Heinrich J.G., Aldinger F. (Eds.) Ceramic Materials and Components for Engines

Подождите немного. Документ загружается.

This Page Intentionally Left Blank

TBC CONSISTING

NEW METAL-GLASS COMPOSITES

M. Dietrich*, V. Verlotski,

VaDen, D. Stover

Institut fur Werkstoffe und Verfahren der Energietechnik,

Forschungszentrum Jiilich GmbH,

D-52425

Jiilich, Germany.

ABSTRACT

A new concept for thermal barrier coating (TBC)

systems is presented, based on a metal- glass composite

(MGC). In this TBC system the composition of the

MGC is chosen in such a way that the thermal

expansion coefficient of the composite is close to the

one of the substrate. This leads to reduced thermal

stresses and hence improved thermal cycling life times.

Another advantage of the gas tight composite coatings

is their ability to protect the bondcoat from severe

oxidation. Correspondingly, longer life times have been

found for these

TBCs

in oxidation tests. In this paper

measurements of thermal properties as well as results of

oxidation experiments at high temperatures and thermal

cycling tests will be presented. Additionally some

aspects of the evolution of the microstructure during

thermal aging is shown.

INTRODUCTION

TBCs find an increasing number of applications to

protect high-temperature metallic components; for

example, TBCs are deposited on components of gas

turbines and diesel engines either to increase the inlet

temperature with a consequent improvement of the

efficiency or to reduce the requirements for a cooling

system [l] or to enhance the lifetime. The selection of

TBC materials is very specific [2] and some basic

requirements are low thermal conductivity, no phase

transformation accompanied by volume change during

heating, chemical inertness, low thermal expansion

mismatch with the metallic substrates and good

adherence [3].

The best compromise among these requirements is

presently offered by partially stabilized zirconia as

7-8

wt% Y,O,-ZrO, (YSZ) on top of a MCrAlY bond coat,

deposited either by air plasma spraying (APS) or by

electron beam physical vapor deposition (EB-PVD)

[4].

However, the application

YSZ-TBCs is limited by

two

factors. At higher temperatures than

1200"C,

phase

transformations from tetragonal (t') to tetragonal and

cubic (t

c) and then to monoclinic (m) occur, giving

rise to the formation of cracks in the coating

[5].

An other factor for failure is the insufficient oxidation

resistance of the substrate which has to be protected by

the gas tight MCrAlY-bondcoat (BC). This BC is

forming a thermally grown oxide (TGO) layer [6]. The

stresses associated to the growth of this Al,O,-layer are

presumed to induce spalling of the coating and

to be

one

the most important sources of failure in TBC-

systems.

A way to overcome this problem is to improve the

thermal expansion match with the substrate material and

to obtain a good adherence. This could be obtainable

with a metal-glass composite (MGC).

In this material it is possible to adjust the coefficient of

thermal expansion in a wide range [7], by varying the

ratio metayglass. In layer systems like TBCs the

adaptation of its thermal expansion to that of the

substrate material reduces the stresses, which are

induced by thermal mismatch.

An other advantage of the material is the absence of

open porosity, permitting the protection of the substrate

material and the bondcoat from corrosive gases8. Up to

now

TBCs

are permeable for these gases, exposing the

bond coat to oxidation and leading to the formation of a

TGO. Corresponding to such a gas tightness, longer

lifetimes for TBCs could be expected during thermal

aging in oxidizing atmosphere.

EXPERIMENTAL

For the preparation of metal-glass-composit coatings

(MGC), industrial available bondcoat powder and

ordinary white container glass were chosen. In order to

use plasma spraying for the deposition of metal-glass

composites, the powders had to be adapted. Taking into

account the metallic component of the composites, only

vacuum plasma spraying

(VPS)

was a possible spraying

technique for these powders.

Data of raw materials for the preparation of metal glass

powder mixtures are compiled in Table 1.

The thermal expansion coefficient of composites with

different metal-glass ratio has been determined by

dilatometer measurements.

The preparation of the powder starts by ball milling

glass- and metal powder with zirconia balls for 24 hours

in ethanol. After drying, the powder is homogenized by

ball milling with steel balls and subsequently annealed

under HJAr-atmosphere at

700"

for 2h.

