Descriptor-Driven Prediction of Adsorption Energy of Oxygenates on Metal Dioxide Surfaces
Chen Chen, Zhihui Li, Jia Yang, Haifeng Wang, De Chen
TL;DR
This paper identifies key factors that predict how strongly oxygenates bind to metal dioxide surfaces, helping to design better catalysts.
Contribution
A new predictive model for adsorption energy based on oxygen effective charge and metal electron affinity is developed.
Findings
Adsorption energy depends on the effective charge of oxygen atoms in oxygenates.
Metal dioxide surfaces with lower electron affinity provide stronger adsorption stability.
A descriptor-based model accurately predicts adsorption strength across different surfaces.
Abstract
Adsorption is a critical factor in heterogeneous catalysis, as the interaction between adsorbate and adsorbent significantly impacts catalytic efficiency and selectivity. In this study, we utilized density functional theory (DFT) to comprehensively analyze the adsorption behavior of various oxygenates on the surfaces of metal dioxide (MO2) catalysts. Our findings reveal a strong dependence of adsorption energy (Ead) on two primary descriptors: the effective charge (eeff) of oxygen atoms in oxygenates and the electron affinity (EA) of the surface metal atoms in MO2. We observed that oxygenates with more negative eeff exhibit stronger adsorption, while MO2 with lower EA offer greater adsorption stability. Using these two descriptors, a predictive Ead scaling relationship was developed and validated across different MO2 surfaces. This descriptor-based model establishes an efficient…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6- —European Commission10.13039/501100000780
- —Norges ForskningsrÃ¥d10.13039/501100005416
- —National Natural Science Foundation of China10.13039/501100001809
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCatalytic Processes in Materials Science · Machine Learning in Materials Science · Gas Sensing Nanomaterials and Sensors
Introduction
1
Metal oxides are essential catalysts for converting oxygenates and facilitating the sustainable transformation of feedstocks into valuable chemicals and fuels.^1−4^ Their high thermal stability allows them to endure rigorous reaction conditions, supporting efficient and reusable catalysis.^5−8^ Among these, metal dioxides (MO_2_) exhibit tunable acid–base and redox properties that drive selective reactions, enhancing yield and reducing byproducts.^9,10^ Redox-active MO_2_, such as iridium dioxide (IrO_2_), titanium dioxide (TiO_2_), tin dioxide (SnO_2_), and cerium dioxide (CeO_2_), facilitate crucial electron transfer, thereby accelerating reaction rates.^11−16^ These attributes underscore the significant potential of MO_2_ in catalytic applications.^17−19^
The adsorption of oxygenates on MO_2_ surfaces is crucial for understanding reaction mechanisms, controlling selectivity, and preventing catalyst deactivation. The redox properties of MO_2_ enable the sustainable conversion of biomass oxygenates into fuels and chemicals. Moreover, adsorption determines active site distribution and reaction pathways, and understanding the adsorption properties is key to tuning catalysts and improving performance.^20−22^ In particular, during the conversion of oxygenates (such as alcohols, aldehydes, ketones, and acids), adsorption influences the kinetics of catalytic reactions, thereby having a major impact on reaction rates and efficiency.^23−26^ For example, Mallesham et al. significantly enhanced the adsorption of acetone on sulfate-modified SnO_2_ surfaces, accelerating the acetalization reaction and achieving a conversion rate of up to 99%.^27^ Research by Tsukamoto et al. demonstrated that inhibiting the adsorption of aldehydes on TiO_2_ surfaces is an effective strategy for increasing the selectivity of alcohol oxidation, promoting the conversion of alcohols to aldehydes.^28^ Additionally, Wei et al. discovered that the stable adsorption of aldehydes on the CeO_2_–ZrO_2_ support is the key factor underlying its exceptional catalytic hydrogenation performance.^29^ Therefore, a thorough understanding of the adsorption mechanisms of oxygenates on MO_2_ surfaces is crucial for rationally tuning the reactivity of adsorption sites and maximizing catalytic performance.
