hello every one

I try to simulate a fuel assembly 17 X 17 using k-code in Monte Carlo code, with the fuel type being UO₂, enriched to 2.82% U-235 and having a density of 10.286 g/cm³. Two approaches were used to define the material composition. In the first case, the density was input into the fuel cell card with atom fractions of U-235 = 0.0282, U-238 = 0.9718, and O-16 = 2.0. In the second case, the atoms/barn.cm unit was calculated using international constants, where Avogadro’s number was taken as 0.6022 × 10²⁴ atoms/mol, and atomic masses were U-235 = 235.0439299 g/mol, U-238 = 238.05078826 g/mol, and O-16 = 15.9994 g/mol, with the formula n = ρ × N_A / M. In the third case, values from the MCNP 6.2 manual and XDIR data were used, where Avogadro’s number was 0.59703109 × 10²⁴ atoms/mol, and atomic masses were U-235 = 233.024800 g/mol, U-238 = 236.005800 g/mol, and O-16 = 15.857510 g/mol. Each approach led to different k-effective values, with a full-core simulation showing differences exceeding 1000 pcm, raising the question of which methodology is correct. Since the problem statement provides fuel density and enrichment, why MCNP using Avogadro number different in value is that means the atom density from MCNP is not correct? it should be have the same value when using atomic density or atom fraction.

reference IN CHAPTER 2 SECTION 2 PAGE 2.2 the value of Avogadro No. MCNP:https://mcnp.lanl.gov/pdf_files/TechReport_2017_LANL_LA-UR-17-29981_WernerArmstrongEtAl.pdf