Skip to content

Paper 1 — Systematic Evidence Review

Search date: 2026-09-04

Protocol: EVIDENCE_REVIEW_PROTOCOL.md

Screening records: EVIDENCE_REVIEW_RECORDS.csv

Status

The accessible-source search, deduplication, and initial screening pass are complete, including all metadata returned by the anonymous IEEE Xplore interface. The owner has confirmed that institutional access to Scopus, Web of Science, and INSPEC is unavailable; IEEE full text is likewise not institutionally available. These are permanent coverage limitations rather than outstanding project actions. Four relevant records still require sufficient full text.

The result below is therefore the completed accessible-source scope review, not a comprehensive institutional-database review or an absolute novelty claim.

Search execution

The protocol's five concept blocks were executed in OpenAlex, Crossref, and IEEE Xplore. OpenAlex returned relevance-ranked metadata; the first 25 records were inspected, or all records when fewer than 25 existed. Crossref's bibliographic search was much broader and returned large approximate totals; the first 20 relevance-ranked records were inspected for each query. IEEE Xplore's anonymous interface returned complete metadata result sets, which were retrieved in pages of 100 and screened by title and available abstract.

Source Search Total reported Inspected
OpenAlex S1 industrial converted-X-ray shielding 1,576 25
OpenAlex S2 converter and source verification 792 25
OpenAlex S3 photonuclear transport 8 8
OpenAlex S4 radiation-transport surrogates 404 25
OpenAlex S5 exact contribution intersection 2 2
Crossref S1 industrial converted-X-ray shielding 2,128,252 20
Crossref S2 converter and source verification 1,575,629 20
Crossref S3 photonuclear transport 1,119,329 20
Crossref S4 radiation-transport surrogates 4,486,226 20
Crossref S5 exact contribution intersection 4,608,279 20
IEEE Xplore S1 industrial converted-X-ray shielding 214 214
IEEE Xplore S2 converter and source verification 493 493
IEEE Xplore S3 photonuclear transport 140 140
IEEE Xplore S4 radiation-transport surrogates 291 291
IEEE Xplore S5 exact contribution intersection 0 0

Crossref totals demonstrate low query specificity and are not counts of eligible studies. Screening used titles and available abstracts rather than treating the reported totals as evidence volume.

Targeted web and official-source searches were used to resolve candidate records and inspect IAEA, PHITS, publisher, laboratory, and repository pages. These searches also recovered the PRISMA 2020 and PRISMA-S reporting guidance used by the protocol.

Access log

Source/interface Access result on 2026-09-04
Scopus web interface Scopus Preview exposed sign-in and access-check controls but no searchable institutional result interface. No query was recorded as a zero-result search.
Web of Science Core Collection Redirected to the Clarivate account/institution sign-in page before search.
IEEE Xplore Anonymous metadata search succeeded for S1–S5; no institutional full-text entitlement was present.
INSPEC through Engineering Village Authorization redirect failed with authorization_request_not_found; no search interface or valid result set was available.

The blocked sources require an authenticated institutional browser session; the review does not treat an authentication failure as evidence absence.

Deduplication and screening flow

OpenAlex records inspected                          85
Crossref records inspected                         100
IEEE Xplore records inspected                    1,138
                                                    -----
Database records before deduplication             1,323
Duplicate database records removed                   79
Unique database records                           1,244
Additional unique records from targeted searches      4
                                                    -----
Unique records screened                           1,248

Excluded at title or available-metadata screening 1,185
Included as direct component evidence                 17
Included as indirect methodological evidence          42
Awaiting sufficient full text                          4
                                                    -----
Total                                              1,248

DOI was the primary deduplication key. Records without a DOI used normalized title and year. The CSV preserves each record's search provenance, decision, and primary reason.

IEEE Xplore contributed 1,073 unique records, one of which duplicated an existing Crossref record. A deterministic first pass searched each title and available abstract for the protocol's accelerator/source, converter, shielding/photonuclear, transport-code, and surrogate concept combinations. The 110 records meeting a physical-system or transferable-method combination received manual candidate review; 31 were retained (12 direct and 19 method). The other 1,041 new IEEE records lacked a qualifying system-method combination in the available metadata and were excluded. This machine-assisted first pass is a review limitation; it reduced screening volume but may miss records whose relevant concepts are absent from their IEEE metadata.

