Download Development of oxide dispersion strengthened (ODS) bondcoat for

Development of oxide dispersion strengthened (ODS) bondcoat for gas-turbine
components
Introduction
Gas-turbine components work in very severe conditions, caused by mechanical stresses, high temperatures, oxidation
and hot corrosion. To protect the base material the state of the art suggests the use of a multilayered thermal barrier
coating, based on an external ceramic layer and an inner bondcoat. The ceramic layer (usually Yttria-stabilized
zirconia, YSZ, deposited by Air Plasma Spray, APS) acts as a thermal barrier to protect the component against the high
temperatures, whereas the bondcoat (usually a Co,Ni-CrAlY coating deposited by High Velocity Oxygen Fuel HVOF)
protects the material from oxidation and hot-corrosion. During the working life the formation of a thermally grown
oxide layer (TGO) between the topcoat and the bondcoat take place.
This system usually fails due to the complete spallation of the ceramic layer, caused by the mismatch between the
ceramic layer and the TGO grown above the bondcoat.
Therefore in order to improve the behavior of this system new coatings have been investigated, in order to enhance
the oxidation and hot corrosion resistance and to reduce the mismatch among those different layers.
The present investigation aims to improve the bondcoat behavior at high temperatures by using oxide dispersion
strengthened (ODS) bondcoat to replace the traditional M-CrAlY coatings (M=Ni and/or Co).
An ODS material consists of nano or microstuctured oxide particles embedded in a matrix based on a ductile material.
The present investigation deals with the study of ODS-materials based on a metal matrix (standard CoNiCrAlY powders
for thermal-spraying processes) and alumina powders as a strengthening media. The ODS-rich particles are generally
obtained by mechanical milling (alloying). Oxide particles and metal powders were mixed together and then milled for
several hours, using a milling machine. During the milling process, due to the ball-powder collisions, the hard phase is
embedded inside the metal matrix, forming a homogenous powder.
The particle size distribution and the morphology of the particles depend on the milling parameters and on the
features of the starting powders.
The first part of this work deals with the optimization of the milling procedure in order to obtain suitable powders for
the thermal-spraying process. Different kind of Al 2O3 powders and milling parameters are used in order to achieve a
homogenous microstructure and a proper particle size distribution.
The second part deals with the coating deposition and with the characterization tests, necessary to investigate the
effect of the oxide dispersed particles on the high temperature behavior of the bondcoats.
ODS systems are usually designed to improve the mechanical properties of the selected component, such us the wear
resistance or the high temperature creep resistance. The novelty of this approach lies in the use of ODS composite
materials to improve the high temperatures behavior of gas-turbine components.
Methods
For the production of the dispersion strengthened powders, a commercial CoNiCrAlY powder (Amdry 995C, -90 + 45
μm, Sulzer-Metco), with a nominal composition (in wt. %) of Co–32Ni–21Cr–8Al–0.5Y, was used as starting matrix
material.
The strengthening material employed in this investigation was
aluminum oxide. In order to investigate the effect of the particle
size on the microstructure of the milled powders and in order to
reach an appropriate distribution of the hard phase, different kinds
of alumina were used: AKP-30 (α-Al2O3, -0.5 + 0.3 μm, Sumitomo
Chemicals Co.,Ltd., Japan), Martoxid MR-70 (α-Al2O3, -3 + 0.1 μm,
Martinswerk GmbH, Germany) and Amperit 740.0 (α-Al2O3, -22 + 5
Figure 1 - scheme of the milling process
μm, H.C. Starck GmbH, Germany).
Different amount of alumina (10, 20 and 30 wt. %) were added to
the standard bondcoat powder, to evaluate the high temperature behavior and properties of coatings based on
different formulations.
A Simoloyer CM01 (Zoz, Wenden, Germany) high-energy milling machine equipped with a 0.7 l chamber was used for
the milling process.
The milling media consisted of 100Cr6 stainless steel balls, with an approximate diameter of 5 mm.
The ball-to-powder mass ration used in this study was 10:1, selected according to previous investigations. The
chamber was filled with 1400 g of milling balls and 140 g of powder mixture.
In order to achieve a high relative velocity between the milling media within the mill and in order to cool down the
machine, the rotation speed was varied during the process: every 4 minutes the chosen revolution speed was
decreased down to 800 rpm for 1 minute.
