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

This Page Intentionally Left Blank

EFFECT

OF

A1

COMPONENT ON MECHANICAL PROPERTIES

IN Al-PENETRATED ALUMINA

Sintering

Temp(C)

A1700

Sang-Woo Hong* and Sung-Churl Choi

Department

of

Ceramic Engineering, Hanyang University, Seoul

133-791

South Korea

Density Apparent porosity Reaction

(dcm')

("/)

(Y

es//No)

3.004

0.293 Yes

ABSTRACT

D1600

AIlO;/AI composites were fabricated by reacting

mullite preforms with aluminum, and their mechanical

properties were investigated according to the content of

metal phase

in

the composite and the direction of metal

penetration.

We controlled the metal content in the composite by

varying sintering temperature of mullite

performs(l700C, 1650C, 1625'C and 1600°C). Two

kinds of composites were prepared for the tests to

investigate the effect of penetration direction. One is

parallel, and the other is perpendicular to the direction of

Al

penetration. The reaction of Al with mullite preform

provided the enhanced mechanical properties to the

A1203/AI composite, such as hardness and fracture

toughness, showing mutually interconnected

microstructure. The microstructure of the composite and

the mechanical properties were affected by the

penetration direction. The mechanical properties were

also affected

by

Al

contents

in

the composite.

2.637 12.749

No

INTRODUCTION

Metal reinforced ceramics have many applications due

to the improved toughness of the inherently brittle

material.

'

But application of ceramic-metal composites

for industrial fields has been limited in the point of

machinability and near-net-shape capability, resulting in

high manufacturing costs. Recently, in an attempt to

reduce manufacturing costs,

in-situ

synthesis techniques

for composites such as directed metal oxidation

(DIMOXTM),'-4 pressureless metal infiltration

(

PRIMEXTM),' reaction bonding,s4 and liquid-solid

displacement have received considerable

attention.

Reactive metal penetration

(RMP)

processing is a kind

of liquid-solid displacement and has much merit in the

near-net-shaping of composites." Reactions between

molten

aluminum and oxides have been known

for

many

years particularly

in

relation to aluminum refining and

these aren't used as a technique to synthesize composites,

especially reaction synthesis of composites

in

the

amorphous Si02/AI system. However, it is appropriate to

fabricate ceramic-metal composites because of good

physical, and mechanical properties of products. The

study of this kind is recently made by Breslin et

a1,*

Loehman et al

,9

and Matsuo and Inabe.'

For the application of ceramic-metal composites under

real condition, it is necessary not only to optimize

fabrication processing but also to control the mechanical

properties ofthe composite.'."-'2

In this study 4 kinds of mullite performs were

prepared by varying sintering temperature to control

A1

content in the composites. Two kinds of composites were

prepared for the tests to investigate the effect of

Al

penetration direction. One is parallel and the other is

perpendicular to the direction of Al penetration. Air

atmosphere was also used for its convenient.

EXPERIMENTAL PROCEDURE

Preparation

of

AVAlzOJ

Composites

Starting materials were commercial mullite powder

and aluminum powder(S.

P.

C.

I,

Korea). Mullite powder

was attrition-milled for 24hours and uniaxially pressed.

The pressed compacts were cold isostatically pressed at

200MPa and sintered at 16OO"C, 1625"C, 1650°C and

1700C for 4 hours to vary the densities of the performs.

Table

1

lists their density, apparent porosity and

reactionability. Mullite preforms were reacted with

Al

bar, which was prepared by pressing A1 powder. A box

furnace was used for the reaction. The reaction between

mullite and Al was carried out at

1

100 'C for

5

hours in

an air atmosphere. An air atmosphere was chosen for its

convenience. The mullite perform sintered at 1600

'C

didn't react with Al, since an air atmosphere and A1 bar

used in this study restricted the wetting of A1 to mullite

perform. To investigate the effect of A1 penetration

direction, two kinds of composites sintered at 1700

"C

were prepared for the tests. One is parallel and the other

is perpendicular to the direction of Al penetration.

B1650

I

2.897

I

0.529

I

Yes

I

C1625

I

2.803

I

2.922

I

Yes

I

To

compare these composites, we made another

sample via conventional powder metallurgy. Alumina

(A1203,

99.9%

purity, Alcoa A16SG, The Netherlands)

having 0.3-0.4

W

particles was ball-milled for 24 hours

483

and then dried. The dried powder was sintered at

1600°C

for

2

hours.

