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

Подождите немного. Документ загружается.

EXPERIMENTAL PROCEDURE

The powder characteristics used

fabrication of the

composites are shown

Table 1. Several Ni,Al alloys

were included and are referred to by the IC- designation

given

alloy development studies. Note the large size

of the Ni,Al and NiCr powders produced by inert gas

atomization compared to the Tic powder. These size

differences have an effect on the densification behavior

as will be discussed later. The samples were fabricated

by milling fine Tic powder, with the metal powders at

appropriate levels to produce composites with

vol.

binder phase. The milling was done in

isopropanol for 19 h using WC-Co milling media and 1

wt.

polyethylene glycol (Carbowax

8000,

Union

Carbide, New York) added as a binder. Media wear

during milling contributed

-0.5

wt.

to each of the

compositions. The mixtures were dried and screened to

100 mesh. Specimens were uniaxially pressed in either

mm diameter steel dies at

-100

MPa

(15

ksi).

Sintering was done

a graphite element furnace at

temperatures from 1400°C to 1500°C. The heating

schedule consisted of a ramp of 10"Clmin from room

temperature to 1200"C, a

0.5

h hold at 1200°C to allow

for any degassing, and another ramp at 10"Clmin to the

final sintering temperature all under vacuum. The

temperature was maintained at the sintering temperature

for

0.5

h under vacuum followed by an argon gas

pressurization to

MPa

(150

psi) in 10 minutes and a

hold under pressure for

minutes. The total time at the

sintering temperature was

minutes.

For all of the test samples, densities were determined by

the Archimedes' method. For mechanical property

testing, selected samples of high density were machined

into bend bar specimens with nominal dimensions of 3

4 mm

mm. Flexural strength testing was

done in four point bending with innerlouter spans of 20

md40 mm. Fracture toughness was determined by both

an indentation and indentationlfracture method

[

12,131.

Hardness testing was done with a Vickers diamond

indenter at a load of

kg. Scanning electron

microscopy (SEM) was done on polished sections using

back-scattered electron (BSE) imaging and with energy

dispersive x-ray analysis (EDAX). The corrosion

resistance was determined by measuring the weight loss

during immersion

a 1N acid solution at 25°C after

100

Table 1. Physical characteristics of powders.

Powder SuppliedGrade Ave. Particle

TY Pe Diameter

(PI

Tic Kennametal (Latrobe, PA)/Grade 2000

1.3

NiCr Alfa Aesar (Ward Hill, MA)/ Ni:Cr, 80:20 wt.% 44-90

NiCrFe Alfa Aesar (Ward Hill, MA)/ Ni:Cr:Fe, 72:14-17:6-10 wt.% <44

<44,<75

<44

<65

Ni,Al Homogeneous Metals (Clayville, NY)

IC-50 (Ni-

1.3 A1-0.6

Zr-0.02

IC-218 (Ni-8.7 A1-0.2 Zr-8.1 Cr-0.02

IC-264 (Ni-8.4 Al-1.7 Zr-7.8 Cr-0.02

IC-396M (Ni-8.0 A1-0.9 Zr-7.7 Cr-3.0 Mo-0.01

RESULTS AND DISCUSSION

DensiJication

The densification results are summarized

Fig.

As indicated, high densities were obtained for

most compositions at temperatures of 21400°C. These

types of composites are densified by liquid phase

sintering

(LPS)

and the relatively high liquid contents

allows particle rearrangement to proceed rather easily.

The effect of different Ni,Al starting particle sizes on

densification can be seen by comparing the 40 vol.

IC-

composites

Figs.la and

lb.

The Ni,Al powders

had particle sizes of

<44

pm (Fig. la) and

<75

pm (Fig.

b). The larger sizes inhibited densification at 1400"C,

but at 1450°C the effect was negligible. Prior work has

shown that as the Ni3A1 melts, the liquid is drawn into

the surrounding Tic powder by capillary action. This

results

the formation of a large void that is difficult to

eliminate. With the larger metal particles, this effect is

exaggerated.

Both the IC-50 and IC-218 based composites exhibited

high densities with the V-LPHIP firing schedule. On the

other hand, the composites with IC-264 and IC-396M

612

alloys, exhibited decreased densification. Both these

alloys contained higher

contents than the other alloys

and, in the case of IC-396M, Mo was also present.

These amounts of

should have only a minor effect on

the solubility of Tic and diffusion in the liquid phase and

thus should not seriously affect

LPS.

