copper graded material

Int. J. of Refractory Metals & Hard Materials 12 (1993-1994) 243-250

Elsevier Science Limited Printed in Great Britain 0263-4368/94/$7.00 ELSEVIER

Fabrication of Tungsten/Copper Graded Material M. Takahashi, Y. Itoh, M. Miyazaki, H. Takano* & T. Okuhata** Heavy Apparatus Engineering Laboratory, Keihin Product Operations* and Yokohama Facility Administration Center,** Toshiba Corporation, 2-4 Suehiro-cho Tsurumi-ku, Tokohama 230, Japan (Received 23 June 1993; accepted 16 August 1993)

Abstract: A tungsten/copper graded material is developed as a new material

against plasma, ion beams, electron beams, etc. The fabrication process of the tungsten/copper graded material is newly established by sintering and an infiltration technique. It is very useful for combining two materials with a considerable difference in melting point, such as a combination of tungsten and copper. In this paper, the process conditions for preparing the tungsten/copper graded material is established. The ability to reduce thermal stresses in the tungsten/copper graded material is confirmed by heat-cycle testing in comparison with a tungsten-to-copper lamination prepared by brazing.

technique of graded material is therefore thought to be excellent for the reduction of the thermal stress. The purpose of this study is to establish the technique for fabricating tungsten/copper graded material with the ability to reduce the thermal stress. First, the process conditions for preparing the tungsten/copper graded material are established. It is confirmed by heat-cycle testing that the tungsten/copper graded material developed is useful for reducing the thermal stresses.

INTRODUCTION Functionally gradient materials (FGMs), in which the composition and microstructure vary continuously from plate to plate, are proposed as a method to reduce thermal stresses caused by a difference in the thermal expansion of the materials used. 1 It is expected that the FGMs are applied as system components used at elevated temperature, for example, spacecraft, fusion reactors, and gas turbines. 2 Materials exposed to high heat flux, such as the first wall and diverter of fusion reactor, whose surface is heated by plasma, ion beams, electron beams etc., and the remainder is forcibly cooled by water, are required to have the characteristics of both a high thermal resistance' and a good thermal conductivity for the enhancement of the cooling property. At present, a tungsten-tocopper lamination produced by brazing is mainly applied to the materials exposed to a high heat flux. This is because the tungsten is a refractory metal with a high melting point and copper has a good thermal conductivity, as shown in Table 1. However, some cracks can be observed in the tungsten and at the tungsten/copper interface. This is due to the thermal stress that is caused by the considerable difference in thermal expansion between tungsten and copper. The fabrication

Table 1. Physical properties of tungsten and copper used

Melting point

(K) Specific heat (J/kg K) (at 293 K) Density

Tungsten (w)

Copper (cu)

3653

1356

134

385

19.3

8'9

( X 10 3 k g / m 3)

(at 293 K) Thermal conductivity (W/m K) (at 293 K) Thermal expansion coefficient ( × 10-~/K) (at 273-373 K) 243

167 4.5

393 17.1

244

M. Takahashi et al.

PROPOSAL FOR A NEW PREPARATION PROCESS

Physical and chemical vapor deposition, thermal spraying, sintering, and a self-propagating hightemperature synthesis process are mainly proposed in order to fabricate the FGMs. 3-6 However, a self-propagating high-temperature synthesis process can be applied only in the case when a combustion reaction occurs. Physical and chemical vapor deposition and thermal spraying are basically coating processes, and these can then be applied only in the case of preparing thin graded films. The sintering process is effective for fabricating thicker graded plates of the order of a millimeter. It is very difficult to fabricate the FGMs in two materials with a considerable difference in melting point because layers with a graded composition are generally laid and sintered. The sintering and infiltration technique has been recently developed to fabricate an FGM in a combination of tungsten and copper, in which the melting points are largely different. The new process consists of six steps as shown in Fig. 1. 7 The first step is a laying process of tungsten powders ranging from small to large size. In the second step, the green compact is obtained by cold-pressing the laid tungsten powders in the laid direction. The third step is a sintering process using a molybdenum furnace in an atmosphere of hydrogen gas. The sintered tungsten with graded pores is fabricated in the third step, because the relative density of sintered tungsten depends on

the size of the tungsten powders used. The fourth step is a capsule-free HIP (hot-isostatic-pressing) process for reducing closed pores, which cause a decrease in thermal conductivity. The fifth process in an infiltration process. Molten copper is infiltrated into the open pores of the sintered tungsten with graded pores by an HIP treatment. Finally, a mechanical finish is applied to the tungsten/copper composite after infiltrating the molten copper. The tungsten/copper graded material can be fabricated by using the above six processes. EXPERIMENTAL Materials

