The role of zinc in formalin-based fixatives was covered in detail in 
my review article published in Biotechnic & Histochemistry 
68(2):7582, 1993.  To summarize very briefly, zinc functions in at 
least 2 ways:

1.  it prevents or at least inhibits crosslinking by formaldehyde, and
2.  it apparently holds macromolecules in their native 3-dimensional 
conformations.

Together, this creates superior morphological detail at the level of 
the light microscope, and allows the use of nearly any antibody 
without having to resort to antigen retrieval or epitope recovery. 
Tissues fixed in NBF take 24-48 hours to fix properly for good 
morphology, but zinc formalin will produce the same level of quality 
in 8 hours (for surgical specimens, less for biopsies).

Tissues can be stored in zinc formalin for weeks or months without 
significant loss of immunoreactivity.

There are several formulations, and they do make a difference.  The 
most common is unbuffered zinc sulfate formalin which may produce 
formalin pigment artifact in bloody tissues.  Published formulations 
generally cause precipitation of zinc salts inside tissues (making 
them somewhat harder to section), and frequently make a mess inside 
tissue processors (some processor manufacturers recommend against 
zinc formalin for this reason).  However, there are commercial 
products that have gotten around both of those problems (see Anatech 
and Shandon for further information) and these can be used in 
processors.

Buffered zinc formalin avoids the formalin pigment problem, but it is 
tricky to make since zinc precipitates at about pH 6.0.  Commercial 
products are generally stable.

Zinc chloride is sometimes used.  It is faster, hardens tissues more 
and cannot be used for storage.  It is corrosive to metal and must 
not be used on tissue processors.

Zinc salicylate has been used rarely, offers no advantage over other 
zinc formulations, and is expensive (homemade or commercial).  It is 
available from Polysciences.

Further details, both purely techni