You can if you wish model the fields in a transformer as photons in a magnetic medium but their wave impedance will be much lower than that of free space due to the mu of the transformer. The coherence length of a 50Hz photon is massive, thousands of km, and the energy of each one is impractically low, so rather like modelling gas pressure by trying to count individual atomic collisions, the particulate view is adding a lot of complication for almost no additional insight, as you will be dealing with very small phase changes across the device, and the classical psuedo-static field approach will be much easier.

I suggest if you are feeling confident, you plug in some real values and calculate particle densities and their probability density funtions for a bell transformer with say a 1 inch cube of soft iron core, taking mains in and driving a 6V 1 amp load and see how much the quantum mech view does not allow you to calculate anything you could not have predicted with several sides less paper and lower risk of algebraic error using the classical approximations. I guarantee the differences you calculate in the output currents will be several tens of orders of magnitude below the threshold of detection.  You will need to set aside a good few hours.

However, if you repeat the exercise for a thin layer of semiconductor a few micorns accross amplifying a signal at 10GHz, then the effecets start to have a modest effect - hence HEMTs and so on work the way they do and can be modelled for (with a lot of effort, and we normally bother just for the active region) as a 2 dimensional  quantum gas (or a very nearly 2D, ) . If the QM theory was wrong, the satellite TV receivers would not work as predicted.

You can if you wish model the fields in a transformer as photons in a magnetic medium but their wave impedance will be much lower than that of free space due to the mu of the transformer. The coherence length of a 50Hz photon is massive, thousands of km, and the energy of each one is impractically low, so rather like modelling gas pressure by trying to count individual atomic collisions, the particulate view is adding a lot of complication for almost no additional insight, as you will be dealing with very small phase changes across the device, and the classical psuedo-static field approach will be much easier.

I suggest if you are feeling confident, you plug in some real values and calculate particle densities and their probability density funtions for a bell transformer with say a 1 inch cube of soft iron core, taking mains in and driving a 6V 1 amp load and see how much the quantum mech view does not allow you to calculate anything you could not have predicted with several sides less paper and lower risk of algebraic error using the classical approximations. I guarantee the differences you calculate in the output currents will be several tens of orders of magnitude below the threshold of detection.  You will need to set aside a good few hours.

However, if you repeat the exercise for a thin layer of semiconductor a few micorns accross amplifying a signal at 10GHz, then the effecets start to have a modest effect - hence HEMTs and so on work the way they do and can be modelled for (with a lot of effort, and we normally bother just for the active region) as a 2 dimensional  quantum gas (or a very nearly 2D, ) . If the QM theory was wrong, the satellite TV receivers would not work as predicted.