However, the

would act as a powerful oxygen getter in the system due

to the high free energy of formation and stability of

ZrOz.

This oxygen gettering could significantly

influence the wetting between the Tic and the liquid

phase and thus affect the solution-reprecipitation

kinetics. Small amounts of surface oxygen are well

known to adversely affect the wetting between carbides

and molten metals and in a similar fashion, minor

additions of Mo to Ni have been shown to reduce the

contact angle with TIC from

17"

0".

Previous results

have suggested inhibited sintering with

additions to

Ni,Al alloys [9,10]. In the case for the composite

utilizing IC-396M as the binder phase, the larger size of

this powder (<65 pm) may have also been a factor in the

decreased densification behavior. The non-aluminide

alloys (NiCr and NiCrFe) produced composites with

high densities even at moderate temperatures.

100

1400

1450

Sintering Temperature

1500

1400°C

1450°C

1500°C

Sitering

Temperature

(“C)

(b)

Fig. 1. Summary of densification results: (a) composites

at different Ni,Al contents using IC-50 (<44 pm); (b)

composites containing 40 vol.

of various binders.

Microstructure

Representative microstructures of

composites containing 40 vol.

binder are shown

Fig.

The Tic particles exhibit a ‘core-and-rim’

morphology caused by reaction of the liquid metal with

the Tic and the precipitation of secondary Tic during

grain growth. Such ‘core-and-rim’ structures are

typically observed

Tic-Ni cermets. Energy dispersive

x-ray analysis of the core-rim structure revealed several

observations that were similar to previous results on TiC-

Ni-Mo cermets. The central core region consists of the

original TIC with no detectable other components. In the

rim region, other carbide formers are observed, such as

Cr, Mo, and

The

is from the wear of the milling

media during processing, whereas the Cr and

come

from the constituents

the various binders. Preferential

accumulation of the carbide formers

the rim structure

on the Tic has been observed in other studies [14]. As

shown in Fig. 2b, rounding

the grains was observed

with the sample containing the NiCrFe binder.

However, such rounding behavior was also associated

with all of the Ni,Al alloys with Cr

Mo additions A

typical fracture surface of the composites is shown

Fig.

Necking and deformation of the Ni,Al around the

Tic grains can be seen on the fracture surface. In

addition, debonding along the TiC-Ni,Al interface was

predominant.

(b)

Fig.

BSE

images of microstructures of (a) Tic-40

Ni,AI (IC-50)

(b) Tic-40

NiCrFe. Both sintered at

1500°C. Tic particles exhibit a ‘core-and-rim’

morphology.

Fig.

Fracture surface of Tic-40

Ni,Al (IC-50)

composite sintered at 1450°C.

Mechanical Properties

The mechanical properties were

determined for composites that achieved a density

295%

A summary of the room temperature flexural

strength results is shown in Fig.

As indicated, the

strength generally increases with increasing binder

content. The IC-218 alloy sample exhibited slightly

higher strength at comparable binder contents and

densities probably due to the slightly finer grain size of

these composites with some contribution from solid

solution hardening

the matrix. The majority of the

strength limiting flaws for all the composites in the

present study was the occurrence of metal “pools” as

shown

Fig.

These metal accumulations are most

likely the result

the large particle size of the starting

613

raw materials. The high temperature strength of the

composites containing IC-50 Ni3Al alloy are shown in

Fig.

As indicated, the strengths are retained

actually increase up to 500°C. This behavior is a direct

effect of the increase in yield strength of the Ni3Al

binder.

Like the strength, the fracture toughness generally

increased with increasing Ni,Al volume content as

shown in Fig.

Such behavior would be anticipated

based on the ductility of the binder phase. In all cases,

the fracture toughness was

K,,

12 MPadm and for the

composite with 50 vol.

Ni3Al and fabricated with

prealloyed Ni,AI powders was as high as

MPadm.

These values are exceptional and were one of the reasons

for choosing these types of composites for development

for diesel engine applications. The high toughness

values are a result of the plastic deformation and crack

bridging effects of the Ni,Al binder. These effects are

illustrated

Fig.

(and also shown

Fig.

previously). Fig.

shows the crack behavior at the tip of

the an indent introduced during hardness and toughness

testing. Crack bridging is readily evident in Fig.

and

the crack length is surprisingly small considering the

indent load was 50 kg.

The indent hardness was measured on several specimens

sintered at different temperatures and is summarized in

Fig.