The ground revolution speed was varied from 1200 to 1600 rpm, to investigate the effect of the milling speed on the
particle size and microstructure (fig. 1).
The progress of the milling process was investigated by taking powder samples every hour for a maximum milling time
of 6 hours.
After the milling process powders were investigated by means of optical microscopy and particle size distribution
measurement, in order to select the optimal milling parameters and powder formulations. Scanning Electron
Microscopy (SEM), Energy Dispersive Spectroscopy (EDS) and X-Ray
Diffraction (XRD) will be also performed, to detect chemical
composition and crystalline phases of the powders.
The most promising powders were then deposited on steel substrates
by using 2 different thermal-spraying techniques: High Velocity OxyFuel (HVOF) (DJ-2600, Sulzer Metco) and Air Plasma Spray (APS) (Triplex
Pro, Sulzer Metco).
The best spraying parameters were used to deposit a thin ODSbondcoat layer on a standard bondcoat (based on CoNiCrAlY Amdry
9954 powder) previously deposited by HVOF (and heat treated in a
Figure 2 - topcoat + thin ODS-bondcoat +
standard bondcoat
vacuum atmosphere) on an Inconel substrate.
XRD, SEM (and Energy Dispersive X-Ray Spectroscopy, EDS), optical microscopy and roughness measurements were
performed.
Subsequently half of the surface of the samples based on the thin ODS-bondcoat layer deposited on the traditional
bondcoat was coated with a standard YSZ topcoat (fig. 2).
These samples were then put inside a oven set at 1100°C (air atmosphere) and cooled down for 1 hour by taking them
out from the furnace each 23 hours, to investigate the oxidation and the thermal shock resistance.
Results and discussion
CoNiCrAlY Amdry 995C (80 wt.%) + Al2O3 MR-70 (20wt.%), effect of the milling time
In order to investigate the effect of the milling time on the particle morphology, the powder batch based on the
CoNiCralY powder Amdry 995C with the addition of the MR-70 alumina powder, was milled up to 6 hours, using the
standard cycle previously described (fig. 1).
The powder was analyzed every hour by means of optical microscope and particle size distribution measurement.
Figure 3 shows the percentile results (D10, D50, D90) of the six powder batches (ConiCrAlY-Al2O3 80-20 wt. %) as a
function of the milling time. It can be noticed that the particle size decreases increasing the milling time, up to 3
hours. After this time the trend is less regular, because
fragmentation and agglomeration phenomena occurred at the
same time: after 4 hours the agglomeration phenomena are
more relevant, whereas after 5 and 6 hours powders are more
brittle (cause the metal matrix is work-hardened) and
fragmentation phenomena prevail. Increasing the milling time
made also the particle size distribution curve more narrow.
SEM images (fig. 4) show the evolution of the particle
microstructure during the milling process. The lamellar
structure of powders milled for 1-2 hours was gradually
Figure 3 - Percentile results (D10, D50, D90 values) as
a function of the milling time.
converted into a more homogeneous microstructure, in which
the hard phase is better embedded inside the metal matrix.
Figure 4 - Optical microscope images (cross-section) of the powder 995C-M20, after 1 hour milling (left) and after 6 hours (right).
CoNiCrAlY Amdry 995C (80 wt.%) + Al2O3 MR-70 (20wt.%), effect of the milling speed
In order to evaluate the effect of different milling parameters on the particle size and morphology new powders were
milled varying the rotation speed (1200 and 1400 rpm, up to 4 hours).
The particle size distribution curves of powders milled at 1200 and 1400 rpm show that the variation of the milling
speed affect the particle size, because the energy generated by the ball-particle collisions is lower. In both batches
even after 3 hours the curves are bimodal, indicating the presence of alloyed and unalloyed particles.
The optical microscope images confirm what previously assumed revealing a high amount of pure metal lamellae and
partially alloyed particles even at the end of the milling process (4 hours).
CoNiCrAlY Amdry 995C (80 wt.%) + Al2O3 MR-70 (10, 30wt.%), effect of the Al2O3 amount
The comparison among samples containing
different amount of alumina (10,20 and 30 wt. %)
reveals that a higher amount of the hard phase
brought to a finer powder (after a 4 hours milling
cycle),
because
increasing
the
alumina
concentration made the particles more brittle and
promoted the fragmentation phenomena (fig. 5),
whereas the higher fraction of ductile phase
within the powder batch containing only 10 wt.%
of
alumina
promoted
the
agglomeration
phenomena during the milling process.