Current experimental and computational methods provide insights into adsorption on catalyst surfaces, but they are often complex and computationally expensive.^30−34^ Addressing these challenges, Nørskov et al. introduced descriptors such as electrostatic interactions, density of states, and d-electrons to predict adsorption and catalytic activity on metal surfaces.^35^ Calle-Vallejo et al. highlighted the outer electron count as a predictor of adsorption energy (Ead) for reaction intermediates on transition metals and their oxides.^36^ Although these approaches offer promising strategies for predicting adsorption activity, studies of oxygenates adsorption on MO_2_ remain limited, which restricting the optimizing and design of MO_2_ catalytic.
In this study, we systematically investigated the adsorption behavior of seven oxygenates—alcohols, aldehydes, ketones, acids, ethers, esters, and phenols—on the surfaces of various MO_2_ catalysts, including IrO_2_, SnO_2_, TiO_2_, PtO_2_, CeO_2_, and ZrO_2_, using density functional theory (DFT). Our findings reveal that the effective charge (eeff) of oxygen atoms in oxygenates and the electron affinity (EA) of MO_2_ are key descriptors for predicting Ead. Based on these, we developed a new formula for calculating Ead and validated its accuracy on five additional MO_2_ (VO_2_, PdO_2_, TaO_2_, HfO_2_, and NbO_2_) catalyst surfaces. This work enhances our understanding of adsorption mechanisms, reduces computational and experimental burdens, and provides theoretical guidance for efficient catalyst design and screening.
Computation Methods
2
The simulations were performed with the Vienna ab initio simulation package (VASP).^37,38^ All density functional theory (DFT) calculations were conducted using the Perdew–Burke–Ernzerhof (PBE) functional within the generalized gradient approximation (GGA).^39^ To account for core–valence electron interactions, the projector-augmented wave (PAW) method was employed. Valence electronic states were described using plane wave basis sets with an energy cutoff of 450 eV. Structural optimization was completed when the maximum forces on relaxed atoms fell below 0.05 eV/Å.^40^ A 1 × 1 × 1 k-point mesh was applied because of the large supercell dimensions (30 × 30 × 30 Å^3^). According to previous reports, the on-site Hubbard U term (DFT + U) was applied to the Ir 5d, Sn 5s, Ti 3d, Pt 5d, Zr 4d, Ce 4f, Hf 5d, Nb 4d, and Ta 5d orbitals, with U values of 2, 3.5, 4.2, 4, 5, 5, 6, 2, and 2 eV, respectively.^41−49^ For a more detailed explanation regarding the selection of the U value, please refer to Supporting Note 1.1 in Supporting Information.
The formula for calculating adsorption energy (Ead) in the gas phase is given by Ead = EX/surf – Esurf – EX + ΔZPE
- TΔS.^50^ In this expression, EX/surf is the total energy of the adsorbed molecule X on the surface; Esurf is the total energy of the surface without the adsorbed molecule; and EX is the total energy of the molecule X in the gas phase. ΔZPE = ZPE_X/surf_ – ZPE_surf_ – ZPE_X_, where ZPE_X/surf_, ZPE_surf_ and ZPE_X_ represent the zero-point energy of the adsorbed molecule on the surface, the bare surface, and the molecule in the gas phase, respectively. T is the temperature (typically 298 K, room temperature). ΔS = SX/surf – SX, where ΔS is the change in entropy before and after adsorption; SX/surf is the entropy of the molecule in the adsorbed state and SX is the entropy of the molecule in the gas phase. The specific effects of ΔS and ΔZPE are discussed in Supporting Note 1.2 in the Supporting Information. After performing static calculations with VASP, vibrational frequency analysis is conducted. VASPKIT is used to extract entropy and zero-point energy information for correction, and this method has been validated for its accuracy.^51^ The exchange-correlation interactions were described using the vdW-DF method, which combines a nonlocal correlation functional with a consistent exchange functional to account for van der Waals forces accurately.^31,52^ To more accurately describe the strength of chemical bonds, bond energies are calculated using the Integrated Crystal Orbital Hamilton Population (ICOHP) method.^53^ All calculations are performed using the Lobster program in conjunction with VASP.^54,55^ Additionally, the electron affinity (EA) is calculated using the formula EA = Eneutral– Echarged, where Eneutral is the total energy of the neutral system, and Echarged is the total energy of the negatively charged system after introducing an extra electron and structural relaxation. This difference reflects the EA of the system. The above DFT setup affords good accuracy, proved by previous work.^56,57^
Results and Discussion
3
Structure of MO2 Surfaces and Oxygenates
3.1
Based on their superior catalytic performance and widespread application in biomass conversion reactions, we selected IrO_2_, SnO_2_, TiO_2_, PtO_2_, CeO_2_, and ZrO_2_ as the MO_2_ model adsorbents.^58,59^ We also chose widely studied active crystal phases and facets from the literature, including rutile-IrO_2_(110), rutile-SnO_2_(110), rutile-TiO_2_(110), tetragonal-PtO_2_(111), tetragonal-ZrO_2_(111), and tetragonal-CeO_2_(111),^60−65^ to investigate the adsorption stability of oxygenates on the surfaces of MO_2_.
