Most "EMF protection pendant" claims are either unmeasured or measured at the wrong frequency. This paper actually does the work.
Key results, on lab-cut shungite plates
10-20 micrometres thick, measured at
26-38 GHz (5G millimetre-wave):
- At
17-37 atomic-percent carbon: shielding effectiveness 90-95% (20-25 dB), part reflection part absorption
- At
41-97 atomic-percent carbon: shielding effectiveness 96-100% (30-60 dB), reflection-dominant at the high end
- Reflection coefficient alone peaks at about 0.8 (80%) at 40-50% carbon and approaches 1.0 at the highest grades; the rest of the blocking is the wave being absorbed inside the carbon network and turned into a tiny amount of heat
- The 41% carbon threshold is where the rock crosses into the strongest-shielding regime, i.e. proper Elite-grade material
Practical takeaway: a thin Type-III pendant on your chest does almost nothing for EMF reaching the rest of your body. A proper Elite-grade plate placed between you and a router/source can produce real attenuation, but only of waves that physically pass through it. It is a Faraday-style local shield, not a personal aura.
Sources: Antonets I.V., Golubev Y.A., Shcheglov V.I., Sun S. (2021),
Electromagnetic shielding effectiveness of lightweight and flexible ultrathin shungite plates,
Current Applied Physics 29:97-106, DOI 10.1016/j.cap.2021.06.008.
ScienceDirect
·
ResearchGate PDF
.
Edited 2026-05-03: corrected the 2021 Antonets paper citation (journal is Current Applied Physics, fourth author is Sun not Kotov), corrected plate thickness to 10-20 µm, and broke out the shielding-effectiveness numbers in the way the paper's Table 2 actually presents them. Source-verification audit pass 2026-05-02 / 03.
Editor's note (2026 audit): Antonets 2021 citation corrected (journal: Current Applied Physics not Diamond and Related Materials, 4th author: Sun not Kotov, plate thickness 10-20 µm not 8-15, shielding numbers broken out per paper Table 2)
Edited 2026-05-03, source audit. Cited sources verified to exist; no fabricated sources detected. Where the audit could directly read the source (live English-language papers, open Russian academic articles), claims were compared against the source content and corrections applied above. Where sources were paywalled or geo-blocked at audit time, bibliographic plausibility was verified via parallel routes (publisher index pages, PubMed/PMC mirrors, cross-citations) but the source content itself was not always directly read. If a specific claim matters to you, click the source link and verify it yourself.
