Review papers with raw data transparency

Visual summary — binding energies & ZPVE (paper table VIII)

Visual diagnostic: functional dependence for H‑bonded vibrations

Detailed critique and reproducibility checklist

1. Methods clarity: paper clearly states functional(s) (BP86 with Becke/Perdew gradient corrections for monomers; B3PW hybrid for dimers), basis set (6-311G), program (Gaussian 94), geometry optimization thresholds and Berny algorithm; this supports reproducibility at the same level of theory (Santamaria et al. 1999 (methods transparency)).
2. Reproducibility score (practical): good — a competent quantum chemistry user can reproduce the harmonic frequencies and energies; caveats: (a) Gaussian 94 is old — reproducing exactly may require same program version/keywords; (b) small basis-set and missing polarization functions will change numbers by systematic offsets; (c) anharmonic corrections not included — expect differences when adding them (Santamaria et al. 1999 (basis set rationale)).
3. Energetics & H‑bond strength: authors compute pair binding energies and extract per-H‑bridge averages (~-0.015 au per H‑bond for AT; ~-0.018 au for GC) and ZPVE corrections; these values are internally consistent but sensitive to functional and basis — compare with high-level MP2/CCSD(T) benchmarks for definitive energetics (Santamaria et al. 1999 (binding energies & ZPVE)).
4. Transferability claims: the paper's Table V catalogs similar modes across bases/dimers and shows mode-by-mode mapping (useful for fingerprinting). This is valuable but must be used cautiously because environment (matrix vs gas vs crystal) and anharmonic coupling can change which modes remain 'transferable' (Santamaria et al. 1999 (Table V transferability)).
5. Missing/limited elements:
   • No anharmonic frequency corrections — later literature shows these can shift X–H and intermolecular modes substantially and change intensity patterns (see Barone/Anharmonic work) (Anharmonic corrections matter for NABs).
   • Basis set: use of 6-311G (no polarization functions) is a pragmatic compromise but underestimates dipoles and may misplace intensities and rotational constants versus polarized basis sets; authors acknowledge this (Santamaria et al. 1999 (basis set caveat)).
   • Environment/anharmonic: comparisons mix gas-phase DFT with matrix/crystal IR/Raman/NIS experiments — authors discuss this and caution interpretation; modern best practice is explicit solvent/periodic or QM/MM + anharmonicity for condensed-phase matches (Anharmonic+solvent importance).

Actionable recommendations (for readers planning replications / follow-ups)

• Recompute key H‑bridge stretching modes with: (a) larger basis sets including polarization and diffuse functions (e.g., aug-cc-pVTZ or 6-311++G**) and (b) anharmonic corrections (VPT2 or vibrational SCF) to assess real shifts vs harmonic BP86 values (Anharmonic correction recommendation).
• Benchmark energetics with higher-level correlated methods (MP2, CCSD(T) extrapolated) on fragment models to quantify functional/basis biases in H‑bond energies (Intermolecular mode benchmarks (HF/MP2/DFT)).
• When assigning experimental spectra in condensed phases, explicitly simulate environmental effects (PCM, cluster solvent, QM/MM, or periodic crystal) and report expected shifts compared to gas-phase harmonic results.

Conclusions — balanced, evidence-weighted

Santamaria et al. 1999 supplies a thorough, pragmatic harmonic DFT atlas of NAB monomer and Watson–Crick dimer vibrations using BP86/6-311G and complementary B3PW hybrid checks. It is high-quality for its time: method transparency, detailed per-mode tables, and an explicit dimer analysis (mode recognition, shifts, and binding/ZPVE energetics) make it a durable reference for identifying spectral fingerprints and for force‑field parameterization. However, for precise H‑bond stretch frequencies, absolute energetics, or condensed-phase matching, the harmonic/BP86+6‑311G framework used has systematic limitations (functional/basis dependence and neglect of anharmonicity/solvent). Follow-up work should add anharmonic corrections and larger basis / correlated benchmarks when precise absolute values are required.