Polarization imaging for gemstone identification and metrology
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Scientific Reports |
| Published | 2026-08-08 |
| Authors | Alaa Hamdoh, Yufei Gao, Linan Jiang, Stanley Pau |
| DOI | 10.1038/s41598-026-66027-w |
Full-Stokes polarization imaging separated natural diamond samples from chemical vapor deposition diamond samples examined in transmission. The natural samples produced symmetric angular patterns and a secondary degree of linear polarization peak at 0.46. The CVD samples showed fragmented angle patterns, noisy polarization histograms, and asymmetric circular polarization. The authors interpret these differences as consistent with stress-induced anomalous birefringence associated with synthetic growth. These measurements support polarization imaging as a supplementary, non-destructive test. The limited sample set does not define a general classification boundary.
Polarimeter setup and calibration
Section titled “Polarimeter setup and calibration”The authors used a Lucid Vision Labs Triton division-of-focal-plane polarimeter, model TRI050S-PC. Its Sony IMX250MZR sensor provided 2448 by 2048 pixels, 5.0 MP resolution, and 3.45 by 3.45 μm pixels over 400 to 900 nm. The division-of-focal-plane camera measured S0, S1, and S2. An achromatic quarter-wave plate before the lens enabled S3 measurement. A 100 mm Minolta Auto Bellows Rokkor lens was set to infinity focus and f/4.
The analysis reports DoLP for linear polarization strength, AoLP for its orientation, and DoCP for circular polarization strength.
Broadband halogen fiber-optic light passed through a Hoya UV-IR Cut filter transmitting 390 to 700 nm. Calibration used unpolarized light, a high-extinction linear polarizer, and an achromatic quarter-wave plate on separate rotation stages. The plate moved from 0° to 170° in 10° steps, and images were acquired with and without it. The procedure reconstructed all four Stokes components and reduced linear-to-circular channel crosstalk.
All stones had brilliant round cuts. Natural diamonds were 4.26 to 4.50 mm in girdle diameter and 2.71 to 2.77 mm deep. CVD diamonds were 6.83 to 7.23 mm across and 4.23 to 4.33 mm deep. Moissanite had a 10.0 mm diameter, while cubic zirconia measured about 16.0 to 17.0 mm.
Unpolarized light supplied transmission, reflection, and side-view measurements. Side-view imaging captured lateral scattering and internal pavilion reflections.
Ideal diamond ray tracing
Section titled “Ideal diamond ray tracing”Polaris-M modeled an ideal brilliant round diamond as isotropic with refractive index 2.42 and critical angle 24.4°. The simulation launched 400 unpolarized rays parallel to the z axis through the crown. At pavilion angles of 39.3° to 41.6°, many rays exceeded the critical angle and underwent total internal reflection. Some rays reflected twice before leaving the crown. A ray below the critical angle refracted through the pavilion and leaked from the stone.
The simulated reflected DoLP was 0.6 or greater over much of the star-shaped detector pattern. Transmitted DoLP was mostly 0.6 to 0.8. Reflected DoCP covered -0.5 to 0.5, while transmitted DoCP was concentrated within about ±0.3. These results establish a geometry-driven reference, not an experimental measurement of the natural samples.
Transmission separates natural and CVD diamond
Section titled “Transmission separates natural and CVD diamond”Transmission produced the clearest diamond separation. Natural diamonds shared a symmetric central AoLP structure and clean interfacet contrast. CVD Diamond-1 appeared fragmented, and CVD Diamond-2 had a nearly shattered AoLP appearance. Natural histograms formed distinct angular spikes, while synthetic histograms formed a broad 0.004 to 0.007 baseline.
| Sample | AoLP result | DoLP histogram | DoCP result |
|---|---|---|---|
| Natural diamonds | Symmetric central structure with clean facet contrast | Primary peak at 0.06 and secondary peak at 0.46 | Symmetric base within ±0.20 |
| CVD Diamond-1 | Fragmented angular distribution | Primary peak at 0.07 and irregular hump near 0.40 | Split at -0.02 and +0.01, with a base from -0.25 to +0.20 |
| CVD Diamond-2 | Nearly shattered angular appearance | Peak at 0.06 with normalized histogram frequency about 5.86, but no secondary peak | Split at -0.02 and +0.01 |
Both natural DoLP curves overlapped. CVD Diamond-1 shifted its primary peak and replaced the clean secondary structure with an irregular hump. CVD Diamond-2 flattened after its sharp primary peak. The authors associate the natural secondary peak with clean internal scattering.
Reflection and side view reveal different signatures
Section titled “Reflection and side view reveal different signatures”In reflection, both diamond groups had low DoLP. Natural samples peaked near 0.10, with histogram frequency about 5.5. CVD samples formed broader peaks from 0.05 to 0.08, with frequencies of 4.5 to 5.0. Neither group had the transmission secondary peak, and all reflection distributions tapered by about 0.40. The authors attribute the transmission feature to internal pavilion scattering rather than surface reflection.