The

slightly

densified and agglomerated powder is then dry coarse

milled in a mortar mill and finally a powder fraction of

40-90 pm is sieved. For the deposition of the composite

powder, a VPS equipment from Sulzer Metco with a F4

gun was used. The powder was sprayed on Inconel 617

and Inconel 783 substrates under 60 mbar Ar-

atmosphere. The thickness of the MGC-TBC varied

between

500

and 600

nun.

513

NiCoCrAlY

Silicate glass

Grain size

powder powder

Content in

theMGC

[wt.%]

Composition

[MolYo]

29.7% Ni

31% Co

30%

0.7%

71% SO2

14%

NqO

K20

10% CaO

Table 1. Composition of the MGC

Two types of oxidation tests were carried out in air

isothermal exposure of samples in a furnace and

oxidation in a thermal gradient across the sample

thickness. The first test was carried out at 1000°C for

1000 or 1500 h. The samples were taken out of the

furnace every 24h, natural cooled, controlled and then

taken back

the furnace. The temperature was chosen

in order to simulate the working conditions at the

TBC/BC Interface.

In the gradient test TBC surface temperatures of 1200°C

were achieved with a stable substrate temperature level

below 1000°C. Taking into account the thickness and

the thermal conductivity of the substrate, the

temperature gradient across the TBC was about 200K.

With this setup the thermal conductivity of the MGC-

TBC were estimated using the equation

h(T)

(dq/dA)

(dx/dT)

The temperatures were measured directly on the TBC

surface and the substrate back-side with thermocouples;

dq was obtained by a measurement with a

YSZ

standard

sample with

known

thermal conductivity.

Further more a burner rig test was carried out using a

methane flame. The sample was cycled on air by

raising the surface temperature from room temperature

to 1200°C within

seconds and with

minutes dwell

time while the back surface was cooled by air.

Subsequently cooling to room temperature for 2

minutes was enforced by cooling air jets from

both

sides.

The microstructure of the TBC was found to change

during thermal aging. This change was studied with

seven samples prepared in an identical manner and

exposed during an oxidation test to 1150°C in air. The

exposure time was varied from 1 to 300 h. The changes

were followed by scanning electron microscopy (SEM),

energy dispersive X-ray microanalysis

(EDX)

and X-

ray powder diffraction (XRD).

RESULTS AND DISCUSION

Dilatometer measurement showed the strong influence

of the metal content on the thermal expansion

coefficient of the MGC. Fore the pure glass it is

9,5*

10-

K-' in the temperature range of 20°C to 300°C. This

value is rising with the metal content and reaches

12,3*10-6

K-'

the temperature range of 20°C

7OOOC

wt%

corresponding to

~01%. Higher metal

contents were not used in order to guaranty complete

covering of the metal particles by the glass matrix.

An optical micrograph of the cross-section of an as

sprayed MGC-TBC shows the typical lamellae structure

of plasma-sprayed coatings where the light component

is metal, the gray one is glass and black represents pores

(Figure 1).

ooum

Figure 1. As sprayed MGC-TBC

The uppermost layer does not show any microcracks,

neither were inclusions or oxides found at the interface.

The porosity is a closed porosity of about

During annealing at 1000°C which is the BC

temperature in technical applications, grain coarsening

without formation of cracks or spalling was observed

Figure 2.

Ohm

Figure 2. Plasmasprayed MGC-TBC after annealing at

1000°C for 300h

Furthermore a metal-deficient zone of 10-40

Several

XRD

measurements, made during removing

thickness is formed on the MGC-TBC surface, within

layer by layer in steps of 30pm, showed the growth of

the

first

hour of exposure (Figure 3).

a-Al,O, mainly from the interface TBC/BC into the

The thickness of this depletion zone does not increase glass matrix. An other phase found in the glass matrix

with further annealing. Until now, there is no was Anorthite (CaAl,Si,O,)

explanation for its formation.

