On the Interaction between Superatom Al12Be and DNA Nucleobases/Base Pairs: Bonding Nature and Potential Applications in O2 Activation and CO Oxidation.

The interaction between quasi-chalcogen superatom Al12Be and DNA nucleobases/base pairs has been explored by searching for the most stable Al12Be-X (X = DNA bases and base pairs) complexes. Our results reveal that Al12Be prefers to combine with guanine by two Al-O and Al-N bonds rather than the other DNA bases, no matter in free state or base pair. The formed Al-N and Al-O bonds between Al12Be and DNA bases proved to be strong polar covalent bonds by the Wiberg bond index, nature bond orbitals, atoms in molecules theory, localized molecular orbitals, and electron localization functions analyses. More importantly, it is found that the formed global minimum of Al12Be-G has the ability to activate an oxygen molecule into a peroxide dianion 1O2 2-, which can further catalyze the CO oxidation via the Eley-Rideal mechanism with a small energy barrier of 7.78 kcal/mol. We hope that this study could not only provide an in-depth understanding on the intermolecular interaction between metallic superatoms and DNA at the molecular level but also attract more interest in designing and synthesizing superatom-based heterogeneous catalysts with DNA/nucleobases as basic building blocks.

Introduction

n class="Chemical">Nanoclusters pan> class="Chemical">comprised of several to hundreds of atoms have found diverse applications in the fields of optics, sensors, electronics, energy, environment, biology, and catalysis because of their unique and unexpected properties, such as the quantum size effect and superparamagnetic relaxation, which are quite different from the corresponding bulk part.[1−4] Hereinto, of particular interest is the application of clusters in catalysis because nanoclusters can act as individual active sites, and minor changes in size or composition could have a substantial influence on their activity and selectivity for a certain catalytic reaction.[5] As a result, extensive studies on the design and synthesis of atomically precise metal nanoclusters for catalysis have been carried out in the past decades.[6−8]

Among various clusters, superatoms[9,10] are a class of special clusters mimicking the chemistry of atoms in the periodic table, which could be regarded as man-made “elements”.[11] Till now, a lot of well-designed superatoms, including superalkalis,[12−14] superhalogens,[15] alkaline earth superatoms,[16] multiple valence superatoms,[17] magnetic superatoms,[18] aromatic superatoms,[19] and so forth, have been reported and studied by experimental and theoretical chemists. Such specific clusters can play an important role in material science because they can maintain their identities similar to atoms when assembled into an extended nanostructure[20] and thus are regarded as excellent building blocks of novel nanomaterials with highly desired and tunable properties.[21−23] More importantly, these intriguing superatoms should also have great potential to be used as ideal model catalysts because they possess well-defined compositions and structures. However, to the best of our knowledge, the applications of “special superatoms”,[9] namely, bare metallic superatom clusters in chemical catalysis are regrettably scarce because they tend to coalesce during the practical applications.[10] To resolve this problem, the best strategy is extending these special superatoms into “general superatoms”[9] by introducing ligands to protect them. The famous general superatoms are ligand-protected gold superatoms,[24] which have been successfully synthesized and utilized to catalyze some special reactions, such as ambient CO oxidation.[25−28] Nevertheless, all of these reported superatomic catalysts are transition-metal-based clusters; therefore, it is of significant interest to explore novel superatom-based catalysts composed of only main group elements.

To search for proper ligands to protect bare pan> class="Chemical">metallic superatoms, the DNA nucleobases, as the building blocks of DNA, have attracted our great attention because that the interaction of DNA with metal nanoclusters has exhibited versatile applications in the field of nanobiotechnology.[29,30] As a merging point for material science and biological science, the DNA-based nanobiotechnology has undergone unprecedented advancements in recent years. Nonetheless, for all these diverse applications in this field, understanding the binding of DNA nucleobases [adenine (A), thymine (T), guanine (G), and cytosine (C)] and Watson–Crick base pairs (BPs) with metal clusters is of critical importance.[31−34] For instance, Jena et al.[31] have suggested DNA base-gold cluster complexes to be attractive and efficient catalytic model systems for the CO oxidation process by studying the interaction between DNA bases and BPs with the Au3 cluster. Hence, one interesting question emerges: could DNA nucleobases be used as unique ligands to modify the electronic structures of superatoms for catalysis applications? To this end, the investigations on the interaction nature between superatoms and DNA nucleobases/BPs are meaningful and also highly desired.

