Corrigendum to “Reaction paths for hydrodeoxygenation of furfuryl alcohol at TiO₂/Pd interfaces” [J. Catal. 377 (2019) 28–40] (Journal of Catalysis (2019) 377 (28–40), (S0021951719303161), (10.1016/j.jcat.2019.07.012))

The authors regret an error made in an input parameter in a density functional theory calculation that led some numbers in this publication to be erroneous. In a paper recently published in Journal of Catalysis [1], we investigated the role of TiO₂ encapsulated Pd interfacial sites towards dictating hydrodeoxygenation (HDO) product selectivity for furfuryl alcohol using Density Functional Theory (DFT) calculations. A rutile TiO₂ (1 1 0) nanowire over a Pd (1 1 1) surface is used as the interfacial model. Interfacial TiO₂/Pd sites were found to provide a bifunctional role that accelerated a direct-deoxygenation reaction path. TiO₂ generates reduced oxide sites that activate the C O bond of furfuryl alcohol, with the alcohol group re-oxidizing the reduced site. The Pd surface activates H₂ and enables hydrogenation to the final product. Consequently, our results showed that deoxygenation is significantly accelerated over a TiO_2-x/Pd oxygen deficient site, by a factor of ~10⁸ at 443 K, altering the Pd selectivity from the undesired furan product to the desired 2-methylfuran. Unfortunately, we found that the DFT input parameters used for calculating the energy of the oxygen deficient TiO_2-x/Pd interface surface were inconsistent with the other reported DFT calculations. Consequently, adsorption energies or other energies relative to this surface were incorrectly reported in the paper. The major qualitative conclusions of the paper are not affected by this error. This errata corrects values in the text, figures, and Supplementary information that are impacted by the error. Also, all the corrected energetics reported in this paper are tabulated in the updated Supplementary Information, Table S1. In section “2.3.2 Rutile-TiO₂ nanowire over the Pd (1 1 1) surface”, the oxygen vacancy formation energy reported for the interfacial oxygen atoms, in terms of the energy to desorb an oxygen atom as ½ O₂ as TiO₂ → TiO_2-X + ½ O₂, ranges between 3.11 and 3.18 eV, correcting the erroneous previously reported range of 3.67 and 3.74 eV. The updated range remains similar to that for the extended 3 × 1 rutile TiO₂ (1 1 0) surface (3.56 eV) [1], indicating interfacial vacancies are somewhat stabilized relative to those on the single crystal TiO₂ surface. Hence the constructed nanowire is a reasonable model to examine the impact of interfacial reducible sites on catalytic chemistry, without overestimating the role of O vacancies due to an unrealistic enhancement of their stability at the interface. Furthermore, a 0.28 eV more stable state than originally reported was found for the singly water-adsorbed state c in the reaction schematic for hydrogen adsorbed at the interface leading to an oxygen vacancy (Fig. 4 in the section “3.1.2. TiO_2-x/Pd Reduced Surface (O Vacancy)” [1]). Consequently, the relative energies for states c and d are updated (also since these states involve the TiO_2-x/Pd interface) and the Figure is updated as Fig. 1 here. Removal of the water molecule from TiO_2-x/Pd states c (Fig. 1) now costs 0.83–1.11 eV, leaving behind an oxygen vacancy (states d) (Table S1, step Xb). The energy cost is still consistent with the previous estimates reported over a rutile TiO₂ (1 1 0) surface [2]. Also, as with the change in O vacancy formation energy, the cost to create a vacancy from the hydroxylated surface (state d relative to state a, Fig. 1) is 1.06 eV (without water co-adsorbed) against the earlier reported value of 1.62 eV. The corresponding cost with a water molecule co-adsorbed at the interface is still 0.96 eV. The updating of the energetics in this figure does not affect any conclusions or discussion in the original manuscript. From section “3.2.1. Direct Deoxygenation (DDO)” of the paper, potential energy surfaces (PES) of DDO paths across DFT models in Fig. 5 [1] have been corrected for the adsorption energy of furfuryl alcohol at the TiO_2-x/Pd interface to −2.06 eV (Fig. 2), correcting the earlier reported value of −2.66 eV. The Figure has also been corrected for the adsorption of the molecule oriented vertically to the surface (labelled “TiO_2-x/Pd (O Vacancy), upright”). Similarly, in the section “3.2.3. Decarbonylation”, the potential energy surfaces (PES) of DCO paths across DFT models in Fig. 10 [1] have been corrected (Fig. 3) for the adsorption energy of the acyl intermediate C₄H₃O(CO)* relative to the energy of gas phase furfural minus ½ H₂ and the new (corrected) energy of the bare TiO_2-x/Pd surface. These changes do not affect the DDO or decarbonylation activation barriers or the reaction energies over this surface as neither reaction states involve a bare TiO_2-x/Pd surface. As the conclusions reached in the paper regarding the rate of the DDO sequence were based on surface reaction activation barriers rather than adsorption energies, the changes to these results have no impact on these conclusions. Finally, in section “3.3.2 DDO Mechanism: TiO_2-x/Pd Reduced Surface (Oxygen Deficient)”, Fig. 13 [1] showing a reaction energy diagram for the DDO of furfuryl alcohol has been corrected for the relative energies of states a, b, c and the later stages relative to gas phase furfuryl alcohol, hydrogen and the bare TiO_2-x/Pd surface (C₄H₃O(CH₂OH)(g) + H₂(g) + 3* + **vac) in Fig. 4. Consistent with the corrections elsewhere, there are no changes in activation barriers or desorption energies during the entire reaction schematic. As such, the major conclusion from our results is still valid, that in the presence of reduced Ti³⁺ sites and oxygen vacancies, the dissociation of the COH bond (direct deoxygenation) is thermodynamically as well as kinetically far more feasible than that observed over the Pd (1 1 1) surface. Regeneration of these reduced active sites is also energetically feasible, completing the catalytic cycle. The authors would like to apologise for any inconvenience caused.