Characterization

Scanning electronic microscopy(SEM) was used to

examine the microstructures of AI/A1203 composites.

A1/Al2O3 composites were etched by dissolving A1 with

NaOH solution. X-ray diffractometry

(XRD)

analysis

was carried out on representative specimens to determine

the phases which were present.

Bulk

densities of fabricated specimens were measured

by using Archimedes immersion method. Four point

bending tests were conducted with universal testing

machine (UTM)

in

which the specimens were placed

in

a

four-point bend fixture with inner span

10

mm and outer

span

20

inm and a crosshead speed of O.Smm/min.

Fracture toughness was measured, based on single edge

notched beam

(SENB)

method, and Vickers indentations

with

10

kg

load using pyramidal diamond indenter were

made on all materials to measure microhardness.

RESULTS

AND

DISCUSSIONS

Phase Analysis and Microstructure

Fig.

I

showed mullite, A1 and AI2O3 peaks detected by

X-ray diffractometry. These results indicate that mullite

preforms reacted with the A1 bar at

1

IOO'C

and created

A1203.

At

this temperature, it is sufficient for reactive

metal to penetrate into mullite preform because

aluminum's wetting angle to mullite is less than

90

degree'.' and Gibbs free energy is negative.' But the

mullite perform sintered at

1600°C

just partially reacted

at

the contact surface with

A1

because Al bar and an air

atmosphere restricted the coiiiact area between mullite

and

AI."

Contrary to other works, small Si peaks were

detected by XRD.

It

is supposed that Al bar used in this

study restricted transport of Si to the Al bar. Residual Si

did harm on AI/A1203 composites since the

Kc

of silicon

is

z

1

MPa.m"'.''

20

40

60 80

213

Fig

1.

XRD patterns

of

mullite preform and A1203/AI

composites

The metallic phase was leached out with sodium

hydroxide (NaOH) solution to observe the surface

microstructure of the fabricated AVA1203 composites. In

Fig.

2,

the microstructure of fully reacted composites

shows mutually-interconnected ceramic and metal phases,

and consists of a brighter phase indicating 3-dimensional

A1203 and darker phases occupied by Al. (a)

is

the upper

part of reacted surface, which is the plane normal to the

direction of A1 penetration. (b) is the side part of reaction

surface on SEM photograph. Alumina has a longly

connected channel shapes like pipes and the rest is

occupied with aluminum. During total reaction,

aluminum's penetration occurs very well, specially to the

one direction.

I

20kV

lkx 1O.OG

Fig

2.

SEM photographs of AI/A1203 composites (a) the

upper part of reaction (b) the side part of reaction surface

I

60

55

A

pl

50

E

%

45

2

40

35

A1

700

B1650

C1625

Fig 3. Bending strength and fracture toughness of

Al/A1203 composites

Mechanical Properties

Fig

3.

shows the bending strength and fracture

toughness of AI/A1203 composites. A

1700

sample had

the best bending strength and the lowest fracture

toughness. The strength of the composites increased with

decreasing Al contents in the composites. Fracture

toughness of the composites was better than that

of

484

alumina monolith prepared by powder metallurgy. The

reason for this increase in fracture toughness can be seen

in

the fracture surfaces from the bending tests. Fig

4.

shows

SEM

photographs of the fracture surfaces of the

composites. The composites show many projection-like

ridges and some basins which result from grain pull-put.

It

can be inferred from this result that plastic stretch

prohibited cracks and resulted in the increase of fracture

toughness composite.

Fig

5

shows the Vickers hardness. Hardness value

increased with the decrease of ductile Al phase in the

composites.

Fig

4.

SEM

photographs of the fracture surfaces

(a)

B

I650

(b)

C

1625

6.0

1

5.8

-

0

$

5.6

-

u)

Q

a

e

5.4-

m

I

$

5.2-

Y

0

5

5.0

-

4.8

J

‘

A1700

0t65Q

C1625

Fig

5.

Vickers hardness of AI/AI2O3 composites.