As shown in Figure 1a, IrO_2_(110), SnO_2_(110), and TiO_2_(110) surfaces are exhibit a terrace-step structure, exposing two-coordinated bridging oxygen sites (O_2c_) and five-coordinated metal sites (M_5c_). Additionally, the PtO_2_(111), CeO_2_(111), and ZrO_2_(111) surfaces are characterized by a zigzag structure. Moreover, CeO_2_(111) and ZrO_2_(111) surfaces exposes three-coordinated oxygen sites (O_3c_) and seven-coordinated metal sites (M_7c_). The main active centers of the adsorption reaction are the metal sites, and the adsorption behavior depends to a large extent on the geometry and electronic structure of these sites. The subsequent results and discussion in this study are based on these extensively studied MO_2_ surfaces, especially the terrace-step or zigzag active surfaces of rutile and tetragonal phases. To systematically study the adsorption behavior of oxygenates on the MO_2_ surface, we selected seven representative categories of compounds, including alcohols, aldehydes, ketones, acids, ethers, esters, and phenols. Using different categories of C3 molecules (e.g., propanol, acetone, propionic acid, etc., excluding phenol) as model compounds to analyze their adsorption stability and interaction mechanisms on the surfaces of MO_2_ (Figure 1b).
(a) Structure of various MO2 surfaces, including rutile-IrO2(110), rutile-SnO2(110), rutile-TiO2(110), tetragonal-PtO2(111), tetragonal-ZrO2(111), and tetragonal-CeO2(111). Insets show the local coordination environments of the metal sites. (b) Molecular structures of oxygenates, including C3-alcohol, C3-aldehyde, methyl propionate (ester), C3-ketone, C3-acid, propyl methyl ether, and phenol. Color scheme: dark blue (Ir), light green (Sn), light gray (Ti), green (Pt), light blue (Zr), light yellow (Ce), red (O), gray (C), and white (H). This color scheme is used throughout the paper.
Adsorption Energy
3.2
The adsorption behavior of seven oxygenates on six different MO_2_ surfaces was investigated, and the corresponding adsorption energies (Ead) and configurations are presented in Figures 2a–c and S4. Overall, the interactions are primarily governed by the bonding between oxygen atoms in the oxygenates and the exposed metal sites on the MO_2_ surfaces.^66^ DFT calculations indicate that oxygenates bearing hydroxyl (−OH) groups, such as alcohols and phenols, generally undergo dissociative adsorption, consistent with earlier reports in the literature.^26,67,68^ By contrast, acids preferentially adsorb via the C=O bond, which proves energetically more favorable than the dissociative pathway (−OH dehydrogenation). For instance, on the IrO_2_ surface, Ead of the acid via the C=O group is −1.11 eV, while −OH dehydrogenation (−O) adsorption is only exothermic by 0.73 eV. Therefore, the most stable C=O-bonded structure was used for subsequent discussions on acid adsorption.
(a) Ead of oxygenates on various MO2 surfaces, including rutile-IrO2(110), rutile-SnO2(110), rutile-TiO2(110), tetragonal-PtO2(111), tetragonal-ZrO2(111), and tetragonal-CeO2(111). (b, c) Adsorption structures of oxygenates on the rutile-IrO2(110) and tetragonal-CeO2(111) surfaces. Green highlights the Ir and Ce adsorption sites, while pink indicates the adsorbed oxygen atoms. Black markers denote bond lengths.
