Tokitani DOI:10.1088/1741-4326/aa6bb3 FIP/P4-37

Fabrication of Divertor Mock-Up with ODS-Cu and W by ImprovedBrazing Technique

M. Tokitani1, Y. Hamaji1, Y. Hiraoka2, S. Masuzaki1, H. Tamura1, H. Noto1, T. Tanaka1,T. Muroga1, and A. Sagara1

1National Institute for Fusion Science (NIFS), Toki, Gifu, Japan2Department of Applied Physics, Okayama University of Science, Okayama, Japan

Corresponding Author: M. Tokitani, tokitani.masayuki@lhd.nifs.ac.jpIt is considered that copper alloy could be used as a divertor cooling tube in not only helical

reactor FFHR-d1 but also tokamak DEMO reactor, because of utilizing its high thermal conductivity.This study focussed on using an oxide dispersion strengthened copper alloy (ODS-Cu) of GlidCop®(Cu-0.3wt%Al2O3) as the divertor heat sink material of FFHR-d1. This alloy has superior hightemperature strength over 300 MPa even after an annealing up to „1000˝C. This characteristicprovides two important advantages. The first advantage ensures the redundancy of the temperaturemargin of the divertor operation even when an unexpected temperature excursion occurs. Thesecond comes during the brazing procedures itself, since rapid cooling down phase is not neededat the final stage of the brazing heat treatment. The rapid cooling down phase would be give anundesired thermal stress to the material.

The changeable material properties of pure-Cu, GlidCop® and CuCrZr by neutron irradiationare summarized in this paper. A primary dose limit is the radiation induced hardening/softening(„0.2 dpa/1–2 dpa), and it has a temperature dependence. If the temperature of the GlidCop®is maintained at 300˝C, radiation induced hardening/softening would be moderated. Accordingto such an evaluation, GlidCop® can be selected as the current best candidate material in thecommercial base of the divertor heat sink, and its temperature should be kept at 300˝C whenpossible during operation.

Joining tungsten armour to the GlidCop® heat sink was successfully performed by using animproved brazing technique with BNi-6 (bal.%Ni, 11%P) filler material (W/BNi-6/GlidCop®). Thejoint strength was measured by three-point bending test, and was around 200MPa. Surprisingly,several specimen showed obvious yield point. This means that BNi-6 brazing layer caused relaxationof the applied stress.

The small divertor mock-up of the W/BNi-6/GlidCop® was successfully fabricated by theimproved brazing technique. The heat loading test was carried out by electron beam device ACT2 inNIFS. The highest temperature under 8 MW/m2 is „350˝C in the tungsten plate. The temperatureprofile is quite reasonable by modelling calculation by a finite element method. The design ofthe W/BNi-6/GlidCop® showed an excellent heat removal capability for using in the FFHR-d1divertor.

Published as a journal article in Nuclear Fusionhttp://iopscience.iop.org/article/10.1088/1741-4326/aa6bb3

1 FIP/P4-37

Fabrication of Divertor Mock-up with ODS-Cu and W by Improved Brazing Technique

M. Tokitani 1, Y. Hamaji 1, Y. Hiraoka 2, S. Masuzaki 1, H. Tamura 1, H. Noto 1, T. Tanaka 1,

T. Muroga 1, A. Sagara 1 and FFHR Design Group 1

1 National Institute for Fusion Science, National Institutes of Natural Sciences, Toki 509-5292, Japan 2 Department of Applied Physics, Okayama University of Science, Okayama 700-0005, Japan