It was seen that A1 atoms diffuse from the BC and from

the metal particles of the MGC into the glassy matrix. It

is presumed that the difference of aluminum solubility

in the glass and the alloy, combined with the oxidation

of the aluminum in the glass are the driving force for

this diffusion. A

XRD

study of the surface of a MGC-

TBC after lOOh oxidation at 1100°C in air shows

reflections belonging to a-AlzO3. This indicates that at

least a fraction of A1 is oxidized to a-Al203. The

addition of A1203 to the glass corresponds to a shift in

the composition leading to a preferential crystallization

the two phases..

Plasma-spraying is similar to fast heating-up and

cooling down. During processing there is no time for A1

diffusion into the glassy matrix. Diffusion, formation of

Figure 3 Metal deficient layer at the surface of an

annealed MGC-sample

alumina and crystallization of the glass can take place

only during subsequent long-time aging at high

temperature in air.

MGC and may contribute to the formation of the

intermediate Al-rich layer at the interface

MGC/bondcoat. The growth of this intermediate layer

starts quickly and seems to stop within the first 100h.

reason could be the crystallization of the glass, slowing

down diffusion.

After calibration of the gradient furnace setup the

specific thermal conductivity of a MGC-TBC has been

determined to 1.4f0.2

W/mK.

comparison, the corresponding value for porous

YSZ-TBC is 0.8

W/mK.

follows that in order to

achieve equal thermal barrier properties the thickness of

a MGC-TBC must be nearly twice that of a traditional

The results of oxidation and thermal cycling tests can be

summarized as follows

Plasma-sprayed MGC TBC on a IN738 alloy substrate

did not show spalling during isothermal exposure to air

at 1000°C for 1500h. YSZ-TBC failed due to strong

spalling after 900h under identical experimental

conditions. Plasma-sprayed MGC-TBC withstand

isothermal oxidation tests

air at about 1200°C for

more after 300h before failing. Plasma-sprayed MGC-

TBC do not show spalling during oxidation test in

thermal gradient furnace, even above

300h

1250/1O0O0C while YSZ-TBC fail already after 200h.

Similar to YSZ-TBC, plasma-sprayed MGC coating did

Y SZ-TBC.

Figure

X-mapping of annealed MGC-TBC:

Evolution of tne interlayer between TBC and BC

Quantitative EDX mappings of A1-K radiation

(Figure

shows the evolution of this layer between

TBC and BC depending on the annealing time. EDX

measurements of the layer revealed that one major

element of the layer is Al.

not fail during ther&al cycling until

1000

cycles-The

good cycling behavior even with

600

pm thick coatings

can be explained with the high thermal expansion

coefficient which is 12.3*10-6

K-'.

Burner rig tests without BC showed very poor results

for YSZ as well as for MGC. This indicates, that the gas

tightness of the MGC alone is not sufficient for the

515

oxidation protection of the substrate. The results are

listed in Table

YSZ

MGC

YSZ

MGC

1000

900

1000 >1500

1200 85

1200

300

MGC

YSZ

without

1200/1000 >loo0

ooooc

Table 2. Results from thermal tests

ooooc

MGC

without BC

CONCLUSION

250

Metal-Glass Composite (MGC) was also found to be

very promising for the use as

TBC

material. During

thermal aging tests, the gas tight MGC coating seems to

protect the substrate material better from oxidation than

classical YSZ TBC does, leading to longer life times in

isothermal oxidation tests. Even with a thickness up to

600 pm the adherence of the coating and its spalling

resistance in thermal cycling was demonstrated to be

comparable to the corresponding results for YSZ

coatings

A better understanding of the microstructural

and phase evolution during thermal aging and

measurement of thermal and mechanical properties are

subject of ongoing work. Up to now no experiments

have been done

excess of 1250OC. Passing to higher

temperature is the aim of further development.

(1) P.Hancock, M. Malik, Materials for Advanced

Power Engineering. Part 1 .D.Coutsouradis et al.

(eds.), Kluwer Acad. Publishers, Dordrecht, (1994)

685

Advanced Gas Turbine Engines.

Therm. Spray

Tech.,

[l] (1996)

31-34

Aircraft Turbine Airfoils: Thermal Challenge and

Materials. Rev. MCtallurgie

CITI Science et GCnie

des MatCriaux, 96 [5] (1999) 585-597

Y203 Thermal Barrier System. Adv. Cer. Mat., 1

[

(1986) 68-76

(5) R.A. Miller,

J.L.