The direction of

Al

penetration slightly affected the

mechanical properties of the composites. The samples

parallel to the penetration showed better fracture

toughness and bending strength. It

is

supposed that the

growth of AI/Al2O3 is nearly parallel to the direction of

Al

penetration,

so

this orientation can make differences in

the mechanical properties in the composites.

CONCLUSION

The reaction of aluminum with the oxide preform

provides the enhanced fracture toughness to the resulting

composites compared with alumina. By varying the

sintering temperature of mullite, A1203/AI composites

with different Al contents can be fabricated.

1.

The reaction between mullite and

A1

depends on the

density of the mullite preform. Air atmosphere restricted

the wetting of A1 to rnullite.

So,

rnullite preform with

higher density is needed for the reactive metal

penetration.

2. A1203/AI composites with different

Al

contents were

fabricated and showed mechanical properties depend on

the

Al

contents.

3. The direction of Al penetration affected the

mechanical properties of Al2O3/AI composites. The

normal direction of the composite showed better bending

strength and fracture toughness.

able 2. The comparision of properties of,

Composites

3.43

-

3.64

253

-

280

5.14

-

5.61

,

A1203 and

..

AI/AI2O3 composites.

REFERENCES

(1) B.R. Lawn, Fracture of Brittle Solid, 2nd Ed.,

Cambridge Solid State Science Series, Cambridge Univ.

Press (1 993).

(2) A.

W.

Urquart, Novel Reinforced Ceramics and

Metals: A Review of Lanxide’s Composite Technologies,

Mat. Sci. and Eng., A144, (1991) 75-82.

(3) C.R. Kennedy, “Reinforced Ceramics via Oxidation

of

Molten Metals”, Ceram. Ind.,

12,

(1991)

26-30.

(4) Saburo Hori, Net Shape Manufacturing of CMC and

MMC by Lanxide Proocesses, Jpn. Ceramics, 32[2],

(5)

N. Claussen, T. Le and

S.

Wu,

Low-Shrinkage

Reaction-Bonded Alumina,

J.

Europe. Ceram. SOC.,

5,

(6)

S.

Wu,

A.J. Gesing, N.A. Travitzky and N. Claussen,

“Fabrication and Properties of Al-infiltrated

RBAO-based Composites”,

J.

Europe. Ceram. SOC., 7,

(1997) 93-97.

(1989) 29-35.

(1991) 277-81.

485

(7)

S.

Matsuo and T. Inabe. Fabrication

of

AI-Alz03

composites by Substitutional Reaction in Fused

Aluminum,

Jpn.

Ceramics, (1991) 222-23.

(8)

M.C. Breslin,

J.

Ringnalda,

J.

Seeger, A.L. Marasco,

G.S.

Daehn and

H.L.

Fraser, Alumina/Aluminum

Co-continuous Ceramic composite(C4) Materials

Produced by Solid/Liquid Displacement Reactions:

Processing Kinetics and Microstructure, Ceram. Eng. Sci.

(9)

R.E.

Loehman,

K.G.

Ewsuk and A.P. Tomsia,

Synthesis

of

A1203/AI Composites by Reactive Metal

Penetration,

J.

Am. Ceram. SOC., 79[

I],

(1996)

27-32.

(10)

W.G.

Fahrenholtz,

K.G.

Ewsuk, D.T. Ellerby, and

R.E.

Loehman, Near-Net-Shape Processing

of

Metal-Ceramic Composites by Reactive Metal

Penetration,

J.

Am.

Ceram. SOC., 79[9],

(1996)

2497-99.

(I

1)

Hiroaki Makino and Shigetaka Wada, Effects

of

Microstructure of Ceramics on Microstructure Induced

by Round Indenter, Jpn. Ceramics, 27[ 101,

(1

992)

943-47.

(12)

K.S.

Lee,

A

Study

on

the Contact Damage in

Silicon Nitride Bilayer, Ph.

Ds

thesis, Dept.

of

Mat. Sci.

and Eng., KAIST, Taejon, Korea (1998).

(1

3)

V.

Laurent,

D.

Chatain and N.Eustathopoulos,

Wettability

of

Si02 and Oxidized Sic by Aluminum,

Mater. Sci

&

Eng., A135, (1991) 89-94

(1

4)

T.

Watari, T. Torikai, W.-P. Tai and

0.