E-mail contact of main author: tokitani.masayuki@LHD.nifs.ac.jp

Abstract. Copper alloy has been considered as a divertor cooling tube or heat sink in not only helical reactor FFHR-d1 but also tokamak DEMO reactor, because it has a high thermal conductivity. This work focused on applying an oxide dispersion strengthened copper alloy (ODS-Cu), GlidCop® (Cu-0.3wt%Al2O3) as the divertor heat sink material of FFHR-d1. This alloy has superior high temperature strength over 300 MPa even after an annealing up to ~1000˚C. The changeable material properties of Pure-Cu, GlidCop® and CuCrZr by neutron irradiation are summarized in this paper. A primary dose limit is the radiation-induced hardening/softening (~0.2 dpa/1-2 dpa) which has a temperature dependence. If the temperature of the GlidCop® is maintained at 300˚C, radiation-induced hardening/softening would be moderated. According to such an evaluation, the GlidCop® can be selected as the current best candidate material in the commercial base of the divertor heat sink, and its temperature should be maintained as close as possible to 300˚C during operation. Joining between W armour and GlidCop® heat sink was successfully performed by using an improved brazing technique with BNi-6 (Ni-11%P) filler material. The joint strength was measured by three-point bending test, and reached to approximately 200 MPa. Surprisingly, several specimens showed obvious yield point. This means that the BNi-6 brazing layer caused relaxation of the applied stress. The small divertor mock-up of the W/BNi-6/GlidCop® was successfully fabricated by using the improved brazing technique. The heat loading test was carried out by the electron beam device ACT2 in NIFS. The highest temperature under 15 MW/m2 is ~650˚C in the W plate. The temperature profile is quite reasonable for modelling calculation by a finite element method. The design of the W/BNi-6/GlidCop® showed an excellent heat removal capability for use in the FFHR-d1 divertor.

1. Introduction

According to the current design of the reactor divertor, the average heat flux is estimated to be around 10 MW/m2 [1-3]. Under heavy neutron irradiation environment in a fusion reactor, a Reduced Activated Ferritic/Martensitic (RAFM) steel, such as F82H, is one of the candidate materials for a divertor heat sink or a cooling pipe [4], because it has high tolerance against neutron irradiation. However, RAFM steel cannot withstand being in a reactor divertor under heat loading of around 10 MW/m2 because of its high self-induced internal thermal stress [3]. Under such a condition, copper alloy with high thermal conductivity has a large advantage. From the recent assessment of the neutronics environment, copper alloys could be available in a divertor heat sink in both FFHR-d1 and DEMO reactor [1,2,5]. The CuCrZr is a precipitation hardened copper alloy (PH-Cu), which was selected as a cooling pipe of the ITER divertor. In the selection of the ITER divertor material, the lower fracture toughness of the oxide dispersion strengthened copper alloy (ODS-Cu) after neutron irradiation at 250°C for a dose of ~2.5 dpa was a concern [6]. On the other hand, the toughness of the CuCrZr at a high temperature over 450˚C is dramatically decreased [7 ], but GlidCop® (Cu-0.3wt%Al2O3) has superior high temperature strength over 300 MPa even after an annealing up to ~1000˚C [7]. Such characteristics provide two important advantages. The first advantage is the redundancy of the

2 FIP/P4-37

temperature margin of the divertor operation even when an unexpected temperature excursion occurs. The second advantage is the brazing procedure itself, since rapid cooling down phase is not necessary at the final stage of the brazing heat treatment. The rapid cooling down phase would induce an undesired thermal stress to the material. According to the above facts, the GlidCop® can be selected as the current best candidate material in the commercial base of the divertor heat sink.

Tungsten (W) block is being considered as the divertor armour material. W has large advantages, such as low hydrogen isotope retention and low sputtering yield. In our previous work, a direct brazing test between W and GlidCop® was carried out by using three kinds of filler materials as listed in TABLE I. In that work, BNi-6 showed the best joint strength of around 200 MPa among three filler materials. However, its stress-strain curve showed brittle fracture properties [3]. If the brittle property is nicely changed to the ductile property, a reliability of the bonding layer will be greatly improved. The brazing technique in this work was the direct brazing between armour and heat sink without any intermediate materials. This leads to the large advantages for not only maintaining a good thermal conductivity from armour to heat sink but also reducing the manufacturing cost of the divertor components. In the present work, the improved brazing technique was applied to W/BNi-6/GlidCop® bonding combination, and we determined the better brazing condition. Then, the small-scale divertor mock-up of the W/BNi-6/GlidCop® was fabricated. Finally, the heat loading test was carried out by using the electron beam device Active Cooling Test-stand 2 (ACT2) at the National Institute for Fusion Science (NIFS) [8].