Smialek, R.G. Garlick, Phase

Stability in Plasma Sprayed Partially Stabilized

Zirconia-Yttria. Science and Technology of

Zirconia, Advances in Ceramics,Vol.3, A.H. Heuer

and L.W. Hobbs (eds.), Am. Cer. SOC., Columbus,

(2) G.A. Kool, Current and Future Materials in

(3) V. Arnault et al., Thermal Barrier Coating for

(4)

Stecura, Optimisation of the Ni-Cr-Al-Y/ZrO,-

OH, USA (1981) 241-251

(6) R. Miller, Oxidation-Based Model for Thermal

Bamer Coating Life.

Am.Cer.Soc., 67 [8] (1984)

517-521

(7) V. Verlotski, D. Stover, H.P. Buchkremer,

VaBen, Warmedtimmende Glas-MetalVKeramik-

Schichten. German Patent No. 198 52 285, Date:

May 3rd, 2000

(8) R. VaBen,

X.Q.

Cao, V. Verlotski, H. Lehmann, M.

Dietrich, D. Stover, Two new candidates for thermal

barrier coatings. To be published in

Surf.

Coat.

Tech.

516

ASPECTS ON SINTERING

EB-PVD TBCS

Klaus Fritscher, Uwe Schulz, Christoph Leyens and Manfred Peters

DLR

German Aerospace Center, Institute

Materials Research,

D-5

170

Koln, Germany

INTRODUCTION

Thermal barrier coatings

(TBCs)

of partially yttria

stabilized zirconia (PYSZ) on turbine airfoils are one

of the most promising recent approaches to improve

the overall economy of fossil fuel dependent energy

conversion systems. The benefits result from re-

duced fuel consumption by higher efficiencies, pro-

longed service intervals as well as lifetimes. Today

the most durable

TBCs

on rotating turbine parts are

PYSZ coatings that are applied by electron-beam

physical vapor deposition (EB-PVD). The EB-PVD

process procures the most superior

strain

and

ther-

moshock tolerant coatings due to their unique co-

lumnar microstructure.

order to integrate these

TBCs into the design of modem aero engine parts

the manufacture of so-called “designed-$

TBCs

quantitative understanding of the correlation be-

tween processing, microstructures, and the resulting

lifetime of the coatings is advised. Mechanisms

which affect the microstructures have to be under-

stood.

Sintering of the ceramic top layer is a mechanism

that can impair vital thermal and mechanical proper-

ties in TBCs. Sintering contributes to a reduction in

thermal isolation [l] and an increase of the in-plain

modulus of elasticity [2]. Sintering phenomena were

identified during thermocyclic exposures. During

cycling oxide columns were observed to crack intra-

columnarly, and sintering of the crack faces oc-

curred [3]. Sintering needs to be defined in order to

clanfy what

can

happen in PYSZ

TBCs.

pure

single components like PYSZ it takes place com-

pletely in the solid state. Besides solid state reactions

also vapor formation and re-condensation as means

transport may occur. Sintering may be accompa-

nied by shrinkage, leading to densification. But sin-

tering and shrinkage are by no means identical. Ap-

parently no definition of sintering exists that takes

full account of all theoretical and practical aspects

and the various stages of sintering. Sintering is a

fairly complicated process that involves several

mechanisms of material transport.

basic definition

sintering can be given like that: it is a thermally

activated material transport in porous matter that

aims at a reduction of the specific surface

area.

Macroscopic shrinkage was observed by thermo-

analytical dilatometry on free-standing plasma-

sprayed PYSZ

TBCs

>9OO0C

[2] and

1038°C

[4]

but will scarcely happen on

fully

adherent

TBCs

this

temperature range. Direct evidence

sinter-

ing of

TBCs

on their substrates can be obtained by

(electron) microscopic observations. Indnect evi-

dence is feasible by specific surface analysis meth-

ods. Mercury intrusion,

gas

permeametry and gas

adsorption techniques are commonly used analytical

methods for the characterization of porous media,

such as powders and green or sintered compacts.

