Construction Classics

Character of the Radiation Field and Shielding at the 184-Inch Cyclotron

Every other shielding document in this collection tells you how to calculate a shield. This one measures one. Six pages, published the year the 184-inch started producing beam, and it is the collection’s first source where the attenuation numbers were taken rather than derived — with the cyclotron itself as the neutron source, which is the only way to measure a spectrum you cannot otherwise specify.

The lead author matters for what came later. B. J. Moyer gave his name to the Moyer model, still the standard analytic method for accelerator shield design, and this 1947 paper is where that line of work starts. He also turns up inside another document in this collection: the US Army’s Accelerator Radiation Protection cites “Moyer, B. J., unpublished data” as reference 10 of its neutron chapter, twenty-five years later.

Why this document can be hosted

AECD-2149 is a declassified US Atomic Energy Commission report, issued by the AEC’s own Technical Information Division at Oak Ridge — the same publication basis as the MDDC memoranda already hosted here. Its title page carries the standard legend: manuscript 19 June 1947, declassified 9 July 1947, and Its issuance does not constitute authority for declassification of classified copies of the same or similar content and title and by the same authors. All six pages were searched for copyright notices, residual classification markings and distribution restrictions: none appear.

One point checked explicitly, because it is the trap that made another document in this wave cite-only. The work was presented at an American Physical Society meeting, and APS meeting material is copyrighted. This is not that: it is the AEC’s own report of the work, typeset and issued by the Commission under its own report number, not a reprint of an APS publication. Determination made August 2026, by reading the scan.

What was measured

A 200 MeV deuteron beam striking a probe target throws a forward cone of fast neutrons with an upper energy somewhat above 100 MeV, the cone measuring about 5.5° from axis to half-height on a high-atomic-number target. The team put slabs of candidate shielding material in front of a detector recessed in a cubical cavity inside a concrete “igloo”, with a separate beam monitor, and measured what came through.

Two detectors were used and they disagree, which is the useful part. An aluminium-walled ionization chamber responds to everything; carbon discs reading the C¹²(n,2n)C¹¹ reaction respond only above about 20 MeV, so they see the penetrating component alone and ignore the low-energy scattered background. The carbon detectors gave cleaner exponential fits, and for the high-Z materials they give a much shorter half-value than the ionization chamber does — the difference between the two columns below is a measurement of how much soft, scattered radiation the chamber is counting.

These are 190–200 MeV deuteron numbers and they do not transfer to a small machine. A sub-MeV proton cyclotron produces no such field at all. What transfers is the method — machine as source, threshold detector for the penetrating component, geometry stated with the result — and the ranking of materials, which is physics rather than a property of this beam.

Concrete itself was measured separately: the carbon-detector fit gives a half-value thickness of 9.9 inches over roughly six feet of concrete (Fig. 2b), a straight line on semi-log paper across the whole range.

The result worth knowing even if you never build this machine

Adding a hydrogenous layer behind a dense one made the reading worse. Moyer measured a paraffin transition increase of 60% following iron and 100% following lead in similar geometry — the neutron beam has to reach equilibrium with the secondaries it generates in the absorber before attenuation becomes exponential at all, and a hydrogenous layer is efficient at converting neutron flux into ionizing protons that the chamber then counts. Lead alone shows a negative transition effect, because it yields fewer such secondaries than the iron tank wall ahead of it.

That is the opposite of the intuition that more material is more shielding, and lead is exactly the material an amateur is most likely to wrap paraffin around. It is why the shielding deep dive now says layer order is a design variable and that a composite shield cannot be sized by adding up its layers.

A dose map, normalised to beam current

The paper ends with something rarer than the attenuation table: the shield’s as-built performance, tied to the beam that produced it. At a deuteron beam of about 0.2 µA — roughly 24 r/hr in the centre of the neutron cone one foot outside the tank wall — the reading just outside the shield in the centre of the beam is 10 mr/hr, general building areas run 0.5–1.5 mr/hr, and above the two-foot concrete roof the level averages 25 mr/hr. Skyshine and roof backscatter are estimated to contribute not more than 10% of the general building level. Slow-neutron flux outside the concrete, through the access openings, is of order 10³ cm⁻² s⁻¹.

Two things in that paragraph are worth copying. The numbers are stated per unit beam current, so a reader can scale them; and the survey separates the forward cone from the general building level and from the openings, which is how you find out that penetrations, not the bulk shield, set the field outside. The paper also notes a second source the designers had not planned for: beam grazing the interior of the dee sprayed fast neutrons in considerable intensity throughout 180° of azimuth, quite apart from the target.

Design rules from this document

10 Design Guide rules come from these six pages — a high yield for the length, because almost every paragraph is a measurement or a method. They are the first measured shielding rules in the guide, alongside the methodology from ORNL-3540 and the operational practice from AD-755510.

• dg-1115 — measure shield attenuation with the machine itself as the source
• dg-1117 — expect transition (buildup) effects at the front face of any shield
• dg-1118 — layer order matters — hydrogenous material after a high-Z layer can raise the reading
• dg-1119 — threshold-activation detectors give cleaner half-values than ionization chambers
• dg-1120 — state the geometry with any published attenuation number
• dg-1121 — choose fast-neutron shielding for high density combined with low atomic number
• dg-1122 — shield for machine-generated loss points, not just the target
• dg-1124 — publish shield performance as a dose map normalised to beam current

Used by the shielding deep dive. Its companion in this collection is residual radiation and personnel dose, which measures what is left after the beam stops rather than what escapes while it runs.