Skip to Content

The SI unit of temperature, the kelvin, is defined as 1/273.16 of the temperature of the triple point of water and, in principle, this defines the kelvin exactly.

The triple point of water is the temperature at which water in solid, liquid and vapour forms all coexist in thermal equilibrium. However, there are a number of small effects giving rise to uncertainties as large as several hundred microkelvin. MSL’s research into several of these effects has led to changes in the way cells are made and a significant improvement of the worldwide reproducibility of cells. Prior to 1996, cells were manufactured with maximum differences between high-quality cells of about 250 μK. Although not a major contribution to the uncertainties in temperature measurements, this level of uncertainty in the triple-point temperature is a nuisance when carrying out research into improvements to the temperature scale. Now, partly as a result of MSL’s research, cells made from fused-silica and with known isotopic composition are used to realise the triple point with a range of less than 50 μK.

Isotopic composition

In 2006, the Comité International de Poids et Mesures (CIPM) clarified the definition of the kelvin by specifying the isotopic composition of the water to be used in triple-point-of-water cells. At the same time, the CIPM’s Consultative Committee on Thermometry recommended values for the constants used to calculate the corrections for cells for which the isotopic composition of the water was not exactly as specified. Both of these changes were triggered by research carried out at MSL.

In 1996, while preparing our submission for an international comparison of water triple-point cells Dr John Nicholas at MSL became concerned about the magnitude of the possible influence of the isotopic composition of the water in the cells. Although the existence of the isotope effect was well known, it had been thought that the effect would be less than 40 μK. However, cells are usually made from fresh water, and John found that the natural variations in isotopic composition due to fractionation that occurs naturally with evaporation and condensation leads to fresh waters with a very wide range of compositions. Such data is readily available because geologists and hydrologists use isotopic composition as a signature for bodies of water so they can track subterranean flows. In New Zealand, the natural composition of the water leads to triple-point cells more than 60 μK below the definition. Further, distillation and degassing of the water during the manufacture of the triple-point cells can cause still more changes leading to cells as much as 90 μK below the definition.

To investigate the magnitude of the effect and determine the values of the correction constants, we manufactured a series of cells with differing 2H and 18O concentrations. The cells were also circulated to the National Institute of Standards and Technology (NIST, USA) and the National Physical Laboratory (NPL, UK). As expected, the measurements established that the effect of the wide variation in the composition of fresh waters was two to three times larger than anticipated.

The clarification of the definition of the kelvin by the CIPM, combined with the use of the isotopic correction constants, has reduced the uncertainty due to isotopic effects to a few microkelvin. Because of the limited resolution in measurements of the triple-point temperature, the uncertainties in the values of the isotopic correction constants are quite large, and work continues at MSL to determine more accurate values for the constants.


Glass dissolution

Around the same time that we started to investigate the isotopic effect in the water triple-point cells, research at NRC in Canada suggested that the temperature realised by triple-point cells drifted with time as the water in the cell steadily dissolved the glass of the cell. While the drift was not large (typically less than 20 μK per year) over the many years that cells were used, the drift could easily exceed 200 μK. A few years earlier, NMIA in Australia published a non-destructive method for measuring the conductivity of water in triple-point cells, with the purpose of verifying the purity of the water. We speculated that it might be possible use the NMIA method to measure the concentration of the components of glass dissolved in the water, and apply a correction.

While the correction method did not prove to be particularly satisfactory, the increased understanding of the dissolution processes that we gained from the study suggested that the cells should be made from pure silica, which is much less soluble than the borosilicate glass (pyrex) typically used for triple-point cells. Borosilicate was used because it is easy to work, and some basic treatments of the glass significantly reduces the amount of leaching and dissolution that takes place. However, as time passes, our needs for improved accuracy increase and older technologies must be replaced. Both of the World’s major manufacturers of triple-point cells now produce fused-silica cells (with a report on the isotopic composition). Recent work at NIST (USA) shows that the drift of cells made from pure silica is at least 20 times less than that of borosilicate cells.


Other smaller effects

The temperature of the triple point of water is almost exactly 10 mK higher than the normal freezing point of water (the ice point). About 7.3 mK of this difference is due to the influence of the atmospheric pressure on the melting point. The remaining 2.7 mK is an impurity effect due to air dissolved in the water. If, during the manufacture of a cell, the water is not completely degassed, the measured triple point temperature will be a little lower than expected. So it is important to be able to check the amount of air in a cell. Traditionally the test was to listen for the loud click as water sloshed backwards and forwards in the cell. A loud sharp click indicated that there was no air in the cell to cushion the impact of the water with the glass. To provide an improved measure, one well-known cell manufacturer built a tubular extension on their cells to allow the trapping of a bubble, from which we could infer a value for temperature difference due to the residual air. MSL’s contribution has been to expand the method so that it is suitable for cells without the tubular extension.

The combination of the pressure effect on melting ice and the buoyancy of ice leads to a very small effect that gives rise to anomalies in some triple-point measurements. Ice floating against the bottom of the thermometer well in the cell induces an increased pressure on the melting ice and a very small reduction in the temperature. MSL has provided a simple model of the effect and explained the anomalies.

More information

For further information on the water triple point, see MSL Technical Guide 3, or contact Rod White

Key publications

J V Nicholas, T D Dransfield, D R White, “Isotopic composition of water used for triple point cells, Metrologia, 33, 265-267, 1996.

J V Nicholas, D R White, T D Dransfield, “Isotope influences on the triple point of water and the definition of the kelvin”. Proc. TEMPMEKO ’96, Ed. P Marcarino, Levrotta & Bello, Torino, pp 9–12, 1997.

D R White, T D Dransfield, G F Strouse, W L Tew, R L Rusby, J Gray, “Effects of heavy hydrogen and oxygen on the triple-point temperature of water” in Temperature: its Measurement and Control in Science and Industry, Vol 7, Ed. D C Ripple, AIP, pp 221–226, 2003.

D R White, “Measuring the residual air pressure in triple-point-of-water cells”, Meas. Sci. Technol., 15, N15-N16, 2004.

D R White, T D Dransfield, “Buoyancy effects on the temperature realised by triple-point-of-water cells”, Proc. TEMPMEKO 2004, Eds Zvizdic, Bermanec, Stasic, Veliki, Faculty of Mechanical Engineering and Naval Architecture, Zagreb, pp 313–318, 2005.

D R White, C J Downes, T D Dransfield, R S Mason, “Dissolved glass in triple-point-of-water cells”, Proc. TEMPMEKO 2004, Eds Zvizdic, Bermanec, Stasic, Veliki, Faculty of Mmechanical Engineering and Naval Architecture, Zagreb, pp 251–256, 2005.

W L Tew, D R White, “Comment on freezing point mixtures of H216O with H217O and those of aqueous CD3CH2OH and CH313CH2OH solutions”, J. Soln. Chem., 34, 1191–1196, 2005.