Their general capability to the specific surface area

determination in adherent coatings on substrates,

however, is limited. Gas adsorption

suited best

and was used to characterize plasma sprayed coat-

ings

[

51.

High-rate EB-PVD PYSZ

TBCs

on alumina sub-

strates were investigated in this study. The specific

surface area of the respective TBCs was determined

by the BET method via inert gas adsorption de-

scribed in [6]

a function of sintering temperature

and time.

EXPERIMENTAL

6mm Qameter and 20mm long cylindrical recrys-

talized fully-dense alumina tubes were EB-PVD

coated with PYSZ

TBCs

of approximately 220pm

thickness. Recrystallized alumina was taken instead

of metallic substrates to minimize diffusive con-

tamination of the

TBC

and to preclude similar ef-

fects of thermally grown oxides on the heated me-

tallic substrates from the inner surface

TBCs

[7].

The TBCs were applied by high rate EB-PVD in a

rotary mode, as

conventionally used for

TBC

deposition on turbine parts [8]. The denomination

“TBCs”

th~s

paper will exclusively indicate

this

type of EB-PVD

TBCs.

The total weight

the

TBCs

on the alumina substrate was typically

0.25g.

The specific surface

the samples was measured in

the as-coated condition. They were then heated iso-

thermally in air at

700,

800,

900,

1000 and 1100°C.

They were taken from the furnace after a definite

time for specific surface measurements and re-

deposited into the furnace.

This

procedure was re-

peated several times to obtain a consistent plot

data for each temperature.

The specific surface area of as-deposited and of air-

annealed samples was determined by the volumetric

BET method

(ASAP

2000,

Micrometrics, Georgia).

Nitrogen or krypton

adsorbent

gas

was used de-

pending on the range

surface area expected. Spe-

517

cific surface area values were obtained by multi-

point determination (five points) from the adsorption

isotherm. The pressure range was 0.1 to

0.3

atm.

RESULTS AND DISCUSSION

Mechanisms

The variation of the specific surface area

PYSZ

TBCs versus time at temperatures between

700

and

1100°C is given in Fig.

concluded that on sintering bigger pores will grow at

the expense of the thinner pores to lower the inner

specific surface area. And the resultant bigger pores

will be attracted more and more by the surface of the

columns for the same reason of overall surface re-

duction. There they become converted into open po-

rosity

and

contribute to some increase of the specific

surface area. The effect of open porosity formation

on sintering follows classical sintering sequences but

was not quantitatively addressed in this study.

700"

PYSZ

TBCs

alumina substrate

10 100

Time

hours

1000

Fig.

Specific surface area of EB-PVD processed

PYSZ TBCs on recrystallized alumina

sub-

strate between

700

and

100°C versus time

It shows that at higher temperatures a lower specific

surface area is attained, and the time required for

fading down to lower specific surface areas is re-

duced. Vice versa, at lower temperatures, the spe-

cific surface areas remain higher and the times to

come up to low values are longer. After short

an-

nealing times at lower temperatures of

700

and

800OC a transient increase in the specific surface

area was observed.

This

effect seems to be in some

agreement to initial sintering stages of some powder

compacts that increase in volume due to rearranging

grains.

rearrangement of

grains

or a

shift

in the

crystallographic planes of the columns appears

un-

likely

they are essentially monocrystals. They are

still in a low-energetic state, and therefore no driv-

ing force is likely to alter the pertinent stage of en-

ergy. The following explanation is more likely.

Microscopic observations of sintered columns have

shown the presence of numerous sub-micron open

pores in the surface. Similar open pores were not ob-

served on columns of as-received columns,

they

obviously formed on sintering.

Instead,

the forma-

tion of a finely dispersed closed porosity mainly

along the center of the columns

well as in line

with the joints between the feather

arms

(in the fol-

lowing the feather

arms

will be denominated "sec-

ondary columns") is confirmed by transmission

electron-microscopic observations

[9].

The closed

pores are typically between

only

and

50nm.

It is

100

1000

Time

in hours

0,l

Fig.

