131

DOI: 1 557/op 0130.1 l.2 .

Effect of high temperature heat treatment on the microstructure and mechanical

properties of third generation SiC fibers

Dominique Gosset

, Aurélien Jankowiak

, Thierry Vandenberghe

, Maud Maxel

, Christian

Colin

, Nicolas Lochet

and Laurence Luneville

CEA-Saclay, DMN-SRMA-LA2M, LRC CARMEN, 91191 Gif/Yvette, France

CEA-Saclay, DMN-SRMA-LC2M, 91191 Gif/Yvette, France

CEA-Saclay, DMN-SRMA-LA2M, 91191 Gif/Yvette, France

CEA-Cadarache, DER-SRJH-LEDI, 13108 St Paul-lès-Durance, France

CEA-Saclay, DMN-SRMA-LTMEX, 91191 Gif/Yvette, France

CEA-Saclay, DM2S-SERMA-LLPR, LRC CARMEN, 91191 Gif/Yvette, France

ABSTRACT

SiC fibers (High Nicalon S -HNS and Tyranno SA3 -Ty-SA3) submitted to heat treatments

in neutral atmosphere up to 1900°C were studied by X-ray diffraction (XRD) and TEM

observations then submitted to tensile tests up to 1800°C. The microstructural changes in both

materials were determined by XRD using a modified Hall-Williamson method introducing an

anisotropy parameter taking into account the high density of planar defects of the SiC-3C

structure. HNS fibers exhibit significant modifications in the CDD size which drastically

increases from 24 nm to 70 nm in the range 1600°C to 1800°C and in the microstrains which

decrease from 0.0015 to 0.0005 between 1750°C to 1850°C. Concerning the Ty-SA3 fibers, no

evolution of CDD size and microstrains has been observed. The mechanical properties of single

fibers were investigated after the heat treatments showing decreases in the tensile strength

reaching up to 20% for Tyranno SA3 and 50% for High Nicalon S. The Weibull moduli were

also significantly affected. These results are correlated to the fiber structural and microstructural

evolutions.

INTRODUCTION

SiC

/SiC

ceramic matrix composites are considered as promising materials for nuclear reactor

applications [

1,2].Indeed, the third generation of SiC fibers have significantly improved their

thermo-mechanical properties due to their near stoichiometric composition, high mechanical

strength and thermal stability. Furthermore, these fibers contain a small amount of oxygen

(<0.2%) [3]. The SiC

/SiC

composites exhibit interesting intrinsic features such as a higher

operating temperature in comparison to metallic alloys and a low activation level under

irradiation. They also provide a very high irradiation stability with a saturation swelling value

lower than 0.2% at the aimed operating temperature of about 1000°C [4]. In addition, the

fiber/matrix interface which is a key component in these materials significantly increases both

388

132

the tensile strength and the fracture toughness which can exceed 20 MPa m

1/2

. It should be

mentioned that these properties are not affected by the temperature up to the fiber stability limit.

However, their performances strongly depend on their microstructure which is modified

when operating [5,6,7]. In this context, fibers are key components since the mechanical

properties of the composite such as creep and tensile strength are greatly influenced by their

microstructure [8,9,10,11]. As a consequence, the microstructural stability of the fibers in

various environments (i.e. temperature, irradiation, oxidation,…) has to be investigated.

The purpose of this work was to study the mechanical properties for two types of third

generation SiC fibers (Hi-Nicalon type S -Nippon Carbon Co. Ltd and Tyranno SA3 -Ube

Industry Ltd; respective mean diameter of 14 and 7.5 μm) after heat treatments up to 1900°C and

to correlate these results with the observed microstructural evolutions in the same temperature

range obtained first with X-ray diffraction (coherent diffraction domains -CDD- and

microstrains) and second with transmission electron microscopy [12].

The structure of the fibers consists in a dense packing of equiaxed nanometric β SiC grains

resulting in a relative density over 97% [3]. Both fibers contain a small amount of oxygen

(<0.2%). Tyranno SA3 fibers exhibit a non uniform concentration in C and Si on the fiber cross

section with a larger amount of free carbon in the core of the fiber; they contain aluminum

segregating at grain boundaries and a carbon-rich phase is observed at the surface. For Hi-

Nicalon type S fibers, free carbon with a turbostratic structure is also found and located between

SiC grains.

