Skip to content

D3/D5 source-head review — primary-source synthesis

Research input and decision record. This note supplies the primary-source synthesis. Decisions explicitly approved by the owner during the subsequent D3/D5 grilling are labelled below and reflected in RESEARCH_SCOPE.md; remaining literature findings and proposed methods do not amend the research scope.

1. Scope and status boundary

OWNER INTENT: Paper 1 concerns conventional industrial LINAC converted-X-ray operation in the 5–7.5 MeV range and the amendment-open broad planar Ta–water–Type-304-stainless scanned-plate archetype. The normalized source is a one-dimensional scan with a local Gaussian electron spot.

APPROVED DECISION: D1 fixes the conventional industrial LINAC converted-X-ray family and the 5–7.5 MeV envelope. D2 fixes the primary response as H*(10) with the project’s PHITS multiplier -200 and output normalization pSv/primary-source-electron, plus identifiable combined/photon/electron/positron/neutron result pages. The owner subsequently approved the planar Ta–water–SS304 candidate-family boundary, the non-numerical source and scan method, the minimal closed component-scoring layout, and provisional built-in photonuclear treatment for preliminary D3 calculations. Numerical convergence thresholds, histories, and the final construction remain open.

SUPERSEDED CONTEXT: The former provisional plate and scan values were not used to seed this review. A paper may independently report a numerically similar value for its own accelerator or objective; such a value remains source-specific and is not transplanted into Paper 1. The owner has since approved bounded screening coordinates; their representative values and the final construction remain conditional on the approved preliminary procedure.

2. Search record and stopping rule

VERIFIED REVIEW RECORD: The review used four iterative primary-source search rounds. Search surfaces were Crossref/DOI records; publisher or repository full text; JACoW/EPAC/APAC proceedings; IAEA and JAEA official publications/manuals; Google Patents/USPTO records; and first-party accelerator/manufacturer pages where technical details were public. Secondary summaries were used only as discovery leads and were not treated as evidence. Search date recorded for the project log: 2026-09-12.

Round Representative queries and scope New evidence found Stopping result
1 — industrial source-head families "industrial electron accelerator" tantalum converter water cooled target 5 MeV 7.5 MeV; bremsstrahlung converter scanned tantalum stainless steel water; Simulation Tool for Scanning X-Ray Beams Irradiator; Bremsstrahlung Converter for High Power EB Radiation Processing; ILU-10 X-ray converter Direct Ta–water–SS planar/composite family; Ta-coated Al industrial plate; multilayer Ta with internal coolant; direct-contact recessed target; moving/window-separated target; distributed disk/plate stacks Candidate families were still being added; continue
2 — D5 and engineering constraints Gaussian electron beam spot FWHM bremsstrahlung target; beam scanning dose uniformity industrial; finite target edge bremsstrahlung scan; PHITS source weights relative error histories; H*(10) official ICRU sphere; targeted citation chaining from the round-1 sources Gaussian spot versus field/scan-width definitions; pulse overlap and trajectory; finite-edge and edge-angle effects; local/global thermal limits; dose-map and DUR measures; PHITS weighting and statistical-error controls; H*(10) definition context Design constraints were still being added; continue
3 — backward/forward citation and supplier/standards closure Citation chains from Petwal, Lazurik, Auslender, Zimek, Esplen, IAEA-TECDOC-2051, PHITS manual, and cited patents; exact-title searches for industrial converter, Ta–water–SS, Ta–Al/Cu, and scan-dose sources; official standards/manual searches No new source-head family and no design constraint beyond the families and constraints already recorded in rounds 1–2 No-new round 1
4 — repeat saturation check Repeated owner-primary searches with alternate wording (planar tantalum converter cooling, industrial X-ray target scan horn, bremsstrahlung target finite edge, spot-weighted scan PHITS, electron beam scan width qualification) and citation-chain checks for the same source families No new source-head family and no design constraint beyond rounds 1–3; only duplicate/less applicable records or inaccessible secondary summaries No-new round 2; saturation stopping rule met

UNRESOLVED: Public manufacturer pages confirm industrial X-ray systems but do not expose production converter drawings, exact joins, coolant pressure limits, or maintenance tolerances. Those are not inferred.

3. Executive result

EXTERNAL LITERATURE: The directly relevant industrial evidence supports a Ta high-Z conversion surface on a long/planar industrial assembly, scan correction or shaping, cooling, and a downstream layer or structure that controls transmitted-electron contamination. Auslender et al. describe a 5 MeV/50 kW ILU-10 with a vertical aluminium plate coated by tantalum, a 90-degree beam transport, a long scanned converter, edge-angle correction, and continuous water cooling of beam-channel components (Auslender et al. 2004, pp. 1–3, official PDF). Petwal et al. explicitly study a 5/7.5 MeV industrial composite Ta–water–SS converter and calculate the emerging field and deposited heat (Petwal et al. 2007, p. 1, official PDF).

APPROVED DECISION: The planar Ta–water–SS304 plate is the bounded D3 candidate family. The sources do not establish its dimensions, optimum layer thicknesses, cooling-channel geometry, finite extent, joining method, or acceptance criteria. The literature shows competing engineering families; final construction selection requires the approved PHITS comparison procedure and an owner decision.

EXTERNAL LITERATURE: The industrial papers do not support treating one reported target thickness or scan width as universal. Zimek’s industrial model optimizes thickness and reports scan nonuniformity for its own energy, beam, conveyor, foil, and objective assumptions; Petwal et al. use a different cylindrical model and objective (Zimek 2020, pp. 2–4; Petwal et al. 2007, p. 1). Their numerical values are source-specific evidence, not Paper 1 candidate values.

4. Primary-source evidence ledger

4.1 Direct industrial converter and scan evidence

Petwal et al., APAC 2007 — direct Ta–water–SS archetype

EXTERNAL LITERATURE: Petwal et al. describe a radiation-processing facility planned for electron and X-ray modes, with 5 and 7.5 MeV electron beams. Their composite target consists of tantalum for bremsstrahlung, water for heat removal, and stainless steel for the water channel and for stopping unwanted electron contamination. They varied tantalum in a study-specific range and calculated the field, spectrum, depth/radial dose, and deposited heat. Cite: V. C. Petwal et al., “Bremsstrahlung Converter for High Power EB Radiation Processing Facility,” APAC 2007, paper THPMA088, p. 1, official proceedings PDF.

