In the literature, there is the idea that steps and kinks are the active si
tes fbr chemical reactions, but the experimental data are far from convinci
ng. In this article we see if there is a correlation between step atom dens
ity, van Hardeveld and Hartog coordination numbers, or the electronic coord
ination number and reactivity for a number of simple decomposition and hydr
ogenolysis reactions on platinum as measured by temperature-programmed deso
rption. We have examined reactions of ethylene, nitric oxide, and methanol
on (111); (110)-(1X1), (110)-(2X1), (100)-hex, (100)-(1X1), (210), (511), a
nd (331) platinum surfaces. We have done a statistical analysis of our data
to see if any of the correlations are non-negligible. We find that, in gen
eral, stepped surfaces have different reactivity than close-packed planes,
but some stepped surfaces are more active than Pt(111) while others are les
s active than Pt(111). There are negligible correlations between step atom
density and catalytic activity for our reactions. Similarly, there are negl
igible correlations between reactivity and the lowest or average van Hardev
eld and Hartog coordination numbers. Calculations were done to understand t
he variations. There are correlations between the lowest electronic coordin
ation number and reactivity for only the reaction of methanol to carbon dio
xide, and between the average electronic coordination number and reactivity
for the reaction of methanol to methane. The first correlation is not surp
rising considering the special nature of the rate-limiting step of this rea
ction. The second correlation raises many yet-unanswered questions. We find
that, in general, stepped surfaces relax to more stable geometriest The re
laxation process raises the electronic coordination number of the surface a
toms. After relaxation, there is not a large difference between the coordin
ation of atoms of platinum stepped surfaces and Pt(111). (C) 1999 American
Vacuum Society. [S0734-2101(99)08404-3].