In this paper, first, we would like to review the degradation of the material properties under neutron irradiation of the copper alloys. Then, fabrication of a divertor mock-up with ODS-Cu and W by improved brazing technique is presented.

2. Operational Window of ODS-Cu under Neutron Irradiation

The possible candidate copper alloys for fusion reactor use are mainly categorized as an ODS-Cu (e.g., GlidCop®) and a PH-Cu (e.g., CuCrZr) [7,9-16]. The precise assessment of the radiation threshold for safe use of copper alloys under neutron irradiation is difficult, because not only energy spectrum and dose of the neutron but also material temperature are complicatedly related to the degradation of the material properties. As reported in the previous work [3], the changeable material properties by neutron irradiation are (1) radiation-induced hardening/softening [7,9-11], (2) embrittlement by transmuted helium [12,13], (3) degradation of the thermal conductivity by transmutation products [14], and (4) void swelling [15,16]. Dose

TABLE I: Chemical composition [%] and solid/liquid phase temperature of the tested filler materials [3].

Filler materials

Solid phase temperature

Liquid phase temperature Cr Cu Mn Ni P Si Fe B

BNi‐6 875˚C 875˚C bal. 11

MBF‐20 969˚C 1024˚C 7 bal. 4 3 3

Nicuman37 880˚C 925˚C 52.5 38 9.5

TABLE II: Dose limit of four changeable material properties by neutron irradiation and threshold temperature of radiation-induced hardening/softening for GlidCop®, CuCrZr and pure-Cu [7,9-16].

Cu and Cu alloys

Yield strength at room temp.

Threshold temp. of radiation‐induced hardening/softening

Dose limit for radiation‐induced hardening

Dose limit for radiation‐induced softening

Embrittlement by transmuted He

Degradation of the thermal conductivity by transmutation products

Void swelling properties

GlidCop®(ODS‐Cu)

>400 MPa 300˚C  ~0.2 dpa 1~2 dpa(slowly)

6 dpa@350˚C

(assumed at 40appm limitwith7appm/dpa)

20%(10 dpa)

100dpa @400˚Cbelow 1%

CuCrZr(PH‐Cu)

>400 MPa 280˚C ~0.2 dpa ~1 dpa 20%(10 dpa)

100dpa @400˚Cbelow 2%

Pure‐Cu ~60 MPa ‐‐‐ ~0.1 dpa ‐‐‐ 20%(10 dpa)

50dpa @400˚C25%

3 FIP/P4-37

limit of four changeable material properties and threshold temperature of radiation-induced hardening/softening for GlidCop®, which are referred from many previous papers [7,9-16], are summarized in TABLE II. Data of the CuCrZr and pure-Cu are also listed for comparison. A primary dose limit of GlidCop® is the radiation-induced hardening/softening (~0.2 dpa/1-2 dpa), and it has a temperature dependence. In the case of the radiation-induced hardening, the saturated radiation-induced dislocation loops and stacking fault tetrahedras (SFTs) heterogeneously prevent a motion of the dislocations. Then, the ability of the homogenous elongation is lost. On the other hand, in the radiation-induced softening, some of the dispersed (ODS-Cu) or precipitated (PH-Cu) particles in the Cu matrix are dissociated by neutron irradiation. Therefore, original dislocations and voids are terminated due to the radiation-induced recrystallization. Consequently, the obstacles for the movable dislocations disappear and, finally, materials are softened [3,10]. As shown in TABLE II, if the temperature of the GlidCop® is maintained at 300˚C, radiation-induced hardening/softening would be moderated. On the other hand, same dose limit (radiation-induced hardening/softening) and its threshold temperature of CuCrZr comes at ~0.2 dpa/~1 dpa and 280˚C, respectively. GlidCop® has a small margin for radiation-induced softening and high temperature use compared with CuCrZr. According to the above evaluation, the GlidCop® can be selected as the current best candidate material in the commercial base of the divertor heat sink, and its temperature should be maintained as close as possible to 300˚C during operation.

