Temperature compensated niobate microwave ceramics with the columbite structure, M²⁺Nb₂O₆

Abstract

Of the niobate ceramics with the formula M²⁺Nb₂O₆, several compounds (M²⁺=Zn, Mg, Ca and Co) have Qf values (at f=1–10 GHz) between 40 000 and 90 000 GHz, offering potential in dielectric resonator applications. However, their temperature coefficient of resonant frequency (τ_f) values are too high for commercial development, at between −50 and −90 ppm. This paper details the doping of these materials with dielectric ceramics having a large positive τ_f, in an attempt to reduce the overall τ_f to zero, whilst maintaining a high quality factor (Q). It was also found that doping increased the relative permitivity (ε_r) of the niobates. Several materials have been made with near-zero τ_f, such as 90% CoNb₂O₆/10% CaTiO₃ (τ_f=+2.0 ppm, ε_r=25.2 and Qf=21 700 GHz), and 94% CoNb₂O₆/6% TiO₂ (τ_f=+4.4 ppm, ε_r=29.6 and Qf=20 300 GHz)

Introduction

With the continuing proliferation of wireless communications technologies operating at microwave frequencies, there is an ever-increasing demand for cheap, but nonetheless high performance, dielectric ceramics. To be effective dielectric resonator materials they should have a sufficiently high relative permittivity to allow miniaturisation of the component (ε_r > 10), low dielectric losses at microwave frequencies to improve selectivity (Q > 5000, where Q=1/tan δ), and a temperature coefficient of resonant frequency near zero for temperature stability (τ_f <±20 ppm per °C).¹

Microwave ceramics are currently available with low τ_f and Qf > 100 000 (Qf=Q × f_r). However, these are usually made from complex perovskites, such as the mixed metal tantalate perovskites BaZn_0.33Ta_0.67O₃ (BZT) and BaMg_0.33Ta_0.67O₃ (BMT).2, 3, 4 These complex perovskites require fairly high sintering temperatures (> 1400 °C), and the structures and properties of the complex perovskites (often with four or more cations) are proving difficult to predict, and depend strongly upon the degree of ordering.⁵ Furthermore, tantalum is a relatively expensive metal, the ore tantalite (60% Ta₂O₅) costing $150 per kg,⁶ whereas niobium is over 20 times cheaper, with the mineral columbite costing just $8 per kg.⁶

The binary niobate ceramics, with the formula MNb₂O₆ where M is a divalent cation, are one of the end members of the perovskite BaM_0.33Nb_0.67O₃ group (the other being BaO), and they are mostly isostructural with the orthorhombic mineral columbite (ZnNb₂O₆, space group=Pnca (60)).⁷ The transition metal columbite niobates sinter between 1100 and 1200 °C, much lower than the perovskites,7, 8, 9, 10 and Q of the columbite niobates is higher than that of the M²⁺Ta₂O₆ compounds, which do not have the columbite structure.⁸ As niobium is so much cheaper than tantalum, and because the chemistry of the binary compounds should be easier to investigate than that of the complex perovskites, a study was made of these binary niobates, and these results have been previously reported.11, 12 Amongst the niobates investigated, four in particular exhibited potential for commercial application: these were ZnNb₂O₆ (ZnNO), MgNb₂O₆ (MgNO), CaNb₂O₆ (CaNO) and CoNb₂O₆ (CoNO). The properties of these materials are shown in Table 1, and it can be seen that these materials have good quality factors, especially considering the low cost and simplicity of the materials compared with the complex perovskites. The τ_f values, whilst being fairly small compared with materials like titania or calcium titanate, were still too large for many resonator applications. Therefore, the effects of using dopants in an attempt to lower the τ_f of these niobate compounds were investigated. Because all of the niobates have negative τ_f values, dopants with large positive τ_f values and good microwave properties were added. The dopants used were commercially available forms of doped TiO₂ (rutile), which has τ_f=+420 ppm K⁻¹ and Qf=∼48 000 GHz,13, 14 and CaTiO₃ which has a τ_f=+800 ppm K⁻¹ and Qf=∼3600 GHz.¹⁵