Skip to content

Stress management of large size (001) and (111) diamond wafers for quantum-optical applications

FieldValue
JournalJournal of Physics D Applied Physics
Published2026-01-13
AuthorsJan Engels, Jürgen Weippert, Jan Kustermann, Sven Mägdefessel, Niklas Mathes
DOI10.1088/1361-6463/ae3778

Stress management produced 2 inch (111) diamond wafers with near-single NV-center alignment across the wafer, and one stack reached an ODMR (T_2) coherence time of 26 µs. The same work produced freestanding, double-side-polished 300 µm (001) wafers. Cyclic growth and etch controlled (001) stress, while a two-mask epitaxial lateral overgrowth stack reversed (111) curvature but cracked during its final 30 µm.

The authors used chemical vapor deposition on Ir/YSZ/Si templates in 915 MHz and 2.45 GHz ellipsoid reactors. An H-plasma treatment lasted 13 min. Most growth occurred at 840 to 860 °C, with O2/H2 at 0.15%. Six routes tested cyclic growth and etch, columnar growth, ELO masks, and a two-mask (111) stack.

Curvature was measured with a Surfcom 1800D. Raman spectroscopy used 532 nm excitation at 1 mW, and PL used 532 nm excitation. Stress came from Raman peak shifts. SEM, EDX, and HRXRD examined defects, mask structures, and crystal quality.

The (001) cyclic process alternated 1.5 h nitrogen-doped growth with 2 h oxygen etching at 840 to 860 °C. The etch used 2% oxygen in hydrogen and formed negative pyramids about 1 µm on a side with {111} facets. Growth used 3040 ppm nitrogen in hydrogen and 1.75% methane in hydrogen.

Repeated cycles produced a 300 µm layer at an effective growth rate of 2.2 µm h−1. Raman mapping found tensile stress over etch pyramids and compressive stress over smooth (001) areas. The measured magnitudes reached 0.7 GPa tensile and 0.6 GPa compressive.

Alternating signs compensated the usual compressive (001) growth stress. The wafer had an 11 m curvature radius and was classified as planar and stable. Raman FWHM averaged 2.2 cm−1, while the 004 rocking-curve width was 0.13 to 0.15°. The final 300 µm (001) wafer was freestanding without Si and double-side polished.

Columnar growth reduced methane to 1.25% and kept the process below 750 °C to retain pyramidal {111} facets. A 10 min biased enhanced nucleation step produced small islands, compared with 30 min for standard growth. The columns formed freestanding planar wafers with a 16 m curvature radius.

The authors suggested that carbon compounds at small-angle grain boundaries likely caused the opacity. Raman FWHM was 4.5 cm−1, and the 004 rocking-curve width was 0.74°. This route was not tested on (111).

ELO patterns expose orientation-dependent stress

Section titled “ELO patterns expose orientation-dependent stress”

ELO used lithographic Ir masks on the initial template. The pixel design had an 8 µm pitch and 5 µm circles. The net design had a 30 µm line pitch and 5 µm lines. Methods 3 and 4 grew (001) wafers, while method 5 grew (111) wafers.

Method 5 grew nominally undoped diamond for 18 h at 840 to 860 °C, then a nitrogen-doped layer for 2.5 h. The functional layer used 0.5% methane and 5000 ppm nitrogen. It was 2 µm thick, and the ELO-grown layer reached 9 µm.

Raman maps showed opposite stress in nucleation and lateral overgrowth regions for both orientations. The (001) stress was dominated by compressive nucleation regions. For (111), lateral overgrowth dominated the stress, with average compressive stress of 3.3 GPa at the ELO interface. Top-surface averages included 0.75 GPa compressive stress for (001) and 1.1 GPa for (111).

Method 5 changed the (111) wafer from the usual concave curvature to convex and avoided cracks. Methods 3 and 4 did not sufficiently compensate (001) curvature and cracked below 40 µm growth height.

Defect density fell in lateral overgrowth regions. The net pattern produced an ELO area of about 20 × 20 µm² with a Raman FWHM of 2 cm−1. Method 5 reduced FWHM from 28 cm−1 to 4 cm−1.

PL maps showed higher relative negatively charged NV density in lateral overgrowth than in nucleation regions. The authors linked this to lower defect density. Because PL data showed both positive and negative correlations between NV concentration and stress, they excluded nitrogen incorporation as the stress mechanism.

Method 6 added a structured Ir mask 2 over the 9 µm layer from method 5. Mask 2 used 1.5 µm holes, with 1.1 µm spacing within groups and 3 µm between groups. Mask 2 blocked growth over nucleation and coalescence defects from mask 1. Its openings acted as nucleation sites for a second ELO step.

The 140 nm Ir film was patterned with a 100 nm Ni mask. The pattern was etched with an Ar, Cl2, and O2 plasma for 5 min at 1.5 Pa and 160 W RF power. Flows were 35 sccm Cl, 5 sccm Ar, and 10 sccm O2.

The stack reached 100 µm total thickness. At 70 µm, stylus roughness measurements found no cracking. During the final 30 µm, complete layer cracking appeared, likely because the wafer had become convex. The final curvature radius was 6.5 m.