PROPOSED METHOD: This is the strongest direct provenance for the broad Ta–water–stainless family and the 5–7.5 MeV envelope. Its geometry and layer values remain the authors’ model; it does not establish Paper 1 plate dimensions, a Type-304 grade requirement, a normalized 1D scan, or a universal optimum.

Auslender et al., EPAC 2004 — industrial Ta-coated plate and edge correction

EXTERNAL LITERATURE: The ILU-10 report describes an industrial accelerator operating up to 5 MeV and 50 kW, a 90-degree beam bend, a vertical long X-ray converter formed by an aluminium plate coated with tantalum, scan-magnet motion along the converter, and a correcting magnet that makes incidence near the converter edges approach normal. It reports water cooling for beam-channel components, finite pulse duration/repetition, and a measured surface-dose distribution. It also warns that the RF accelerator’s electron spectrum is not monochromatic and that low-energy electrons add target heating. Cite: V. L. Auslender et al., “Electron Accelerator for Energy up to 5.0 MeV and Beam Power up to 50 kW with X-ray Converter,” EPAC 2004, pp. 1–3, official proceedings PDF.

PROPOSED METHOD: This is direct industrial evidence for planar/long-plate construction, finite scan extent, edge-angle correction, beam-energy spread, and source-head cooling. The report uses an Al substrate rather than the D3 SS304 layer and does not provide a transferable D5 Gaussian convention.

Lazurik et al., 2004 — scan waveform, dose mapping, and thermal model

EXTERNAL LITERATURE: Lazurik et al. describe industrial RT-Office/XR-Soft models containing a high-power electron accelerator, scanner, converter with cooling, product on a moving conveyor or a rotating turntable, and dose/temperature calculations. Their input model includes electron spectrum, diameter and spatial intensity, scan waveform/repetition, exit window, converter layers, coolant, material, conveyor speed, geometry, and the window-to-converter distance. Their 5 MeV example compares scan behavior and dose mapping in product geometry. Cite: V. T. Lazurik et al., “RT-OFFICE FOR OPTIMIZATION OF INDUSTRIAL EB AND X-RAY PROCESSING,” Problems of Atomic Science and Technology (2004), pp. 1–3, official PDF.

PROPOSED METHOD: The source supports treating scan waveform, conveyor motion, product boundaries, and target/coolant geometry as coupled variables. Its reported dose ratios and converter dimensions are that model’s results, not Paper 1 thresholds.

Zimek, 2020 — industrial thermal envelope and scan nonuniformity

EXTERNAL LITERATURE: Zimek describes high-power industrial X-ray conversion with a high-Z target, water cooling, a cover selected for water-pressure requirements, and energy-dependent target-thickness optimization. The paper evaluates 2.5, 5, and 7.5 MeV cases and reports that a scanned field can be nonuniform along the scan path; it discusses changing converter/current patterns and multi-pass operation to improve uniformity. Cite: Z. Zimek, “Economical evaluation of radiation processing with high-intensity X-rays,” Nukleonika 65(3), 167–172 (2020), DOI 10.2478/nuka-2020-0027, pp. 2–4.

PROPOSED METHOD: This supports energy-dependent comparison, finite scan-path response, scan weighting/current shaping, and a cooling/cover tradeoff. Its numerical beam, foil, water, cover, scan, and conveyor values are not Paper 1 candidates; no provisional value is imported.

Petwal et al., Pramana 2010 — pulse overlap and moving-product superposition

EXTERNAL LITERATURE: The publisher abstract states that MCNP stationary beam profiles were compared with measurements, then used to optimize overlap of successive scan pulses for approximately uniform surface dose; a mathematical model incorporated conveyor speed and was compared with measurements. Cite: V. C. Petwal et al., “Dosimetric measurements and Monte Carlo simulation for achieving uniform surface dose in pulsed electron beam irradiation facility,” Pramana 74, 457–468 (2010), DOI 10.1007/s12043-010-0041-8, publisher abstract and pp. 457–468.

PROPOSED METHOD: This supports modelling delivered dose as a weighted/successively displaced sum of local stationary responses and checking scan/conveyor overlap. The facility is a 10 MeV pulsed EB system, so no pulse count, scan frequency, or tolerance is transferred to Paper 1.

4.2 Engineering comparators outside the exact Ta–water–SS archetype

Esplen et al., 2022 — Ta–Al explosion-bonded flange and thermal/dosimetric tradeoffs

EXTERNAL LITERATURE: Esplen et al. optimize a Ta–Al explosion-bonded converter for a TRIUMF electron linac. Their source model uses a monoenergetic electron beam with an axisymmetric Gaussian spot; they explicitly distinguish electron beam size from downstream photon field size. The study varies beam energy, target thickness, and Gaussian spot size; it uses a stopping condition based on the combined Ta and Al CSDA ranges so primary electrons do not dominate the phantom surface. Phase-space transport is followed by water-phantom depth-dose scoring and uncertainty reporting. FEA uses conservative reductions of heat-transfer/contact parameters, cooling channels, and thermal/structural/fatigue response. Cite: N. Esplen et al., “Design optimization of an electron-to-photon conversion target for ultra-high dose rate x-ray (FLASH) experiments at TRIUMF,” Phys. Med. Biol. 67, 105003 (2022), DOI 10.1088/1361-6560/ac5ed6, pp. 2–14, especially §§2.1.1–2.2, Figs. 3–9 and Tables 1–6; open repository PDF.

PROPOSED METHOD: This is an engineering comparator for Ta/low-Z heat removal, primary-electron interception, Gaussian spot convention, spot-size/penumbra/thermal tradeoffs, and MC-to-FEA handoff. It is a medical UHDR 8–10 MeV design using Al, not an industrial 5–7.5 MeV Ta–water–SS304 design. Its numerical values are not D3/D5 acceptance criteria.

IAEA-TECDOC-2051, 2024 — disk stacks, composite converters, cooling and corrosion

EXTERNAL LITERATURE: The IAEA review states that bremsstrahlung converters generally use high-Z Ta or W plates/disks, often multiple disks with coolant gaps. It explains that forward yield has an energy/material/thickness optimum because increased high-Z material also increases scattering and absorption. It documents composite high-Z/low-Z converters, including Ta/Al explosion-bonded designs, and notes alternative liquid-metal converter concepts. Its cooling section discusses water radiolysis, corrosion, oxidation, stainless-steel cooling loops, cleaning/deionized water/chemistry control, and helium cooling as an alternative. Cite: IAEA, Photonuclear Production of Radioisotopes, IAEA-TECDOC-2051 (2024), §§3.3–3.3.3, pp. 37–44, official PDF.