As shown, the hardness decreased with

increasing Ni,Al content. The IC-218 alloy sample

exhibited slightly higher hardness probably due to some

contribution from solid solution hardening of the matrix.

The corrosion behavior of selected composites in the

study is summarized in Fig. 10. The NiCr-based binders

are currently used

corrosion resistant cermets and,

general, their behavior was better than the Ni,Al-based

composites

the nitric and sulfuric acid solutions.

However,

the hydrochloric acid solution, Ni,Al-based

composites showed slightly better resistance Thermal

expansion behavior of different composites is shown

Fig 11. As expected, increasing the Ni,AI volume

content increases the thermal expansion coefficient.

Comparison with the NiCr sample, reveals that the

Ni3Al-based composite has a lower expansion coefficient

for the same binder content. The expansion coefficients

for the composites

the study are similar to those of

steel which is

-1

1.5X10-6.

1200

, , , , ,

1000

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

800

600

400

200

IC-50140 IC-21W40 NCrFd40 IC50/30

Fig. 5. Fracture surface of Tic-40 vol.

Ni,A1

composite sintered at 1450°C.

1300

, , , ,

900

........

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

800

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

......

700

200

400

600

Temperature

(“C)

Fig.

Summary of flexural strength results on TiC-

Ni,Al composites sintered at 1450°C.

IC-5W40

IC-218140

NiCrl40 NiCrFd40 IC-5W30 IC-50150

Binder

TypeNd. Content

(%)

Fig.

Fracture toughness results on TIC-Ni,A

composites sintered at 1450°C and 1500°C.

Binder TypeNol.

Conlent

(%)

Fig. 4. Summary of flexural strength results on TiC-

Ni,Al composites sintered at 1450°C.

Fig.

SEM image of microstructure of Tic-50 vol.

Ni,Al sintered at 1500°C. Crack bridging by the binder is

evident at indent tip (top

photograph).

614

(I,

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

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

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

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

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

IC-50

IC-218 NiCr

Binder

Type

NCrFe

Fig. 9. Summary of indent hardness results on Tic-Ni,AI

composites sintered at 1450°C and 1500°C.

100

........

I,,

...

......

0.1

IC-50

IC-218 NiCr NlCrFe

Binder

Type

Fig. 10. Weight

loss

for Tic-40 vol.

composites immersed

1N acid solutions.

binder

vol.

IC-50

NiCr

Birder

Canlent

and

Type

Fig. 11. Thermal expansion coefficients

for

selected TIC

composites sintered at 1450°C.

CONCLUSIONS

Tic-Ni,AI composites are under development for

application in diesel engines because of desirable

mechanical and physical properties,

addition to good

corrosion and wear resistance. Liquid phase sintering of

the composites with 30 to

vol.

Ni,AI results in high

densities being achieved for sintering temperatures

21400°C. The final microstructures consist of Tic

grains and a surrounding matrix of a (Ni,Ti),AI. The

Tic grains are made up of a core region of pure TIC and

a rim region of a Ti(Cr,Mo,W)C solid solution

precipitated from solution during liquid phase sintering.

Room temperature flexural strengths ranged from >800

MPa up to 1150 MPa for volume contents from 30 to

50%.

These strengths were retained

actually

increased up to temperatures of at least 500°C. The

fracture toughness values were all >12 MPadm and as

high as 27 MPadm at Ni,AI volume contents from 30 to

vol.

respectively. The thermal expansion

coefficient of the Tic-40 vol.

Ni,AI composite is

similar to that for steel.

ACKNOWLEDGMENTS

Research sponsored by both the Propulsion System

Materials Program, DOE Office of Transportation

Technologies under contract DE-AC05-000R22725

with

UT-Battelle. The research also used the ORNL SHaRE

User Facility supported by the Division of Materials

Sciences, U.S. Department of Energy.

REFERENCES

T. N. Tiegs, et al, Mater. Sci. Eng.,

Vol.

A209, No.

2. T. N. Tiegs, et al, pp. 211-218 in Internat. Conf. on

Powd. Metall., Met. Powd. Indus. Fed., Princeton, NJ

(1995).

P. Plucknett, et al, Ceram. Eng. Sci. Proc.,

4. T. N. Tiegs, et al, pp. 339-357

Internat. Symp.

Nickel and Iron Aluminides, ASM International, Metals

Park,

OH (1997).

Schneibel, et al, pp. 329-338

Ref.4.