Specific surface

area

recession

(So

S)/So

of EB-PVD processed PYSZ TBCs on re-

crystallized alumina substrate between

700

and 1100°C versus time

In Fig.

the values for the specific surface

(de-

nomination of the actual speclfic surface is

the

specific surface of the starting

TBC

material is

So)

of Fig. 1 are used to calculate the

(So

S)/So

ratio to

bring about the specific surface area recession versus

time. The values are fairly good representatives for

the initial stages of sintering mainly at

700

and

800°C

where shrinkage and densification cannot be

noted.

the transport processes in sintering have

stabilized straight lines in the double log graph are

expected

will be briefly explained

[

101.

the

early stages of sintering, preferably of small pow-

ders, the parameter

(So

S)/So

can be used to iden-

the sintering mechanism

((So

S)/S"?

(1)

where C is a kinetic term that includes

mass

trans-

port constants,

is the sintering time and

is an ex-

ponent that

close to

n/2

for small powders, where

n is the exponent

the initial-stage sintering equa-

tion (the sintering stage that is controlled by neck

growth). n depends on the mechanism of mass

trans-

port. E.

n is

for surface-diffusion controlled

mechanisms, and other mechanisms provide lower n

exponents.

The inclination of the line plots at 1 100°C in Fig.

is shallow and

uniform.

The number of data, how-

ever,

restricted, and their accuracy depends on the

mass of the TBCs of 0.25g

that

is at least one order

of magnitude too small

compared to recommended

518

mass weights for giving values with

small

standard

deviations. Moreover, the accuracy drops more and

more on repeated sintering of the samples due to

their reduction of the specific surface area.

some

of these cases of BET values below 1mVg the ad-

sorbent had to be changed from nitrogen to krypton

for the sake of better resolution. These values turned

out to be of the order of

to 10% lower

than

with

nitrogen, which, however, was estimated to be

within the range of scatter of these data and was

taken here to be of minor concern. But the low incli-

nation of the plots in Fig. 2 still suggests

that

in-

dicative of dominating surface diffusion control. The

value for

2n will then give the best fit. It must be

born in mind that the dependencies of equation (1)

are applicable to initial sintering of small-grained

powder compacts. The basic question in

this

context

is whether the dependencies can be transferred to an-

isotropic columnar particles that are apart from each

other having secondary minute columnar protrusions

all over their surfaces. And only

few of these sec-

ondary columns are contacting others.

This

situation

is really different from the one as for small particles.

the conclusion of predominant surface-dfision

control is still speculative.

Mate- Annealing

rial temperature

mz/g

During sintering the specific surface area declines

rapidly from an initial value

to an intermediate

value

is given as an average value in Fig. 1 and

is more detailed by the values of the respective

sam-

ples in Table 1. These

data

also show the scatter of

of individual BET measurements on identical

samples from the same EB-PVD batch.

(s"-s)/s"

after

100

PYSZ

PYSZ 1000 4.619 0.914

PYSZ

1100

The results on sintering show a dramatic reduction

in surface area in Fig. 1 or, alternatively, an increase

of the surface area reduction pamneter

(So

S)/So

Fig. 2 at

all

temperatures. The values are most

prominent for the highest temperature of 1100°C.

equivalent effect in dilatation analytical work

was obtained on sintering in the temperature range

900°C [2]. These quantities for specific surface

changes are unexpectedly high with regard to the

minor macroscopic effects. The flux of matter dur-

ing sintering must have been directed along superfi-

cial short-circuit diffusion paths because volume-

related effects were not observed at

900°C. These

observations give some more support for the sug-

gestion of dominating surface diffusion control.

TBC

on alumina

. . . . . . . . . . . . . . .

. .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . .

100

Time

hours

0,1

Fig.

Specific surface area recession

(So

S)/So

of EB-PVD processed PYSZ TBCs on re-

crystallized alumina substrate compared to

free-standing identical microstructures at

1 100°C versus time

Fig.

shows the difference of sintering of

TBCs

1100°C on

alumina

substrates and without substrates

free standing coatings. The differences are appar-

ent during the first hour. Then the values become

equivalent. It indicates that the straight positions of

the columns on a substrate accelerate sintering only

at the beginning. Later the columns show no differ-

ent behavior. Some columns apparently exhibited an

activated sintering mode probably due to forced

contact

with

neighboring columns and have arranged

afterwards. The same situation does not happen with

free standing coatings where stress relief on contact

points before sintering

feasible.

