Clausius Implies That Nearly Anything Can Be A Thermometer

Abstract: There are three types of thermometries. One is a proxy, such as the purely phenomenological resistivity. More fundamental are those based on thermodynamics, as in the Carnot cycle, and those based on statistical mechanics, such as the ideal gas law. With heat flow $Q$ and temperature $T$, a temperature scale in principle (but not in practice) can be based on the simple Carnot cycle relation $Q/T+Q'/T'=0$, with a temperature $T_{0}(p_{0},V_{0})$ specified. More generally, a thermodynamics based temperature scale may be determined by the Clausius condition $\oint dQ/T=0$ for every closed path in a given region $\Omega$ of $p$-$V$ space. Taking a discretized grid $i$ (from which such closed paths can be composed), for some parametrized model temperature function $T_{n}$ a root-mean-square minimization of $\sum_{i}(\oint_{i}dQ/T_{n})^{2}$ yields the best set of model $T_{n}$'s parameters. Thus any stable material -- even one not described by a known statistical mechanical model -- can be used as a thermometer. If, because of inaccuracy of $dQ$ measurement, the Clausius condition method gives a temperature scale of lower accuracy than the best proxy temperature scale, then that proxy temperature scale can be employed with the rms Clausius condition method to improve the accuracy of (i.e., raise the standards for) the $dQ$ measurements to the accuracy of the proxy-based temperature scale.