PROPOSED METHOD: The source supports treating parallel-disk/plate, composite low-Z-backed, and coolant-isolation families as engineering context and treating water chemistry/thermal hydraulics as constraints. Its main energy range is higher-energy isotope production; no 5–7.5 MeV geometry or value is inferred.

US6463123B1, 2002 — multilayer high-Z/low-Z internal-channel target

EXTERNAL LITERATURE: This patent describes sequential high-Z layers, a low-Z thermally conductive substrate, coolant passages through the substrate, beam deflection to avoid repeated local heating, and filtration/absorption of unwanted low-energy X-rays. It states that coolant can be kept out of direct contact with the Ta layer and claims variants with multiple Ta layers and conductive substrates. Cite: Steris Corporation, “Target for production of X-rays,” US6463123B1 (2002), “Detailed description,” Figs. 2–4 and claims 1, 4, 9, 11–19, Google Patents; PDF.

PROPOSED METHOD: This is a distinct multilayer/internal-channel family and a comparator for direct-water corrosion, heat spreading, and spectral filtering. It is not evidence that the D3 planar Ta–water–SS304 construction should be replaced or that the patent’s claimed efficiency is independently verified for Paper 1.

US4323780A, 1982 — direct-contact recessed water-cooled target

EXTERNAL LITERATURE: The patent describes a target metal plate in a recess, coolant channels/chambers directly exposing the target to coolant, a stainless-steel base/plate option, and channel geometry intended to increase coolant velocity and heat dissipation. Cite: Siemens, “Target assembly for a linear accelerator,” US4323780A (1982), “Detailed description” and claims 1–14, Google Patents; PDF.

PROPOSED METHOD: This is an explicit direct-contact water-cooled target family and a comparator for channel placement, pressure/thermal design, and SS base construction. It is a fixed medical target with a different geometry and does not establish an industrial scanned planar candidate.

US8098796B2, 2012 — separate electron/photon windows and movable target

EXTERNAL LITERATURE: The patent describes a target assembly with separate electron and photon windows, a stainless-steel photon window, a copper substrate, adjacent water channels, oxidation/vacuum controls, and target motion to reduce peak local power and temperature. Cite: Varian, “Target assembly with electron and photon windows,” US8098796B2 (2012), “Summary” and “Detailed description,” Figs. 1–4 and claims, Google Patents.

PROPOSED METHOD: This is a distinct window-separated/movable-target family. It is useful for oxidation, window attenuation, serviceability, and local thermal-load comparisons, but it is a medical multi-energy assembly rather than a D3 5–7.5 MeV industrial Ta–water–SS304 plate.

4.3 Official process, response, and PHITS evidence

IAEA TRS-481, 2015 — spot/scan qualification and dose mapping

EXTERNAL LITERATURE: The IAEA manual states that accelerator qualification includes beam energy, beam spot profile, and scan width; scan width may be defined at a specified fractional level. It states that surface dose uniformity depends on spot profile and scan width, and that scan coverage must include the product sides relative to conveyor motion. It treats dose mapping as specific to the irradiator, path, loading configuration, and product, recommends enough dosimeters for statistical significance and at least triplicate mappings, and requires uncertainty to be considered when interpreting extrema and setting process parameters. Cite: IAEA, Manual of Good Practice in Food Irradiation, Technical Reports Series No. 481 (2015), §§3.3.2, 4.3, 4.5–4.5.7, 5.3–5.4, 10.4, pp. 29, 34–47, 74, official PDF.

PROPOSED METHOD: This is process-qualification guidance, not a source-head design selection. It supports recording the spot definition, scan-width convention, finite-edge coverage, repeat mapping, uncertainty, and change control in D5/D3 verification.

PHITS User’s Manual 3.37 — Gaussian source and transforms

EXTERNAL LITERATURE: PHITS §5.3.5, Tables 5.3.40–5.3.50, defines s-type=3/6 independent Gaussian coordinates with x1, y1, and z1 widths specified as FWHM. PHITS §5.3.7, Tables 5.3.63–5.3.72, defines s-type=13/14 for an x-y-plane circular Gaussian with r1 as the FWHM, and defines dir relative to local z and phi as local azimuth. PHITS §5.8 documents rotation/translation order and states in §5.8.3 that transformed tally coordinates use the body-fixed relation opposite to ordinary source/object transformation. Cite: JAEA, PHITS User’s Manual 3.37, §§5.3, 5.3.5, 5.3.7, and 5.8, source index, independent Gaussian, x-y-plane Gaussian, transform.

APPROVED DECISION: Use the documented circular x-y Gaussian as the smallest native model because no ellipticity evidence is supplied. Declare its physical reference plane and transformation explicitly; do not copy an object/source transform into a tally without checking the opposite tally-coordinate relation. A phase-space source remains a phase-space source: preserve its energy, angular, and spot metadata rather than silently replacing it with a Gaussian.

PHITS User’s Manual 3.37 — source weights and statistical precision

EXTERNAL LITERATURE: PHITS §5.3.1 permits multiple source definitions with relative <source> weights and a global totfact; §5.3.2 specifies common source-weight parameters and that multiple sources use totfact rather than factor. PHITS §5.2.2 documents maxcas histories per batch, maxbch batches, history- or batch-based variance options, and relative-error reporting; it recommends more than ten batches for reliable batch uncertainties and states that this statistical quantity does not estimate systematic uncertainty from nuclear-reaction models. Cite: JAEA, PHITS User’s Manual 3.37, §§5.3.1–5.3.2 and 5.2.2, multi-source, common source parameters, number of history and bank.

APPROVED DECISION: Separate-position runs plus an external weighted sum are the audit default. A single PHITS multi-source run is a checked alternative only when every relative weight and totfact is recorded and its result is compared with the separate-position sum. Transport statistical error remains separate from geometry, material, source-profile, nuclear-model, engineering, and measurement uncertainty.