As shown in Figures 3a and S5, we compared the Ead values for oxygenates with varying carbon chain lengths (C3, C4, and C6) on MO_2_ surfaces. Tables S9–S14 present the Ead of oxygenates with C3, C4, and C6. In all cases, the standard deviations are below 0.032. Therefore, at least within this range of C3–C6 linear molecules, the effect of steric hindrance is minimal, and variations in carbon chain length have a negligible impact on Ead. These findings are consistent with previous reports.^69,70^ Accordingly, we chose C3-based molecules as representative adsorbates to optimize computational efficiency without compromising accuracy.
(a) Error bar chart of Ead of oxygenates with varying carbon chain lengths (C3, C4, and C6) on rutile-IrO2(110) and tetragonal-CeO2(111) surfaces. In the case of phenol, the C3, C4, and C6 designations refer to the addition of alkyl groups (−C3H7, −C4H9, and −C6H13) to the benzene ring, corresponding to phenol derivatives with different carbon chain lengths. (b) Bader charges of oxygen atoms (eO) in oxygenates, and the black markers represent the effective charge (eeff) of oxygen atoms in oxygenates. (c, d) Correlation between the eeff and Ead on rutile-IrO2(110) and tetragonal-CeO2(111) surfaces, along with M–O bond energies (-ICOHP). The M–O bond refers to the bond formed between the metal site of MO2 and the oxygen atom of the adsorbed oxygenates. The red and blue lines represent the best-fit regression models for Ead on rutile-IrO2(110) and tetragonal-CeO2(111) surfaces, respectively. The green line shows the correlation between the eeff and M–O bond energies. The shaded regions around the lines represent the 95% confidence intervals, indicating the uncertainty in the predicted mean values. The outer shaded regions represent the 95% prediction intervals, showing the expected range of future observations with 95% confidence.
Moreover, our Ead results align well with prior studies. For instance, Giuliano et al. reported an Ead of 0.60 eV for alcohols on TiO_2_, matching closely with our calculated value of 0.58 eV.^69^ Similarly, Alfredo et al. obtained an Ead of 0.33 eV for phenol on CeO_2_, compared with our result of 0.29 eV.^71^ The computed adsorption configuration was also in good agreement, featuring a Ce–OC_6_H_6_ bond length of about 2.5 Å. These comparisons confirm the reliability of our DFT results.
As shown in Figure 2a, a clear trend in Ead is observed across different adsorbates, in the order: esters > ketones > phenols
acids aldehydes > alcohols > ethers. The relatively sparse data points in the range of −1.55|e| to −1.75|e| stem from the selection of commonly studied oxygenates, which do not naturally occupy this charge region. Moreover, the bonding between the double-bonded oxygen in esters and ketones and the metal sites enhances the exothermic nature of the adsorption process, leading to greater adsorption stability. According to reports by Balajka et al., the adsorption of oxygenates containing double-bonded oxygen is more stable, which aligns with the conclusions of our study.^70^ For instance, on IrO_2_ surfaces, the strongest adsorption is observed for methyl propionate, with an Ead of −1.27 eV, corresponding to an Ir–O bond energy of 1.77 eV and a bond length of 1.95 Å. In contrast, for the weakest adsorption—methyl propyl ether—the Ead is only −0.89 eV, with an Ir–O bond energy of 0.45 eV and a bond length of 2.29 Å. This variation in adsorption strength reflects the differing interactions between adsorbates and metal sites. The Ead of oxygenates on the surfaces of MO_2_ are presented in Tables S15–S17. Similar adsorption trends were observed across the five catalyst surfaces, demonstrating a high level of consistency in adsorption stability.
Dependence of Adsorption Energy on the Oxygenate
Properties
3.3
The essence of adsorption bonding lies in electron transfer and electron cloud overlap, making the electronic properties of the adsorbate crucial for adsorption stability. To investigate how the molecular properties of oxygenates affect adsorption strength, various properties, such as C–O and C–H bond energy and bond length, as well as the charge on oxygen and carbon atoms, were estimated and correlated with Ead. As shown in Figure 3b, we performed a quantitative analysis of the oxygen atom charge in oxygenates using the widely applied Bader charge analysis.^72−75^ The results indicate significant differences in the charge of the oxygen atoms across different molecules. Notably, oxygen atoms in ester molecules carry the most negative charge, while those in ether molecules carry the least negative charge. This distinction arises from their electronic structures: in esters, the carbonyl double bond increases oxygen’s electronegativity and enhances the π-electron delocalization effect, shifting the electron cloud toward the oxygen atom and reducing the availability of lone pair electrons for sharing. Conversely, in ether compounds (e.g., R–O–R’), the oxygen atom is bonded to two carbon atoms via single bonds, resulting in a weaker electron-withdrawing effect and a more uniform electron distribution, which gives the oxygen atom a more positive charge.
