The kelvin redefined and its implications

Professor Graham Machin (National Physical Laboratory)

In May 2019 the International System of Units (the SI) underwent what was its biggest change since its introduction when the definition of four of the seven SI base units were changed to be based on defined values of fundamental physical constants. Since the change, the kelvin is now defined in terms of the Boltzmann constant, the ampere on the electron charge, the kilogram on the Planck constant and the mole on the Avogadro constant.

The redefinition of the kelvin has opened several new possibilities for traceable thermometry direct to the kelvin definition. These could include using primary thermometry to calibrate sensors at National Measurement Institutes (NMIs) and, in the medium term, in calibration laboratories dispensing with traceability to the defined scales (ITS-90, PLTS-2000) and so disseminating thermodynamic temperature. In the longer term these changes could lead to the rise in potential, paradigm changing approaches to temperature sensing such as traceability at the point of measurements both through self-validating thermometers and more radically by the deployment of practical primary thermometry based on fundamental physics and where temperature sensor itself will, unlike today, no longer need calibrating to provide traceability.

In this talk an introduction to how the kelvin was redefined and to the mise en pratique for the definition of the kelvin (MeP-K) will be given. How traceable temperatures are attained will be discussed, both presently, through the defined scales and how, in the medium and long term, this is likely to change; with thermodynamic temperature approaches becoming increasingly prevalent. The talk will end by introducing novel approaches to temperature traceability including provision of NMI like uncertainty thermodynamic temperatures in calibration laboratories and the rise of in-situ/in-process traceability and the implications, particularly in the context of digitalisation and the need for "point-of-truth" in for example autonomous sensor networks.