Matsuda,

Fabrication and Mechanical Properties

of

c(-AI20~/P-Al20~/AI/Si Composites by Liquid

Displacement Reaction,

J.

Mat. Sci., 35, (2000)

5

15-520.

(15)

I.

J.

McColm, Ceramic Hardness, Plenum Press,

New York

(

1990)

P~oc.,

15[4], (1994) 104-12.

INFORMATION

Department of Ceramic Engineering

Hanyang

University, Seoul 133-791 sung dong-gu

haeng dang-dong, South

Korea.

TEL

:

82

+

2

+

2290

-

0505

FAX

:

82

+

2

+

2291

-

6767

e-mail

:

sangwoo@dreamwiz.com

486

CAVITATION CREEP

IN

THE NEXT GENERATION SILICON NITRIDE

P.

Lofaj#

’,

S.M. Wiederhorn? G.G. Long:

P.R

Jemian? M.K. Ferber4

(l)

Institute of Materials Research of SAS, 043

53

Kosice, Slovakia

(*)

National Institute of Standards and Technology, Gaithersburg, MD 20899, USA

(3) University of Illinois at Urbana-Champaign/ Argonne National Laboratory

IL 60439, USA

(4) Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

ABSTRACT

The tensile creep behavior of a commercial grade

of

silicon nitride,

SN

281,

has

been studied in

the

temperaturerange 1350°C

-

1550”C, within the periods

of up to 10,000 h. The corresponding creep damage

has

been investigated by electron microscopy and by

anomalous ultra small-angle X-ray scattering (A-

USAXS). Failure strains were around

0.5

%.

The

creep

resistance of

this

grade was found

to

be several orders

d

magnitude greater than that of previously examined

silicon nitrides. The corresponding stress exponents

were higher than

6.

Creep damage consisted

CE

multigrain junction cavities. Strain rate

data.

fit the

cavitation creep model of Luecke and Wiederhom,

which is based on a dilatation of the secondary phase

pockets. Evolution of the pockets containing secondary

phases was

confiied

by A-USAXS. High

creep

resistance in the current material was attributed to the

suppression of cavitation via restriction of the

redistribution of the secondary phases (crystalline

secondary phases and residual glass) with higher

effective viscosity. High effective viscosity seems to

result fiom new type of sintering additives containing

Lu. Lutetium-based secondary phases may be essential

for the next generation of silicon nitrides with

the

operating temperatures up to

1500°C.

INTRODUCTION

The development of more efficient gas turbines is

ultimately constrained by the maximum operating

turbine inlet temperature (TIT) and by the availability

of the structural materials with sufficient creep and

oxidation resistance. Over the last decade, structural

ceramics, particularly silicon nitride, were considered to

be the most promising materials for the structural parts

in gas turbines subjected to high stresses and

temperatures up to 1400°C

[I].

The tests of the CGT

302 developed within the Japanese national project

“Research and Development of Ceramic Gas Turbines

(300

kW class)” demonstrated a feasibility of a low-

power ceramic gas turbine operating at 1400°C with the

thermal efficiency of 42% [2], which is twice that of

the

corresponding metallic turbines with lower TIT.

Additionally, CGT 302 exhibited a significant

reduction in the level of pollutant emissions. However,

the long-term reliability of ceramic gas turbines was not

proven. One of

the

principal limitations which

ultimately controls the lifetime, creep resistance,

is

sufficient in the current generation of silicon nitride

ceramics for operation periods over 10,000 h only to

temperatures of below 1350OC (Fig. 1) [3]. However,

the maximum TIT in the advanced high power metallic

gas turbines with thermal barrier coatings and special

cooling

of

the blades already exceeded this temperature

[4].

Thus,

the

potential advantage provided by ceramics

was significantly undermined, especially when the

reliability and price of the ceramics were considered.

Ceramics

can

be competitive only

if

they provide

significantly higher TIT which superalloys cannot

principally sustain.

The aim of

the

current work is to investigate tensile

creep behavior and creep mechanisms in a silicon nitride

grade with creep resistance of the magnitude needed

6s

the next generation of materials for structural

applications.

LW

10-7

10-8

10-9

1

n-10

-.,

1100

1200

1300

1400

Temperature,

“C

Fig.

1.