Among the different molecule properties studied, a correlation between the Ead and the Bader charge of the oxygen atom (eO) in these molecules was observed. However, using IrO_2_ as an example, directly correlating eO with the Ead yields a relatively weak fit (R^2^ = 0.80), primarily due to deviations from acids and ketones. Moreover, this weakened correlation resulting from acids and ketones is consistently observed across all investigated MO_2_ surfaces. This phenomenon can be attributed to the polarization and conjugation effects of acids and ketones. Specifically, in acid molecules, the high polarity of the carboxyl oxygen induces electron cloud migration toward the carbonyl (C=O) double bond, leading to a higher negative charge on the oxygen atom. In contrast, the conjugation effect in ketone molecules delocalizes the electron density of the oxygen atom, reducing its local negative charge and making the oxygen atomic charge more positive.^76,77^ However, the Bader charge analysis method cannot fully capture these intricate electronic effects, thereby affecting the accuracy of the correlation.
To address this limitation, we introduced an empirical correction to the descriptor, namely the oxygen atom’s charge for oxygenate molecules in acids and ketones.^78,79^ Notably, these correction factors are intrinsic molecular properties and therefore remain constant across different metal dioxide surfaces. The corrected effective charge (eeff) is expressed as
where eO is the original oxygen atom charge and a is the correction factor. For acid and ketone molecules, the correction factors are 1.05 and 0.95, respectively.
The eeff of oxygen atoms in oxygenates are shown in Table S18. These corrections effectively account for polarization and conjugation effects, yielding corrected oxygen charges of −1.86|e| for acids and −1.79|e| for ketones. As shown in Figures 3c,d and S6, applying these corrections significantly improves the linear correlation between eeff and the Ead with an R^2^ value exceeding 0.94 (IrO_2_). Furthermore, the strong correlation between eeff and M–O bond energy confirms the descriptor’s reliability. The more negative the corrected oxygen charge, the stronger the electron transfer, leading to enhanced adsorption interactions with metal sites. These results demonstrate the robustness of the empirical correction method and its applicability across various MO_2_ surfaces.
As shown in Figure 4a, the fitting lines for Ead and eeff across different MO_2_ surfaces are nearly parallel, with an average slope of 0.922. This results in the following expression for Ead
Where bx is the fitting constant reflecting the intrinsic properties of MO_2_. The values of k_x_ (average slope), bx (intercept), and R^2^ (goodness-of-fit) are shown in Table S19. This linear relationship implies that for a given MO_2_ surface, once the constant bx is determined, the Ead of oxygenates can be predicted based on eeff. Thus, eeff serves as an effective descriptor of Ead, providing a theoretical basis for predicting and controlling adsorption behavior on MO_2_ surfaces.
(a) Correlation between the Ead and the eeff. (b) Correlation between the EA and the Ead. (c) Three-dimensional plot illustrating the relationship between Ead and the two descriptors (eeff and EA). (d) The linear fit between bx and EA.
Dependence of Adsorption Energy on the Metal
Dioxides
3.4
The Ead of oxygenates on MO_2_ depends not only on the properties of adsorbate but also closely relates to the properties of the adsorbent. The Ead are shown in Table S15. The same adsorbate exhibits a consistent adsorption trend on different MO_2_ surfaces: IrO_2_ > SnO_2_ > TiO_2_ > PtO_2_ > ZrO_2_ > CeO_2_. This phenomenon is determined by the geometric and electronic structure of the adsorption active sites on the catalyst surface. First, the platform-step structured surfaces (rutile-IrO_2_(110), rutile-SnO_2_(110), rutile-TiO_2_(110)) exhibit better adsorption stability for oxygenates compared to the zigzag surfaces (tetragonal-PtO_2_(111), tetragonal-ZrO_2_(111), and tetragonal-CeO_2_(111)). The flat surfaces provide more space for molecular adsorption, while the increased steric hindrance on the zigzag surfaces lead to a decrease in adsorption stability. Additionally, oxygen atoms in oxygenates usually carry a strong negative charge, which facilitates electron transfer to the metal sites on MO_2_ surfaces. Consequently, the ability of MO_2_ to accept electrons significantly influences adsorption stability: the stronger the electron acceptance capability, the more stable the adsorption.