For fabricating tungsten/copper graded materials, six kinds of tungsten powders (Toshiba Corporation), ranging from 0.49 kLmto 9"15/~m in average particle size and an oxygen-free copper plate are used. The physical properties of the tungsten powders and the copper plate used are shown in Table 2. The tungsten powder of 0-49-pm particle size is over 99"0% in purity, and the other tungsten powders are over 99"9% in purity. There is a tendency for the density of tungsten powder to be low and the specific surface area high when the particle size of tungsten powders is small, and this tendency is remarkable in the case of smaller powders. Evaluation

Laying of tungsten powders

l

~

ParticlesizeA ~ cB

Pressing

Cold pressing Sintered tungsten Sintering

t Capsule-free HIP I

~

Some experiments were performed to establish the process conditions. First, the influence of the tungsten-particle size on the relative density after sintering is examined in the sintering process shown in Fig. 1 in order to fabricate the sintered tungsten with graded pores. The technique of con-

OCIosedp o r e s OOpen pores

Table 2. Tungsten powders and copper plate used in preparation

Reduction of closed pores

Purity (%)

1

Infiltration of I Tungsten/copper graded material copper by HIP

1

Black : Cu White : W Network sbucture

I Mechanical finish ] Fig. 1.

Sintering and infiltration technique for preparing tungsten/copper graded material.

W-1 W-2 W-3 W-4 W-5 W-6 Cu

> > > > > > >

99.0 99.9 99.9 99.9 99.9 99.9 99.9

Particle size (lam)

Density ( x 10 3 kg/m 3)

0-49 1-02 2-27 2.99 4.39 9.15 --

8-1 14.3 20.6 21.4 29.2 33-2 --

Specific surface area (m2/kg) 629 306 153 105 71 34 --

Fabrication of tungsten/copper graded material trolling the sintering shrinkage of tungsten powders is investigated because large differences in shrinkage cause cracks and a large deformation to occur. In the capsule-free HIP process, the effect of a capsule-free HIP treatment on the reduction of closed pores, which remain as closed pores until the end of the treatment, is investigated. In the infiltration process, the infiltration behavior of molten copper into sintered tungsten is examined and is confirmed by the observation from its microstructure that open pores of sintered tungsten are full of the copper. The ability to reduce thermal stresses in the tungsten/copper graded material is confirmed by heat-cycle testing in comparison with a tungstento-copper lamination prepared by brazing. The brazing is performed at 1173 K under vacuum conditions by using an Ag-Cu-Ti filler film. Specimens used in this experiment are shown in Fig. 2. The diameter of both the specimens is 25 mm, and the thicknesses of the 100% copper layer and the 100% tungsten layer are, respectively, 10 mm and 3 mm. The tungsten/copper graded material consists of four layers, whose composition changes from 100% tungsten at the surface to 100% copper at the other side. The experimental conditions of the heat-cycle testing are shown in Fig. 3. Experiments are conducted 25¢ 25¢ 100% -Cu-Ti 100%

(a) Tungsten/copper graded material

(b) Tungsten to copper lamination prepared by brazing

Fig. 2.

245

by using a graphite-resistance furnace under 0.1 MPa in an argon-gas atmosphere, and the heating and cooling are repeated 10 times in a temperature range from 1073 K to 573 K.

RESULTS AND DISCUSSSION Investigation of process conditions

(a) Sintering process Figure 4 shows the relation between the relative density and the shrinkage when various particle sizes of tungsten powders were used in a sintering process. These results were obtained in sintering conditions of a heating temperature of 2073 K, a holding time of 8 h, and a hydrogen-gas atmosphere. It was clear that the relative density of sintered tungsten became higher by the use of tungsten powders of a smaller particle size and that the relative density could be changed in the range from 60 to 97% by the use of tungsten powders of a particle size ranging from 0.49 to 9.15 /zm. This means that the sintered tungsten with graded pores can be fabricated when the tungsten powders are laid from a small particle size to a large particle size. But, in the case of preparing the sintered tungsten with graded pores, cracks and large deformation can sometimes be observed, as is shown in Fig. 5, if there is a wide range in the shrinkage of each tungsten layer in the sintering process. It was also clear from Fig. 4 that the particle size of tungsten powders had an influence on the shrinkage properties and that the shrinkage tended to be larger in the sintered tungsten with a high relative density when the smaller particle size of tungsten powder was used. The variable range of the relative density is sintered tungsten with the graded pores can be controlled by selecting the particle size of tungsten powders that are laid.