Subramanian and

Schneibel,

Metals, 49[8]

7. T. N. Tiegs, et al, Ceram. Eng. Sci. Proc.,19[3] 447-

455 (1999).

P. Plucknett, et al, J. Mater. Res., 12 [lo] 2515-

2517 (1997).

9. T. N. Tiegs, et al, Ceram. Eng. Sci. Proc.,Am. Ceram.

10. T. N. Tiegs, et al, Adv. Powd. Met. Partic. Mater.-

1999, Metal Powder Industries Fed., Princeton, NJ

(1999).

11. D. E. Wittmer, et al, pp. 237-248

Adv. Powd. Met.

Partic. Mater.- 1999, Metal Powder Industries Fed.,

Princeton,

(1999).

12.

Anstis, et al,

Am.

Ceram.

SOC.,

[91 533-

538( 1981).

13. P. Chantikul, et al,

Am.

Ceram.

SOC.,

[9] 539-

543 (1981).

2,243-47 (1996).

17[3]314-321 (1996).

50-54 (1997).

SOC., 21[3]0721-728 (2000).

615

This Page Intentionally Left Blank

CERAMIC JOINTS BETWEEN Sic BODIES:

MICROSTRUCTURE, COMPOSITION,

AND

JOINING STRENGTH

Woltersdorf,

Pippel, P.Colombo*

Max-Planck-Institut flir Mikrostrukturpbysik, Halle, Germany, *Universita di Bologna, Italy

Introduction

Sic-based ceramic composites are promising

structural materials because of their excellent heat

resistance and mechanical properties. Moreover, silicon

carbide fiber (SiCf) reinforced composites are of

specific interest for future thermonuclear fusion reactor

applications as they exhibit a low radioactivity by

neutron transmutations, a high stability after the reactor

shut-down, good high- temperature properties, a low

plasma contamination and a low specific weight

order to fabricate large pieces or complicated

structures, suitable joining techniques have to be

developed. While the joining of monolithic Sic has

been accomplished by using different techniques as,

e.g., active metal brazing, solid state joining and

reaction bonding

the joining of SiC/SiCf composites

has

not been extensively reported in the literature

3,4.

The used joining techniques often require

conditions which are not compatible with the upper

temperature of about 1 200”C, which currently limits

the operation of SiC/SiCf composites owing to the

degradation of the Sic fibers used

far’ and with

respect to the requirements of fusion technology.

Brazing with metallic fillers is not applicable because

of the heavy nuclear transmutability of the fillers.

Therefore, we developed an alternative joining method

using preceramic polymers like polycarbosilane or

polysiloxane

6,7,

which yield a Sic or SiOC ceramic,

respectively, when heated in an inert atmosphere

8,9.

The present paper reports the results of

microstructural and microchemical investigations of

the interlayers in such polymer derived joints between

SiC/SiCf bodies as revealed by transmission (TEM)

and analytical (AEM) electron microscopy in addition

to some data which characterize their mechanical

properties including those of joined monolithic

The study was performed in order to clarify the

interlayer mechanisms determining the joining process.

Experimental

Monoliths of sintered Sic were provided by

Associazione EURATOM-ENEA, Italy, and the two-

dimensionally reinforced SiC/SiCf composite material

was from Societ6 Europ6enne de Propulsion, France.

The surface structure

the latter material was rather

complex (total roughness

-25

pm,

average peak-to-

valley height

130 pm).

improve the surface

quality all specimens were mechanically polished on

one side, using a

pm Sic granulation, and

ultrasonically cleaned in acetone prior to the joining.

Another feature of the SiC/SiCf composite material is

that it possesses a thick

100

pm)

over-coat of CVD

Sic which was not completely removed by the

mechanical surface preparation. Thus the fibers were

not exposed and did not participate in the joint

formation. The preceramic polymers used for the

joining experiments were a

methyl-hydroxyl-siloxane

(SR350, General Electric Silicone Products Div.,

Waterford,

NY)

and polycarbosilane (PCS, Dow

Coming X9-6348) which were homogeneously applied

to the surface of the specimens to be joined.

Subsequently, some samples were loaded with an axial

pressure of 9.8.10” to 4.10-2 MPa, and heated at 200°C

for 2 hours in order to achieve the complete

crosslinking of the silicone resin or to cure the

polycarbosilane. Then, the samples were heated in an

argon (99.99

YO)

flux for 1 hour at temperatures

between 900 and 1200°C. The heating and cooling

rates were very slow (1Wmin)

to minimize possible

residual stresses due to the thermal expansion

mismatch. In some experiments, commercial titanium

(Cerac Inc. T- 124 1, average grain diameter 15

pm)

p-Sic

(H.