The comparison of the creep performance in different

generations

of

silicon nitride

ceramics

at

150

MPa

[3].

EXPERIMENTAL

PROCEDURE

Material Characterization

silicon nitride designated as

SN

281

*

(Kyocera Corp.,

The material studied is the hot isostatically pressed

#/

Current address: NIST, Gaithersburg, MD

20899,

USA;

e-mail: lofaj@nist.gov.

*

The use of commercial designations is for identification only and does not indicate endorsement by NIST.

487

Kyoto, Japan). It consists of SisN4 matrix grains with

the mean diameter of about 1

pm

and aspect ratio 2 to

4, and a small number of the large grains of length up to

30

pm

and diameter

of

3 -6 pm, often containing pore-

like defects and nuclei fiom seeding (Fig.

2).

Secondary

phases in the pockets

are

crystalline and they contain

Lu, which was also reported in the gas-pressure sintered

version of this grade, SN 282 [5]. The mean value

of

the four-point bending strength (3040 mm) at room

temperature is 687d26 MPa; at 1200°C it is

approximately

680

MPa, and 580 MPa at 1400°C [6-71.

Fig.

2.

The microstructure

of

the silicon nitride studied

with typical bimodal grain size distribution.

Tensile Creep Testing

The tensile tests were performed on flat, dog-bone

specimens (SR51 type [8]). The gage size of the

specimens is 2 mm x 2.5 mm and the gage length is 15

mm. They were loaded via single pin Sic pull rods,

lever

arm

and dead weight. Tensile strain was measured

in

situ

using a pair

of

silicon carbide flags suspended by

their own weight on the specimen and laser-

extensometry system [8]. The raw data were recorded by

PC in the interval of

5

min

or

15 min and averaged over

a time period corresponding to 3 to 7 data points.

Details of the testing procedure are described elsewhere

Twenty-one specimens were tested

in

air at 1350"C,

1400"C, 1450"C, 1500°C and 1550°C at the stresses

ranging fiom 150 MPa to 380 MPa and periods up to

10

000

h. Two specimens broke during loading. Three

specimens broke prematurely and the measurement

system failed in three of the tests. Data fiom these

specimens were excluded

from

consideration.

[8-

101.

Microstructure Characterization

The phase composition

of

the specimens was

investigated by X-ray difhction at 40 kV (Cu

K,,

h

=

1.54046

A)

fiom different zones

of

the bulk samples.

The interior of the specimens, revealed by grinding

away half of the specimen thickness, was used

as

a

representative composition for the evaluation of the

changes in the bulk.

The samples for the transmission electron

microscopy (TEM) studies were prepared

hm

fm

creep tested specimens. They were cut from the core

of

gage section parallel to the direction of the applied

stress. After hand grinding and polishing to -100 pm

thickness and dimpling, the foils were thinned by ion

milling at

5

kV using Ar gas until foil perforation.

Transmission electron microscopy investigations

on

carbon coated specimens were carried out at 200 kV and

400 kV.

Creep damage was investigated by a scanning

electron microscope (SEM) on the secondary

fmcttm

surfaces produced at room temperature after creep and on

the polished and plasma etched cross sections.

Anomalous Ultra SmalCAngle X-ray Scattering

Cavity size distribution and the evolution of the

secondary phase pockets were investigated using

anomalous ultra small-angle X-ray scattering on the

beam line 33ID-D at the Advanced Photon Source at

Argonne National Laboratory. The beam incident on the

sample was monitored by an ionization chamber. The

beam scattered by the sample was detected by a silicon

PIN photodiode operating in an unbiased mode. The

details of the USAXS instrument were reported

elsewhere

[

1

I].

The samples with the dimensions of 2.5 mm x 4

mm x

0.15

mm were prepared fiom the as-received

material and hm the gage and grip zones of three

specimens after creep. The grip-gage pairs of specimens

were used to eliminate the effects of initial porosity and

possible changes in the microstructure during prolonged

heat treatment on cavity and pocket size distributions.

Data collection fiom USAXS scan for each specimen

required approximately 20 min with a

5

s

counting time

per data point. The iterative method of Lake was used

to desmear the data [12].