As shown in Figure S7, we calculated the electron affinity (EA) of metal sites on various MO_2_ surfaces, which shows a certain scaling relationship with the electronegativity of the metal elements. The formula for EA is EA = Eneutral – Echarged, where Echarged is the total energy of the system with an additional electron localized at the surface metal site (Figure 5a), and Eneutral is the total energy of the neutral system. As shown in Figure 4b, the EA shows a strong exponential correlation with Ead (R^2^ > 0.99), and the lower the EA, the stronger the adsorption stability. Therefore, EA is also a valid descriptor for Ead. For example, the EA of IrO_2_ is 0.21 eV, requiring less energy for electron transfer, which allows it more easily to accept electrons from the adsorbate, resulting in stronger interactions with oxygenates. In contrast, CeO_2_ has a much higher EA of 2.04 eV, making it more difficult for oxygen atoms to transfer electrons to the Ce sites, leading to relatively weaker adsorption stability with oxygenates. Moreover, the adsorption trends across different MO_2_ surfaces are consistent, with the fitted curves being nearly parallel, and the average slope is −1.302. Therefore, the Ead can be expressed as
where cx is the fitting constant that reflects the properties of the oxygenates (detailed data shown in Table S20).
(a) Spin state density plots of an electron localized on the MO2 surface. (b) The EA of MO2 and the bx in eq 2.
It is important to note that both EA and the fitting constant bx characterize the adsorption strength based on the intrinsic properties of MO_2_, and thus they are relevant. As shown in Figure 4d, EA and bx exhibit consistent trends and demonstrate a strong linear correlation (R^2^ = 0.95). Consequently, the unknown coefficient bx in the Ead calculation formula eq 2 can be substituted with the expression for EA
Combining eqs 2 and 4, Ead can be expressed in terms of the properties of the adsorbed molecule (eeff) and the properties of MO_2_ (EA) as follows in eq 5
This equation accounts for the combined effects of oxygenates and surface interactions, providing a more comprehensive explanation of the adsorption mechanism from the perspective of electronic properties. As illustrated in Figure 4c, Ead is plotted as a function of two descriptors: one associated with the adsorbate (eeff) and the other with the adsorbent (EA). Although eeff and EA are derived from DFT calculations, they represent intrinsic material properties that remain consistent for a given system. Table S18 presents the eeff values for seven oxygenate molecules and the EA values for 11 MO_2_. Due to their intrinsic nature, these parameters can be broadly applied to various adsorption scenarios without necessitating repeated adsorption energy calculations. In contrast to traditional adsorption simulations, which require full DFT optimizations for each adsorption configuration, our approach substantially reduces the computational cost and effort.
Validation
3.5
To further validate the accuracy and universality of the proposed Ead prediction formula based on the eeff and EA descriptors, we compared the Ead from DFT-calculated Ead with scaling relationship based-prediction on another five different MO_2_ surfaces: rutile-PdO_2_(100), rutile-VO_2_(100), rutile-TaO_2_(100), tetragonal-HfO_2_(−101) and rutile-NbO_2_(100) (Ead as shown in Table S21). As illustrated in Figure 6a, the Ead calculated using eq 2 exhibits a strong correlation with eeff (R^2^ > 0.92), with a consistent fitting slope (σ = 0.02). Similarly, the Ead predicted using eq 3 shows a strong exponential correlation with EA (Figure 6b), with fitting coefficients comparable to those obtained for previously analyzed MO_2_ systems (σ = 0.04).
(a) Correlation between the Ead of oxygenates on rutile-PdO2(100), rutile-VO2(100), rutile-TaO2(100), tetragonal-HfO2(−101) and rutile-NbO2(100) and the eeff (eq 2). The gray fitting line represents the six MO2 previously calculated. (b) Correlation between the EA of rutile-PdO2(100), rutile-VO2(100), rutile-TaO2(100), tetragonal-HfO2(−101) and rutile-NbO2(100) and the Ead of oxygenates on their surface (eq 3). Hollow spheres represent previous calculations and solid spheres represent predictions. (c) Error bar chart of Ead error obtained from DFT and prediction eqs (eq 5).
