Skip to content

Structural effects of boron doping in diamond crystals for gamma-ray light-source applications - Insights from molecular dynamics simulations

FieldValue
JournalDiamond and Related Materials
Published2026-05-09
AuthorsMatthew D. Dickers, Felipe Fantuzzi, N. J. Mason, Gennady B. Sushko, Andrei V. Korol
DOI10.1016/j.diamond.2026.113714
Citations1

Boron substitution produced a relative lattice expansion of about 2.5 × 10⁻³ at 5% boron and 300 K. The response was linear across 14 crystal geometries. Published data agreed reasonably below about 0.5% doping, while the deviation from Vegard’s law grew at higher concentrations.

Why (110) spacing matters for light-source geometry

Section titled “Why (110) spacing matters for light-source geometry”

Gamma-ray crystal light sources direct ultra-relativistic charged particles through crystal planes. Channelling radiation forms as the particles oscillate within the planar potentials. A periodically bent crystal also produces undulator radiation. Its bending amplitude and period control the emitted spectrum and angle.

The proposed boron-graded design varies boron along the axial [100] direction. Larger substitutional boron expands the (110) plane spacing and bends the channelling path. The (110) spacing is therefore the primary structural target, while the (100) spacing tracks axial expansion.

Defects and dislocations disrupt the periodic potential and cause dechannelling. Longer dechannelling lengths indicate higher crystal quality. Excessive doping raises lattice strain and the risk of defect formation. Manufacturing must control dopant level and substrate quality, not just nominal composition.

The authors modelled C1-xBx with periodic boundaries in the transverse Y and Z directions and a free axial X surface. A three-atom carbon layer at negative X was fixed to mimic mechanical support from an HPHT growth substrate. The simulation box extended 15 Å beyond the crystal along X to prevent periodic-image interactions.

Carbon sites were randomly replaced by boron from x = 0 to 0.05 in steps of 0.01. The 5% target corresponded to 8.82 × 10²¹ cm⁻³. Every structure was idealised as fully substitutional, with no interstitial boron. The set included cubic, extended and slab geometries, with 50 to 200 random distributions per concentration. The dataset contained 1940 unique structures.

Each structure underwent 10,000 velocity-quenching optimisation steps with a fixed box. Molecular dynamics then ran for 1 ns at 300 K in the NVT ensemble, using a Langevin thermostat with a 0.2 ps damping time. Atomic interactions used a multicomponent Tersoff potential.

Analysis averaged the final 10 frames, corresponding to the last 10 ps. Atoms were projected onto the (110) and (100) planes, and successive plane positions gave the spacings. A geometric correction addressed suppressed transverse expansion under periodic boundaries.

Lattice expansion is linear but differs from Vegard’s law

Section titled “Lattice expansion is linear but differs from Vegard’s law”

Across all geometries, the lattice constant rose linearly with boron concentration. The trend held across the full 0 to 5% concentration range. Its slope was systematically lower than the covalent-radius and atomic-volume modifications of Vegard’s law.

The nominal 300 K reference values were 3.5669 ± 0.0001 Å for the lattice constant, 1.2611 Å for the (110) spacing, and 0.8917 Å for the (100) spacing. Selected simulation results show the absolute change for one cubic and one extended geometry. Uncertainties are one standard deviation.

MeasurementConfiguration0% B5% B
C1 lattice constant21.402 Å cube3.5673 ± 0.00018 Å3.5756 ± 0.00027 Å
E1C bulk (110) spacing178.350 × 21.402 × 21.402 Å1.2612 ± 2.6 × 10⁻⁵ Å1.2639 ± 0.00011 Å
E1C bulk (100) spacingE1C crystal0.8918 ± 2.2 × 10⁻⁵ Å0.8940 ± 0.000079 Å

The C1 lattice constant had a relative expansion of 2.4008 × 10⁻³ at 5% boron. The fitted MD expansion coefficient was 0.05093 across the crystal sizes.

Static lattice optimisation gave 3.561 Å for pure diamond and 3.690 Å for hypothetical cubic boron. Applying Vegard’s law to these values yielded a coefficient of 0.03623. The corresponding atomic-volume coefficient was 0.03760. Because the optimisation excluded thermal effects, the authors state that thermal expansion would reduce the difference from the MD result.

These tests suggest that the 7.28 ų/atom boron volume used in prior literature overestimates the optimised volume of 6.28 ų/atom.

Plane spacing changes with position and crystal size

Section titled “Plane spacing changes with position and crystal size”

The substrate region covered seven atomic layers above the fixed substrate. The bulk contained the remaining interior and ranged from 8 to 184 layers. The free-edge definition covered ten layers, but three surface layers were excluded, leaving seven analysed.

At 1% boron, the (110) and (100) spacings were broadly uniform across sizes near the substrate. This indicates that the fixed substrate constrained the local response. In the bulk, (110) spacing increased with axial length, while (100) remained nearly uniform.

At 5% boron, E1A and E1C had bulk (110) spacings of 1.2636 ± 0.00013 Å and 1.2639 ± 0.00011 Å. Both had a bulk (100) spacing of 0.8940 Å within the reported precision. The authors attribute the (110) trend to finite-size and anisotropic-geometry effects, even after correction.

