[Guest]

Concerning of a variation of Appel reaction (R-OH + Ph3P + imidazole + I2 -> R-I + Ph3P=O + imid*HI):

"A neat trick to remove Ph3P=O is to load you material onto celite, dilute with hexanes, and then filter, using hexane washes. The Ph3P=O will crash out and the celite will keep you material from becoming trapped in the Ph3P=O, allowing you to isolate clean iodide from your hexane washes."

quoted from:
http://totallysynthetic.com/blog/?p=520

[Guest]

Another option is to use Me3P or Et3P instead of Ph3P (when possible, of course). Correpsponding phosphine oxides are soluble in water
(see e.g. 10.1002/ejoc.200900634)

According to MSDS from Aldrich, Me3P and Et3P are not toxic!

Me3P (b.p. 38-40 °C) is Highly Flammable and Irritant
Et3P (b.p. 127-128 °C) is highly Flamable and Corrosive

Probably, Et3P is safer as it has higher b.p.

[kilomentor]

If one can get the desired reaction product into acetone solution there is a procedure that would probably work, but I have not used this method. the suggestion is based on the physical properties of compounds in the literature. Both sodium bromide and sodium iodide are reported to for high melting comlexes with five! molecular equivalents of triphenylphosphine, [J. Chem. Soc. (A) 449 (1968)]. an acetone solution containing the triphenyl posphine is mixed with a saturated aqueous solution of the correct chemical equivalents of the alkali-metal salt at 50 C.
It is reported that the adduct precipitates as large, well-formed crystals.

kilomentor