Overall, the Ead predicted using eeff and EA closely match the DFT-calculated values, with a standard error below 0.08 (Figure 6c). This indicates that the proposed predictive model exhibits broad applicability across different MO_2_ surfaces and effectively captures the energy variations during the adsorption process. By validating the model on these five MO_2_ surfaces, we further confirm that eeff and EA serve as reliable descriptors for assessing the adsorption stability of oxygenates on different MO_2_ surfaces. This finding provides a solid theoretical foundation for the rational selection of catalysts based on adsorption properties.
Conclusions
4
In conclusion, using density functional theory (DFT), we have developed an efficient method for predicting adsorption energy (Ead) based on descriptors. The effective charge (eeff) of the oxygen atoms involved in adsorption accurately describes the adsorption stability and interaction strength of various oxygenates, from esters with the strongest adsorption to ethers with the weakest. This phenomenon is attributed to the more negative eeff in esters compared to ethers. Another descriptor, electron affinity (EA), also shows a strong exponential correlation with Ead: lower EA facilitates charge transfer, thus enhancing adsorption stability. Based on the descriptors eeff and EA, we propose a scaling relationship for predicting Ead. The accuracy of this method is further validated by comparing its predictions with results obtained from DFT calculations. This model reduces the demand for extensive computational resources, providing a practical approach for screening and designing metal oxide catalysts with optimized adsorption properties.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Liu Z.-P.; Gong X.-Q.; Kohanoff J.; Sanchez C.; Hu P. Catalytic Role of Metal Oxides in Gold-Based Catalysts: A First Principles Study of CO Oxidation on Ti O 2 Supported Au. Phys. Rev. Lett. 2003, 91, 26610210.1103/Phys Rev Lett.91.266102.14754070 · doi ↗ · pubmed ↗
- 2Wang L.; Shi C.; Wang L.; Pan L.; Zhang X.; Zou J.-J. Rational design, synthesis, adsorption principles and applications of metal oxide adsorbents: a review. Nanoscale 2020, 12, 4790–4815. 10.1039/C 9NR 09274 A.32073021 · doi ↗ · pubmed ↗
- 3Yuan S.; Duan X.; Liu J.; Ye Y.; Lv F.; Liu T.; Wang Q.; Zhang X. Recent progress on transition metal oxides as advanced materials for energy conversion and storage. Energy Storage Mater. 2021, 42, 317–369. 10.1016/j.ensm.2021.07.007. · doi ↗
- 4Ahmad S.; Sohail K.; Chen L.; Xu H.; Din H. U.; Zhou Z. Type-II van der Waals heterostructures of Ge C, Zn O and Al 2SO monolayers for promising optoelectronic and photocatalytic applications. Int. J. Hydrogen Energy 2023, 48, 25354–25365. 10.1016/j.ijhydene.2023.03.268. · doi ↗
- 5Lang R.; Du X.; Huang Y.; Jiang X.; Zhang Q.; Guo Y.; Liu K.; Qiao B.; Wang A.; Zhang T. Single-Atom Catalysts Based on the Metal–Oxide Interaction. Chem. Rev. 2020, 120, 11986–12043. 10.1021/acs.chemrev.0c 00797.33112599 · doi ↗ · pubmed ↗
- 6Li Y.; Zhang Y.; Qian K.; Huang W. Metal–Support Interactions in Metal/Oxide Catalysts and Oxide–Metal Interactions in Oxide/Metal Inverse Catalysts. ACS Catal. 2022, 12, 1268–1287. 10.1021/acscatal.1c 04854. · doi ↗
- 7Yu K.; Lou L.-L.; Liu S.; Zhou W. Asymmetric Oxygen Vacancies: the Intrinsic Redox Active Sites in Metal Oxide Catalysts. Adv. Sci. 2020, 7, 190197010.1002/advs.201901970.PMC 697494131993288 · doi ↗ · pubmed ↗
- 8Lee J.; Orilall M. C.; Warren S. C.; Kamperman M.; Di Salvo F. J.; Wiesner U. Direct access to thermally stable and highly crystalline mesoporous transition-metal oxides with uniform pores. Nat. Mater. 2008, 7, 222–228. 10.1038/nmat 2111.18223653 · doi ↗ · pubmed ↗