a comparative analysis of surface morphologies on diamond Schottky diodes, highlighting the effects of surface smoothness, hillocks, and polishing on device performance and reliability.

Findings

Smooth surfaces exhibit low reverse current density up to 2.2 kV.

Hillock-rich surfaces show increased reverse current and reduced breakdown field.

Polished surfaces perform similarly to smooth ones, indicating suitability for fabrication.

Abstract

Vertical diamond Schottky diodes with blocking voltages $V_{\text{BD}} > 2.4 \text{ kV}$ and on-resistances $R_{\text{On}} < 400 \text{ m}\Omega \text{cm}^{2}$ were fabricated on homoepitaxially grown diamond layers with different surface morphologies. The morphology (smooth as-grown, hillock-rich, polished) influences the Schottky barrier, the carrier transport properties, and consequently the device performance. The smooth as-grown sample exhibited a low reverse current density $J_{\text{Rev}} < 10^{-4} \text{ A}/\text{cm}^{2}$ for reverse voltages up to $2.2 \text{ kV}$. The hillock-rich sample blocked similar voltages with a slight increase in the reverse current density ($J_{\text{Rev}} < 10^{-3} \text{ A}/\text{cm}^{2}$). The calculated 1D-breakdown field, however, was reduced by $30 \text{ } \%$, indicating a field enhancement induced by the inhomogeneous surface. The polished sample demonstrated a similar breakdown voltage and reverse current density as the smooth as-grown sample, suggesting that a polished surface can be suitable for device fabrication. However, a statistical analysis of several diodes of each sample showed the importance of the substrate quality: A high density of defects both reduces the feasible device area and increases the reverse current density. In forward direction, the hillock-rich sample exhibited a secondary Schottky barrier, which could be fitted with a modified thermionic emission model employing the Lambert W-function. Both polished and smooth sample showed nearly ideal thermionic emission with ideality factors $1.08$ and $1.03$, respectively. Compared with literature, all three diodes exhibit an improved Baliga Figure of Merit for diamond Schottky diodes with $V_{\text{BD}} > 2 \text{ kV}$.