Consequenpan>tly, the quasi-chpan> class="Chemical">alcogen superatom,[35] that is, Al12Be was selected as the precursor to study its interaction with DNA nucleobases/BPs in this work. As Al12Be lacks two extra electrons to achieve an electron-closed shell, it can serve as a Lewis acid, whereas all the DNA bases containing nitrogen and oxygen with lone pairs act as Lewis bases. Thereby, it is highly expected that Al12Be can form stable complexes with DNA bases. To verify this hypothesis, the Al12Be cluster was combined with nucleobases [adenine (A), thymine (T), guanine (G), and cytosine (C)] and BPs [guanine–cytosine (GC) and adenine–thymine (AT)] to obtain the global minima of Al12Be–X (X = DNA bases and BPs) by considering all of the possible interacting modes. Our objective for this work is to illuminate (1) the intermolecular interaction between Al12Be and DNA bases at the molecular level and (2) the mechanism of O2 activation and CO oxidation on the resulting DNA base-superatom complexes. Understanding the bonding interaction between metallic superatoms and DNA bases can be taken as the first step toward extending the application of superatoms in the emerging DNA-based nanobiotechnology. However, the larger perspective that we try to bring about from the current study is to critically evaluate the potential of such DNA base-superatom systems as model catalysts for O2 activation and CO oxidation. Considering the unique attributes of DNA as a “programmable assembler” for the bottom-up synthesis of materials,[36] we hope that this study could intrigue more interest in achieving such superatom-based heterogeneous catalysts with DNA as a basic building block or an active support.

Results and Discussion

Geometric Structure

Initin class="Chemical">ally, the geometric structures on class="Chemical">f bare Al12Be, isolated DNA bases, and BPs are optimized at the M06-2X/6-311+G(d) level and are shown in Figure S1. It is observed that the optimized Al12Be cluster exhibits an icosahedral structure with D3 symmetry, which is consistent with previous study.[35] The optimized nucleobases and BPs have a C point group, except for guanine with a low C1 symmetry. By combining Al12Be with DNA nucleobases/BPs, a series of Al12Be–X [X = guanine (G), cytosine (C), adenine (A), thymine (T), and BPs (GC and AT)] complexes were obtained. The global minima of Al12Be–X are shown in Figure , while the corresponding low-lying isomers are shown in Figures S2–S7. For convenience, the resulting isomers of Al12Be–X were named by the involved DNA bases or BPs, whose relative energy increase in the order X-1 < X-2 < X-3 < X-4...

Figure 1

Global minimum structures of Al12Be–X [X = DNA bases (G, C, A, and T) and BPs (GC and AT)] at the M06-2X/6-311+G(d) level. Selected bond lengths (in Å) and symmetry (in the parentheses) are also shown.

Global minimum structures opan> class="Chemical">f Al12Be–X [X = DNA bases (G, C, A, and T) and BPs (GC and AT)] at the M06-2X/6-311+G(d) level. Selected bond lengths (in Å) and symmetry (in the parentheses) are also shown.

As shown in Figure , the inpan>teraction mode n class="Chemical">between Al12Be and DNA is either monodentate for C-1, A-1, T-1, and AT-1 or bidentate for G-1 and GC-1. To be specific, Al12Be preferentially binds to both of O6 and N7 sites of guanine, the N3 site of cytosine and adenine, and the O4 site of thymine in these Al12Be–X complexes. As for Al12Be–G, it is found that Al12Be can be combined with any nitrogen atom or oxygen atom with the lone pair of guanine, resulting in various isomers, as shown in Figure S2. Among them, the strongest interaction occurs in G-1, in which guanine tends to synchronously bind with Al12Be via its O6 and N7 atoms, yielding the Al–N and Al–O bonds of 1.973 and 1.871 Å, respectively. When Al12Be is only linked to one N7 or O6 atom, the next two low-lying isomers G-2 and G-3 are obtained, which are higher in energy than G-1 by 9.27 and 10.75 kcal/mol, respectively.

Turning to cytosine, the nonprotonated N3 atom can contribute electrons to the aluminum atom more easily than other atoms in this DNA base, and thus, the most stable isomer C-1 generates when N3 combines with an Al atom of Al12Be, yielding the Al–N bond of 1.950 Å. The same binding mode has been also found in the previous study on the interaction between aluminum atom/ion and cytosine.[37] The next isomer C-2 with a relative energy of 13.78 kcal/mol is obtained by combining Al12Be with the other nonprotonated O2 atoms of cytosine. Furthermore, it is observed that the Al12Be superatom can also combine with the heterocyclic carbon in cytosine, forming the Al–C bonds in isomers C-3 and C-4.