PHITS User’s Manual 3.37 — tally capabilities and multiplier

EXTERNAL LITERATURE: PHITS §7.2 defines t-cross surface scoring for current and flux, with crossing contribution 1 for current and 1/cosθ for flux, and provides energy/time/angle and forward/backward pages. PHITS §7.1 defines t-track track-length fluence in regions. PHITS §7.5 defines t-deposit deposited energy/dose and warns that mixed-material region dose is not a simple mass-weighted average; §7.7 does not recommend t-heat after the stated version. PHITS §7.8 and §7.9 provide residual-nucleus yield and produced-particle diagnostics. PHITS §5.26 identifies -200 as the H*(10) multiplier and allows it for t-track, t-cross, and t-point. Cite: JAEA, PHITS User’s Manual 3.37, §§5.26, 6/Table 6.1, and 7.1–7.2, 7.5, 7.7–7.9, tally format, t-track, t-cross, t-deposit, t-yield, t-product, multiplier.

APPROVED DECISION: Use the fixed component layout in §7. t-cross is for surface current/flux, t-track is a regional fluence diagnostic, and t-deposit is the component deposition tally. Omit component H*(10) and the project -200 multiplier from D3 selection; D2 remains the authority for complete-room H*(10) in D4.

IAEA GSG-7, 2018 — H*(10) definition context

EXTERNAL LITERATURE: The IAEA defines the ICRU sphere, expanded/aligned field, and ambient dose equivalent at the reference depth of 10 mm for strongly penetrating radiation. Cite: IAEA, Occupational Radiation Protection, General Safety Guide GSG-7 (2018), paras. 2.39–2.43, pp. 34–35, official PDF.

APPROVED DECISION: Paper 1 D2 already selects H*(10) and the project multiplier. This source supplies definition context only; it does not reopen D2 or define the project multiplier.

5. Selected candidate family and contextual comparators

APPROVED DECISION: Numerical candidate selection remains within the planar Ta–water–SS304 candidate family labelled C-A in the comparison table, using the same approved source/scan method, energy cases, finite plate/edge treatment, and fixed component tally layout. C-B–C-F remain literature context for construction risks and do not enter numerical comparison without an explicit amendment.

Reference Construction family What it tests or explains Main risk/gap
C-A Finite coextensive planar Ta–water–SS304 layers in vacuum, with no frame or channel geometry Owner candidate family; transmitted-electron suppression, photon output, finite edges, scan response Joining, pressure, cooling implementation, fabrication, and Type-304 batch properties remain outside the transport model
C-B Ta-coated low-Z conductive plate with embedded or adjacent coolant, represented by the ILU-10 and Ta–Al evidence Planar manufacturability, heat spreading, beam-window/plate coupling, energy spread and edge-angle behavior Al is not SS304; the industrial report does not supply a portable local Gaussian convention
C-C Multiple parallel high-Z plates/disks with coolant gaps Distributed heat deposition, lower local load, multi-stage attenuation/filtering Not a single broad planar SS304 plate; spacing and disk geometry must not be inferred
C-D Ta high-Z layer with internal coolant passages kept out of direct target contact Water chemistry/corrosion, thermal interface, serviceability, spectral filtering Patent claims and medical implementations are not industrial 5–7.5 MeV validation
C-E Direct-contact recessed water-cooled target plate Cooling/channel concept and direct-water corrosion comparison Fixed medical-target geometry; pressure/erosion/oxidation behavior unresolved for the scanned industrial plate
C-F Window-separated or movable Ta target assembly Window attenuation, oxidation isolation, local thermal dwell, maintenance Different medical multi-energy architecture; likely outside the minimal D3 construction

Applicability: The planar Ta–water–SS304 family is the owner archetype to be compared, not an approved final construction. C-B–C-F explain alternative water, backing, edge, and thermal arrangements; they are not Paper 1 candidate families. No source-specific published dimension is transplanted as a Paper 1 optimum.

5.1 CandidateBounds evidence — material and construction bounds

EXTERNAL LITERATURE: NIST ESTAR lists pure tantalum (Z=73) density as 16.6540 g cm^-3 and default-grid CSDA mass ranges of 3.682 g cm^-2 at 5 MeV, 4.260 g cm^-2 at 6 MeV, 4.796 g cm^-2 at 7 MeV, and 5.297 g cm^-2 at 8 MeV. NIST states that user-entered energies calculate stopping powers and density effect rather than a new tabulated range; therefore a 7.5 MeV value is not an ESTAR output. Cite: NIST ESTAR tantalum material record and output definitions, tantalum ESTAR, composition, method, version/history, definitions.

PROPOSED METHOD: Linear interpolation of the adjacent default-grid ranges gives 5.0465 g cm^-2 at 7.5 MeV and, using the listed density, approximately 3.030 mm of Ta. This is an interpolation for dimensionless screening, not an ESTAR result, a projected range, a penetration guarantee, or an approved thickness. ESTAR’s CSDA range is an average continuous-slowing-down path length; multiple scattering makes projected depth a separate quantity, so PHITS must establish the actual deposition and escape response.

EXTERNAL LITERATURE: Petwal et al. model a cylindrical composite target with radius 5 cm, beam-facing Ta followed by a 2 mm water channel and 2 mm stainless steel, vary Ta from 0–20 mm, and report 5 and 7.5 MeV cases. Their reported useful transmitted-energy fractions, 9.3% at 5 MeV and 14.2% at 7.5 MeV, belong to their optimized models. Cite: Petwal et al. 2007, p. 1, official proceedings PDF.

EXTERNAL LITERATURE: Zimek’s model uses Ta foil 0.8 mm at 5 MeV and 1.3 mm at 7.5 MeV, a 20 mm water layer, and a 0.8 mm steel cover. Its common scan model reports a 3 cm beam diameter, 2 cm width at half maximum, 10 Hz scan frequency, 100 cm scan width, 200 cm scanner height, 60 cm scan-to-conveyor distance, and a 10 cm by 100 cm Ta foil. The paper’s separate 2 MeV example reports 0.5 mm Ta, 20 mm water, and a 0.8 mm stainless sheet. These values are the source’s own controls and geometry, not a Gaussian convention or Paper 1 candidate grid. Cite: Zimek 2020, pp. 2–3 and Table 2, publisher PDF, DOI.

UNRESOLVED: Petwal’s stainless grade is not shown as Type 304, and Zimek reports steel rather than Type 304. The reviewed public sources provide no defensible public Type-304/water/channel construction bound or complete SS304 thickness grid for a planar 5–7.5 MeV scanned plate. Their exact values cannot be combined into a candidate grid because they belong to different geometries, objectives, materials/grades, and response definitions. Former project values were not used as candidate seeds.

