Skip to content

On the fracture toughness and structural integrity assessment of MPA CVD diamond in nuclear fusion devices

FieldValue
JournalRepository KITopen (Karlsruhe Institute of Technology)
Published2026-07-30
AuthorsG. Aiello, P. Estebanez, B. Gorr, A. Meier, T. Scherer
DOI10.5445/ir/1000195806

The first structural assessment covered a 70 mm, 1.11 mm optical quality polycrystalline MPA CVD diamond disk. It found a maximum calculated stress below 150 MPa under the worst modeled ITER load. A preliminary crack calculation gave K of about 2.6 MPa m^1/2 for an approximately 320 µm flaw. The paper compares this with a literature KIC of about 8 MPa m^1/2 cited from Davies in 2004. These are modeled results rather than a measured fracture toughness result. The planned double torsion experiments had not yet been performed.

The authors treat the optical disks in the ITER EC H&CD system as safety important. They identify disk fracture as a failure mode that must be assessed against crack propagation. Published MPA CVD values cover different thicknesses, specimen shapes, and test methods. They therefore do not represent a common specimen or test condition.

ReferenceMaterial and thicknessKIC (MPa m^1/2)Test methodSamples
1991MPA CVD, 400 µm5.3 ± 1.3Indentation11
1995MPA CVD, 150 to 200 µm6.3Tensile test, E-3992
1995MPA CVD, 150 to 200 µm5.6 ± 0.4Indentation8
1998MPA CVD, 880 µm8.7 ± 0.3Double torsionNot listed
2004MPA CVD optical diamond, 1000 µm8.3 ± 0.4Double torsion5
2004MPA CVD mechanical diamond, 1000 µm8.5 ± 1Double torsion2

The 8.3 ± 0.4 result for 1 mm optical diamond came from five 13 x 18 mm rectangular specimens tested by double torsion. The table’s 2004 rows also report 8.5 ± 1 for 1 mm mechanical diamond using two specimens.

The survey also reports 6.5 ± 1.2 and 7.6 ± 1.8 for arc discharge CVD. Its arc plasma jet and hot filament values were 4.6 and 6.8 ± 1.1. Natural type Ia and IIa diamond gave 3.4, while cobalt phase PDC gave 13. The authors note limited fracture toughness literature for MPA CVD.

Double torsion was selected because its specimen and loading arrangement are relatively simple. The method has been applied to ceramics, glasses, composites, steels, and diamond, but it is not yet standardized.

At failure load, the analysis assigns K equal to KIC. The planar configuration makes symmetry and alignment important.

ParameterSpecimen value
Thickness, t1.11 mm
Width, W15 mm
Length, L30 mm
L/W2.0
W/t13.5

The selected disk thickness is 1.11 mm, while specimen width W is 15 mm and length L is 30 mm. This gives L/W of 2.0 and W/t of 13.5. The authors set the crack length ratio a/L at 30% to 40% as the best compromise for sample size.

Preliminary laser studies assessed cuts from the top and from the side. The frequency doubled Nd:YAG laser was green and operated at 13 W. One trial used a cutting speed of 10 mm/s and set the cut length to 8 mm.

Later trials used speeds of 0.05 mm/s and 1 mm/s with different focus positions. The authors describe laser precracking as promising for diamond. The presentation reports no completed precrack validation.

The characterization program covers loss tangent, EBSD, Raman, XRD, and polarimetry. Reported grain size examples came from a Ø150 mm disk described as thermal and a Ø80 mm disk described as optical.

The growth side of the thermal disk had an average grain size of about 50 µm by number and 200 µm by area. The nucleation side of the optical disk averaged about 6 µm by number and 9 µm by area. These examples show different grain statistics for the thermal and optical disks and for the sampled sides. The authors report that characterization activities for the diamond samples were ongoing.

The microscopy procedure was developed for 60 ITER disks, each 70 mm in diameter and 1.11 mm thick. Extended Focal Imaging produced a single image containing microfeatures over the disk surface. The workflow includes the brazing area and uses an Equivalent Diameter Criterion.

For aperture area S-T30, the reported equivalent diameter was 322 µm. That value was used as the flaw size in the first integrity estimate. Continued microscopy of the ITER disks remained part of the work plan.

The worst load scenario was an NO plus 2 bar event on the torus side. The authors used CFD conjugated heat transfer with structural analysis. The HE11 beam power was 1.31 MW, and the absorbed power Pabs was 346 W.

The analysis placed the maximum disk stress below the 150 MPa allowable limit. The integrity calculation then assessed the growth side. This model covers thermal loading and stress in the disk. It assesses the load case separately from the planned crack propagation tests.

The first estimate used a local stress of about 84 MPa and a flaw size of about 320 µm. It set the geometry factor β at about 1 in the first order expression K = βσ0(πa)^1/2. The resulting stress intensity was K of about 2.6 MPa m^1/2.

The paper then compared it with KIC of about 8 MPa m^1/2, cited as Davies in J. Mater. Sci., 2004. Its literature table reports 8.3 ± 0.4 for 1 mm MPA CVD optical diamond and 8.5 ± 1 for 1 mm mechanical diamond from that year. The calculated K remained below the selected reference for the modeled case. The integrity criterion was K no greater than KIC.

The cited values came from 13 x 18 mm rectangular specimens, not from the 70 mm optical disk. The paper identifies mechanical characterization against crack propagation as required work. Double torsion tests, setup drawings, and continued disk microscopy remained unfinished.

The numerical result checks one load and flaw case against a literature reference. Device engineering still needs fracture data for the actual disk grade before this case can support qualification.

The paper’s ITER EC H&CD application uses an optical quality polycrystalline MPA CVD diamond disk with a 70 mm diameter and 1.11 mm thickness, plus a defined brazing area and surface inspection. The 6CCVD diamond materials catalog includes polycrystalline diamond grown by MPCVD in plates up to 125 mm, so the disk’s lateral size falls within the listed range. The paper’s thickness, optical quality, surface condition, and brazing interface should be confirmed through engineering support before a material is specified.

View Original Abstract

Nuclear fusion power plants require electron cyclotron (EC) heating and current drive (H&CD) systems to deploy MW-class, mm-wave radiation, generated by the gyrotrons, to the plasma for heating and control of instabilities. Diamond windows are used as confinement and vacuum barriers on the torus side of these systems and as vacuum inserts only at the output of the gyrotrons. The optical grade polycrystalline diamond is produced by microwave plasma assisted (MPA) chemical vapor deposition (CVD) process and it is used in shape of disks with thickness between 1 and 2 mm and diameter in the range of 70 to about 110 mm. Failure due to crack propagation is the main failure mode of these large diamond disks, which have the primary safety function on the torus side to confine tritium and other hazardous materials such as radioactive dust. Microcracks are indeed generated in the disks by the CVD process. In addition, beyond operational loads, the torus windows may be subject to off-normal events like overpressure acting on the disk. This work shows the path to measure the fracture toughness of diamond to use in nuclear fusion and to check the structural integrity of the diamond disks by fracture mechanics. The experimental setup, the samples and their first characterization are presented. Considering the ITER EC torus diamond windows, numerical analyses have been carried out to provide the stresses generated in the disks for the worst load conditions. High resolution microscopy activity is ongoing to detect, measure and classify the microcracks in specific areas of about 60 ITER disks, produced by the same manufacturer and in the same growth conditions. The information on the microcracks size shall be combine with the stress values to check the structural integrity of the disks against failure to fracture.