Skip to content

Expanding the TM01-Mode MPCVD Reactor Based on Electromagnetic Mode Amplification for Potential 4-Inch Diamond Deposition

FieldValue
JournalProcesses
Published2026-02-13
AuthorsJialiang Yang, Yang Yong, Pan Tang, Chengshu Shen, Xiaoshan Peng
DOI10.3390/pr14040645

An expanded TM01-mode MPCVD reactor design generated a simulated 4-inch-diameter plasma sphere while retaining TM01 single-mode transmission with and without plasma. The model covered cylindrical chamber radii of 75 to 85 mm. At 5000 Pa and 1000 W, the modeled plasma nearly covered a 110 mm substrate after 100 s. The result is a reactor simulation, not an experimental 4-inch diamond film.

Mode amplification preserves the TM01 field

Section titled “Mode amplification preserves the TM01 field”

Simply enlarging a circular reactor introduces higher-order modes. At 2.45 GHz, the free-space wavelength is about 122.45 mm. Calculated mode-existence radii are 46.74 mm for TM01, 74.66 mm for TM11, and 107.41 mm for TM02. Chamber radii of 75 to 85 mm therefore permit TM11 propagation by geometry alone.

The authors placed a profiled trumpet waveguide between a 107 mm inner-diameter circular waveguide and the larger reaction chamber. Its contour selectively amplified TM01 and suppressed TM11 and TM02. A simulation at 1 W input showed a TM01 electric field with no significant higher-order content at an 80 mm reactor radius.

The chamber also used a concave substrate holder with a 55 mm radius and 10 mm height. It strengthened the field above the substrate without changing the chamber’s TM01 distribution.

They interpret the final results as tolerance to machining variation, thermal deformation, and substrate-height changes. Their evidence is numerical, based on field and plasma simulations.

The hydrogen plasma model tests discharge behavior

Section titled “The hydrogen plasma model tests discharge behavior”

The authors added a coupled hydrogen-discharge model because electromagnetic simulations describe the reactor before ignition. The Helmholtz equation supplied the microwave field at 2.45 GHz. Transport equations tracked charged and heavy species. An energy balance calculated electron and heavy-species temperatures.

The authors summarize the chemistry as 35 electron-collision reactions, seven heavy-species reactions, and six surface reactions. The pure hydrogen model was used because methane content during diamond growth is typically below 5%.

A two-dimensional axisymmetric domain reduced computation based on TM01 field symmetry. The quartz window had a relative dielectric constant of 3.78 and was positioned 158 mm from the substrate surface. Metal walls used perfect electric conductor boundaries, while the heavy-species wall temperature was fixed at 350 K.

The model used COMSOL Multiphysics 6.3 and 11,101 grid cells. Each run took about 1 to 2 h on a computer with an i7-10700K CPU and 128 GB RAM. Time advanced from (1 \times 10^{-9}) s to 100 s, covering breakdown through a near-steady discharge.

Operating sweeps established how the expanded reactor responds after discharge begins. The table summarizes comparisons at the 100 s simulation state.

Operating caseGas pressureMicrowave inputSimulated response
Power sweep5000 Pa1000, 1200, and 1400 WThe plasma ball shifted toward the quartz window as power increased.
Pressure sweep5000, 7000, and 9000 Pa1000 WThe plasma sphere diameter decreased as pressure increased.

The authors attribute the power response to higher electron density, reduced skin depth, and stronger microwave reflection. The reflected wave raises the standing-wave peak above the plasma, where stronger ionization shifts the ball toward the quartz.

At higher pressure, shorter molecular mean free paths increase collision frequency and restrict particle motion. These trends agreed with experimental observations and prior simulations cited by the authors.

Geometry changes leave the 4-inch plasma nearly unchanged

Section titled “Geometry changes leave the 4-inch plasma nearly unchanged”

The geometry sweep compared chamber diameters of 150, 160, and 170 mm at 5000 Pa and 1000 W. Plasma extent was defined by the hydrogen-atom contour at (I_b = I_{\max}/e), with (e = 2.718). The maximum hydrogen-atom number density is (I_{\max}).

As chamber diameter increased, hydrogen-atom density at the plasma center decreased. The plasma size remained almost unchanged, while its distribution became more uniform. The authors interpret this as improved uniformity with a larger chamber.

Across this comparison, the plasma nearly covered the 110 mm circular substrate. This established a 4-inch sphere by the paper’s contour method.

Electromagnetic simulations retained TM01 transmission before discharge and at the steady plasma state. Varying substrate height raised the center hydrogen-atom density but left the plasma size relatively unchanged. The authors use these results as evidence of geometric robustness, not as proof against all manufacturing or operating tolerances.