Starck B10, average grain diameter

0.3

0.5

pn)

powders were added to the polymers, in a weight

ratio of

0.5

0.7

to 1 to act as fillers to lower the

polymer shrinkage during the pyrolysis.

Microstructural and nanochemical investigations

were performed on joints between sintered SiC/SiCf

composites, processed with

SR350

preceramic polymer

and applying optimum fabrication conditions (cf. next

section). The examinations of the interfaces between

composites and joining material were carried out by

using a Philips CM 20 FEG (TENSTEM) electron

microscope operated at 200 kV and equipped with a

Voyager EDX system, which enables an atomic lattice

plane imaging (HEM-mode) and a chemical analysis

(element mapping) with a spatial resolution

a few

nanometers. The TEM cross section specimens were

prepared by standard techniques with final Ar-ion

milling to electron transparency. These specimens

allow an observation of the composites, the joining

layers, possible reaction layers and interfaces side by

side in one and the same specimen. For details

concerning the special problems of TEM investigations

of interfaces in ceramics and Si-C-0 composite

materials including the elucidation of structure/property

relations, see, e. g., refs. 12-19.

Results

All the images reported

are

to be considered typical

and representative of the interfacial region of each

617

sample. To get an overview of the interfaces to

examined and of their special features

mentioned

above, Fig.

shows a light microscopic image of joint

SiC/SiCf composites. Clearly the fibre structure (above

and below), the

thick over-coat of CVD Sic

(light gray) and the almost flat interface joing layer can

seen.

(right). Contrary to this, the Sic is free of defects at a

distance of

away

from

the interface. While the

joining layer is amorphous in its interior, the high

resolution image reveals a precipitation of small

amounts of

carbon

near

the

interface (cf., the

0.34

graphite atomic planes in the almost amorphous joining

layer of Fig.

right).

Fig.l: Cross-section

joint SiC/SiCy (light

microscopic image).

The amorphous joining layer

SiC/SiCf composites

investigated

free

of pores and

has

a thickness

5-6

pn.

shown in th TEM image of Fig.

the interface

between layer and SIC matrix of the composite is very

flat again, with a

thin zone of a destroyed

Sic structure next to the interface.

Fig.

Flat interface between Sic (upper IeB) and the

joining layer with

25-40

nm zone

destroyed Sic

structure next to the inte$ace

The latter is demonstrated in the high resolution lattice

plane image of Fig.

while in the interior of Sic (left)

the

{

1 1

atomic planes of Sic regularly propagate

over large areas, a lot of crystal defects (stacking faults

and/or microtwins, characterized by an interruption and

slight change in the direction

the atomic planes)

have formed near the interface to the joining layer

'Fig.

HREM

image

the interface between Sic (lej)

and joining layer with crystal defects in the

Sic,

notice

the small graphitic particles in the amorphous joint

(right)

In the present stage of investigation it is difficult to

infer that the SiC/SiCf composite surface has any effect

on the development

a weak graphitic boundary layer

13,

it is well

known

that silicon oxycarbide glasses

contain some turbostratic carbon

Together with the CM

electron microscope the

Voyager I1 EDXS-equipment allows the recording

X-ray spectra and element line scans from selected

specimen areas with a resolution of some nanometers

depending on the specimen thickness.

For

a proper

evaluation of the following X-ray measurements it

should be considered, however, that fluorescence

phenomena and beam spreading effects in thicker

specimen regions

at contamination hillocks will

deteriorate the real spatial resolution of the chemical

analysis although the electron probe has an actual

diameter of only 1.5

nm.

further uncertainty in

recording the elemental profiles may

due to a minor

overlapping of the materials at the interface. Regarding

this, in all the specimens investigated the chemical

composition

the joining interlayer proved to be

nearly the same

with

only

silicon,

oxygen

and

a small

amount of carbon, i.e., a silicon oxycarbide glass has

formed. Fig.

demonstrates a linescan across the

interface of a SiC/SiCf joint presenting a situation

similar to that of Fig.

a small electron probe was

stepped along a definite line of

discrete points, with

each recording

complete EDX-spectrum. The

corresponding X-ray emission of carbon, oxygen and

silicon from Sic (left) to the joining layer (right) is

shown, with an overall line length of

226

(in the

linescan the position of the interface,

visible by

TEM, is marked by a perpendicular line). While some

part of the carbon signal results from contaminations in

the microscope, the silicon X-ray intensity roughly

618

represents the specimen thickness decreasing from Sic

toward the interlayer owing to preparation. The oxygen

linescan, however, clearly reveals almost no diffusion

from the joining layer into Sic (expect within a range

-Si

Fig.