5.2 Recorded derivation of the approved screening coordinates

OWNER-APPROVED METHOD: For material density rho, reference mass scale S, and dimensionless factor f, calculate the layer thickness as t = f S / rho. The factors define the screening coordinates; the rounded millimetre values are display labels, not independent inputs.

PRELIMINARY CALCULATION RECORD: The approved labels use tantalum radiation length S = 6.82 g cm^-2 with rho = 16.654 g cm^-3; water 7.5 MeV ESTAR CSDA mass range approximately 3.7875 g cm^-2 with rho = 1.000 g cm^-3; and custom-composition SS304 7.5 MeV ESTAR CSDA mass range approximately 4.7648 g cm^-2 with rho = 8.000 g cm^-3. These scales produce the recorded approximate Ta values 1.024, 2.048, and 3.071 mm; water values 1.894, 4.734, and 7.575 mm; and SS304 values 2.978, 5.211, and 7.445 mm. PDG reports the tantalum radiation length as 6.82 g cm^-2 (PDG tantalum properties); NIST STAR material 276 defines liquid water density 1.000 g cm^-3 and mass fractions H 0.111894, O 0.888106 (NIST composition).

PRELIMINARY CALCULATION RECORD: The custom SS304 ESTAR composition is the PNNL-15870 Rev. 1 material-298 nominal composition normalized to unity: C 0.000400, Si 0.005000, P 0.000225, S 0.000150, Cr 0.190000, Mn 0.010000, Fe 0.701725, and Ni 0.092500, at 8.000 g cm^-3 (PNNL-15870 Rev. 1). This is a reproducible reference composition, not a manufactured batch specification.

PROVENANCE GATE: Before generating the first preliminary Case, retain the raw water and custom-SS304 ESTAR outputs, exact full-precision reference scales, formula operands, and derived centimetre values in the Project record. The approximate values above are insufficient as immutable Case inputs.

6. Approved D5 source, scan, and quadrature method

6.1 Coordinate and PHITS-native source definition

APPROVED DECISION: Preserve the Paper 1 frame: plate +Y up, beam travelling toward global −Y, one-dimensional scan along global X, and remaining transverse direction along global Z.

APPROVED DECISION: Define each local circular electron spot on the Ta entrance plane, before transport through Ta, in the right-handed local basis

[ (u,v,w)=(+\mathrm{global\ }X,+\mathrm{global\ }Z,-\mathrm{global\ }Y). ]

Use PHITS-native s-type=13/14 when the spot is circular. Its documented r1 is the circular FWHM in the local x-y plane, so use FWHM F as the canonical PHITS input and report the equivalent standard deviation

[ \sigma=\frac{F}{2\sqrt{2\ln 2}}, \qquad g_F(u,v)=\frac{4\ln 2}{\pi F^2} \exp!\left[-4\ln 2\frac{(u-u_i)^2+(v-v_0)^2}{F^2}\right]. ]

The conversion is the usual one-dimensional Gaussian relation FWHM = 2√(2 ln 2) σ; it is a reporting conversion, not permission to substitute a photon-field width, scan width, 2σ, or fractional-level width. The density is normalized on the unbounded plane. Keep the mathematical Gaussian unbounded and retain finite Ta, water, SS304, and surrounding void geometry; do not clip and renormalize tails at an edge unless a later physical source definition explicitly requires that. The approved sensitivity coordinates are 2.36, 5.89, and 11.78 mm FWHM; 5.89 mm remains provisional until the approved retention check passes.

EXTERNAL LITERATURE: PHITS documents dir relative to the local z axis for s-type=13/14; under the basis above, dir=+1 points toward global −Y. PHITS also documents independent Gaussian coordinates in s-type=3/6, whose widths are FWHM, but those types do not replace the circular x-y-plane primitive when the intended source is circular (JAEA, PHITS User’s Manual 3.37, §§5.3.5 and 5.3.7, independent Gaussian, x-y-plane Gaussian).

APPROVED DECISION: If evidence later requires an ellipse, define both principal-plane widths and orientation explicitly or use an appropriate phase-space/discrete source; do not relabel an isotropic circular source. Esplen et al. distinguish electron beam size from downstream photon-field size, while Bazalova et al. show that off-axis steering can defocus a beam and produce an elliptical spot; neither establishes an ellipse for Paper 1 (Esplen et al. 2022, §§2.1.1–2.1.3, pp. 4–7; Bazalova et al. 2012, §2.3.1 and Discussion).

EXTERNAL LITERATURE: PHITS §5.8 documents rotation/translation order, including the sign-dependent order of rotation and translation, and §5.8.3 states that transformed tally output uses the body-fixed coordinate relation opposite to ordinary object/source transformation. Verify transform signs and tally coordinates independently; a source transform must not be copied blindly into a tally transform (JAEA, PHITS User’s Manual 3.37, §5.8, official transform section).

6.2 Fixed-normal scan weighting

APPROVED DECISION: For a stationary-position response R(r;x_i) and nonnegative weights w_i, define

[ R_{\mathrm{scan}}(\mathbf r)= \frac{\sum_{i=1}^{N}w_i R(\mathbf r;x_i)}{\sum_{i=1}^{N}w_i}. ]

For a continuous sweep x(t) with delivered current or charge rate I(t), the spatial density is

[ p(x)=\frac{I(t(x))\,|dt/dx|}{\int_a^b I(t(x))\,|dt/dx|\,dx}, \qquad R_{\mathrm{scan}}(\mathbf r)=\int_a^b p(x)R(\mathbf r;x)\,dx. ]

Thus p(x) follows delivered charge/current and spatial speed. Uniform spatial weighting is valid when current per time and spatial speed are constant over the active interval, or when uniform density is explicitly adopted as the idealized source definition. It is not valid merely because the positions are evenly spaced if current, speed, pulse overlap, fly-back, or turnaround dwell makes charge per unit X nonuniform. Uniform source density also does not imply a uniform converter field when finite edges and local spot tails alter R.

APPROVED DECISION: For Paper 1 use the idealized uniform normalized one-dimensional scan with fixed normal incidence Ω=−global Y. Do not infer a position-dependent angle law, turnaround dwell law, or endpoint dwell from industrial examples. Such laws require explicit beamline geometry, measured waveform, or a later owner amendment. Auslender, Lazurik, and Bazalova document machine-specific scan angles, waveforms, dwell positions, fly-back effects, or off-axis responses; those facts establish what can matter, not Paper 1 laws (Auslender et al. 2004, pp. 1–2; Lazurik et al. 2004, pp. 2–3; Bazalova et al. 2012, §2.3.1 and Discussion).

