The wear and friction behaviour of silicon nitride against bearing ste
el was investigated under lubricated and dry fretting conditions as a
function of amplitude and test duration. Tests were performed on a hig
h frequency fretting tester. Silicon nitride bearing balls were used a
s the upper oscillating specimens while the lower stationary Bats were
standard specimens of bearing steel. Amplitudes in the intermediate 5
to 50 mu m range and a test duration from 10 to 360 min were studied.
In lubricated conditions a commercial lubricant, ISO VG 220, was used
. Light microscopy, scanning electron microscopy (SEM), energy dispers
ive spectroscopy (EDS), Auger spectroscopy (AES) and transmission elec
tron microscopy (TEM) were employed to determine the wear mechanisms.
Under lubricated conditions transition from high to low wear volumes w
as recognised with increasing amplitude. At lower amplitudes and in th
e early stage of fretting tests at moderate amplitudes, mechanical wea
r dominated. Cracks on the stick-slip boundary and spalling of a thin
tribolayer was observed, Under these conditions the highest wear in lu
bricated fretting was obtained. In the final stage of fretting tests a
t moderate amplitudes, and from the beginning at higher amplitudes, tr
ibochemical wear is suggested as the dominant wear form. A 0.2 mu m th
ick tribolayer was observed on the contact, containing inclusions with
different Fe and Si contents. A very high concentration of carbon, fo
rmed by oil degradation, was also determined in this layer, confirming
the critical influence of oil on the wear behaviour. Quite a differen
t wear mechanism is proposed for dry fretting conditions. Results of A
ES analysis showed a layer an order of magnitude thicker than in lubri
cated fretting, also having a remarkably different chemical compositio
n. TEM analysis confirmed that the reaction layer consisted of a silic
a-rich amorphous phase containing small inclusions of Fe2O3 and Fe3O4.
In contrast to lubricated conditions, where the layer created was duc
tile, in the case of dry fretting the layer was brittle. The continuou
s process of forming and spalling the brittle tribolayer caused much h
igher wear rates and wear losses than under lubricated fretting condit
ions. No transition in wear behaviour was observed as was the case in
lubricated fretting. (C) 1997 Elsevier Science S.A.