DoCP spread increased the separation. Natural reflection bases extended from -0.36 to +0.32 and within ±0.29. CVD bases reached ±0.40 and ±0.39, with asymmetric peak splits. Side-view DoLP gave similar primary peaks at 0.18 for natural diamond and 0.19 for CVD diamond. The secondary peak was sharper at 0.72 for natural diamond, compared with a broad CVD hump near 0.64. Side-view DoCP bases were symmetric within ±0.23 for natural diamond. The CVD base ranged from -0.23 to +0.20.
Crystal symmetry separates moissanite and cubic zirconia
Section titled “Crystal symmetry separates moissanite and cubic zirconia”Moissanite produced the strongest linear polarization. The paper cites ordinary and extraordinary refractive index ranges of 2.59 to 2.76 and 2.60 to 2.80 for moissanite. Diamond had 2.39 to 2.46, while cubic zirconia had 2.11 to 2.21.
Moissanite’s transmission AoLP pattern was ordered around 45° and 135°, with a DoLP tail reaching 0.86. Reflection showed a primary peak near 0.13, a plateau at 0.60 to 0.66, and a hump near 0.93. The paper assumed the moissanite c-axis was perpendicular to the table.
Cubic zirconia behaved as an isotropic medium. Transmission produced the symmetric “Maltese cross” AoLP pattern, with DoLP peaks at 0.065 and 0.43. Reflected DoLP was broader and peaked between 0.15 and 0.25. Its isotropic response followed the cut geometry, while moissanite showed facet-dependent polarization from its directional refractive indices.
Within-class differences track manufacturing consistency
Section titled “Within-class differences track manufacturing consistency”Both cubic zirconia stones had transmission DoLP peaks at 0.065 and 0.43. CZ-1 had a sharper primary peak, while CZ-2 had a broader distribution and a tail extending to 0.85. Their DoCP bases differed, at ±0.20 for CZ-1 and ±0.23 for CZ-2. The authors interpret this spread as a difference in facet alignment or surface polish.
Both moissanite samples had a reflection DoLP peak at 0.125. Moi-1 had a plateau near 0.60. Moi-2 had a secondary peak at 0.655 and a tertiary plateau reaching 0.93. Small angular differences were attributed to mounting. Both samples kept their DoCP base within ±0.22.
Limits and engineering use
Section titled “Limits and engineering use”The main limitation is sample size. The study examined two natural diamonds, two CVD diamonds, two moissanite stones, and two cubic zirconia stones. It did not test Type Ia and Type IIa diamonds or HPHT-grown diamond. Other simulants such as YAG, GGG, and synthetic spinel were also absent. Natural and CVD stones differed in diameter, and the paper did not isolate that variable.
The ray-tracing model assumed a perfect, stress-free, homogeneous diamond with a limited ray count. It excluded growth strain, inclusions, facet imperfections, and higher-order internal scattering. The paper cites earlier work showing that small geometry errors can shift simulated polarization patterns. Strain-optic tensors and measured surface topography remain future model inputs.
For gemstone authentication and quality assessment, full-Stokes imaging can supplement geometric cut grading and established gemological methods. The paper does not report classification accuracy against Mohs, thermal conductivity, UV-Vis-NIR, XRD, or Raman. The measured signatures can map facet scattering and circular-polarization asymmetry associated with growth-related internal strain. Reference distributions should be collected for each growth method and cut before polarization limits are used for acceptance. The supplied article is an accepted manuscript marked article in press, so its text may change before final publication.
View Original Abstract
Abstract This study applies polarization raytracing and full-Stokes polarization imaging using transmission, reflection, and side-view configurations to analyze the polarimetric signatures of four brilliant round-cut gemstones: cubic zirconia (CZ), natural diamond, chemical vapor deposition (CVD) synthetic diamond, and moissanite. Raytracing simulations provide a theoretical baseline for ideal light-facet interactions, while experimental measurements of the DoLP, AoLP, and DoCP reveal distinct optical behaviors. Isotropic CZ exhibits symmetric, geometry-driven polarization patterns, while birefringent moissanite shows extreme linear polarization responses dictated by its optical axis. Differences in polarization signatures are observed between the natural and CVD diamond samples examined, despite their identical chemical compositions. The natural diamonds in this study exhibit more symmetric polarization patterns and a distinct secondary DoLP peak (0.46) in transmission, while the CVD samples show more spatially fragmented AoLP distributions and asymmetric DoCP distributions, consistent with stress-induced anomalous birefringence associated with synthetic growth. These trends highlight measurable polarization features that correlate with growth origin in the samples examined. Symmetry features of each gemstone are further evident in the extracted histograms of the polarization parameters. Overall, our results demonstrate the potential of full-Stokes polarization imaging as a non-destructive tool for probing optical structure in gemstones, with possible applications in identification and quality assessment. The observed distinctions between natural and CVD diamonds motivate future investigation to establish the broader applicability of polarization-based characterization.
Single Crystal Cvd Diamond Plate
View Specs →
Customized Single Crystal Cvd Diamond Plate
View Specs →