Let's say you've got a xenon atom. It likes to absorb energy and emit photons. You know, xenon lamps etc.
OK, so lets ask a real simple question: When a tube filled with xenon gas has some energy pumped into it and the electrons go to higher orbitals -- yes this happens for a very short period of time before photons are emitted but let's talk about just the short period of time. The diameter of the atoms presumably increases. Does the gas pressure increase during that interval? Now lets say that the energy is sufficient to actually strip the electrons away and form an ionized gas for a short interval. Does the ionized gas pressure increase during that interval? Now lets talk about really-simple magnetic confinement (say a magnetic mirror <http://en.wikipedia.org/wiki/Magnetic_mirror> type bottle) used in conjunction with a solid tube so that the non-conducting (because non-ionized) gas phase is confined by the solid tube and the conducting (because) ionized gas phase is confined by the magnetic bottle: When the electrons fall back into their ground states we can comfortably assert that the photons emitted will equal the energy input. However, what if the plasma has expanded during the high pressure phase, ie: done work against the magnetic confinement (like, oh, I don't know, generating an electrical power spike in a conductor associated with the magnetic field). Does that mean the "free" electrons of the plasma no longer want to return to their ground states and give up exactly the same amount of energy that they would have in the absence of having done work? If not, where did the electrons go and where do the xenon atoms get electrons to substitute for them?