APPROVED DECISION: In a numerical integral, half weights at the two interval endpoints are quadrature weights. They are not physical turnaround dwell. If turnaround or endpoint dwell is later included, represent its finite charge/time explicitly in p(x) or a declared point-mass contribution rather than changing trapezoid weights by interpretation.

6.3 Nested composite-trapezoidal refinement and convergence

APPROVED DECISION: For the approved uniform density on [a,b], use uniform X nodes and the normalized composite trapezoidal rule

[ R_N=\frac{h}{b-a}\left[\frac{R(a)}{2}+\sum_{i=1}^{N-2}R(a+ih)+\frac{R(b)}{2}\right], \qquad h=\frac{b-a}{N-1}. ]

Refine by the nested rule N' = 2N − 1, halving h while preserving every previous position. Keep the total normalized scan charge/current integral fixed across refinement. Endpoint half weights remain integration weights only. This is an auditable policy choice; no source establishes a mandatory rule, node count, or tolerance for this D5 problem.

EXTERNAL LITERATURE: Wang et al. vary uniformly spaced source positions in a linear transport dose calculation and observe a sharp error improvement followed by leveling; Tian et al. formulate sub-source contributions as a linear weighted sum. These support measuring position-refinement behaviour and preserving source weights, but they do not mandate a PHITS node count, quadrature rule, or tolerance for Paper 1 (Wang et al. 2019, §§2.A–2.D and §3.B; Tian et al. 2017, §§2.A.1–2.B).

APPROVED DECISION: Run each stationary position with the same physical geometry, PHITS-native spot definition, fixed normal direction, energy, finite bounds, tally layout, and per-primary normalization. Separate-position weighted sums are the audit default: retain every position response, weight, coordinate, and source metadata. PHITS multi-source is a checked alternative only with recorded relative <source> weights and totfact, and with at least one comparison against the separate-position sum (JAEA, PHITS User’s Manual 3.37, §§5.3.1–5.3.2).

APPROVED DECISION: Convergence is observable- and statistics-aware. At each nested refinement compare at least downstream forward-photon current/fluence, energy spectrum and angular distribution; transmitted electron/positron and neutron components; deposited energy per primary electron in Ta, water, and SS304; and interior/edge profiles, extrema, edge-to-interior ratios, and outside-plate escape. Account for the combined PHITS statistical standard error rather than using a file-level difference, node count, or apparent visual match alone. Keep scan-quadrature refinement separate from tally-mesh refinement and transport-history convergence. UNRESOLVED: Absolute and relative tolerances, histories, node count, and run budget require preliminary evidence and owner approval.

6.4 Finite interception and edge treatment

APPROVED DECISION: Keep Ta, water, SS304, and surrounding void finite in X/Z. Do not conditionally clip and renormalize the unbounded Gaussian at each edge. Include scan positions near both finite ends and record primary/secondary escape, side crossings, and edge-versus-interior response. Evaluate k=3,4,5,6 under the approved common-margin formula; the retained margin remains unresolved pending response convergence at all three energies.

7. Fixed component scoring layout and comparison outputs

7.1 Fixed P0/P1/edge layout

APPROVED DECISION: Use the same minimal closed component layout for every Ta–water–SS304 candidate and the finite Ta-only limiting case:

  1. P0, the upstream accounting face: place it just upstream of the Ta entrance plane. Outward crossings measure upstream backscatter escape; inward crossings are re-entry diagnostics. The normalized monoenergetic source defines incident energy analytically.
  2. Per-material deposition regions: use t-deposit for every present non-void material, reporting deposited energy per primary electron before operating scaling. Full candidates require separate Ta, water, and SS304 deposition; the Ta-only limit requires Ta alone. Use t-track only for regional track-length fluence diagnostics.
  3. P1, the downstream accounting face: place it just downstream of the complete assembly and off coincident boundaries. Use t-cross for particle-resolved current/flux, energy spectrum, angular distribution, and directional pages.
  4. Four finite side faces: together with P0 and P1, form one closed six-face accounting boundary. Record inward and outward crossings separately for photons, electrons, positrons, and neutrons.
  5. Excluded from D3 selection: no frame/backing region, optional P2 stand-off plane, component H*(10), production/yield diagnostics, activation gate, occupancy criterion, or cooldown assumption. The latter diagnostics require a later D6-specific authorization.

EXTERNAL LITERATURE: PHITS defines t-cross current and flux differently (1 for current and 1/cosθ for flux), t-track as track-length fluence, and t-deposit as deposited energy/dose. The manual warns against interpreting a mixed-material region as a simple mass-weighted dose and does not recommend t-heat after the stated version. Its t-product and t-yield sections document the production diagnostics excluded from D3 above (JAEA, PHITS User’s Manual 3.37, §§7.1–7.2, 7.5, 7.7–7.9).

APPROVED DECISION: Omit the project -200 multiplier and component H*(10) from D3 candidate selection. D2 remains the authority for the complete-room H*(10) response and normalization in D4.

7.2 Response vector, not an invented scalar score

APPROVED DECISION: Report each candidate/energy/spot/scan case with units and fixed normalization, including:

  • forward-photon current/fluence, spectrum, and angular distribution at P1;
  • transmitted electron, positron, and neutron contributions;
  • deposited energy per primary electron in each material region;
  • lateral/depth profiles, interior/edge extrema, edge-to-interior ratio, and outside-plate escape;
  • weighted scan-line mean, minimum, maximum, standard deviation, percentile range, and any declared DUR = max/min with region and normalization;
  • sensitivity to scan weighting, node placement, finite interception, and spot convention;
  • successive-refinement differences with PHITS statistical error separated from model, material, geometry, engineering, and measurement uncertainty.

APPROVED DECISION: Select the simplest eligible candidate under strict dominance, not an invented scalar score. Maximize forward-photon current at P1; minimize transmitted electron/positron current, unintended side/edge escape, and complete-assembly deposited energy under equal incident normalization. Report neutron contribution without minimizing it. Keep spectra, angular distributions, spatial profiles, and numerical-stability evidence undirected. Retain the centre unless one candidate dominates it and every other eligible replacement; incomparable replacements require an amendment before detailed response-map inspection.

