It is well known that the dielectric constant of two-dimensional (2D) electron system goes negative at low electron densities. A consequence of the negative dielectric constant could be the formation of the droplet state. The droplet state is a two-phase coexistence region of high density liquid and low density "gas". In this paper, we carry out energetic calculations to study the stability of the droplet ground state. The possible relevance of the droplet state to recently observed 2D metal-insulator transition is also discussed. PACS numbers: 71.30.+h, 73.40.Hm The recent discovery of two-dimensional (2D) metalinsulator transition (MIT) by Kravchenko et al. [1] has challenged the scaling theory of localization [2,3] in which a 2D MIT is forbidden. A noticeable character of the electron system in these experiments is that r s , the parameter measures the strength of the Coulomb interaction, is fairly large. We suspect that the electron system may be unstable against phase separation at these large values of r s . We demonstrated in our previous paper [4] that this assumption alone is sufficient to provide a theoretical description that is consistent with all the known experiments. For a two-dimensional (2D) electron system, there believed to be two phases: a high density Fermi gas phase and a low density insulating Wigner crystal phase. The dielectric constant of the liquid phase becomes negative when r s ≃ 2 [5], which indicates that the liquid phase is unstable. At lower densities, the Wigner crystal phase appears around r s ≃ 37 in the absence of disorder [6]. This critical value of r s appears to be reduced to around r s ≃ 10 with disorders [7]. In the intermediate values of r s , we believe that there is a liquid phase which we think is responsible for the observed MIT.In this paper, we propose that a droplet state of the electron system resulted from the phase separation of the electrons into this new liquid phase and a low density "gas" phase. Here we call the low density phase "gas" purely for the reason that its density is low. In fact, in the presence of impurities, the "gas" phase is disordered Wigner crystal. To investigate our proposal, we have studied the energetics of such a droplet state. We find that both electron-electron interaction and potential fluctuations are crucial for the formation of the droplet state.An obvious condition for the droplet state is that the electron gas is unstable. To investigate what possibilities of the instability leads to, we study a simple but physically motivated model. Let us consider electrons in the disc of radius b with positive background. Imagine that the electron system is shrunk to a new radius a < b while the positive background remains intact. Clearly the charging energy due to the separation of the electrons from the positive background increases the energy of the system. However there can also be energy gained (decreasing total energy): since for a uniform electron gas the ground state energy E g is at its minimum when r s ≃ 2 [5], for r s > 2 the...