Although previous answers are correct in connection with atomic, and classical systems, I would like to give a more general and complementary point of view, which may help in improving the conceptual understanding of what is behind the concept of temperature.
The really misleading concept one has to remove in order to understand why temperature does not change during a first order phase transition is the proportionality between internal energy and temperature, which is true for ideal gases but does not have universal validity, even far from phase transitions.
For classical systems which have kinetic energy, it is possible to show (as a theorem, it is the content of the so-called equipartition theorem) that the average kinetic energy is proportional to the temperature of the system. Therefore an explanation of the constancy of the temperature at a phase transition in terms of an increase of potential energy while kinetic energy remains the same is possible (see Bob D's and Kraig's answers).
However, there are two limitations of such a point of view:
- it holds only for classical systems, because, as soon quantum effects
enter into play, equipartition theorem is not valid anymore and there
is no proportionality between kinetic energy and temperature, thus
phase transitions at low temperatures would become ununderstandable.
- Even for classical systems, there are thermodynamic systems where the
kinetic energy does not play a direct role in the thermodynamic
description. A simple example is the case of magnetic phase
transitions.
Moreover, the explanation in terms of interplay between kinetic and potential energy does not really justify why temperature remains exactly constant and it does not happen that there is simply a different share between increase of kinetic and potential energy.
A more general point of view, which sheds some additional light on the concept of temperature is the following.
From a purely thermodynamic point of view, the temperature can be defined as the inverse of the slope of the entropy surface $S$ as a function of the internal energy $E$:
$$
T = \frac{1}{ \frac{\partial S}{\partial E} }
$$
where all variables of $S$ different from energy have to be kept constant when taking the partial derivative.
On the other hand, the thermodynamic definition is consistent with the Boltzmann's statistical mechanic definition of entropy as proportional to the logarithm of the number of microscopic states available to the system at given energy:
$$
S(E) = k_B \log \Omega(E)
$$
These two ingredients help to understand, on a completely general base, why temperature remains constant at a first-order phase transition.
Such a transition is characterized by the physical phenomenon of phase coexistence. I.e., there is an interval of energies (say $[E_1,E_2]$)where the system cannot stays anymore in a homogenous one-phase state, but becomes inhomogeneous, with coexistence of macroscopic regions made of one phase at the lower energy $E_1$, and others mode of the second phase, at the upper energy $E_2$.
As a consequence of this fact, entropy in the whole interval $[E_1,E_2]$ is simply a linear combination of the boundary entropies:
$$
S(E)=S(E_1)+ \frac{E-E_1}{E_2-E_1} \left( S(E_2)-S(E_1) \right)
$$
Therefore, it is an immediate consequence of the thermostatistical definition of temperature as derivative of the entropy, to obtain the constance of $T$ all along the phase coexistence interval. The physical intuition one can build on top of this description, just adding the extensiveness of the entropy, is that temperature remains constant because of the linear increase with the energy of the spatial region, in the coexisting system, filled by the phase at higher energy.
All that is not in contradiction, but as a complement to the explanation in terms of kinetic energy, valid for the classical systems where kinetic energy plays a role.
Best Answer
Correct. You're converting energy in one form to energy in another form. A simpler example would be burning wood inside the system.
There's no violation. Energy is converted from one form (chemical or nuclear potential energy) to another form (thermal energy).