7.3 Thermal and activation context

APPROVED DECISION: Deposited energy supports an engineering-plausibility screen and explicit linear operating scaling; it does not establish a universal beam power, current, duty factor, scan utilization, cooling qualification, or construction safety. Water is a transport coolant region, not a validated cooling system. Flow, pressure, temperature, boiling margin, fatigue, corrosion, fabrication, constructibility, commissioning suitability, and regulatory approval remain outside the claim.

APPROVED DECISION: Use PHITS built-in photonuclear treatment provisionally for preliminary D3 calculations, retain visible particle-resolved neutron output, and label it model-dependent. The conditional JENDL-5, ENDF/B-VIII.1, and built-in D6 comparison remains mandatory on the selected construction before D4. No production/yield diagnostic, activation gate, occupancy criterion, or cooldown time is approved for D3.

8. Disagreements and non-transferable results

  1. Target thickness: Industrial papers optimize different configurations and objectives; a reported optimum in one facility cannot be treated as a Paper 1 optimum. Differences arise from energy, spectrum, spot, target material, cover/backing, cooling, and response plane (Petwal 2007, p. 1; Zimek 2020, pp. 2–4; Esplen et al. 2022, pp. 5–12).
  2. Direct versus isolated water: Direct-contact cooling, internal channels, low-Z substrates, and separate windows trade thermal performance against attenuation, corrosion, oxidation, and manufacturability (US4323780A; US6463123B1; US8098796B2; IAEA-TECDOC-2051, pp. 42–44).
  3. Spot width versus scan width: Gaussian electron-spot size, photon-field width, FWHM, 2σ, and fractional scan width are not interchangeable. Every result must name the convention and plane (Esplen et al. 2022, pp. 5–7; IAEA TRS-481, p. 29; PHITS §5.3.7).
  4. Static versus moving delivery: A stationary response can be shifted/summed only after trajectory, dwell/current weights, conveyor motion, and finite boundary are explicitly represented (Petwal et al. 2010; Lazurik et al. 2004).
  5. Statistical versus systematic uncertainty: PHITS relative error describes sampled transport variance; it does not cover nuclear-model, geometry, material, source-profile, cooling, or measurement uncertainty (PHITS §5.2.2; IAEA TRS-481 §4.3).
  6. Efficiency claims: Conversion/yield values are objective- and geometry-specific, and patent claims are not independent validation. Report the fixed response vector instead of ranking candidates by an unqualified efficiency number (US6463123B1; Petwal 2007; Zimek 2020).

9. Unresolved gaps before D3/D5 selection

UNRESOLVED: The primary-source review and subsequent owner decisions do not resolve:

  • the final construction, retained spot, scan span, and plate margin, which remain conditional on the approved preliminary screens and gates;
  • a defensible public Type-304/water/channel engineering bound;
  • coolant flow, pressure, chemistry, fabrication, inspection, or replacement assumptions, which remain outside the transport-archetype claim;
  • exact electron energy spectrum or source angular distribution beyond the approved monoenergetic fixed-normal idealization;
  • quadrature node count, convergence tolerances, histories, or maximum run budget;
  • acceptance tolerances for photon output, charged contamination, edge leakage, energy residual, statistical precision, mesh refinement, margin convergence, and normalization equivalence;
  • production/yield diagnostics, activation gates, occupancy criteria, and cooldown assumptions, which remain excluded from D3 pending a later D6-specific authorization;
  • independent edge/spot physical validation against measurements or a validated surrogate, which remains unresolved under D8/D9 and must not be conflated with transport verification.

10. Decision handoff

DECISION HANDOFF: The owner has approved the planar Ta–water–SS304 candidate-family boundary; circular PHITS-FWHM source; uniform fixed-normal scan; nested trapezoidal refinement and separate-position audit; minimal closed P0/P1/four-side layout; provisional built-in photonuclear treatment for preliminary D3; staged 2^3 geometry screen with centre and Ta-only limit; 5, 6.25, and 7.5 MeV; centre-default strict-dominance selection; strict preliminary/formal separation; the declared layer, spot, span, and k=3,4,5,6 margin coordinates; expected 38-preliminary/conditional 30-formal Case structure; history-variance default with recorded fallback; independent recorded seeds; the approved calibration order and qualitative-change definition; analytic incident-energy and closed-boundary accounting; complete formal eligibility; directed dominance responses and amendment-on-tie rule; one targeted failed-centre refinement per implicated factor; one fixed geometry across energy; the strict engineering-claim boundary; and the finite coextensive Ta–water–SS304 geometry in vacuum.

UNRESOLVED: Node counts, histories, tolerances, acceptance thresholds, maximum run budget, and the final construction require preliminary evidence and later owner gates. D3 and D5 remain amendment-open.