3. Improved brazing technique of W/BNi-6/GlidCop®

Based on the previous work [3], we sought to improve the direct brazing technique between W and GlidCop® by using BNi-6 filler material with the thickness of 38 m. The main objective of this improvement is that mechanical property of a bonding layer is changed from a brittle to a ductile property with maintaining bonding strength of over 200 MPa without any intermediate materials. The size of the W and GlidCop® for brazing are 20 × 20 × 18 mm3 and 20 × 20 × 38 mm3, respectively. Each 20 × 20 mm2 plane is the brazing surface. The W block was naturally placed on the BNi-6 filler material without any forces.

FIG. 1-(a) and -(b) show the heat treatment procedures in the previous work [3] and in the improved present work, respectively. We introduced three improvement points as shown in (1), (2) and (3) in FIG. 1-(b). The points are (1) treatment temperature was decreased from 980˚C to 960˚C, (2) Treatment time was reduced from 45min to 10min, and (3) Slow cooling rate (natural cooling) was adopted. Point (3) is the most advantageous of this improved brazing

FIG. 1. Procedures of the brazing heat treatment in the previous work (a) and the present work (b).

60min 10min

960˚C

860˚C

N2 Gas cooling

Natural cooling to 100˚C

(1)

(2)

(3)

60min 45min

980˚C

860˚C

N2 Gas cooling

(a)

(b)

FIG. 2. Photograph of the brazed specimen of W/BNi-6/GlidCop®.

4 FIP/P4-37

technique for making the reliable bonding. Slow cooling rate cannot be applied for CuCrZr because the rapid cooling down phase by gas cooling from near the melting temperature is required for maintaining the original strength. FIG. 2 shows the brazed specimen of W/BNi-6/GlidCop®. The BNi-6 brazed layer is located in the center of the photograph. A very thin brazed layer without any cavities can be observed.

After the heat treatment procedures, the brazed blocks were fabricated as small size specimens of 36 × 5 × 1.5 mm3. Then, a three-point bending test was carried out by using the SHIMADZU Autograph at Okayama University of Science. FIG. 3 shows six sets of stress-strain curves of W/BNi-6/GlidCop® brazing specimens. Most of the specimens achieved the maximum strength of around 200 MPa, and several others showed obvious yield point. Since the yield strength of GlidCop® is over 300 MPa even after the heat treatment of FIG. 1-(b), such a ductile fracture property was achieved by local deformation of the bonding layer itself. Namely, the bonding layer caused relaxation of the applied stress. This result demands attention because it could be generally considered that the bonding layer itself is brittle. Such a relaxation capability of the applied stress by the bonding layer contributes greatly to the reliability of the divertor components when they accept the unexpected thermal stress during reactor operation. The low fracture strength below 150 MPa (see No. 2 curve in FIG. 3) seems to be due to the positional dependence in the strength of the interface between filler and bulk material.

4. Fabrication of Divertor Mock-up and Evaluation of Heat Removal Capability

The small-scale divertor mock-up of the W/BNi-6/GlidCop® was successfully fabricated by the improved brazing technique as shown by the photograph in FIG.4. Taking into account the applicability for the three-dimensional helical divertor, the flat plate W armour with the thickness of 5 mm has been considered. BNi-6 filler materials of 38 m thick were inserted between W armour and GlidCop®. During the brazing heat treatment procedures, the W armours were naturally placed on the BNi-6 filler materials without any forces.

For evaluating the heat removal capability under the reactor relevant condition, the heat loading test was carried out by using the electron beam device ACT2 with water cooling. Pressure, flow rate and temperature of the cooling water were ~0.3 MPa, ~52 L/min and ~25˚C, respectively. FIG. 5 shows the temperature profile of the small scale

FIG. 3. Stress-strain curves of W/BNi-6/GlidCop® brazing specimens.

0

50

100

150

200

Str

ess

[MP

a]

Strain [%]

No. 1

No.2

No.3No.4

No.5No.6

0.25%

FIG. 4. Small-scale divertor mock-up fabricated by W/BNi-6/GlidCop® brazing.

FIG. 5. Temperature profile of the W/BNi-6/GlidCop® small-scale divertor mock-up in FIG. 4 during a steady state heat loading.

0 3 6 9 12 15

100

200

300

400

500

600

700

Tem

pe

ratu

re [

deg

.]