The free edge produced the largest spacings and the largest run-to-run variation. It showed no consistent size trend, so the authors treat those results cautiously.

Why the deviation from Vegard’s law is large

Section titled “Why the deviation from Vegard’s law is large”

Boron has a covalent radius of 0.88 Å, compared with 0.77 Å for carbon. Random substitution therefore causes local steric expansion. It also produces a structure closer to ideal than layer-by-layer MPCVD growth.

At 5% boron, at most about 6.2% of boron atoms formed B2 dimers. This equals about 0.31% of all crystal atoms. Across 1940 structures, the authors counted 3444 B2 clusters, 674 B3 clusters, 17 B4 clusters, and no clusters larger than B4.

The authors attribute the departure from reference curves to several factors. Experimental MPCVD crystals can contain clusters, vacancies, interstitials, dislocations and residual stress. The MD model omits the negative lattice strain associated with electronic free carriers. The optimised cubic boron volume was also below the value used by the atomic-volume reference.

The authors therefore interpret their curve as an upper bound for fully substitutional, stress-free crystals rather than a universal growth law.

Limits on prediction of manufactured crystals

Section titled “Limits on prediction of manufactured crystals”

The MD model does not reproduce MPCVD growth kinetics or temperature gradients. It starts from random substitution and ends at an idealised post-growth equilibrium. The authors state this may underestimate lattice distortion because surface diffusion, vacancies and growth-related defects are absent.

Interstitial boron, electronic effects, graded boron profiles and actual periodic bending were not modelled. The geometric correction for (110) also does not combine contributions from several facets.

The free edge shows surface relaxation, and the 2 × 1 reconstruction of diamond (100) was not modelled. The authors judge its omission unlikely to change the region result because it affects the closest layers. Nanometre-scale defects cannot be analysed in the small simulation cells. The results therefore set a structural reference rather than a manufacturing tolerance.

Implications for diamond selection and device engineering

Section titled “Implications for diamond selection and device engineering”

At boron concentrations of about 1% or lower, the simulated response agrees with published experimental and computational data. It covers the concentrations used in candidate gamma-ray crystals. The linear response can inform calculations of (110) spacing and strain along a graded [100] axis.

Manufacturers still need low-defect material because structural expansion does not capture dechannelling from dislocations or point defects. The paper states that type IIa HPHT substrates give higher-quality crystals than type Ib substrates because substrate defects can propagate into the growth.

Engineers should treat 5% boron as an idealised benchmark, not a recommended operating concentration. Candidate crystals reported to date did not exceed about 1% boron. Actual periodic grading and MPCVD incorporation need kinetic modelling and experimental verification.

For manufacturing tolerances, the paper argues for coupling atomistic modelling with experimental data. Lattice metrology should therefore be paired with defect measurements.

6CCVD lists MPCVD-grown single-crystal and polycrystalline diamond, plates and wafers up to 125 mm, and boron-doped diamond in the diamond materials catalog. For the paper’s MPCVD route to periodically bent gamma-ray light-source crystals, engineers need boron-doped single-crystal material with controlled (110) spacing, low defect content and a defined axial geometry. The linked engineering support can be used to specify those fabrication requirements, while the paper’s defect-free limit remains a simulation benchmark rather than evidence of measured catalog performance.

View Original Abstract

Boron-doped diamond crystals (BDD, C 1 − x B x ) exhibit exceptional mechanical strength, electronic tunability, and resistance to radiation damage. This makes them promising materials for use in gamma-ray crystal-based light sources. To better understand and quantify the structural distortions introduced by doping, which are critical for maintaining channelling efficiency, we perform atomistic-level molecular dynamics simulations on periodic C 1 − x B x systems of various sizes. These simulations allow the influence of boron concentration on the lattice constant and the (1 1 0) and (1 0 0) inter-planar distances to be evaluated over the concentration range from pure diamond (0%) to 5% boron at room temperature (300 K). Linear relationships between both lattice constant and inter-planar distance with increasing dopant concentration are observed, with a deviation from Vegard’s Law. This deviation is larger than that reported by other theoretical and computational studies; however, this may be attributed to an enhanced crystal quality over these studies, a vital aspect when considering gamma-ray crystal light source design. The methodology presented here incorporates several refinements to closely reflect the conditions of microwave plasma chemical vapour deposition (MPCVD) crystal growth. These results enable reliable atomistic modelling of doped diamond crystals and support their use in the design and fabrication of periodically bent structures for next-generation gamma-ray light source technologies.

  1. 2019 - Boron-doped diamond: current progress and challenges in view of electroanalytical applications [Crossref]
  2. 2022 - Boron-doped diamond electrodes: Fundamentals for electrochemical applications [Crossref]
  3. 2022 - Lightly boron-doped nanodiamonds for quantum sensing applications [Crossref]
  4. 2025 - Intervalence plasmons in boron-doped diamond [Crossref]
  5. 2000 - UV photodetectors from B-doped diamond film
  6. 2017 - The investigation of boron-doped diamond absorbance spectrum
  7. 2015 - Towards a crystal undulator
  8. 2017 - Synchrotron bragg diffraction imaging characterization of synthetic diamond crystals for optical and electronic power device applications [Crossref]
  9. 2020 - Crystal-based intensive Gamma-ray light sources [Crossref]