11. References

  1. V. C. Petwal et al., “Bremsstrahlung Converter for High Power EB Radiation Processing Facility,” APAC 2007, paper THPMA088, p. 1. Official proceedings PDF: https://proceedings.jacow.org/a07/PAPERS/THPMA088.PDF.
  2. V. L. Auslender et al., “Electron Accelerator for Energy up to 5.0 MeV and Beam Power up to 50 kW with X-ray Converter,” EPAC 2004, pp. 1–3. Official proceedings PDF: https://proceedings.jacow.org/e04/PAPERS/THPKF050.PDF.
  3. V. T. Lazurik et al., “RT-OFFICE FOR OPTIMIZATION OF INDUSTRIAL EB AND X-RAY PROCESSING,” Problems of Atomic Science and Technology (2004), pp. 1–3. Official PDF: https://vant.kipt.kharkov.ua/ARTICLE/VANT_2004_2/article_2004_2_186.pdf.
  4. Z. Zimek, “Economical evaluation of radiation processing with high-intensity X-rays,” Nukleonika 65(3), 167–172 (2020), DOI 10.2478/nuka-2020-0027, pp. 2–4. DOI: https://doi.org/10.2478/nuka-2020-0027; publisher PDF: https://reference-global.com/download/article/10.2478/nuka-2020-0027.pdf.
  5. V. C. Petwal et al., “Dosimetric measurements and Monte Carlo simulation for achieving uniform surface dose in pulsed electron beam irradiation facility,” Pramana 74, 457–468 (2010), DOI 10.1007/s12043-010-0041-8, publisher abstract and pp. 457–468. DOI: https://doi.org/10.1007/s12043-010-0041-8.
  6. N. Esplen et al., “Design optimization of an electron-to-photon conversion target for ultra-high dose rate x-ray (FLASH) experiments at TRIUMF,” Phys. Med. Biol. 67, 105003 (2022), DOI 10.1088/1361-6560/ac5ed6, pp. 2–14, especially §§2.1.1–2.2, Figs. 3–9 and Tables 1–6. DOI: https://doi.org/10.1088/1361-6560/ac5ed6; open repository PDF: https://dspace.library.uvic.ca/bitstreams/381a065e-8669-4b0d-b8c0-781a35df981b/download.
  7. IAEA, Photonuclear Production of Radioisotopes, IAEA-TECDOC-2051 (2024), §§3.3–3.3.3, pp. 37–44. Official PDF: https://www-pub.iaea.org/MTCD/Publications/PDF/TE-2051web.pdf.
  8. Steris Corporation, “Target for production of X-rays,” US6463123B1 (2002), “Detailed description,” Figs. 2–4 and claims 1, 4, 9, 11–19. Google Patents: https://patents.google.com/patent/US6463123B1/en; PDF: https://patentimages.storage.googleapis.com/87/24/aa/82c1d9be675e7d/US6463123.pdf.
  9. Siemens, “Target assembly for a linear accelerator,” US4323780A (1982), “Detailed description” and claims 1–14. Google Patents: https://patents.google.com/patent/US4323780A/en; PDF: https://patentimages.storage.googleapis.com/21/db/ba/5897c034ca21e3/US4323780.pdf.
  10. Varian, “Target assembly with electron and photon windows,” US8098796B2 (2012), “Summary” and “Detailed description,” Figs. 1–4 and claims. Google Patents: https://patents.google.com/patent/US8098796B2/en.
  11. IAEA, Manual of Good Practice in Food Irradiation, Technical Reports Series No. 481 (2015), §§3.3.2, 4.3, 4.5–4.5.7, 5.3–5.4, 10.4, pp. 29, 34–47, 74. Official PDF: https://www-pub.iaea.org/MTCD/Publications/PDF/trs481web-98290059.pdf.
  12. JAEA, PHITS User’s Manual 3.37, “Source index and Gaussian source definitions,” §§5.3, 5.3.5, and 5.3.7. Official pages: https://phits.jaea.go.jp/manual/PHITS-en/chapters/sections/source/source.html, https://phits.jaea.go.jp/manual/PHITS-en/chapters/sections/source/s-type_3.html, https://phits.jaea.go.jp/manual/PHITS-en/chapters/sections/source/s-type_13.html.
  13. JAEA, PHITS User’s Manual 3.37, “Multiple sources and common source parameters,” §§5.3.1–5.3.2. Official pages: https://phits.jaea.go.jp/manual/PHITS-en/chapters/sections/source/source_1.html, https://phits.jaea.go.jp/manual/PHITS-en/chapters/sections/source/source_2.html.
  14. JAEA, PHITS User’s Manual 3.37, “Transform,” §5.8 and §5.8.3. Official page: https://phits.jaea.go.jp/manual/PHITS-en/chapters/sections/transform.html.
  15. JAEA, PHITS User’s Manual 3.37, “Tally format and tally types,” §6/Table 6.1 and §§7.1–7.2, 7.5, 7.7–7.9. Official pages: https://phits.jaea.go.jp/manual/PHITS-en/chapters/tally-format.html, https://phits.jaea.go.jp/manual/PHITS-en/chapters/tallies/t-track/t-track.html, https://phits.jaea.go.jp/manual/PHITS-en/chapters/tallies/t-cross/t-cross.html, https://phits.jaea.go.jp/manual/PHITS-en/chapters/tallies/t-deposit.html, https://phits.jaea.go.jp/manual/PHITS-en/chapters/tallies/t-yield.html, https://phits.jaea.go.jp/manual/PHITS-en/chapters/tallies/t-product.html.
  16. JAEA, PHITS User’s Manual 3.37, “Multiplier,” §5.26. Official page: https://phits.jaea.go.jp/manual/PHITS-en/chapters/sections/multiplier.html.
  17. JAEA, PHITS User’s Manual 3.37, “Number of history and Bank,” §5.2.2. Official page: https://phits.jaea.go.jp/manual/PHITS-en/chapters/sections/parameters/parameters_2.html.
  18. NIST, ESTAR, “Stopping-Power and Range Tables for Electrons,” tantalum material record and method/definitions pages. Official pages: https://physics.nist.gov/cgi-bin/Star/chooseElement.pl?elName=Tantalum&elNum=73&filename=ESTAR-t.html, https://pml.nist.gov/cgi-bin/Star/compos.pl?matno=073&mode=text, https://physics.nist.gov/PhysRefData/Star/Text/method.html, https://physics.nist.gov/PhysRefData/Star/Text/verhistory.shtml, https://physics.nist.gov/PhysRefData/Star/Text/appendix.html.
  19. M. Bazalova, M. D. Weil, B. Wilfley, and E. E. Graves, “Monte Carlo model of the scanning beam digital x-ray (SBDX) source,” Phys. Med. Biol. 57, 7381–7394 (2012), DOI 10.1088/0031-9155/57/22/7381, §§2.1–2.3.1 and Discussion. Primary full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC4163683/.
  20. A. Wang, A. Maslowski, T. Wareing, J. Star-Lack, and T. Gilat Schmidt, “A fast, linear Boltzmann transport equation solver for computed tomography dose calculation (Acuros CTD),” Med. Phys. 46, 925–933 (2019), DOI 10.1002/mp.13305, §§2.A–2.D and §3.B. Primary full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC6367051/.
  21. Z. Tian, Y. Li, N. Hassan-Rezaeian, S. B. Jiang, and X. Jia, “Moving GPU-OpenCL-based Monte Carlo dose calculation toward clinical use: Automatic beam commissioning and source sampling for treatment plan dose calculation,” J. Appl. Clin. Med. Phys. 18, 69–84 (2017), DOI 10.1002/acm2.12049, §§2.A.1–2.B. Primary full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC5689963/.
  22. IAEA, Occupational Radiation Protection, General Safety Guide GSG-7 (2018), paras. 2.39–2.43, pp. 34–35. Official PDF: https://www-pub.iaea.org/MTCD/Publications/PDF/PUB1785_web.pdf.