Neither OpenAlex nor IEEE Xplore returned an S5 record concerning the complete proposed contribution; IEEE Xplore reported zero S5 results. The top Crossref S5 records included unrelated LINAC, material-shielding, medical, and generic machine-learning studies. Targeted searches located PHITS-based accelerator-shielding surrogates, but these model neutron transport through simplified shields in a heavy-ion accelerator context, not the complete conventional-LINAC electron-to-converter-to-room problem.

No inspected record combined all of:

  1. a conventional industrial LINAC electron source;
  2. a finite converted-X-ray source model;
  3. an incident-electron domain covering 5–7.5 MeV;
  4. complete facility shielding fields generated with PHITS;
  5. whole-configuration separation of model-fitting and final-test cases; and
  6. photonuclear sensitivity or particle-resolved assessment.

This supports the working gap only within the searched and accessible evidence. It does not establish universal priority.

Direct component evidence

  • Hua et al. (2015), Theoretical Study on Bremsstrahlung Transferred From Electron Beam of 7.5 MeV. FLUKA calculations vary converter material and thickness and report transmitted photon and electron efficiency, photon-flux distribution, bremsstrahlung energy distribution, and dependence on beam sweep width. This directly supports treating converter construction and scan distribution as physical source conditions, but it does not validate a Paper 1 converter or room.
  • Design of a 5 MeV electron LINAC based X-ray source (DOI). The record provides a 5 MeV electron-to-X-ray source-design comparator for industrial radiography. It is component evidence rather than a facility-shielding validation.
  • Cleland (2006), Industrial applications of electron accelerators. This supports the industrial accelerator context but does not determine Paper 1 parameters.
  • Lazurik et al. (2003), Simulation tool for scanning X-ray beams irradiator. The abstract models the converter, scanner, product transport, and other components of an industrial X-ray line together. It supports representing scan delivery and finite converter geometry explicitly rather than reducing the source to a fixed photon spectrum.
  • Ziaie and Tahami (2005), Mass throughput rate calculation for X-ray facilities. This is the closest reviewed industrial Monte Carlo precedent for spatial superposition: their EGS4 calculation represents a 100 cm scanned electron-beam Rhodotron and integrates stationary dose distributions to obtain dynamic processing quantities. It supports the principle used by the RAMAL-EBX PHITS quadrature, but not its exact source syntax or node count.
  • Ma et al. (1997), Accurate characterization of Monte Carlo calculated electron beams. Spatial and multiple-source Monte Carlo representations of scanned electron beams agreed closely with full phase-space calculations. This is supporting method evidence from radiotherapy, not direct industrial validation.
  • Measuring and simulation of dose at irradiation by bremsstrahlung gamma beam (2018, DOI). This provides an industrial-irradiation dose-measurement comparator, but not complete room shielding validation.
  • Experimental verification of bremsstrahlung attenuation (1993, DOI). Measurements from 5.5–25.1 MeV agreed with Monte Carlo predictions within the abstract's reported ±9%. This directly brackets most of the approved energy interval, but applies to on-axis attenuation in specified absorbers rather than a complete room.

Photonuclear method evidence

Source-measurement comparators awaiting complete detail

Both are promising spectrum comparators, but the accessible metadata does not establish sufficient geometry, detector-response, normalization, and uncertainty detail for quantitative Paper 1 use.

Two additional records remain unresolved:

  • Development of a Method for the On-Line Spectral Characterisation of Bremsstrahlung Photons from an Electron Accelerator (record). The title is relevant to source measurement, but the accessible record is insufficient for method extraction.
  • Uncertainty-Aware Hybrid Monte Carlo–Machine Learning Surrogate Modeling of Neutron Dose Attenuation in Ferroboron–Concrete Multilayer Shielding (preprint record). This may inform uncertainty-aware surrogate evaluation, but it is a physically different neutron-shield problem and sufficient technical content was not accessible.