Optical microscope pictures (fig. 6) clearly shows
the different amount of aluminum oxide inside
Figure 5 - Percentile results of particle size distribution
measurements performed on
the metal matrix of the two different batches.
Figure 6 - Optical microscope images of powders containing different amount of Al2O3, milled for 4 hours:
10wt.% of alumina (left) and 30wt.% of alumina (right).
Coating deposition
The most promising powders were chosen and sprayed on different substrates. In order to evaluate the effect of the
alumina amount powders containing 20 and 30 wt. % of the selected oxide were deposited with two different torches
(HVOF and APS).
Figure 7 – SEM images (back-scattered electrons detector) of coatings deposited by HVOF:
the sample on the left contains 20 wt% of alumina, the sample on the right contains 30 wt.%.
Sample name
M-12-169-DJ-tc
M-12-173-DJ-tc
M-12-221-tc
M-12-223-tc
Spraying gun
HVOF
HVOF
APS
APS
Al2O3 amount (wt.%)
20
30
20
30
Topcoat (YSZ)
APS
APS
APS
APS
Table 1 – features of the sprayed samples: Inconel 738 substrate + standard bondcoat (Amdry 9954) + thin ODS layer + standard
YSZ topcoat.
The coatings (deposited by HVOF) containing 20 wt.% of alumina are denser and more homogenous than coating
based on a higher amount of this oxide (fig. 7). The thin ODS layers sprayed above the standard bondcoat are less
dense than the previous samples, because due to the lower
number of passes the peening effect (i.e. the hardening and the
increasing of the density caused by the collisions of the new
sprayed particles on the already deposited coating) is less
relevant.
The coating deposited by APS are thinner and more
heterogeneous; compared to the HVOF samples the deposition
efficiency is lower. The thin ODS layers deposited on the
Figure 8 - I topcoat; II thin ODS layer; standard
bondcoat; IV substrate
standard bondcoat have a poor quality and they don’t cover the
whole surface of the sample.
The 4 samples based on the thin ODS bondcoat were coated with a standard topcoat (YSZ) and then their high
temperatures resistance were determined by an oxidation test (fig. 9).
The preliminary results of the oxidation and high temperatures resistance are listed in table 2. The samples based on
the ODS bondcoats deposited by APS failed after a limited number of cycle (a single cycle consists of 23 hours at
1100°C and 1 hour cooling). This probably because since the ODS coatings were not perfectly deposited, oxygen
diffused inward and reacted with the standard bondcoat and with the substrate. Furthermore the weak adhesion
between the topcoat and the bondcoat could have increased
the spallation phenomena, making the topcoat failure easier.
The samples based on the ODS coating deposited by HVOF
exhibited a better resistance, due to their higher thickness
and better microstructure.
This test indicated also that the coating containing 30 wt.%
of alumina (deposited by HVOF) showed a better resistance.
Therefore it may be assumed that a higher amount of
Figure 9 – Coated samples before the isothermal oxidation
alumina improve the behavior of the coatings at high
test: (1) M-12-169-DJ-tc; (2) M-12-173-DJ-tc;
(3) M-12-221-tc; (4) M-12-223-tc
temperatures. The comparison between these samples and
standard
bondcoat without
alumina
will allow the
comprehension of this strengthening mechanism.
Sample name
M-12-169-DJ-tc
M-12-173-DJ-tc
M-12-221-tc
M-12-223-tc
N° of cycles before the first sign
of damage
19
22
4
8
N° of cycles before the
complete failure
24
52
15
15
Table 2 – Results of the oxidation test.
Future perspectives
Future investigations will concern both the optimization of powders and spraying process, in order to improve the
behavior of these ODS coatings at high temperatures.
Further characterization analysis on the milled powders and coatings will be useful to investigate the microstructure
of the new samples and to determine the presence of oxidation phenomena that may occur during the spraying
process.
The most promising candidates will be tested to evaluate their furnace cycling oxidation resistance, in order to
investigate the effect of the oxide phase on the coating properties.