I have crystals, but are they salt?
After examining a series of crystallization screens, crystals may be observed. The next step is to verify that the observed crystals are composed of the target macromolecules.
 Are they salt crystals?
Whenever crystals appear only in conditions containing Mg, Ca, Zn, or Cd, think phosphate contamination. Check to see if a phosphate buffer was used in any of the purification steps. Check to see if crystals grew in almost all conditions containing Mg, Ca, Zn, or Cd. Even a hint of phosphate in the presence of Mg and ammonium sulfate will produce struvite (a type of kidney stone). These tend to be needle crystals in the presence of protein.
Calcium citrate also a relatively low solubility (~15 mM in the cold), and its solubility decreases as the temperature goes up. Thus, getting calcium citrate crystals is a possibility if both components are at a relatively high concentration. However, sodium citrate alone is soluble up to ~1.4 M at least, depending on the pH.
A list of conditions from commercial screens, where people obtained crystals that turned out to be from salt, is at Conditions prone to salt crystallization. This should not be taken to mean that these conditions always produce salt crystals, but they apparently sometimes do.
 What if I have a membrane protein?
Most nonionic detergents do not form crystals in aqueous solution. Check the published phase diagrams to be sure. In fact, few of the alkyl glycoside detergents (like octyl glucoside and dodecyl maltoside) crystallize if the sugar group is a beta anomer, even in organic solvents. This is one of the reasons why they are so difficult to purify by recrystallization. Under some extreme conditions, nonionic detergents can form liquid crystals, but these are rare.
 How can I tell?
1. Izit dye (methylene blue; see Hapmton Research) is a popular tool for distinguishing protein crystals from salt crystals. Bernard Rupp has make a little movie of Izit dye soaking into lysozyme crystals ().
2. If they are harvestable crystals, dissolve up some crystals in SDS-PAGE sample buffer and see if they contain protein.
3. Also, dissolve up a couple of drops containing the most crystals in SDS-PAGE sample buffer and see if the protein is intact. Slow proteolysis of your protein by contaminating proteases can trim it to a "crystallizable" form. If it is, just consider truncating the protein for expression.
4. If the crystals are of reasonable size, just drop them into a low ionic strength buffered solution containing 0.2% - 2% glutaraldehyde. Protein crystals will quickly be fixed (faster than they dissolve) into a golden, gelatinous lump. The color comes from the formation of Schiff bases with the lysines and amino terminii. Sometimes a crystal-like shape is retained, other times just a yellowish, rubbery drop is left. In contrast, salt crystals should dissolve over time and should not be colored.
You could also add a small drop of 2% glutaraldehyde to your protein drop. Protein crystals then turn a light golden color. Crystals fixed like this can then be put into a low ionic strength solution where salt crystals should dissolve. You can easily transfer glutaraldehyde into a protein drop by vapor diffusion by adding glutaraldehyde to the reservoir to make it 2-3%. Glutaraldehyde is quite volatile. If you can smell it, it is slowly fixing your olfactory cells and corneas. Using solutions of glutaraldehyde less than 1% is generally quite safe.
The only caveat to using this method is that there should be no free amines around other than on the protein (i.e., no ethanolamine or Tris buffer, no ammonium ions, etc.). Most protein crystals are easily fixed, although they may end up looking like a yellow gelatinous lump at the end. But there probably is an exception or two to this rule. The use of glutaraldehyde fixation has advantages over Izit and other dye methods.