Scope and methods
This author’s master’s dissertation reports 90 fermented and dried bean samples, homogenized with shell, from four Brazilian states, Côte d’Ivoire and Ecuador, and one Bahia pilot processing batch. Chapter 2 section 2.1 (printed page 45) defines the samples; analytical triplicates do not add independent bean samples. ICP-OES was used after comparison of digestion methods. Table 1B (printed page 50) reports cadmium LOD 0.0005 mg/kg and LOQ 0.0015 mg/kg, and lead LOD 0.007 mg/kg and LOQ 0.022 mg/kg. Collection dates and moisture normalization are not supplied in the admitted extract.
Origin comparison
Chapter 2 Table 2 (printed page 52) reports cadmium (Cd) mean ± SD, mg/kg: Bahia 0.060±0.052(n 33), Côte d’Ivoire 0.038±0.028(n 12), Espírito Santo 0.142±0.017(n 4), Pará 0.129±0.149(n 29), Rondônia 0.063±0.008(n 8), Ecuador 1.153±0.364(n 4). The Côte d’Ivoire group had the lowest observed Cd mean; it is not a dependable country guarantee.
Reported lead (Pb) means ± SD, mg/kg, in the same order are 0.727±0.583, 0.629±0.307, 0.533±0.178, 0.985±0.656, 0.397±0.216 and 0.019±0.024. Ecuador had the lowest reported Pb mean but included results below LOQ 0.022 mg/kg; substitution used for means is unresolved. Country groups were unequal and these are whole-bean-with-shell results. The analyte directions differ and the group means do not establish a universal cleanest origin.
Processing
Chapter 2 section 3.3 (printed pages 55–57) follows a 2 kg Bahia batch, including roasting for 60 minutes at 120°C, grinding, shell separation and pressing. Reported Cd concentrations are 0.113±0.003 mg/kg in fermented/dried starting beans, 0.124±0.004 in shell, 0.060±0.006 in liquor, 0.111±0.010 in powder and <0.0015 in butter. The thesis calls butter not detected at 0.0015, but Table 1B identifies that value as LOQ; retain the numerical bound without calling it a verified LOD. All derivative Pb except shell was <0.022 mg/kg; shell was 0.142±0.006 mg/kg. The study could not quantify Pb partitioning.
These are source-specific concentrations, not proof of elemental destruction or a transferable commercial removal percentage. The single pilot batch does not establish replicated control effectiveness.
Overlap and limitations
The 90-sample origin groups and pilot work overlap the dataset subsequently reported by Ferreira de Oliveira and colleagues in Food Control 119, 107455, 2021, DOI 10.1016/j.foodcont.2020.107455. They remain one dataset family. The dissertation is a primary academic work, not an additional independent peer-reviewed occurrence survey. The original journal full text remains access-blocked; it is not silently substituted for the dissertation. Some regional Cd range bounds in Table 2 are printed <0.015 although Table 1B LOQ is 0.0015 mg/kg; no corrected bound or zero is invented.