The anomalous USAXS technique involves

measurement at -250 eV,

-100

eV, -40 eV and

-10

eV

below the

LIII

absorption edge energy, which is 9244 eV

for Lu. The presence of LuaSi207 as the secondary phase

was assumed in the calculations. The details of the A-

USAXS technique and evaluation method were reported

elsewhere

[

131.

RESULTS

Creep Behavior

Fig. 3 shows the long-term tensile creep behavior

of

the material at 1400°C under the stress of

200

MPa.

The

test

was interrupted prior

to

failure after 10,200 h.

During this period, the test was interrupted once due to

the power outage. The total strain is around

0.5

%,

however, due to the several failures in the data

collection, an uncertainty of

>

0.1% is included.

Despite that, the transient stage exceeds

6000

h azd

the

strain rate after this period is less than

1

x

10-

I/s,

close to the limit of the laser extensometer resolution. A

conservative estimate

of

the minimum strain rate at

these conditions is

7

x

lo-"

l/s. Similar transient creep

was observed at 1450°C. The short-term tests at

1550°C indicated conventional creep behavior with the

primary and secondary stages. However, these data were

affected

by the oxidation of the Sic pull rods which

form glass bubbles which interact with the laser beam.

Because of prolonged lifetime, the number of tests

performed until fmal failure was limited. Stress

dependence of the minimum strain rate fiom the tests

unaffected by premature failure is summarized in Fig. 4.

It can be seen that the stress exponent exceeds 6.

488

t

___---

7

in-11

41-

I-“

0.1

14OO0C/

200

MPa

0

I

I, I

I,

0

2000

4000

6000

8000

10000

Time,

[h]

Fig.

3.

Tensile creep curve obtained at 140OOC under the stress of

200

MPa in air indicate prolonged transient stage and the

minimum rates below

1

x 10”

Us.

IU

-

10-7

u)

\

L

i

10“

1

09

d

3i

10’0

4-

E

.-

E

10’’

.

. . . . .

.

. . . .

.

. .

.

*I

SCG

A2.2

h

*I

SCG

II

=

6.2

..--

135OOC

iooc

*..N;:m--

13OOOC

*

..-I

4OOOC

I

200

300

Applied

Stress,

[MPa]

Fig. 4. Stress dependence of the minimum strain rate in

traditional log-log representation

(*

indicates

data

not

used in the analysis).

Creep

hcture

surfaces were similar to those in

the

other silicon nitride grades with relatively large

creep

damage zone and rough morphology typical for

creep

rupture regime. Fig.

5

illustrates intergranular

!?actme

occurring during secondary

hcture

at room temperature

after creep. The

crack

propagated along the interhce

between the matrix grains and large reinforcing grain.

The shape of smaller matrix grains and the locations

CE

the multigrain junctions are visible due to the residual

secondary phase left on the facet of the large

grain.

TEM observation

of

the

creep

tested specimens

revealed very few cavities compared to the TEM studies

on earlier silicon nitrides which were deformed to larger

strains [7, 10, 14-15]. The cavities observed were

exclusively the isolated multigrain junction cavities

(Fig. 6). Cavity size is hm 100 nm to 400 nm which

is equal to the size of the secondary phase pockets.

Energy dispersive X-ray analysis confmed the

presence of Lu in secondary phase pockets and

subsequent X-ray difE-action analysis revealed the

presence of Lu2Si207

(JCPDS

cards

No.

35-326,

34-509

or

3

1-777) and LuSi2N207

(JCPDS

card

No.

33-847)

as

the dominant crystalline secondary phases in the

as-received material. The same phases were detected by

X-ray diWon

&er

creep

tests at

1400°C

and

1550°C

after

10000 h and

520

h, respectively, indicating that

the lutetium silicates are very stable secondary phase.

Fig.

5.

Intergranular fracture observed on the large grains

on

the

secondary

fracture

surfaces after creep

at

1400OC.

Fig.

6.

TEM

micrograph

of

the multigrain junction cavity

found in

the studied material after creep.

489

The results of

A-USAXS

study for the grip and gage

of

the specimen tested at 1400°C are illustrated in Table

1 and Fig. 7. The size distributions

of

voids in the

corresponding grips (pores) and gages (cavities

+

pores)

are very wide. The volume tiaction of cavities, fv, is

very low and close to the lower limit of the resolution.