Comparison with multimode and high-order reactors

Section titled “Comparison with multimode and high-order reactors”

Conventional TM01 reactors are limited by cavity size and wavelength, so their plasma spheres generally do not exceed 4 inches. Larger TM0n cavities with n ≥ 2 and multimode cavities permit wider plasma coverage. Their fields become more sensitive to cavity dimensions, machining accuracy, thermal deformation, and operating conditions.

The paper cites a TM02 plus TEM reactor that deposited 75 mm diamond. It also cites a TM01 plus TM02 reactor that deposited 80 mm diamond. A three-mode cylindrical design deposited 75 mm diamond on an 80 mm substrate.

An ellipsoidal TM03 design generated a 4-inch plasma over a 100 mm substrate. A separate 2.45 GHz, 10 kW system deposited a 4-inch free-standing diamond film.

The present result matches the 4-inch target but differs in mode control. Its simulation produced a 4-inch plasma across a 110 mm substrate while maintaining TM01 across a 75 to 85 mm radius range. No film thickness, growth rate, defect density, or spatial uniformity was reported for the proposed reactor.

The central limitation is chemistry. The multiphysics model uses pure hydrogen and omits the carbon source methane. It also neglects non-equilibrium plasma behavior at higher pressures. The authors state that these omissions may cause deviations from actual reactor performance.

Other assumptions constrain the result. The calculation is two-dimensional and axisymmetric, uses perfect electric conductor walls, and fixes important thermal boundary conditions. The 100 s state is described as near steady state rather than a measured operating equilibrium.

The paper reports no fabricated reactor, no deposited diamond film from this design, and no direct experimental validation of the 4-inch plasma. The strongest evidence is numerical consistency with known pressure and power trends.

Implications for diamond process development

Section titled “Implications for diamond process development”

The model indicates the potential for larger-area MPCVD deposition without a multimode cavity. The authors suggest that maintaining TM01 could reduce sensitivity to chamber dimensions and substrate position. The simulations support that interpretation because plasma size changed little in the tested geometry sweeps.

For material selection, the paper identifies no diamond grade, film thickness, growth rate, or defect level. Those material properties require growth trials because the study models plasma behavior rather than carbon deposition. Process development must establish gas composition, pressure, microwave power, and thermal conditions in an H2/CH4 mixture.

An experimental program would need to confirm the plasma footprint and quantify film thickness and spatial uniformity across the 110 mm area. The present design is therefore a candidate for 4-inch reactor trials, not a validated deposition process.

The paper’s process needs MPCVD-grown diamond substrate material for a 4-inch deposition target modeled on a 110 mm substrate. 6CCVD’s diamond materials catalog lists single-crystal and polycrystalline MPCVD diamond plates and wafers up to 125 mm, which covers that substrate scale. The engineering support page can help define the substrate geometry and process requirements for reactor trials.

View Original Abstract

Diamond is becoming an increasingly popular substrate material in the semiconductor industry due to its high thermal conductivity and wide forbidden band characteristics. With the development of high-power electronic devices, the demand for large-area single-crystal diamond films is also dramatically increasing. Microwave plasma chemical vapor deposition (MPCVD) technology is the dominant method for producing high-quality diamond films due to its advantages of high controllability, fast deposition rate, and low contamination. Despite the excellent performance of MPCVD reactors in various aspects, the question of how to increase the plasma size while improving its homogeneity remains a challenge for device design and optimization. This paper proposes a method to expand the geometrical sizes of the TM01-mode MPCVD reactor while maintaining the mode’s axisymmetric homogeneity. The size-enlarged TM01-mode MPCVD reactor was first designed and optimized with electromagnetic simulations. A multiphysics model that accounted for the microwave field, hydrogen gas discharge, and energy conservation was proposed to evaluate the performance of the MPCVD reactor afterwards. The results demonstrate that the size-enlarged MPCVD reactor can generate a plasma sphere with a diameter of 4 inches while still maintaining TM01-mode single-mode transmission, either with or without plasma. Its outstanding robustness and adaptability underlie excellent potential for large-area diamond thin-film deposition.

  1. 2018 - Thermal conductivity of high purity synthetic single crystal diamonds [Crossref]
  2. 2021 - Diamond composites of high thermal conductivity and low dielectric loss tangent [Crossref]
  3. 2012 - A promising solution using CVD diamond for efficient cooling of power devices [Crossref]
  4. 2022 - Diamond Semiconductor and Elastic Strain Engineering [Crossref]
  5. 2024 - Effect of growth rate on the incorporation of silicon impurity in single crystal diamond [Crossref]
  6. 2022 - Chemical vapor deposition single-crystal diamond: A review [Crossref]
  7. 2019 - Diamond power devices: State of the art, modelling, figures of merit and future perspective
  8. 2006 - High rate homoepitaxial growth of diamond by microwave plasma CVD with nitrogen addition [Crossref]