X-RAY ANALYSIS OF THE FIBERS

It is for long known that X-ray diffraction can usefully be used to obtain microstructural

parameters of materials: the analysis of the line profiles of the diffraction patterns directly leads

to estimations of the coherent diffraction domains (CDD) size and to the residual microstrains

induced by internal defects such as dislocations. In order to allow a correlation analysis of the

mechanical properties and the microstructure of the fibers, we resume here results we recently

obtained [12].

The X-ray diffraction analyses have been performed on a Bruker D8 Advance

diffractometer. The beam is produced with a classical Cu tube (40kV, 40mA) then a Göbel

mirror, which leads to a flat, parallel, highly intense, monochromatic (CuK

1+2

) beam. The

detection is made with a multichannel Vantek detector, this allows fast analyses with good

angular resolution (channel width = 0.006°). In order to avoid any preferred orientation effect,

the fibers are crunched after the heat treatments to short segments (around 50μm long) then put

in silica capillaries. The capillary holder allows accurate centering of the capillary and

continuous rotation during the analysis.

The diagrams we obtained show monotypic (3C), well crystallized SiC materials. Small

grains and high density of defects contribute to high diagrams distortions (Figure 1).

Microstructural analyses have been performed with the Hall-Williamson method [12]. We then

observed, the distribution of the linewidths correspond to strongly anisotropic materials, this

corresponding to a high density of stacking faults or twins along the (111) planes of the SiC-3C

structure as shown by TEM observations (Figure 3, left). As compared to the classical Hall-

Williamson calculation [13], we then introduced an anisotropy parameter accounting for this

planar defect density: the width of a given (hkl) line is made to depend on its direction according

to a reference direction. We then use the following expression:

133

β.cosθ = (λ/d + 4.ε.sinθ) (1 + δ cos²χ),

()

222

²²²

cos

rrr

lkhlkh

llkkhh

++⋅++

⋅+⋅+⋅

(1)

with  the integral width (surface over height ratio, corrected for instrumental broadening) of a

given (hkl) line, θ its Bragg angle, λ the X-ray (CuK) wavelength, d the coherent diffraction

domain (CDD) size,  the residual microstrains, δ the anisotropy parameter and  the angle

between the (hkl) and the (h

) reference directions. The best agreement is here obtained with

) = (145): Figure 1.This then leads to coherent diffraction domains with apparent platelet

shape, the aspect ratio is directly given by δ+1, and anisotropic microstrains.

The main results are reported on Figure 2. The Ty-SA3 fibers show large domains, around

70 nm, with no modification up to 1800°C and a decrease of the anisotropy parameter above

1600°C. The HNS fibers show small domains, around 25 nm increasing up to 70 nm above

1600°C together with a decrease of the anisotropy parameter, but the microstrains relax only

above 1750°C, this meaning that two annealing processes are involved. At the highest annealing

temperatures, the two materials have the same microstructural parameters values.

20 40 60 80 100 120 140

(°)

Intensity (u.a.)

30 35 40 45 50

(111)

(200)

0,2 0,4 0,6 0,8 1

sin

cos

θ (°)

(200)

(400)

(331)

(111)

(220)

0.2 0.4 0.6 0.8

Figure 1. Left: diffraction diagram of the as-received HNS fibers (insert: (111) and (200) lines,

evidencing the high faults density).

Right: Hall–Williamson analysis of the diagram.: experimental.: calculated with eq. 1

and d = 25 nm,  = 0.0012, δ = 8.5.

100

1000 1200 1400 1600 1800 2000

annealing temperature (°C)

CDD size (nm)

Ty-SA3

HNS

0,0000

0,0005

0,0010

0,0015

0,0020

1000 1200 1400 1600 1800 2000

annealing temperature (°C)

microstrain

0.0020

0.0015

0.0010

0.0005

0.0000

1000 1200 1400 1600 1800 2000

annealing temperature (°C)

anisotropy parameter

Figure 2. Microstructural parameters (size of CDD, microstrains and anisotropy parameter) of

the HNS and Ty-SA3 fibers as a function of annealing temperature.