Surrogate-method evidence

  • Khuwaileh and Metwally (2020), Gaussian process approach for dose mapping in radiation fields. This supports a Gaussian-process spatial baseline using sparse measurements or simulation responses, but it does not concern the Paper 1 source or shielding domain.
  • Chowdhury et al. (2024/2025) and the PHITS Workshop 2025 contribution Shielding Optimization: An Approach for Extending PHITS with Machine Learning confirm PHITS-labelled accelerator-shielding surrogate work. Their neutron-source, shield, and output definitions remain physically distinct from converted-X-ray rooms.
  • Badal and Badano (2019), MCDNet. The study denoises voxel-wise Monte Carlo X-ray dose distributions with a convolutional network. It is transferable evidence for spatial-field surrogate evaluation, not for industrial shielding physics.
  • Rapid Assessment Models for Neutron Beams of BNCT (2025, DOI). This is an additional accelerator-neutron surrogate precedent, but its source, beam-shaping assembly, and clinical output contract differ from Paper 1.

Quality and applicability assessment of new candidates

Adequate, partial, and inadequate refer only to information available in this review, not the intrinsic quality of an inaccessible full paper.

Source Specification Quantity and normalization Uncertainty Verification/validation Reproducibility Paper 1 applicability
Hua et al. 2015 Partial Partial Inadequate in accessible abstract Simulation only Partial Direct converter and scan method
5 MeV LINAC X-ray source Partial Partial Inadequate in accessible record Simulation/design record Partial Direct source component; different application
Cleland 2006 Adequate for context Not applicable Not applicable Not applicable Partial Industrial context only
Rafiei et al. 2023 Adequate for stated source Partial Partial MCNP–FLUKA and theoretical comparison Partial Photonuclear method; different target purpose
Sakha 2016 Partial Partial Inadequate in accessible record Simulation only Partial Converter photon/neutron component
Blideanu et al. 2024 Partial Partial Partial Inter-code performance test Partial Photonuclear verification method
Levy et al. 1976 Inadequate in metadata Inadequate Inadequate Measurement named Inadequate until full text Candidate spectrum comparator
Stritt et al. 1996 Inadequate in metadata Inadequate Inadequate Measurement and simulation named Inadequate until full text Candidate spectrum comparator
Khuwaileh and Metwally 2020 Partial Partial Partial Method demonstration Partial Indirect spatial surrogate method
PHITS Workshop 2025 shielding optimization Partial Partial Inadequate in workshop record PHITS workflow precedent Partial Indirect accelerator-shielding surrogate

The existing Annotated Bibliography remains the source-level assessment for evidence already synthesized by the narrative review.

Methodological evidence map

Method component Current evidence Remaining limitation Consequence for Paper 1
Industrial application IAEA guidance and industrial accelerator reviews Conventional LINACs are not the only industrial architecture Keep the accelerator architecture explicit
5–7.5 MeV range Direct converter, spectrum, attenuation, processing, and activation literature Intermediate energies and energy distributions are sparsely documented Supports the approved interval; simulate justified intermediate energies
Converter/source Petwal et al. support a compact Ta-water-stainless component benchmark; Zimek supports a broad plate, 100 cm scan direction, and finite local beam for industrial radiation processing No located source fixes a universal 5–7.5 MeV coolant/cover construction or measured scan-position density for the intended archetype Reopen D3/D5; verify a bounded broad scanned-plate source, retain Petwal only as a component benchmark, and do not universalize uniform scan weighting
Closed-room shielding Peri–Orion ordinary-concrete attenuation and angular calculations, HJ 979 TVLs and exterior-reference convention, and PNNL transport materials No complete room measurement validates the selected box family or owner-approved dimension bounds Use the D4 five-variable closed-room domain; verify limiting cases, material records, exterior margin, and omitted-structure sensitivity; do not claim construction safety
Radiation quantity IAEA, ICRU Report 95, PHITS -200/-204, and operational-quantity sources Facility-specific Malaysian licence and instrument practice remain to be matched D2 approves $H^(10)$ per D3 primary source electron as primary and $H^$ only as a separate secondary comparison
Photonuclear transport IAEA guidance and code/data comparisons, including ENDF/B-VIII.1 and JENDL-5 Strong isotope, energy, library, and model dependence; local photonuclear tables require installation and verification D6 selects a conditional JENDL-5 reference with ENDF/B-VIII.1 and built-in-model comparisons; retain neutron dose in the combined target
PHITS verification PHITS manuals, benchmarks, reference data, and inter-code studies No complete benchmark matches the final source and room Use a chain of component and limiting-case checks
Physical validation Some converter, spectrum, attenuation, and facility measurements Applicability and uncertainty detail are incomplete Limit validation claims to matched measured components
Dataset design Monte Carlo statistics and leakage-aware evaluation principles Thresholds and sampling domain remain undecided Set criteria through preliminary calculations before Dataset generation
Surrogate modelling PHITS neutron surrogates, photon-shielding surrogates, Gaussian-process dose maps, and spatial ML No located model covers the complete Paper 1 mapping Compare simple baselines first and retain whole-configuration final tests