Specimens used in heat-cycle testing. 50 Particle size

40 .

I 0 cycles

\

t~

~ ~' 41~ 4.50~m

.

1.Sks I, ,I

o.5 7

~) 30

O ~ • 0.49/Lm A ~ • I.(~/Lm V W • 2.27pm [] [] • 2.99pm

........

2073K. 28.8ks, H2 gas 49MPa , ~ ~ /

~

Pa 10

,o73K

k 8 ks

....

RT --

~

A,=so,ere

0

60

70

80

90

100

Relative density (,~)

Fig. 4. Conditions for heat-cycle testing.

I

50

Gas pressure: 0.1MPo

Fig. 3.

--~

Relation between relative density and shrinkage in sintering tungsten.

M. Takahashi et al.

246

50 ParUcle F~e-press 2073K,28.8ks, I-b~as 40

s~e (pro) (MPa) e j 1.02 382

• G) 30 t~

2.27

I I 2.99 n,

2,99

196 1. o

/ 49M~//~/ ~

_

_

•-- 20

I0

!

50

Fig. 7.

Sintering deformation of two-layer tungsten with different pores.

Fig. 5.

30

~

A

~zo

- press OMPa

Pressure 285 MPa

Pre- press 285MR:

0

(O

IO

J

Pressure V Q 49MPa W Q) 98MPo

ve ~

o

Fig. 6.

I

80

Powder size : 1.02#m 2073 K , 28.8ks, H2 gas

[ 196MPoI I

I

90 Relotive Density (%)

Effect of pre-pressing on reduction of sintering shrinkage.

It can be confirmed from Fig. 4 that increasing the applied pressure in a cold-pressing process was effective in reducing the shrinkage of sintered tungsten in the sintering process. The effect of pre-pressing tungsten powders instead of increasing the applied pressure is next investigated. In two cases of the no pre-pressing and pre-pressing at 285 MPa, the change in the relative density and the shrinkage is comparatively shown in Fig. 6. The particle size of the tungsten powder used is 1.02/~m. It was clear that the tungsten powders pre-pressed and crushed by ball milling were effective in reducing the shrinkage. This means that the shrinkage on sintering tungsten can be controlled by the pre-pressing of tungsten powders. The shrinkage of each layer, in which the particle size of tungsten powders used is different, can be almost equal in fabricating the sintered tungsten with graded pores and then the cracking and the large deformation can be reduced by the large difference in shrinkage between the various tungsten layers.

60 70 80 90 Relative density (X)

100

Changes in relative density and shrinkage by using pre-pressed tungsten powders.

The sintered tungsten with graded pores was fabricated on the basis of the above test results. The sintered tungsten consisted of four layers, and the tungsten powders used, the relative density, and the shrinkage of each tungsten layer in sintering are shown in Fig. 7. The sintering conditions are a heating temperature of 2073 K, a holding time of 8 h, and an applied pressure of 98 MPa in a hydrogen-gas atmosphere. It was found from Fig. 7 that the shrinkage of each sintered tungsten layer was almost equal in the range of 16-18% and cracks could not then be observed in sintered tungsten. Simultaneously, it was clear from this result that the relative densities measured by Archimedes' method varied gradually in the range from 71 to 94%.