For the Al12Be–A complexes, Al12Be always tends to bind with the nitrogen atoms of adenine in the five resulting low-lying isomers, as shown in Figure S4. Apparently, Al12Be is more conducive to binding with the nonprotonated N1, N3, and N7 atoms. Among these nonprotonated nitrogen sites, the N3 atom of adenine is the easiest one to be combined with Al12Be because of the steric hindrance effect of the amino group on the N1 and N7 sites.

As for n class="Chemical">Al12Be–T complexes, only four isomers were obtained, where Al12Be tends to combine with the O2 and O4 sites instead of the heterocyclic carbon or nitrogen atoms of thymine. Hence, the T-3 and T-4 are much higher in energy as compared with the first two isomers T-1 and T-2 (see Figure S5). In particular, the O4 site is superior to O2 in combination with the Al12Be superatom because the O4 atom is easier to provide a lone pair to combine Lewis acid than the O2 position.[38] Hence, the most stable T-1 is formed via forming the Al–O4 bond of 1.811 Å.

Likewise, the interaction betweenpan> Al12Be and BPs has been also studied and the resulting equilibrium structures of Al12Be–BP are presented in Figure as well as Figures S6 and S7. Nine low-lying isomers of Al12Be–GC were obtained by approaching the Al12Be superatom to the GC BP. It is observed that, in this GC BP, the guanine base is superior to cytosine in combination with Al12Be, yielding the isomers GC-1 and GC-2. This can be understood by the fact that both of the nonprotonated nitrogen and oxygen atoms in cytosine have participated in forming hydrogen bonds in the GC BP, which hinders its combination with Al12Be. It is interesting to find that GC BP also combines with Al12Be via the O6 and N7 atoms of guanine in the lowest-energy GC-1 complex, which is consistent with the way that Al12Be binds with G in G-1. Moreover, the formed Al–N and Al–O bonds of 1.864 and 1.975 Å in GC-1 are nearly equal to those of 1.871 and 1.973 Å in G-1, respectively. This indicates that Al12Be tends to combine with guanine by the bidentate mode no matter in BP or even in the DNA sequence. When the GC BP combines with Al12Be via only one N7 atom of guanine, the less stable isomer GC-2 with a relative energy of 9.83 kcal/mol is generated. Similarly, an aluminum atom of Al12Be is linked with the GC BP by a single bond in the rest of low-lying Al12Be–GC structures (see Figure S6). In the case of Al12Be–AT compounds, Al12Be tends to bind with the O4 site of thymine of AT BP in the global minimum. In this process, the hydrogen bonds in the AT BP are substantially broken by the introduction of Al12Be, leading to the distortion of the planar structure of AT BP in the AT-1 compound. In contrast, the structure of the AT subunit is well preserved in the AT-2, AT-3, and AT-9 isomers, respectively.

To evaluate the effect of Al12Be on the hydrogen bonds in BPs, we performed the atoms in molecules (AIM)[39−41] theory analysis by calculating the electron density (ρ). It is known that ρ at the bond critical point (BCP) of the H-bond (D–H···A) has a strong linear relationship with the strength of hydrogen bonds.[42−44] In this study, the H-bond lengths, relevant angles, and ρ at the BCP of H-bonds for the lowest-energy Al12Be–BPs are shown in Table S1. It can be seen clearly that the strength of hydrogen bonds nearby Al12Be is weakened, whereas the other hydrogen bonds are reinforced to different degrees in Al12Be–BPs. To be specific, when Al12Be binds with O6 of guanine in GC-1, the length of the hydrogen bond N4(C)–H···O6(G) is elongated from 1.884 to 2.194 Å, the ∠N4(C)–H–O6(G) is reduced from 175.8 to 145.4°, and ρ is decreased from 0.027 to 0.015, implying that the strength of this hydrogen bond is significantly weakened. As for the other two hydrogen bonds [N3(C)···H–N1(G) and O2(C)···H–N2(G)] in GC-1, the increase in ρ and the decrease in the hydrogen bond distances indicate that the strength of these two hydrogen bonds are strengthened upon combining with Al12Be. This can be ascribed to the electron transfer from guanine to Al12Be during the combination between GC and Al12Be. As shown in Table S2, after being attached to Al12Be, the nature population analysis (NPA) charges on guanine in GC are obviously increased from −0.048 |e| to 0.249 |e|, which leads to the decreased electron density and enhanced Lewis acidity of the guanine directly linked to Al12Be and therefore greatly enhances its ability to form hydrogen bonds with cytosine. On the contrary, the strength of N4(C)–H···O6(G) is weakened because the O6 site of the guanine serves as a proton acceptor in this hydrogen bond. The same case is also true for the hydrogen bonds in AT-1. Consequently, it can be concluded that the introduction of Al12Be indeed affects the hydrogen bonds of DNA BPs.