X-ray line scans of Si,

across the interface

in SiC/SiC’, length:

226

nm, obviously no oxygen has

difised into the matrix

nm).

The small 0-signal in the Sic region

(left) probably results from surface oxidation of Sic

during the preparation procedure.

The quality of the joints was determined by

kture

shear stress tests following the ASTM

D905-89

test

procedure, using a crosshead speed

mdmin. Even

if a pure shear

stress

field does not necessarily occur

this testing technique, it is, however, a suitable means

of comparative evaluation. Fig.

represents the

fracture stresses obtained for the joining of a-Sic

monoliths.

50,

I--

€

SR3G

sR35ocn

--c

sR3s

PCS

pcS+SiC

PCS

900

loo0

1100

luw)

Joining

Temperature

(“C)

Fig.

Fracture shear stresss for a-SiC monoliths

joined at diyerent temperatures

It is obvious that using solely pure silicone resin

(SR

350)

as joining material results in a sufficient

mechanical strength of the structures obtained. The

shear strength increased with the joining temperature,

and,

for the samples heat treated at 12OO0C, most of the

failures did not occur in the joint region but in the

Sic monolith. PCS was not effective as a joining

material. At every joining temperature tested, the

specimens exhibited a mechanical strength almost not

measurable, independently of the thicknesses of the

joint layers (ranging from about 20

up to 200 pm

after curing). Because of the mentioned considerable

shrinkage and density change to which preceramic

polymers are subject due to the polymer-to-ceramic

conversion

68,

we expected that the introduction of

fillers would have helped to reduce this volumetric

variation, which could have adverse effects on the joint

and its strength. In fact, for pure

SR350

resin, the joint

thickness

reduced by a factor of five to seven during

the heat treatment at 1200°C. However, using inert

(Sic) or active (Ti) fillers degraded the mechanical

strength of the joints. A possible explanation is that

with the introduction of these powders the joint

thickness (measured after crosslinking) increased from

about 20

to about 32

(for

SR350

Sic powder)

and to about

(for

SR350

Ti powder). For PCS-

joined specimens, the limited increase in strength with

the introduction of powders into the joining material

could be due to a more homogeneous layer present

there. The following table lists the fracture stresses (in

MPa) obtained from the SiC/SiCf joints:

Joining conditions

000°C 1200°C

SR350,

no pressure, unpolished

0.2

1.4 6.3

SR350,

no pressure, polished

SR350,

pressure=200g/cm2

9.9

18.0

SR350,

pressure=400g/cm2 13.5 21.6

SR35O+Ti, unpolished

0.7

3.1

SR350+Ti, polished

3.5

The increase of the joint strength with temperature is

again obvious. Moreover, applying pressure during the

formation of the joint is certainly advantageous. As for

monolithic a-Sic, PCS does not afford the formation

of proper joints. Mechanically polishing the SiC/SiCf

composite was proved to

necessary,

this probably

helped to create flat areas where joining could take

place

’.

The introduction of powders improved the joint

strength only for unpolished specimens, because of the

more effective filling of the surface roughness by the

joining material.

Discussion

The microstructural and nanochemical characteri-

zations show that the pyrolytic decomposition of

silicone resin produces a rather homogeneous silicon

oxycarbide glass layer, the thickness of which seems to

depend on the type of substrate joined. The

investigations of the interfaces between the SiOC glass

and the various Sic bodies indicate that the joining

material is absolutely non-reactive, at least at the

processing temperatures used, as there are neither

reaction layers visible nor a chemical attack of Sic near

the joints. Furthemore, the mechanical polishing of the

substrates produced very flat interfaces. All these

619

particulars allow one to conclude that here the joining

process is similar to adhesive bonding

lo**.

The joining

mechanisms do not seem to involve any mechanical

interlocking. They rather imply a chemical bonding

with the direct formation of bonds between the Sic

bodies and the inorganic adhesive material.