Heat load [MW/m2]

CH1 CH2 CH3 CH4 CH5 CH6 CH7 CH8

[°C

]

5 FIP/P4-37

divertor mock-up measured by thermocouples during steady sate heat loading. The positions of the thermocouples from CH1 to CH8 correspond to the red marker positions in FIG. 4. All thermocouples were embedded to the horizontal direction up to the depth of 10 mm from the red marker. The highest temperature under 15 MW/m2 is ~650˚C in the W plate. The temperature profile is quite reasonable for modelling calculation by a finite element method. This temperature is quite acceptable for this mock-up from the viewpoint of structural reliability. After the 15 MW/m2 heat loading, the mock-up did not show any sign of a visible destruction.

The design of the W/BNi-6/GlidCop® showed an excellent potential for using in the FFHR-d1 divertor.

5. Summary

According to the intrinsic mechanical properties and assessment of the neutron irradiation properties, ODS-Cu (GlidCop®) was selected for the divertor heat sink of fusion reactors in this study. The flat plate W is being considered for the armour material, and it is expected to be bonded to the GlidCop® through the brazing procedures. In this work, the direct brazing between armour and heat sink without any intermediate materials was selected. This leads to the large advantages for not only maintaining a good thermal conductivity from armour to heat sink but also for reducing the manufacturing cost of the divertor components. The reliable bonding layer was able to be obtained by applying an improved brazing technique with BNi-6 filler material. The joint strength of W/BNi-6/GlidCop® was confirmed by three-point bending test, and reached approximately 200 MPa. Surprisingly, several stress-strain curves showed obvious yield point. This means that the bonding layer itself caused relaxation of the applied stress. Such a relaxation capability of the applied stress has great merit from the viewpoint of the reliability of the divertor components even when they accept the unexpected thermal stress during the reactor operation. Then, the small-scale divertor mock-up of the W/BNi-6/GlidCop® was successfully fabricated by the improved brazing technique. The heat loading test under the reactor relevant condition was carried out by using the electron beam device ACT2 in NIFS. The temperature of the mock-up was an acceptable value from the viewpoint of maintaining a structural reliability even when the input power was 15 MW/m2. After the heat loading test, the mock-up did not show any sign of a visible destruction. The design of the W/BNi-6/GlidCop® showed an excellent heat removal capability for use in the FFHR-d1 and tokamak DEMO divertor.

6. Acknowledgements

This study was conducted as a Fusion Energy Research Project (FERP) in National Institute for Fusion Science (NIFS). The authors would like to thank the Metal Technology Co. Ltd. for brazing.

References

[1] A. Sagara et al., Fusion Eng. Des. 89 (2014) 2114-2120.

[2] K. Hoshino et al., Proc. IAEA FEC (2014) FIP/P8-11.

6 FIP/P4-37

[3] M. Tokitani et al., Plasma Fusion Res. Vol.10 (2015) 3405035.

[4] S. Suzuki et al., Fusion Eng. Des. 81 (2006) 93-103.

[5] T. Tanaka et al., Fusion Eng. Des. 89 (2014)1939-1943.

[6] G. Kalinin, R. Matera, J. Nucl. Mater. 258-263 (1998) 345-350.

[7] S.J. Zinkle et al., Fusion Materials (1994) DOE/ER-0313/16.

[8] Y. Hamaji et al., Plasma Fusion Res. Vol.11 (2016) 2405089.

[9] B.N. Singh and S.J. Zinkle, J. Nucl. Mater. 206 (1993) 212-229.

[10] B.N. Sing et al., J. Nucl. Mater. 238 (1996) 244-259.

[11] S.A. Fabritsiev et al., J. Nucl. Mater. 283-287 (2000) 523-527.

[12] S.A. Fabritsiev et al., Fusion Eng. Des. 36 (1997) 505-513.

[13] T. Muroga, H. Watanabe, N. Yoshida, J. Nucl. Mater. 258-263 (1998) 955-960.

[14] S. Ishino et al., J. Nucl. Mater. 283-287 (2000) 215-219.

[15] F.A. Garner and M.L. Hamilton, J. Nucl. Mater. 191-194 (1992) 386-390.

[16] D.J. Edwards et al., J. Nucl. Mater. 191-194 (1992) 416-420.