Energetic Properties

In order to evalun class="Chemical">ate the interaction strength between Al12Be and bases/BPs, the Gibbs free energy changes (ΔG) and binding energies (Eb) for these compounds are listed in Table . Apparently, all the calculated ΔG values are negative (−22.81 to −35.50 kcal/mol), indicating that the combination of Al12Be with bases/BPs is a spontaneous process. Besides, the calculated Eb values are as large as 35.32–50.76 kcal/mol, demonstrating the strong interaction between Al12Be and DNA bases/BPs in these obtained compounds. Also, the Eb values of these compounds decrease in the order GC-1 (50.76 kcal/mol) > G-1 (48.99 kcal/mol) > C-1 (40.70 kcal/mol) > A-1 (39.82 kcal/mol) > AT-1 (37.62 kcal/mol) > T-1 (35.32 kcal/mol), which is just opposite to the increasing trend of GC-1 (−35.50 kcal/mol) < G-1 (−34.96 kcal/mol) < C-1 (−27.25 kcal/mol) < A-1 (−26.71 kcal/mol) < AT-1 (−23.74 kcal/mol) < T-1 (−22.81 kcal/mol) for ΔG, reflecting that these structures with bidentate interaction are more stable than those containing only one linkage Al–O/N bond. In particular, it is noted that the GC-1 and G-1 have much larger Eb and more negative ΔG values than the rest, indicating that this Al12Be cluster is most likely to bind with guanine when it is exposed to DNA sequences. Besides, it is found that the complexes containing one Al–N bond possess higher binding energies than those containing one Al–O bond, indicating that the strength of Al–N bonds is stronger that of Al–O bonds in these compounds.

Table 1

Gibbs Free Energy Changes (ΔG, kcal/mol), Binding Energies (Eb, in kcal/mol), WBI, NPA Charges on Al and O/N Atoms (q, in |e|) Involved in the Linkage Bonds, and Total NPA Charges on Al12Be (Q, in |e|) for the Most Stable Al12Be–X Species

species	binding atom	ΔG	Eb	WBI	qAl	qN/O	Q
G-1	N7	–34.96	48.99	0.309	0.349	–0.646	–0.348
	O6			0.272	0.528	–0.792
C-1	N3	–27.25	40.70	0.340	0.459	–0.812	–0.248
A-1	N3	–26.71	39.82	0.335	0.402	–0.750	–0.214
T-1	O4	–22.81	35.32	0.334	0.589	–0.829	–0.167
GC-1	N7	–35.50	50.76	0.308	0.363	–0.650	–0.357
	O6			0.266	0.522	–0.835
AT-1	O4	–23.74	37.62	0.303	0.668	–0.880	–0.247

Nature of the Interaction between Al12Be and Bases/BPs

Wiberg Bond Index and Nature Population Analysis

As is mentioned above, the interaction betweenpan> pan> class="Chemical">Al12Be and base/BPs in these Al12Be–X compounds is quite strong. This can be further proved by the large Wiberg bond index (WBI) values of 0.266–0.340 for the linkage bonds in the most stable Al12Be–X compounds, as shown in Table . Furthermore, these WBI values were also decomposed to quantify the contribution of atomic orbitals of Al and N/O atoms to the Al–N and Al–O bonds (see Table S3). As shown in this table, all the WBI values are mainly derived from the overlapping between the 3p atomic orbital of Al and sp3 hybrid orbitals (i.e., 2s and 2p orbitals) of N/O atoms. For example, the WBI of 0.340 for the C-1 complex is mainly derived from the overlaps of 3p (Al)–2s (N) and 3p (Al)–2p (N), which donate 0.104 and 0.116, respectively. This overlap between atomic orbitals of Al and N/O atoms validates the covalent characteristics of these Al–N and Al–O bonds.