The most typical microstructural feature of the

interfaces is a 30 to 100

extended zone of enhanced

diffraction contrast in Sic near the interface resulting

from a destruction of the Sic lattice by mechanical

strain and/or crystal defects. According to results of

micro-indentation tests at the joint interfaces we can

exclude the presence of residual stresses in the

materials

that this zone can be attributed to the

mechanical grinding/polishing of the

Sic

parts prior to

the joining. The lack of residual stresses is certainly

related to the low coefficient of thermal expansion

(CTE) mismatch between Sic and SiOC materials

lo.

has to be pointed out that all studied interfaces appear

smooth on the scale observed, but a smootheqing of the

interfaces during the bonding process seems quite

improbable, considering the comparatively low

temperature used for joining. Furthermore,

clearly

shown by the microchemical analyses, no evident

compositional gradient was observed in the glassy

layer and in the Sic material in the proximity of the

interface. Moreover, no growth or dissolution of the

Sic material in the glassy interlayer was observed.

The difference of the mechanical properties

between joints of a-Sic and joints of SiC/SiCf bodies

obviously is primarily not due to different

microstructural

microchemical interfacial features. It

might rather

due to the variation

the thickness of

the joining layers: the SiC/SiCfjoint is thicker than

Sic joint.

Since oxygen does not seem to play an important

role in the joining mechanism, the difference in the

bonding ability between silicone resin and

polycarbosilane preceramic polymers might be

explained by their different decomposition behaviour

upon pyrolysis. First of all, the weight loss during

pyrolysis for pure PCS is considerably higher

than

for

SR350 (PCS ceramic yield

-55%,

SR350 ceramic

yield -75%,

'),

possibly resulting in the formation of

voids in the joint.

Conclusions

Joints between Sic bodies were successfully

produced using a silicone resin. In the case of joining

monolithic a-Sic parts a much higher strength was

attained than for joints of SiC/SiCf composites.

Microchemical investigations showed that the joining

material consisted of a silicon oxycarbide glass, and

that no oxygen diffusion occurred between the glass

and the joined Sic bodies. The flat interface structure

and the lack of any reaction layer suggest that the

joining mechanism involves the direct formation of

chemical bonds between the Sic bodies and the joining

material. The interfaces between joint and Sic material

were similar for a-Sic monoliths and SiC/SiCf

composites. The lower strength of the SiC/SiCf

composite joints is probably due to the uneven surface

morphology of the composites. Polycarbosilane was

not effective

a joining material, possibly because of

its different decomposition behaviour upon pyrolysis.

The introduction of Sic or Ti powders did not improve

the mechanical

strength

of the joints.

References

Donato,

Andreani

Fusion Technology,

29 (1 996) 58-72.

2. T.Iseki

Silicon Carbide Ceramics

Somiya and

Y.Inomata eds., Elsevier Applied Science, London,

3. B.H.Rabin

J.Am.Ceram.Soc.,

75 (1992) 131-35.

4. P. Lemoine, M. Salvo, M. Ferraris, M. Montorsi and

H. Scholz

J.Am.Ceram.Soc.,

78 (1995) 1691-94.

Simon and A.R. Bunsell

J.Mater.Sci.,

19 (1984) 3658-70

6. A.

Donato,

P. Colombo, M.O. Abdirashid,

in:

Ceram. Transactions,

57,43 1436,1995

7. P. Colombo,

M.O.

Abdirashid,

Scarinci and

Donato, in

Fourth Euro-Ceramics, Coatings and

Joinings,

Vol. 9, B.S.Tranchina and A. Bellosi eds.,

Gruppo Editoriale Faenza Editrice S.p.A., Faenza,

Bouillon, F. Langlais, R. Piller, R. Naslain, F.

Cruege, P.V. Huong, J.C. Sarthou, A.Delpuech, C.

LaRon,

Lagatde,

Monthiouxand and A. Oberlin

J.Mater.Sci.,

26 (1991) 1333-45

G.M.

Renlund,

Prochaza

and R.H. Doremus

J.Mater.Res.,

(1991) 2716-34

10.

Pippel, J. Woltersdorf, P. Colombo, A. Donato

Eur. Ceram.

SOC.

17,1259-1265 (1 997)

11. P. Colombo,

Sglavo, E. Pippel, J. Woltersdorf

Mat. Sci.

33 2405-2412 (1998)

12. J. Woltersdorf, E. Pippel

Colloq.

Phys.