(b) Capsule-free HIP process Sintered tungsten with a high density on the heating surface and few remaining pores is necessary to obtain an excellent beam target composed of a tungsten/copper graded material because the existence of pores in the heating surface causes migration of the inner copper out of the heating surface in use and the remaining pores make the thermal-conductivity property worse and then the temperature of the heated surface increases. In this section, the effect of a capsule-free HIP treatment for enhancing the relative density in sintered tungsten is investigated. It is clear from Fig. 8 that the relative density of the sintered tungsten could be enhanced by the capsule-free HIP treatment. This tendency could be observed with both sintering conditions, i.e. 2273 K, 3 h and 2073 K, 8 h. In both cases, the relative density became nearly 100% by using the capsule-free HIP treatment in cases where the relative density of sintered tungsten was more than about 90%, but the change in relative density by the capsule-free HIP treatment

Fabrication of tungsten/coppergraded material was small in cases of less than about 87%. The range from 87% to 90% of relative density is shown as the 'transition range' in Fig. 8. This means that the relative density in the heating surface can be established by the capsule-flee HIP treatment of sintered tungsten over the transition range and graded pores of sintered tungsten have to be used under the transition range. The mechanism for the enhancement of the relative density by capsule-free HIP treatment in sintered tungsten was investigated. Figure 9 shows varieties of open and closed pore ratios and the average pore size before and after capsule-free HIP treatment in two kinds of sintered tungsten with an initial relative density of 88-2% and

• : 2073K,28.~ks,H2gas 100-O:2273K, 10.gks,H2gas~_dO(])

9o

•

o•

80-.

~/

//

@ O //

"

Transition

~'/range

,,"

76.6%. In the case of the sintered tungsten with high relative density, the relative density was broadly increased from 88.2% to 95.5% by the capsule-flee HIP treatment. It was thought that capsule-flee HIP could accelerate the sintering of tungsten, because the average pore radius was also small and ranged from 0.063 ,urn to 0-009 pm. In particular, the large reduction in closed pore ratio from 2.8% to 1.5% means that the capsule-flee HIP treatment is effective for decreasing the closed pores, which act as a source of thermal resistance in the beam target. The relative density showed little change (from 76.6% to 81.2%) in the case of the sintered tungsten with low relative density, but the closed pore ratio was considerably reduced from 5.1% to 1.1%, because the change in open pore ratio was small, i.e. from 18"3% to 17"7%. Simultaneously, the average pore radius showed little change from 0.54 # m to 0.52 /~m. From the above examples, it can be said that the capsule-flee HIP treatment was effective in enhancement of the relative density of the sintered tungsten on the heated surface and the reduction of closed pores, into which the copper cannot migrate.

(c) Infiltration process of copper

80

90

Relative

Fig. 8.

•

density

The infiltration property of the copper into the open pores of the sintered tungsten can be judged by the following function:~

1O0 HIP (~)

before

Enhancement of relative density by capsule-free HIP.

__

.9.:~.~ ...........

p.r> - 2o.cos 0

o.sx

""

}

f:;UI'

\0.5~

:

76.6Jg i 8 1 ~

C3 : ....

: Open pore ~/~j~:

Closed

;,

pore

:W

As s i n t e r e d

Z073K 196MPa 10.gks

A f t e r HiP

O ~

~

100.0

....

As sinteredHiP ' I (Averaged 0.063pm) , . ~ " / l--After

As s i n t e r e d

10.0

1.0

0.1

Pore radius

i==r.

mllW

w

A f t e r HIP

0 ,~,

an3K lg6MPa 10.gks

]

/

if:3

sI o O.

........

""1

O'Ol(Av°r"e° J / /

0.01

0.001

(/Jm)

(a) Sintered t u n g s t e n ~ t h high relative density Fig. 9.

247

o O.

50.0 1 [

10.0

I I

(Averaged 0.54pm) ] After open

HIP|

,1 (~,~ged 0;~Um)/ 1.0

0.1

Pore radius

0.01

0.001

(/~m)

(b) Sintered tungsten with

low relative density

Effect of capsule-free HIP on pores measured by the mercury-infiltration method in sintered tungsten.

(1)

M. Takahashi et al.

248

where p: pressure of infiltration r: radius of pore tr: capillary action of tungsten (= 0.135 N/m at 1373 K)9 0: contact Jangle of moiten copper against tungsten It is clear that the infiltration property depencls on the infiltration pressure of copper under the infiltration conditions, because the radius of open pores, the capillary action of tungsten infiltration, and the contact angle of molten copper against the tungsten are restricted by reason of the sintered tungsten and copper used. However, it is known

W W W W Cu

that the molten copper has good wettability against the tungsten, and the contact angle of molten copper against tungsten is small and about 5 ° at 1473 K. 1° This means that the infiltration pressure of copper may be to small to allow the above eqn (1) to be satisfied. The pressure of the HIP treatment in the infiltration process of copper can then be decided in conditions of 196 MPa, in which it is possible to infiltrate the molten copper in calculation into the open pores over 0.002 gm in radius. As a result, it was observed that the molten copper could be well infiltrated in every sintered tungsten with a different relative density as shown in Fig. 10.

