The thermophysical properties of plasma sprayed coatings depend on numerous
parameters. The molten state and velocity of particles upon impact are con
ditioned by different choices (torch design, powder parameters,...). Moreov
er the substrate material, its surface preparation, its preheating temperat
ure and time control the flattening of the impinging first droplets, the re
sulting splats cooling and solidification. Their layering on already solidi
fied splats and thus the coating generation depend finally on the spray pat
tern and powder mass flow rate, deposition efficiency and cooling during sp
raying (1, 2).
Among the different properties of coatings, their adhesion/cohesion (A/C) i
s a very important one (3). In good correlation with the splat cooling rate
s which increase with substrate preheating temperature T-p, A/C increases w
ith T-p provided the preheating time tp is not too long.
Both (t(p)) and (T-p) must be adapted carefully in order to avoid a too hig
h oxidation state of the substrate and are generally set according to stren
gth tensile measurements. With 304L stainless steel substrates preheated at
500 degrees C with the plasma torch, the adhesion/cohesion is almost tripl
ed compared to that obtained on cold substrates. However over a specific ox
idation level, the metallic surface properties are less favourable to the f
ormation of disk shaped splats on smooth surfaces with an increase of their
cooling rates and a decrease of coating A/C (4).
The current surface state used in plasma spraying technology is the grit-bl
asted surface. However different analytical techniques involved in the stud
y need other surface preparation (e.g. mechanical polishing). The aim of th
e present study is to establish the links between given surface states and
their plasma oxidation results. Using different spectroscopic techniques (s
uch as XRD, Mossbauer spectroscopy and ultraviolet-visible-near infrared sp
ectroscopy), we have characterized oxides layers and metal-oxide interfaces
after an optimized preheating (10 min long for a preheating temperature of
500 degrees C). Three different surface states have been chosen correspond
ing to two different mechanical polishings (A and B) and a grit-blasting on
e (C). More precisely, the samples A are polished by using SiC papers and f
ine diamond paste (1 mu m), whereas substrates B result only from SiC paper
s abrasions (grade 1000).
A plasma oxidation is really atypical compared to usual thermal oxidation.
Such treatment is performed in air by using a custom-built dc. plasma torch
with a 7 mm internal diameter nozzle. The surfaces are exposed directly to
the jet plume at 100 mm from the nozzle exit The disk shaped substrates of
304L stainless are disposed on a rotating cylindrical holder, which axis i
s orthogonal to that of the plasma torch. This latter is translated back an
d forth. The substrate temperature is kept at 500 degrees C thanks to a con
trol system equipped with an I.R. pyrometer and machined slots through whic
h compressed air is blown with flow rates monitored by the pyrometer. In th
e plume of the jet, the temperature has been measured using Rayleigh light
scattering and is about 3000-3500 K close to the substrate (5). The plasma
composition is therefore mostly constituted of monoatomic oxygen. On top of
that, the samples are submitted to thermal cycles, since the surfaces expo
sed to the jet-plume during a very short time (less than one minute) receiv
e briefly a high energy level 0.5 x 10(8) W/m(2).
The topographic investigations on A, B and C surfaces allow to determine th
ree different mean roughnesses, about respectively 0.01, 0.05 and 2.5 mu m.
On the other hand, the 304L stainless steel is austenitic and metastable.
The austenite f.c.c. structure is therefore partially transformed in strain
inducing b.c.c. martensite (SIM, alpha'-martensite) during mechanical surf
ace preparation (e.g. B surface state). XRD and Mossbauer investigations ar
e in good agreement to confirm the presence of martensite upon the austenit
e. The thickness of affected austenite is about 60 nm. Such transformation
results from a'-martensite nucleation at the intersections of mechanical tw
ins due to abrasive polishing (16). On the contrary, surface states A and C
don't exhibit a such surface phase transformation. This can be probably at
tributed to the fact that stresses are in A and C cases more compressive, t
hus hindering the volume expansion required for the a'-martensite transform
ation, which, in turn, suppresses its formation.
After preheating, the oxidation results on A and C samples surface are quit
e similar. The oxides layers are dual and consist mainly of a spinel phase
NixCr2-yFe3-x-yO4 (With x similar to 1 and y similar to 0) mixed to a rhomb
oedral oxide Fe-x,Cr2-x,O-3 (with x' similar to 2). The weak substitution l
evel of iron III for chromium III of Fe-x,Cr2-x'O-3 have been determined us
ing XRD lines positions and hyperfine field values evaluated by Mossbauer s
pectroscopy according to references data (34). In the case of B surfaces, t
he presence of alpha'-martensite modify sensitively the oxides layer compos
ition, since only the spinel phase NixCr2-yFe3-x-yO4 has been identified (s
ee fig. 10). The x and y values are different and evaluated around 0.3 and
1.5 respectively On top of that, the metal-oxides interfaces are generally
constituted of b.c.c, ferrite and austenite. Mossbauer spectroscopy allows
to distinguish without ambiguities this b.c.c. phase from the martensite, s
ince the hyperfine fields values own to this ferromagnetic phase are quite
different. The amount of ferrite is determined by the strain effects before
and after the plasma preheating. The presence of initial alpha'-martensite
seems besides to enhance the ferrite formation under the oxide scales.