Over mask 2, ELO reduced Raman FWHM by more than four, from 12 cm−1 to 2.7 cm−1. Holes showed 0.3 GPa compressive stress, while lateral regions showed 2.3 GPa compressive stress. The 91 µm overgrowth layer averaged above 1.3 GPa compressive stress. No tensile regions were detected.

Quality improved through the growth height. At depth points of 100, 90, and 10 µm, Raman FWHM was 9.3, 6.3, and 3.7 cm−1, respectively. Mask misalignment left defects in the first layer. The authors said full defect-density validation remained pending and proposed TEM for the unexplained FWHM near mask holes.

ODMR used a custom transmission setup with 532 nm laser excitation. Ramsey and Hahn-Echo sequences measured (T_2). The zero-phonon-line PL map covered the 2 inch (111) method 5 wafer. ODMR confirmed near-single NV alignment across the wafer.

The (001) method 1 wafer showed four NV orientations with a nearly equal distribution. The (111) wafer showed one dominant dip pair, consistent with growth-induced alignment.

Method 5 had a reported (T_2) of 9.3 µs, 7.3 ppm nitrogen, and 0.007 ppm NV concentration. Nitrogen was estimated by secondary ion mass spectrometry, and the NV value came from PL. Method 6 reached 26 µs at the top surface over mask 2, with an NV concentration of 0.01 ppm.

Its nitrogen concentration was estimated as identical to method 5 because the CVD parameters and NV concentrations were similar. The 26 µs value was about 2.8 times the method 5 value and reached the theoretical limit for the estimated nitrogen concentration.

(T_2/T_2^*) ratios were about 100 for method 5 and 235 for method 6, versus an expected factor of 5 to 20. The authors suggested strain and tilted subgrains may explain this behavior.

The overgrown facet controlled the stress sign across the reported growths. Overgrown {111} facets produced tensile stress and concave curvature, while growth over {001} facets produced compressive stress and convex curvature. The authors did not establish the microscopic cause.

One hypothesis attributes the difference to facet-dependent hydrogen incorporation and transport during growth and annealing. Validation would require measuring incorporated hydrogen at the diamond surface during growth. The hypothesis remains difficult to test because hydrogen was present throughout CVD.

Freestanding (111) fabrication remains pending. The two-mask process requires polishing between mask layers, and the authors describe multiple-mask ELO as costly and untested globally.

For (111) quantum optics, the demonstrated wafer offers near-single alignment and 26 µs (T_2), but the stack cracked late. For (001) processing, cyclic growth and etch yielded a freestanding, polished 300 µm wafer. Wafers with curvature radii below 6 m had already fractured, so further accurate curvature data were unavailable.

The paper calls for single-crystal (001) or (111) diamond wafers for NV-center quantum optics, including a freestanding 300 µm (001) example. 6CCVD’s diamond materials catalog lists single-crystal diamond grown by MPCVD, plates and wafers up to 125 mm, and thickness from 0.1 to 500 micrometers. Patterned Ir masks and a polished intermediate layer are required, so the linked engineering support page can be used to discuss custom fabrication support.

View Original Abstract

Abstract Large size (001) and (111) diamond wafers are essential for various applications in quantum sensing and quantum computing. Notably, (111) diamond wafers offer the advantage of self-organized single alignment of nitrogen-vacancy (NV) centers, which are crucial for these applications. In this work, we demonstrate the near-single alignment of NV centers across the entire area of 2 inch (111) diamond wafers. However, several challenges arise during the growth process by chemical vapor deposition. When attempting to grow wafers, stress induces a curvature of the wafer. The curvature can lead to crack formation. Here, we propose solutions to prevent cracking such as cyclic growth with growth and etch steps, columnar growth, and epitaxial lateral overgrowth in two layers. We discuss hypothesis that explain why and under what circumstances these solutions are successful, providing evidence through methods such as photoluminescence mapping, Raman spectroscopy, SEM analysis, and x-ray diffraction. Interestingly, all the different growth methods reveal the same systematic characteristics of growth type, related stresses, and radii of curvature. Furthermore, measures are implemented to reduce the defect density. Optically detected magnetic resonance analysis of NV centers in our monocrystalline diamond wafers yield T 2 coherence times of 26 µ s.

  1. 2014 - Perfect alignment and preferential orientation of nitrogen-vacancy centers during chemical vapor deposition diamond growth on (111) surfaces [Crossref]
  2. 2014 - Perfect selective alignment of nitrogen-vacancy centers in diamond [Crossref]
  3. 2014 - Perfect preferential orientation of nitrogen-vacancy defects in a synthetic diamond sample [Crossref]
  4. 2025 - Wafer scale N-doped diamond (111) with mainly nitrogen spin bath limited nitrogen vacancy coherence times from heteroepitexial growth [Crossref]
  5. 2020 - Sensitivity optimization for NV-diamond magnetometry [Crossref]
  6. 2023 - NV-doped microstructures with preferential orientation by growth on heteroepitaxial diamond [Crossref]
  7. 2024 - Nitrogen-vacancy center magnetic imaging of Fe3O4 nanoparticles inside the gastrointestinal tract of Drosophila melanogaster [Crossref]
  8. 2012 - Epitaxial lateral overgrowth of diamonds on iridium by patterned nucleation and growth method [Crossref]
  9. 2021 - Diamond magnetometry and gradiometry towards subpicotesla dc field measurement [Crossref]