CCL: NPT vs. NVT, considerations for crystal simulations and nucleation

["conor d parks" <coparks2012%a%gmail.com>]

 Sent to CCL by: "conor d parks" [coparks2012{}gmail.com]
 Hello everyone,
 I was wondering if someone could potentially help me work through a very
 conceptual problem. I
 recently was trying to run nucleation simulations of water at 1.00 g/cm^3 217 K
 in the NVT
 ensemble. However, I wasn't seeing the types of cluster size fluctuations (I.E.
 any) I was
 expecting, and started to wonder if it was due to my use of the NVT ensemble.
 Now firstly, this
 isn't due to finite size effects I believe, because the simulation was ran with
 2000 water
 molecules.
 My issue with water above made me start to think about how I wanted to handle my
 glycine
 simulations as well. I am interested in running glycine in water nucleation
 simulations. For
 instance, I know glycine's solubility is .338 g/ml at 324 K. So I was going to
 make my initial box
 size based off the 0.338 g/ml, equilibrate a fluid at 354 K, and then crash to
 270 K. Now I can't
 seem to figure out if I should run this simulation in NPT or NVT. The solubility
 data is gathered
 experimentally at 1 atm, obviously. So if I run NVT, my pressure average will
 almost assuredly not
 be 1 atm. However, if I run NPT and fix the pressure at 1 atm, I don't see why I
 should be
 guaranteed to fluctuate around the correct solubility. If I think to my
 undergrad thermo, in the NPT
 ensemble with a constant number of glycine and water molecules (hence mol
 fraction is
 determined), there should be only one equilibrium concentration. I don't see why
 this equilibrium
 concentration hopefully generated from an NPT equilibration would correspond to
 my solubility
 point.
 Sorry if that didn't make sense.
 Conor