The Future of Thermal Analysis

In the past, nature was either studied by experiment and a link to a simple, all-encompassing theory was sought. In the last three decades, however, methods evolved to simulate atomic motion with computers, as described in Sects. 1.3.6-8, and to predict properties by neural network techniques, as summarized in Appendix 4. Both techniques are neither theory nor experiment. The simulations let us see the changes in the microscopic sttucture in slow motion and give a base for the understanding of physical problems which at present are too comphcated to fit into a simple theory. The neural network analysis, in tnm, looks for a method to use implicit functional relationships between the properties of different substances which are too difficult to ascertain explicidy [3]. The method of Appendix 6, produces a computer program which then correlates the properties. 

Heat Capacity as Extracted from Molecular Dynamics Simulations 

Further, the number of double defects is listed in Fig. 2.15. Such double defects lead in Sect. 5.3 to an understanding of defect crystals and their deformation kinetics. The number of defects refers to the whole crystal. Assuming that the increase in heat capacity from 3 to 4.5 R in Fig. 2.14 is due to defect formation, it is possible to calculate the energy of defect formation as 44 kl mol 

These first thermal analysis simulations open the view how future macroscopic experiments may be directly connected to molecular structure and motion. Much progress is naturally needed to establish proper simulation parameters and to expand the low-temperature analyses to quantum-mechanical calculations. 

With neural networks, as described in Appendix 4, the prediction of properties of unmeasured substances on the basis of data of known substances became possible (see introduction to Sect. 3.4). For thermal analysis, it has been possible to extend measured heat capacities at high temperatures into the region of low temperature where measurement is more difficult, as well as predict the theta-temperatures needed for the description of the vibrational heat capacities [4,5]. 

---