The size distributions of the secondary phase pockets

have distinct

peaks

between 300

nm

and 400

nm,

which

approximately correspond to those of the void size

distributions. The total volume of each phase, which is

calculated

as

a

surface

area

below the corresponding

distribution, and the mean size corresponding to the

maximum volume of the both phases with certain size,

are shown in Table 1. The resulting volume fiaction

d

cavities, was calculated as a

difference

between the gage

and grip data.

as-

volume Mon mea size

f

n

SPP

I

voids

SPP

I

voids cavities

0.065

I

0.0027 351

I

378

received

#1

grip

#1

gage

I I I I

I

#2

I

0.057

I

0.0029

I

350

I

430

I

nm nm

0.040

0.0050

421 363

0.033 0.0064 389 360 0.0014

grip

#2

gage

I

I I

I

Table

1.

Summary of the

U-SAXS

data

from the specimens

tested

#1

-

at 1400"C,

150

Ma for 2804

h,

and #2

-

at

14OO0C,

200

MPa for

215

h

(

"SPP" is

used

for the

secondary phase pockets, "voids" are related to the sum of

the initial porosity in the as-received material and

cavitation after creep).

0.046 0.0032 365 377 0.0003

DISCUSSION

Direct comparison

of

the creep data from the material

studied with those from the earlier grades of silicon

nitride is not possible because of higher temperatures

and stresses required to produce measurable creep.

Therefore, strain rates were extrapolated to 150 MF'a and

plotted in Fig. 8 using a temperature

dependence

similar

to

that in Fig. 1. This comparison indicates that

at the strain rate of -3 x 10"' Us under the

stress

of

150

MPa, the strain

of

1% is reached

after

10000

h at the

temperature

of

approximately 1500°C. This temperature

capability is at least by 180°C higher than that of NT

154.

A

classification

of

the materials based on their

mq

resistance is implied from Fig.

8.

The first generation

d

silicon nitride materials containing MgO additives

exhibited sufficient

creep resistance (normalized to 150

MPa) at temperatures below 1000°C. Second generation

of silicon nitrides containing

YAG

secondary phases

was resistant up to approximately 1100"

-

1150°C.

Introduction of pure rare-earth oxide addltives such

as

Y203

and

Yb203,

improved

creep

resistance and

increased the corresponding temperature up to 1300°C

-

1325°C.

-1

h

a"4

iz

0

Fig. 7. The volume fraction size distribution,

f

(D),

of the

secondary phase pockets and voids in the grip

-(A)

and

in

the gage part

-(B)

of the specimen crept at

1400°C

for 215 h.

The current silicon nitride which can operate up to

1500"C, can be conditionally considered as the material

of

fourth

generation. Apparently, the main reason for the

improvement in this case is related to the Lu containing

secondary phases originating from sintering additives

(probably LuzO~). However, an understanding of the

mechanisms controlling creep and of the

Wmca

between different materials are necessary to

confirm

such

assumption.

A number of recent creep studies on

difkent

silicon

nitrides emphasized

the

role of cavitation among the

process contributing to tensile creep strain [6-7, 9-10,

13-18].

A

linear dependence between the strain and

volume density change (volume hction

of

cavities) [7,

9-10, 15, 181, sound velocity change [6], Young's

modulus change

[

171, a method based on the

difference

between tensile strain and compressive strain

[

161 and

direct measurement

of

cavitation by

USAXS

in NT 154

[15] and SN 88 [13, 181 confirmed that cavitation at

multigrain junctions

are

responsible for more than 90%

of the tensile strain. Thus, cavitation was concluded to

be the main

creep

mechanism in silicon nitride and

possibly in other two-phase materials [7,9-10, 161.

High stress exponents in

the

current material (Fig.

4) support a possibility of cavitation creep. TEM

observation indirectly implies a correlation between

strain and cavitation: small strains correspond to very

low concentration of cavities. Similarly to other

materials, only multigrain junction cavities were found

Their size is comparable to the size of multigrain

490

junctions filled with secondary phase. Thus, despite

limited data imposed by low strains, it

can

be

concluded that cavitation is the main

creep

mechanism

in the studied material, similar to that of materials with

lower creep resistance

[7].

This conclusion is supported

by the

A-USAXS

data (Fig.