Fig. 4 shows the impact of annealing in laboratory

air

and vacuum of approximately Pa on the de-

gree of surface reduction. The vacuum atmosphere

clearly retards sintering. The light gray color of the

vacuum-annealed TBC is indicative for loss of

oxy-

gen &om the zirconia lattice.

1000°C,

100

air

vacuum

Annealing atmosphere

Fig.

The influence of the atmosphere

air

and

vacuum during annealing at 1000°C for 100

h on the specific surface area recession in-

dicating less sintering in vacuum

519

The dependency of the migration of matter on the

oxygen

partial

pressure needs some comment. The

predominant point defect in partially stabilized

zir-

mnia are double charged oxygen vacancies

Schottky-type disordered lattice. Oxygen is quickly

transported through zirconia

well

known.

But

migration of matter is controlled by the slowest

moving species. These are the cations. Creep ex-

periments with single-phase zirconia show, however,

that the creep rate is independent on the type and

concentration of the stabilizing element. But

de-

pends on the oxygen pressure

[

ll. From these ob-

servations

was concluded that the

flux

of matter is

controlled by neutral cation vacancies, e.

VZr*,

that are,

course, a minority defect species. Their

concentration can be described

full

analogy

equations with charged ions

by the

mass

law with

reference to the two

main

elements according

[V,,*

[V0*]2

const.

(2)

The constant depends on temperature and on the

oxygen pressure according to p-ln.

This

leads to the

dependency from the oxygen pressure

[V0*]

p-“4

(3)

ionic conduction via oxygen holes in

PYSZ

that is

also experimentally confirmed by

[

121. Low oxygen

pressures will therefore increase the amount of

mi-

gration-retarding neutral Zr vacancies according to

(2)

and hence retard sintering in a low oxygen envi-

ronment.

This

explanation is concluded from indi-

rect and sometimes speculative information in lit-

erature and has to be taken with restraint.

Favorite sites in

TBCs

for sintering:

In-

formation from microscopic observations

on the (scanning) electron microscope

Microstructural investigations are necessary to iden-

the location and to understand the kinetics of

sintering.. Formation of

TBC

microstructure via the

EB-PVD process and service life can be understood

as continuous dynamic processes in high tempera-

ture environments, which have fundamental

simi-

larities.

imagination of what is happening in the

course of vapor deposition

will

make the investiga-

tor sensitive for the mechanisms that rule the pro-

gressive sintering on high temperature exposure.

Three observations are reported from electron

mi-

croscopic examinations:

Inner pores of typically

50nm

diameter are

found within the main columns that have ap-

proximately

30pm

diameter. Many of the

pores are “in line” with the open boundaries

between the secondary columns

[9]

as just

mentioned in the last chapter.

Sub-micron thick secondary columns are ob-

served

all

over the primary columns but not at

the prominent top

the columns.

The “young” secondary columns in close neigh-

borhood

the top with the habit planes on it are

the most peaky having the highest aspect ratio

[

131. They seem to

usually smoother at lower

parts of the columns.

The observations suggest the following conclusions:

The pores in line with the interstices closely re-

semble the well

known

simulation experiments

of pore formation according to the three- or

multi-particle interaction models. It suggests

that neighboring high-aspect ratio secondary

columns come in contact

each other at several

contact points (probably in the course of on-

going deposition of vapor along the

flanks

the secondary columns). The contact points will

then exhibit progressive “neck formation”.

Pores will form.

This

mechanism is suggested to

be also probable

sintering.

The formation

peaky secondary columns is a

dynamic

growth

process due to shadowing and

deposition and simultaneous smoothing of the

peaks by the tendency for surface reduction.

The peaks have

grown

the quickest on top of the

columns and slower apart from the top due to

shadowing by the prominent top of the columns.

The interstices between the secondary columns

are concluded to be most effectively filled up at

the top of the columns by the tremendous vapor

flux

on high-rate vapor deposition. Thus they

form habit planes instead.

