The ITS-90: A review of the evolution of the cryogenic ranges since 1990 and a view of its future after the new kelvin definition

ITS-90 reproducibility, xenon fixed point substitution and new interpolating equations between 13.8033 K and 273.16 K

“…Although deficiencies in the ITS-90 became apparent immediately after its introduction over thirty years ago [5,7,35,36,41,43,44], by now the scale is deeply ingrained in measurement infrastructure worldwide. As such, discussion of potential revision or replacement of the ITS-90 should consider the disruption that may be caused by such changes.…”

“…The resulting CU has a peak value of 0.278 mK: 0.165 mK lower than that of ITS-90 subrange 4, but still 0.073 mK worse than the peak CU when subrange 4 uses TPXe in place of TPHg. TPXe fixed points are likely to be realized using similar technology as TPAr fixed points [5,43], and it may be that some of the clients currently served by ITS-90 subrange 4 thermometer calibrations would instead opt for faster/cheaper calibrations that only required realization of two fixed points (TPXe and TPW), with lower overall scale realization uncertainties below 215 K than what they are getting now.…”

“…Indeed, long-stem SPRT calibration presents challenges for all of the leading alternatives to mercury: in the case of sulfur hexafluoride, suitable small-cell performance [23,25,27] did not easily translate to larger immersion-type cells [24,25], though these difficulties were eventually overcome [30]. It may be that techniques used to realize the triple point of argon are more applicable to the triple point of xenon than are those used to realize the triple point of mercury [5,43].…”

“…The improved versions of subranges 1-4 described above are much more reproducible than their ITS-90 counterparts above the triple point of oxygen, but gains in the problematic regions of subranges 1 and 2 at lower temperatures are more modest. Research avenues that might ameliorate this situation could be explored, such as: (a) Shifting the FP17 and FP20 points of subrange 1 to 15.25 K and 18.7 K (noting that 18.7 K also corresponds to the triple point of equilibrium deuterium [5]), a change that was suggested almost immediately after the creation of the ITS-90 [35,36]; (b) Considering other fixed-point swaps [43]; (c) Shifting all defining fixed points to ideal Chebyshev node positions using pre-interpolation transformations [5]; (d) Constructing new physics-based interpolating equations that better model the underlying physical phenomena driving the divergence of SPRT characteristics at low temperatures [5,35,36] than do the empirical deviation functions of the ITS-90 or of section 6; and/or, (e) Employing a hybrid approach in which physical understanding of the low-temperature behaviour of platinum wire informs the choice of additional functional forms to be included in empirical brute force searches for new deviation functions.…”

“…The recent redefinition of the kelvin has increased focus on primary thermometry [45], and several techniques now exist that can realize thermodynamic temperature between 13.8033 K and 24.5561 K with uncertainties of 0.5 mK or less [46][47][48][49][50], competitive with the defined scale CU values shown for this temperature range in figure 20. Given the difficulty of realizing the FP17 and FP20 points, the select group of NMIs capable of realizing ITS-90 subrange 1 may, in time, choose to directly disseminate thermodynamic temperature below the triple point of neon instead [5,43].…”