134

TRANSMISSION ELECTRON MICROSCOPE OBSERVATIONS OF THE FIBERS

Transmission electron microscope (TEM) observations have been performed in order to

check the XRD microstructural parameters estimations [12]. Fibers have been embedded in resin

then ion thinned (Precision Ion Polishing System, GATAN 691.056). The observations have

been performed on a 200 kV JEOL 2100 TEM with a LaB

cathode and a double tilt sample

holder stage. The observations coarsely confirm the XRD results: no significant modification of

the Ty-SA3 fibers, high grain growth of the HNS fibers. However, the grain size of the Ty-SA3

fibers appears larger than the CDD size. In both cases, a significant decrease of the planar defects

density is observed, this meaning the anisotropy parameter is correlated to this faults density.

200 nm

Ty-SA3 1800°C

50 nm

HNS

50 nm

HNS 1800°C

Figure 3. TEM observation of the fibers. Left: Ty-SA3 as-received (and SAED, showing the

high twins density) and annealed at 1800°C. Right: HNS, as-received then annealed at

1800°C.

From a comparison of the XRD analysis and the TEM observations, it appears the anisotropy

parameter can be related to the planar defects density. On the other hand, the CDD size and the

apparent mean grain size are similar for the small grain materials (e.g. as-received HNS). But for

the large grain materials, the CDD are smaller than the grain size: this is a quite general result,

for the CDD can be bounded by small crystal distortions, CDD size is lower than grain in the

sub-micronic range. Other studies lead to the same estimation: the largest CDD size in high

temperature treated SiC fibers or nanopowders when determined by XRD is generally around

70nm whatever the actual grain size [14].

EFFECT OF THE TEMPERATURE ON THE FIBERS MECHANICAL PROPERTIES

Heat treatments were performed in high purity argon flow in a graphite furnace. They were

followed by mechanical tests performed on single fibers at room temperature using the MecaSiC

apparatus presented in Figure 4 [15]. This tensile test device operates in secondary vacuum

(10

-5

Pa) and allows tensile tests to be carried out up to a 5 N load from 25°C up to 1800°C with

135

an electric self-heating of the fiber and using a dedicated LABVIEW program. The apparatus

consists in a Melles Griot nanostep 1000-100 motorized linear stage for accurate fiber strain

determination combined to a HBM U1A load cell.

Weibull statistics were applied to

evaluate the mechanical behavior of the

fibers at room temperature. In this study,

the as-received fibers were heat treated at

550°C for 30min in helium to remove the

fiber sizing. Desized fibers were considered

as the reference state for the properties

determination. The tensile strength (σ) and

Weibull modulus (m) of desized and

annealed fibers were investigated. The

probability of failure P

was determined for

each tested sample using the classical

approach: P

= (i-0.5)/n with i = 1…n. The

m values were deduced from the slopes of

the ln[ln(1/(1-P

)] against ln(σ) curves for

each set of samples. XRD and TEM analyses

have shown that heat treatment induce a fault (planar defects, microstrains) density decrease. In

addition, for the HNS fiber, an increase in the CDD size is observed whereas it remains stable for

Ty-SA3. From Figure 5, a decrease of 24% in the tensile strength can be deduced for the HNS

fibers heat treated at 1600°C. This loss reaches about 50% at 1900°C. Similarly, heat treatments

of Ty-SA3 fibers also result in a decrease in the tensile strength. However, the strength loss is

only about 13% at 1900°C which is lower than that observed for the HNS fibers.

Figure 5. Weibull plots as a function of test temperature. Left: Hi-Nicalon S, Right: Tyranno

SA3

As it can be seen in figure 6, for HNS fiber, this tensile strength decrease can be linked to the

CDD size increase since this phenomenon initiates at 1600°C and continues at higher

temperatures. The Weibull modulus m of Ty-SA3 fibers is also affected by the heat treatments

Figure 4. Tensile test device for single SiC fiber

136

when its microstructural parameters remain nearly constant. This indicates at least another

process has to be involved to explain the mechanical properties modifications. It could arise from

a modification in the flaws and surface state characteristics occurring during the heat treatment.