7)

where

the

volume

weighted cavity size overlaps with the size distribution

ofthe pockets. The

A-USAXS

data (Table 1) indicates

that cavitation may

occur

via redistribution of the

secondary phase between the secondary phase pockets,

since a correlation between the volume fraction

of

cavities and secondary phase is found an increase in

cavity content results in a corresponding

decrease

of the

secondary phase content (Table

1).

Such mechanism

was assumed in the recent cavitation

creep

model

i?r

silicon nitride proposed by Luecke and Wiederhorn [9].

Luecke and Wiederhorn cavitation

creep

model

i?r

two-phase materials is based on the effect of dilatation

d

the primary phase (silicon nitride) when subjected to

deformation. It occurs via grain boundary sliding and

results

in

expansion of the pockets with

soh

secondary

phase. Hydrostatic tensile stresses generated in some

d

the pockets are high enough for cavity nucleation.

Consequent cavity growth occurs via redistribution

of

the secondary phase fiom the pockets with cavities into

neighboring pockets and it is driven by the stress

gradients between those pockets. The minimum strain

rate, M6t, calculated for such mechanism is [9]

&/at

=

B

(l/q)

(T

exp (-Q/RT)exp(ao)

(1)

where B and

01

are the coefficients of proportionality,

q

is the effective viscosity and

Q

is the activation energy.

This equation assumes exponential dependence on

stress which is different fiom conventional power law.

Such dependence corresponds to the stress exponents n

-

1 at low stresses and n

>>

1 at high stresses [9]

which is indicated in Fig.

4.

Experimental data hm

this Figure are replotted according to the Eq. (1) in Fig.

9.

Because of limited stress range and number of data,

both exponential and power law

fit

the data equally

well. However, the stress exponents of

6

cannot

be

explained based on conventional dihion creep models

while they are directly involved in cavitation

creep

model.

The principal processes involved in cavitation

creep

are schematically shown in Fig. 10

(A)

and Fig. 10

(B).

In idealized case of cavitation at multigrain junctions

without dilatation (Fig. 10

(A)),

cavity formation would

correspond

to

the removing of the secondary phase hm

the pockets and out of the solid. However, matter is

conserved in a real situation and the secondary phase

fiom the cavities is redistributed between the remaining

uncavitated pockets (Fig. 10 (B)). This inevitably leads

to the dilatation of the uncavitated pockets which is

accompanied by a sliding apart

of

the

hard

grains

cf

primary phase. Thus, the redistribution of the secondary

phase hm cavitated to uncavitated pockets is the

ultimate controlling process of tensile

creep

deformation.

Redistribution rate is determined by the

effective

viscosity of the secondary phases,

q,

(see Eq.

(1)).

Physical meaning of the effective viscosity depends on

the possible mechanisms for the redistribution. Since

the term “secondary phase” is related to crystalline

secondary phases and low amount of residual glass,

both solution-precipitation and viscous flow

can

be

considered. However, it is difficult to determine which

of the processes is

rate

controlling

-

dissolution

of

the

crystalline phase, its diffusion via amorphous boundary

phase, its precipitation or viscous flow of the glass.

Thus, the term “effitive viscosity” is related to the

solubility and diffusivity of the crystalline secondary

phases and viscosity of the residual glass as well.

Stepwise increase in the effective viscosity

of

the

secondary phases based on different types of additives

provides an explanation for the variations in

creep

resistance between different generations of silicon nitride

while the other

parameters,

such as activation energy

and stress dependence, remain the same (Fig. 9).

Despite the absence of the experimental data on the

effective viscosity of different rareearth based crystalline

secondary phases, it can be concluded that lutetium

based secondary phases exhibit higher effective viscosity

than

other rare-earth based oxides.

1100

1200

1300

1400

1500

1600

Temperature,

“C

Fig.

8.

A

comparison of the creep resistance for different

silicon nitrides with that of material studied depending

on

the temperature and sintering additives.

3

10-9

G

b)

10-l0

lo-”

150

200

250

300

350

Applied Stress,

MPa

Fig.

9.

Stress dependence

of

the minimum strain rate in

SN

281

described according to the cavitation creep model

of

Luecke and Wiederhorn

[9].

49

1

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

Подождите немного. Документ загружается.