Scope decision from the accessible evidence

The accessible evidence does not justify expanding Paper 1 to the broader 0.2–10 MeV, direct-EB plus converted-X-ray programme. That domain joins multiple source, converter, electron-range, photonuclear, activation, geometry, and output regimes and would weaken the first paper's verification and applicability claims.

The evidence continues to support the narrower working scope:

  • conventional industrial LINAC;
  • converted-X-ray mode;
  • incident-electron energies from 5 to 7.5 MeV with intermediate conditions;
  • one D3/D5-amendment broad planar Ta-water-stainless converter archetype with a fixed, declared, normalized one-dimensional scanned-beam distribution;
  • the D4-approved five-variable ordinary-concrete closed-room family without streaming openings;
  • one predefined three-dimensional radiation-protection quantity;
  • PHITS reference calculations with photonuclear sensitivity; and
  • surrogate evaluation on complete configurations excluded from model fitting.

The accessible evidence supports retaining this scope. The four awaiting-full-text records must still be resolved before relying on them, but unavailable institutional databases are reported as a coverage limitation rather than a prerequisite that blocks the study.

The project owner accepted this evidence-supported boundary as D1 on 2026-09-04. The exact approved decision and justification are recorded in the Methodology.

Industrial-source reassessment — 2026-09-10

The project owner subsequently clarified that Paper 1 must target industrial radiation-processing representativeness. This changes the source-selection criterion but does not rewrite the original search or screening history. Petwal's compact circular geometry remains directly useful for component verification; it is no longer sufficient as the production source archetype. Zimek's broad tantalum plate and scanned-beam evidence is more representative of the intended application, while also exposing unresolved scan-current weighting and 5–7.5 MeV cooling-construction evidence. D3 and D5 are therefore amendment-open. Before revised acceptance, the project must obtain or explicitly bound the coolant/cover construction, define the normalized scan-position density and its quadrature, verify finite-edge interception and source normalization, and compare the scanned plate with the compact component benchmark. No earlier compact Result may enter D7 or the production Dataset.

Review limitations

  1. Scopus, Web of Science, and INSPEC were not searchable without an authenticated institutional session. IEEE Xplore metadata was searchable, but institutional full text was unavailable.
  2. Crossref bibliographic totals were extremely broad; only the first 20 relevance-ranked records per search were inspectable.
  3. OpenAlex inspection was limited to the first 25 relevance-ranked records per search except when fewer existed.
  4. English-language metadata was required.
  5. One researcher performed screening; included and borderline records received a consistency pass but no independent second screener was available.
  6. IEEE Xplore screening used a documented machine-assisted first pass and may miss relevance not expressed in available metadata.
  7. Four records require better full text before a final inclusion decision.

The project may proceed with protocol and verification preparation while reporting these limitations. It must not describe the Paper 1 gap as established by a comprehensive institutional-database review.