In particular, flaws related to surface oxidation cannot be excluded to explain the fibers failure

evolution.

The Weibull modulus decrease is more important for the HNS fiber where a strong decrease

can be observed with temperature. For Ty-SA3 fibers, where only a few microstructural

modifications are noticed, the decrease is lower. For the latter, it should be mentioned that some

non-linearity can be observed especially in the 1900°C tested fibers. This is mainly due to simple

scatters since heat treated fibers are weaker and might be damaged when the specimen (single

fiber) is prepared.

Figure 6. Weibull modulus m (

) and tensile strength (S) of Hi-Nicalon S fiber vs CDD size.

CONCLUSION

Hi-Nicalon S and Tyranno SA3 fibers were submitted to heat treatments from 1600°C to

1900°C in He for 2h. The microstructural changes in both materials were determined by XRD

using a modified Hall-Williamson method and compared to TEM observations. Mechanical tests

after high temperature treatments were performed on individual fibers allowing Weibull modulus

and tensile strength determinations at room temperature.

The XRD analyses are in agreement with TEM observations. This confirms the high interest

of XRD to obtain microstructural parameters of nanometric materials, with good sensitivity, high

statistical meaning and easy samples preparation and diagrams analyses.

In the range 1600°C to 1800°C, HNS fibers exhibit significant microstructural modifications

with CDD size which significantly increases from 24 nm to 70 nm. In addition, the microstrains

decrease from 0.0015 to 0.0005 between 1750°C and 1850°C and the planar defects density

decreases from 1700°C. Concerning the Ty-SA3 fibers, no evolution of CDD size and

microstrains occurred, but the planar defects density decreases in the same temperature range as

the HNS fibers. As a consequence, the Ty-SA3 fibers exhibit a more stable microstructure in

comparison to the HNS fibers which can explain the Ty-SA3 fibers to have a higher tensile

strength after heat treatment. Indeed, the mechanical strength decrease can reach 20% for

137

Tyranno SA3 and 50% for High Nicalon S for the same heat treatment conditions. The Weibull

moduli were also significantly affected. Although a partial correlation between the

microstructure evolution and the mechanical properties can be made, other types of flaws are to

be involved to explain the two types of fibers to have different mechanical properties after heat

treatments at high temperature.

REFERENCES

1 P. Yvon, F. Carré, J. Nucl. Mat., 385-2, 217–222 (2009)

2 Basic research needs for advances nuclear energy systems, Office of Basic Energy Sciences,

U.S. Department of Energy, October 2006

3 C. Sauder, J. Lamon, J. Am. Cer. Soc. 90 [4] 1146–1156 (2007)

4 W. M. Kriven, Hua-Tay Lin, R. H. Jones, Chapter 38. SiC/SiC Composites for Advanced

Nuclear Applications, 27th Annual Cocoa Beach Conference on Advanced Ceramics and

Composites: B: Ceramic Engineering and Science Proceedings, Volume 24, Issue 4

5 K. Ozawa et al., J. Nucl. Mat., 367-370 part A, 713-718 (2007)

6 L. L. Snead, O. J. Schwarz, J. Nucl. Mat., 219, 3-14 (1995)

7 Y Katoh et al., J. Nucl. Mat., 403 [1-3], 48-61 (2010)

8 C. Sauder, J. Lamon, J. Am. Cer. Soc. 90 [4], 1146–1156 (2007)

9 G.E. Youngblood et al., J. Nucl. Mat., 289 [1-2], 1-9 (2001)

10 J. J. Sha et al., J. Nucl. Mat., 329–333 part A, 592-596 (2004)

11 J.J. Sha et al., Mat. Charac., 57 [1], 6-1 (2006)

12 D. Gosset et al., accepted J. Am. Cer. Soc. (2012)

13 G. K. Williamson, W. H. Hall, Acta. Met. 1, 22-31 (1953)

14 Y. Lee, Y. Park, T. Hinoki, ‘Influence of Grain Size on Thermal Conductivity of SiC

Ceramics’, IOP Conf. Series: Materials Science and Engineering, 18 162014 (2011)

15 C. Sauder, J. Lamon, J. Am. Cer. Soc., 90-4, 1146–1156 (2007)