100% 85%~ Cu15% 75%, Cu 25% 70%,Cu 30% 100% 5mm

(a) M a c r o s t r u c t u re

(b) Fig. 10.

I

n

TIT

IV

50 m

Microstructures Macrostructures of cross-section in tungsten/copper graded material.

Fabrication of tungsten/coppergraded material

Tungsten / co p per graded material

249

Tungsten to copper lamination prepared by brazing

0

C

0 0 O. O.

E

2 ¢0 lb.. 4"--

(l) 0 t,..

Fig. 11. Appearances of heated surface and cross-section of tungsten/copper graded material after heat-cycle testing.

Ability to reduce thermal stresses in tungsten/ copper graded material Figure 11 reveals the appearance of the tungsten surface and the cross-section of the tungsten/ copper graded material and the tungsten-tocopper lamination prepared by brazing after the heat-cycle testing. The temperature difference is 500 K, and the number of cycles is 10, as shown in Fig. 3. No appearance changes could be observed in the case of the tungsten/copper graded material. On the other hand, in the tungsten-to-copper lamination prepared by brazing, many vertical cracks could be observed in the tungsten, and detachments could also be observed at the interface between the tungsten and the copper. From these results, it was thought that the grading at the interface between the tungsten and the copper was very effective for the reduction of thermal stress. In addition, it was clear that the newly developed tungsten/copper graded material

could endure sufficiently in the above heat-cycle testing. This advantage in the reduction of thermal stress shows that the tungsten/copper graded material has strong possibilities as the beam target. CONCLUSIONS From the results of this study, the following conclusions can be drawn.

(i) (ii)

The tungsten/copper graded material could be newly fabricated by developing a sintering and infiltration technique. In the sintering process, it was clear that the sintered tungsten with graded pores could be fabricated when the tungsten powders were laid in a range from small to large and that the cracking and much deformation caused by a considerable difference of shrinkage between the various

250

M. Takahashi et al.

tungsten layers could be reduced by using pre-pressed tungsten powders. (iii) The capsule-free HIP treatment was effective in reducing the remaining closed pores of sintered tungsten, which provide thermal resistance in use. (iv) In the infiltration process of molten copper, it was confirmed from the microstructure that the molten copper could be infiltrated into the open pores of the sintered tungsten very well, because the molten copper had good wettability against the solid tungsten. (v) It was found by heat-cycle testing that the thermal stress of the tungsten/copper graded material could be reduced in comparison with that of a tungsten-to-copper lamination.

REFERENCES 1. Koizumi, M. & Tada, I., Metals, No. 4 (1988) 2. 2. Itoh, Y., NewMater., 3 (1992) 70. 3. Sasaki, M. & Hirai, T., In Proceedings of the 1st International Symposium of FGM, Sendai, Japan, 1990, p. 83. 4. Watanabe, R. & Kawasaki, A., In Proceedings of the ist International Symposium of FGM, Sendai, Japan, 1990, p. 107. 5. Shimoda, N., Kitaguchi, S., Saito, T., Takigawa, H. & Koga, M., In Proceedings of the 1st International Symposium of FGM, Sendal, Japan, 1990, p. 151. 6. Miyamoto, M., Takakura, T., Tanihata, K., Tanaka, I., Yamada, O., Saito, M. & Takahashi, H., In Proceedings of the 1st International Symposium of FGM, Sendai, Japan, 1990, p. 169. 7. Takahashi, M., Itoh, Y. & Kashiwaya, H., In Proceedings of the 1st International Symposium of FGM, Sendai, Japan, 1990, p. 129. 8. Harada, Y. & Murama'tsu, Y., J. Japan Inst. Metals, 49 (1985) 272. 9. Metals Data Book, Japan Institute of Metals, Tokyo, Japan. 10. Muramatsu, Y., Harada, Y., Dan, T. & Isoda, Y., J. Japan Inst. Metals, 54(6), (1990) 679.