C1, Suppl. au no

T. 51, 947-956

(1990) (Proc. Int. Congr. on Intergranular and

Interphase Boundaries in Materials, Paris 1989)

13.

Grathwohl, M.

Kuntz,

Pippel, J. Woltersdorf

Phys.

Stat.

Sol.

(a)

146,393414 (1994)

14.

Pippel, J. Woltersdorf, A. Hilhnel, R. Schneider

Cer. Transactions

(American Ceramic Society) 57,

15. A. Hilhnel, E. Pippel, J. Woltersdorf

Microscow (London),

177 (3), 264-271 (1 995)

16. A. Hihel, E. Pippel, R. Schneider,

Woltersdorf,

D. Suttor,

Composites Part

27A, 685-690 (!996)

17. J. Woltersdorf,

Pippel, A. Hihel

angew. Muth.

Mech.

(Wiley) 78, Suppl.1 S81-

S84 (1998)

18. E. Pippel,

Lichtenberger, J. Woltersdorf,

Mat.

Sci.

letters,

accepted

2000

19. A. Hiihnel,

Pippel, J. Woltersdorf

Cryst. Res.

Techn.,

accepted 2000

1991, pp. 239-263.

Italy, 1995, pp. 75-82.

273-278 (1 995)

620

FATIGUE BEHAVIOR

CERAMICS STRESSED NEAR

FATIGUE LIMIT UNDER ROTARY BENDING

H.N.KO

Nakanihon Automotive College

Sakahogi-cho, Kamo-gun, Gifu-ken, JAPAN

505-0077

ABSTRACT

a preparatory experiment to examine the

fatigue damage of ceramics, the fatigue strength of plain

specimens, stressed near fatigue limit, was studied at

room temperature under rotary bending. The main

materials tested were sintered Si3N4 and gas pressure

sintered Si3N4, which were used practically for rotors

turbochargers. These materials had the effect of cyclic

loading, although the degree of cyclic effect was

different.

It was known that the fatigue strength was

not decreased by the pre-loading near fatigue limit. On

the contrary, the pre-loaded specimens seemed to be

stronger than virgin specimens.

Similar tendency was

confirmed on Y-TZP plain specimens.

The effect of

pre-loading, such as understressing effect

coaxing

effect, may appear in some ceramics which satisfy a

certain condition relating to the microstructure and

microcracking.

INTRODUCTION

It is important to know the fatigue behavior of

ceramics when using them as structural components,

since the fatigue strength is generally lower than the

static strength.

The fatigue strength, the fatigue limit

in particular, must be studied on ceramics to make their

reliability higher as structural materials. It also seems

necessary to know the fatigue damage of ceramics for

their reliability as structural materials

[l].

However,

the basic data on fatigue behavior, such as the effect of

cyclic loading, are still not sufficient, and elementary

data of ceramics relevant to fatigue damage are few.

More basic data on fatigue behavior seem necessary to

clarify the fracture mechanism.

for the aspect, the fatigue behavior of two kinds

of Si3N4 plain specimens, whose materials were used

practically for rotors of turbochargers, was studied at

room temperature under rotary bending

[2,3].

The

results obtained were compared with those under static

fatigue, and the effect of cyclic loading was discussed.

The rotary bending test was performed for a fairly long

period, and the presence of fatigue limit was suggested.

Therefore, the fatigue strength of Si3N4 plain specimens,

stressed near fatigue limit, was studied to obtain the

fundamental knowledge on the fatigue damage of

ceramics. Because of the unusual behavior of Si3N4,

the similar test was also performed on Y-TZP plain

specimens, having transformation-toughening behavior.

Based on the results obtained, the effect of pre-loading

near fatigue limit was discussed.

EXPERIMENTAL PROCEDURE

The materials used were sintered Si3N4

(SSN)

obtained

from

Kyocera (Japan) and gas pressure

sintered Si3N4 (GSSN) obtained

from

NTK (Japan).

Each material was fabricated with YzO3 and

A1203

additives.

The microstructure of each etched surface

shown in Fig.1.

The materials had more rodshaped

grains with various aspect ratios. The properties of the

materials are shown in Table

Sintered tetragonal

zirconia polycrystals containing

2.5

mol

YzOs

(Y-

TZP), obtained

from

Toray (Japan), was also used in

comparison with above

two

kinds

Si3N4.

The specimen had a cylindrical shape and the

diameter of the middle part was

mm and each end of

Fig.1 Microstructures

materials

(a:

SSN

material,

GSSN material).

62 1