Choosing a liquid calibration bath and its fluid

• Ethanol (−80 to +10 °C): low viscosity cold, flammable, hygroscopic; absorbs water over time which raises the freezing point.
• Water (5 to 95 °C): cheap, non-toxic, excellent uniformity; evaporation and biological growth are the maintenance issues.
• Silicone oil (−40 to +250 °C depending on grade): low vapour pressure; viscosity rises sharply at the cold end, degrading uniformity.
• Molten salt (200 to 550 °C): high uniformity, hazardous, corrosive to some sheaths.
• Fluidised alumina baths: 50 to 700 °C, dry, poorer uniformity than liquids.

Whatever the fluid, characterise uniformity and stability with the fluid, stirrer speed and depth you will use, and record it. A bath's brochure numbers apply to the manufacturer's test conditions.

Viscosity is the hidden variable

A bath's uniformity comes from circulation, and circulation comes from a stirrer moving a fluid. As a fluid cools it thickens, the stirrer moves less of it, and the gradient grows — so a silicone oil bath that is superb at 150 °C can be markedly worse at −30 °C, in the same bath with the same stirrer. This is why one fluid across the whole range is a convenience rather than a technical choice, and why ethanol remains the fluid of choice for genuinely cold work despite being flammable.

Depth, and the reason people under-buy

Depth sets the immersion available, and immersion is usually the largest avoidable error in the whole calibration. A bath deep enough for a 6 mm probe at 110 mm of immersion is not deep enough for a 10 mm reference at 180 mm, and the difference is a few hundred millimetres of tank that costs real money at purchase and nothing afterwards. Buy the depth for the longest probe you intend to calibrate, not the one on the bench today.

Characterise it as you use it

• Measure uniformity and stability with the fluid, the stirrer speed, the fill depth and the loading you will actually use. Brochure figures apply to the manufacturer's test conditions and typically to an empty bath.
• Repeat the characterisation per temperature range. A bath is not equally good at both ends of its span.
• Record the results per bath, per fluid, per range, and put them in the budget as the gradient and stability terms rather than as a single bath number.
• Re-characterise after a fluid change, a stirrer service or a significant change in how the bath is loaded.
• Watch the fluid age. Hygroscopic ethanol, oxidised oil and contaminated water all change the numbers you characterised, silently.