Numerical analyses of the W7-X ECRH gyrotron diamond output window and overview on the residual stress analysis of MPA CVD diamond disks
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Repository KITopen (Karlsruhe Institute of Technology) |
| Published | 2026-07-30 |
| Authors | G. Aiello, H. P. Laqua, F. Légrand, A. Meier, T. Scherer |
| DOI | 10.5445/ir/1000195808 |
The numerical results support 2 MW operation of the W7-X ECRH gyrotron with a water-cooled CVD diamond output window. The authors found that 5.7E-5 was the maximum loss tangent that might remain tolerable at 2 MW in the modeled configurations. The reference structural case predicted stresses below the 150 MPa allowable limit and a disk bow of about 45 µm. These conclusions depend on temperature-dependent diamond properties and an assumed 250 °C limit.
Operating point and measurement inputs
Section titled “Operating point and measurement inputs”The earlier window design used a Gaussian mm-wave beam with a 20 mm radius and a symmetric CFD model. The modeled coolant flow was 0.2 kg s⁻¹, with a 25 °C inlet and a 0 Pa outlet pressure. Simulations varied absorbed power and compared water with DC 200(R) 5 cSt silicon oil. The window model used a 140 GHz beam, a relative permittivity of 5.67, and a disk thickness of 1.8 mm.
Measured loss tangent values supplied the power-absorption input. KIT used Fabry-Perot resonators in spherical and hemispherical setups to measure the loss tangent of the diamond disks. The authors treated the resulting thermal reserve as protection against degradation factors that could reduce the disk’s microwave transmission.
Cooling and loss tangent limits
Section titled “Cooling and loss tangent limits”The earlier cooling comparison used a 250 °C limit for diamond. Its results slide is labeled 1.5 kW absorption, while the surrounding design material describes 1.5 MW operation. The unit mismatch should be resolved before that comparison is used quantitatively. Oil cooling was rejected for the high-power window, and water cooling was required at high power.
The parametric loss-tangent study found substantially greater temperature reserve with water, even when absorbed power increased. The authors said this reserve was needed to absorb possible losses in microwave transmission caused by disk degradation.
The 2 MW analysis combined different loss tangents and beam radii. It found that a larger beam radius could reduce maximum disk temperature at higher absorbed powers. The resulting upper limit for loss tangent was 5.7E-5 at 2 MW. The window summary supports water cooling and temperature-dependent material data. It also treats beam radius as a design variable for limiting peak disk temperature.
Structural response at 2 MW
Section titled “Structural response at 2 MW”The structural reference case combined the 20 mm beam radius with a 5.0E-5 loss tangent. The model included the metallic housing and used a steady-state plastic analysis. It applied thermal loads, a coolant pressure of 1 bar, and atmospheric pressure on the disk’s lower surface.
Thermal loading produced about 120 MPa of compressive stress. Pressure loading produced about 47 MPa of tensile stress, and the thermal case governed the total stress distribution. Adding pressure loads to thermal loads relaxed stress on the disk’s lower side. The authors judged the stresses acceptable against the 150 MPa allowable limit. They also reported a disk bow of about 45 µm. These values apply to the stated reference case rather than every loss tangent and beam radius combination.
Polarimeter and Raman stress mapping
Section titled “Polarimeter and Raman stress mapping”The residual stress work examined a 1.11 mm thick, 80 mm diameter optical-quality MPA CVD diamond disk. It also used 30 mm diameter optical-quality samples at various thicknesses. Polarimetry estimated the principal stress difference from retardation, sample thickness, and the stress-optical coefficient. It detected the largest stress difference in the outer region and showed an axially symmetric stress distribution that varied with radius.
Raman spectroscopy used a LabRam HR Evolution spectrometer and the diamond line at 1332 cm⁻¹. Initial points covered the center and outer region of a GS sample. The stated settings were 532 nm excitation, 10% laser power, a 2400 g/mm grating, 10x magnification, and a 50 µm aperture. Each point used five 3 s acquisitions and Lorentz fitting.
Polarized Raman images showed large point-to-point changes in peak position. The authors related lower Raman shifts to tensile stress and upper shifts to compressive stress. They observed tensile stress at the center and a mixture of tensile and compressive stress in the outer region. The initial Raman survey covered local center and outer regions, not a full-area thickness series.
XRD results and measurement limits
Section titled “XRD results and measurement limits”X-ray diffraction used the sin²Ψ method on GS and NS samples 300, 440, 800, 1100, and 1820 µm thick. The Empyrean diffractometer used Cu Kα radiation and the (331) peak at 2θ = 140.6°. Measurements covered a 3 mm by 3 mm central area. The authors collected 36 tilt angles spanning -67° to 67°. They fitted peak positions with a modified Lorentzian function. They then obtained stress from the d versus sin²Ψ slope and material elastic constants.
The GS samples produced larger scatter than the NS samples. At GS thicknesses of 1100 and 1820 µm, only about half of the measurement points could be used. The authors related these limitations to columnar growth and large grains. They said the sin²Ψ method might apply only to NS material. They proposed slight sample oscillation to increase the number of contributing grains. They stated that this was feasible at KIT Campus South.
No clear stress trend appeared with thickness or between the two sample types. The reported standard deviation was large, with texture and columnar growth identified as possible causes. The XRD results therefore provide a measurement method, but not a reliable thickness law for these samples.
Implications for material selection
Section titled “Implications for material selection”The window results show that loss tangent, coolant choice, beam radius, and material data must be selected together. The 5.0E-5 structural reference is below the 5.7E-5 calculated limit at 2 MW. The calculation also shows why temperature-dependent properties must accompany the loss-tangent input. The 2 MW analysis therefore requires water cooling and a loss tangent below the calculated limit.
Residual stress characterization must also account for sample surface, radial position, grain structure, and the measurement method. Polarimetry and Raman resolved spatial differences that XRD could not summarize reliably. The authors called for a systematic stress investigation over each sample area. This supports inspection plans that distinguish radial and surface effects instead of treating one measurement as representative.
The 2 MW performance claim remains numerical, and the structural numbers describe one reference load case. The residual stress findings are initial measurements. The assumed 250 °C limit, sample-dependent XRD scatter, and incomplete area coverage remain the main limits on transferring the results to a specific disk.
Materials and support
Section titled “Materials and support”The W7-X gyrotron needs a water-cooled, low-loss CVD diamond output window at 2 MW to remain below the assumed 250 °C limit. The diamond materials catalog lists single-crystal and polycrystalline diamond grown by MPCVD, with plates and wafers up to 125 mm and substrates up to 10 mm. 6CCVD’s engineering support can be used to discuss custom fabrication and engineering requirements for the paper’s disk geometry and output-window cooling interface.
View Original Abstract
Numerical analyses of the W7-X ECRH gyrotron diamond output window and overview on the residual stress analysis of MPA CVD diamond disks
Single Crystal Cvd Diamond Plate
View Specs →