Abstract
Feed-through connectors are used in the electronics industry to route electrical current into isolated environments. Electrically insulating glasses are used to separate metals from each other to form leak tight glass-to-metal seals. The coefficients of thermal expansion (CTE) are controlled to reduce thermomechanical stresses in the seal. The majority of glass-to-metal seals consist of stainless steel housings and pins bonded to silicate-based glasses. Titanium and its alloys are alternative housing materials due to their low density and thus weight saving potential. Silicate-based glasses form weak and brittle interlayers when bonded to titanium, so borate-based glasses are used instead. Glasses are replaced with glass-ceramics due to improved toughness and refractory performance whilst allowing the CTE to be controlled. A knowledge gap is present as few heat treatments are published to manufacture robust glass-ceramic-to-metal seals with a Ti-6Al-4V housing. Hermetic glass-ceramic-to-metal seals were required to incorporate a Ti-6Al-4V housing and an electrically conducting pin, to survive for up to 30 years at standard laboratory conditions (20 ± 10 °C / 30 ± 20 %RH). The manufacture and testing of glass-ceramic-to-metal seals was divided into three parts: seal development, characterisation, and ageing. Hermetic seals were made between Ti-6Al-4V, a strontium boroaluminate glass-ceramic and an oxidised Kovar pin with a helium leak rate less than 1 × 10-9 mbar l s-1. Kovar was selected as the pin material predominantly due to the low CTE. Differential thermal analysis and push-rod dilatometry were used to tailor the CTE of the glass-ceramic (8.4 ± 0.1 × 10-6 °C-1) between the Ti-6Al-4V housing (10.1 ± 0.1 × 10-6 °C-1) and the oxidised Kovar pin (7.4 ± 0.1 × 10-6 °C-1). The CTEs were measured between 30 °C and 600 °C. The Kovar pin was oxidised in air at 700 °C for 10 minutes. The seal was heat treated in an argon atmosphere that peaked with an isothermal hold at 800 °C for 20 minutes, specifically to tailor the CTE of the glass-ceramic. Seals were reproducible in terms of hermeticity, microstructure and mechanical strength. Reproducible leak tight seals were manufactured in batches of 5 seals. A reproducible microstructure of the seal was confirmed with scanning electron microscopy, energy dispersive X-ray spectroscopy and X-ray diffraction. The seals had a load at failure of 1510 ± 10 N. Seals were aged between 20 – 85 °C, 35 – 93 %RH, for up to 32 weeks. There was no change in the hermeticity of the seals as a function of ageing time or temperature. The surface of the glass-ceramic aged at 85 °C / 93 %RH was degraded by contact with water vapour at a rate of 5 ± 1 μm yr-1. The resulting strontium enriched layer was mechanically weak. There was a 1 % change in the load at failure of the seals, aged as a function of time and temperature, which was accounted for by natural variation within the samples. Thus, the seals developed in this study are expected to remain hermetic following tens of years of ageing at standard laboratory conditions.