The time of formation

most peaky structures

at high temperatures is restricted as they be-

come smoothened by diffusional loss. This fact

is related to general observations that surface

atoms have their highest mobility on low radius

concave surfaces by enhanced vacancy injec-

tion. These locations of extensive surface diffu-

sion continuously attract matter from surround-

ing surface

arm

and

mainly from the peaks.

Thus the

peaks

of the secondary columns are the

most prominent points to be susceptible to sin-

tering

contact with

other

surfaces. The more

peaky they are the more effective they promote

sintering. The secondary columns at the lower

part

the main columns are “older”

than

those

near the top, and due to that they were smooth-

ened by off-diffusion over a longer time during

the processing

step.

520

Modeling

the TBC microstructure

The microstructure of EB-PVD TBCs is commonly

assumed to be composed of parallel cylinder shapes

is explicitly suggested by the denomination "co-

lumnar".

the close paclung of identical cylinders

in TBCs is assumed, similar to cigarettes closely

matched in a box, a paclung density of

78.5%

is at-

tained.

Ths

value agrees reasonably well

with

the

real packing density in TBCs. But

a representative

value for the specific area in TBCs of 5mz/g

taken

the diameter

the parallel cylinder shapes

will

only 0.2~.

This

value is one to two orders of

mag-

nitude too low. Thus

this

approach will be rejected.

better fitting conceptual design of a simple model

shape of an EB-PVD

TBC

is based on

pyramidal

forms,

given in Fig. 5b).

Fig.5:

Design of a pyramidal model shape (in 5b.)

for calculating the specific surface area (the

contribution of secondary columns ex-

cluded) of an EB-PVD TBC

in 5.a

(cross-sectional view) and 5.c (top view).

Proposed coating thickness is

400p.m

consisting of six layers of the respective

thickness h, N2,

W4,

N8, N16 and N32, the

basic plane of approximately

lOpm

being

considered solid without porosity. The con-

tribution of the secondary columns on the

overall specific surface area of as-coated

EB-PVD TBCs

twenty-fold. The model

is used to demonstrate the main impact

secondary columns on the specific surface

area.

It addresses the cuboidal habit of the columns (see

Fig. 5.c.)

well

as their tapered formation on com-

petitive growth from the vapor phase (see Fig. 5.a.).

specific surface area for the model shape, how-

ever,

turns

out to be

times less

than

the real spe-

cific surface areas measured. The divergence be-

tween calculated and measured data has to be attrib-

uted to the neglection of the surface area of secon-

dary columns in

this

concept and must be taken into

account.

the secondary columns are also assumed

to be pyramids for simplicity,

a fair coincidence

of real and calculated

data

is obtained

the interme-

diate aspect ratio of the secondary columns of the

order

20 is considered.

CONCLUSIONS FROM MODELING

ON FAVORITE SINTERING SITES

The model clearly suggests that the secondary col-

umns and their aspect ratio cannot be ignored when

discussing sintering of TBCs. They have the main

impact on the overall BET surface.

a decline of

the specific surface area on sintering can

mainly/exclusively be attributed to a reduction of the

aspect ratio of the secondary columns. The aspect

ratios on prolonged sintering can be calculated

from

the BET values and be given as a function of tem-

perature and time.

a decline from

to 0.5m2/g is

taken equivalent to

aspect ratio reduction of the

secondary columns from

to 1

2. There may

be (minor) dfierences between real and calculated

aspect ratios. Among others, they may result from

deviating cross sections

the secondary columns

from pyramidal to more lamellar forms in the as-

deposited condition that transform to more circular

shapes on high temperature exposure.

SUMMARY

Sintering of high-rate EB-PVD TBCs on alumina

substrates was investigated through specific surface

recession measurements between

700

and 1100°C

gas

adsorption methods. The specific surface re-

cession is dependent on time and temperature and

was highest at

1100°C.

700

and

800°C

an inter-

mittent increase before subsequent surface reduction

was observed.

Less

specific surface recession occurs

vacuum. Specific surface calculations suggest that

the specific surface recession is dominated by the

secondary columns that reduce their aspect ratio on

sintering.