Partitioning of trace elements between mineral phases reflects the phy
sical, chemical and kinetic conditions of crystallization. Variations
in environmental conditions during growth often result in complex and
small-scale chemical zoning in minerals. The low abundance of trace el
ements and their spatial inhomogeneity on a mum scale makes their anal
ysis by a microprobe technique essential for addressing many petrologi
cal problems. Mu-PIXE (2-3 MeV) has been successfully applied to many
mineralogical problems and is rapidly becoming a routine analytical to
ol for geologists. High-energy PIXE (40-60 MeV) provides a new dimensi
on in mineralogical analysis. The K X-rays for many petrologically imp
ortant trace elements occur in the 25-90 keV region, here the X-rays a
re not affected by interference from the X-rays of more abundant geoch
emically coherent elements. Furthermore, the K X-ray spectrum for an e
lement is less complex than its corresponding L X-ray spectrum so data
reduction is simplified. The use of high energy protons for elemental
analysis makes high-energy PIGE accessible, here on-line emission of
gamma-rays can be used to provide information on element (or in some c
ases isotope) concentrations. For the analysis of chemically complex m
aterials such as rocks and minerals it is necessary to thoroughly char
acterize the material beforehand such that likely proton induced react
ions can be predicted. Nuclear reactions produced by proton interactio
n with mineral samples occur during on-line exposure of the sample. Th
e by-products of such reactions may have significant half-lives which
will make them amenable to off-line analysis. One such case is where P
t undergoes (p, xn) reactions to form Au which then decays back to Pt
via electron capture. The off-line spectrum after such a run contains
Au X-rays and the background to such spectra is low, which raises the
possibility that this form of analysis will provide low detection limi
ts. This is the proton analogue of neutron activation analysis - proto
n activation analysis (PAA).