The classical mechanics (and it's relativistic enlargements) describes macroscopic objects. This objects consist of a very large number of particles, normally atoms. The properties of the mechanical objects can be exactly defined (mass, speed, position in space) and reproduce.
Differently the particles. It is virtually impossible to measure the properties of a single particle exactly. To describe particles in thermodynamics we already operate with probability. All particles have different speed and positions, this means different energies (and also relativistic masses). Furthermore particles can change their energies continuous. You can say that all particles have variable internal energy. Scientists like Pauli and Heisenberg apply the mechanics to the particles. How reliable can be mechanics in this case?