In addition, from Table , it is pan> class="Chemical">found that Al atoms carry opposite NPA charges (0.349 |e| to 0.668 |e|) to those of −0.646 |e| to −0.880 |e| on the N/O atoms involved in the linkage Al–N/O bonds, suggesting that these newly formed bonds between Al12Be and DNA bases/BPs also have ionic characteristics. This can be further confirmed by the electron density difference (EDD) isosurfaces of Al12Be–X, which illustrate that N/O atoms always gain electrons, while Al atoms donate electrons in the Al–N/Al–O bonds (see Figure S8). Even so, from an overall perspective, there is obvious charge transfer from DNA bases/BPs to Al12Be because the total NPA charges on Al12Be are in the range of −0.167 to −0.357 |e|, reflecting that Al12Be obtain electrons from DNA bases. More interesting is that the changing order of Eb values of Al12Be–X accord well with the increasing trend of charge transfer between Al12Be and bases/BPs, namely, the larger the Q is, the larger the Eb value the complex has (see Figure S9). Hence, the charge transfer plays an important role in the strong interaction between Al12Be and bases/BPs.

AIM Theory

Electron-density topologicn class="Chemical">al anpan>alysis offers another quantitative measure of bonding that differs from the abovementioned orbital-based NPA methods. The AIM theory[39−41] proposed by Bader is such a valuable tool to analyze the nature and strength of bonding interactions based on topological parameters at the BCPs. Herein, several important intramolecular atomic descriptors, including Laplacian (∇2ρr), electronic energy density (Hr), kinetic energy density (Gr), and potential energy density (Vr) at the BCPs of the linkage bonds between Al12Be and bases/BPs are listed in Table .

Table 2

Topological Parameters (in a.u.) at the BCPs of Al–N/O Bonds and Corresponding Laplacian Bond Order (LBO, in a.u.) for Al12Be–X Obtained by Using the AIM Theory at the M06-2X/6-311++G(d,p) Level

species	binding atom	∇2ρr	Hr	Gr	Vr	LBO
G-1	N7	0.335	–0.004	0.087	–0.091	0.174
	O6	0.462	0.005	0.111	–0.106	0.117
C-1	N3	0.363	–0.006	0.097	–0.102	0.176
A-1	N3	0.394	–0.006	0.104	–0.110	0.208
T-1	O4	0.565	0.005	0.136	–0.131	0.144
GC-1	N7	0.335	–0.004	0.087	–0.091	0.170
	O6	0.462	0.005	0.111	–0.106	0.118
AT-1	O4	0.559	0.005	0.134	–0.129	0.130

It is known that ∇2ρr is related to the bond interaction energy by a local expression of the virial theorem as follows[45]

The sign on class="Chemical">f ∇2ρr n class="Chemical">at a BCP determines whether the negative Vr or the positive Gr is in excess of the virial ratio [2Gr/(−Vr)]. The excess Vr at a BCP leads to a negative ∇2ρr, which indicates a sharing of electronic charges between both nuclei that defines the covalent interactions. In contrast, the dominating Gr results in a positive ∇2ρr at a BCP, which shows depletion of the electronic charge along the bond path. This is observed for interactions between closed-shell systems, such as ionic interactions. From Table , it is observed that all the ∇2ρr values at the BCPs of Al–N and Al–O bonds in these Al12Be–X are positive, implying that these linkage bonds possess ionic characteristics.

However, only takin class="Chemical">ng ∇2ρr into acn class="Chemical">count is sometimes not enough to determine the nature of the bond of interest. Thus, Cremer and Kraka[46] have defined the electronic energy density Hr at BCP as follows

Bonds with covalent character always have a BCP with negative Hr. This is to say, if ∇2ρr > 0 and Hr < 0, the bond should be partially covalent in nature.[47−49] Accordingly, the formed Al–N bonds also possess partially covalent character because of their negative Hr values. As for the formed Al–O bonds, they possess more ionic characteristics because they have positive Hr values of 0.005. However, these Hr values are too small to strictly separate ionic interaction from the covalent one. Thus, the contour maps of ∇2ρr that cover the whole bonding regions of Al–O/N bonds for Al12Be–X are also plotted in Figure S10. It is observed that the aggregation of electron density happens in the regions near to the electronegative N/O atoms, whereas Al atoms always donate electrons. This clearly verifies the ionic characteristics of these Al–O/N linkage bonds.

On the other hand, from pan> class="Chemical">Figure S10, it can be seen that the change trends of ∇2ρr along the bond paths of Al–O/N bonds are very similar to those of well-known covalent compound AlCl3 and coordination compound Ni(CO)4. This implies that these Al–O/N bonds are also partially covalent in nature. To confirm this, the Laplacian bond order (LBO),[50] defined as a scaled integral of negative parts of ∇2ρr in fuzzy overlap space, is computed and listed in Table . LBO characterizes the ∇2ρr in the whole bonding region and thus can avoid erroneous conclusions from the irrational position of BCP. As shown in Table , the LBO values of these Al–O/N bonds range from 0.117 to 0.208, which are larger than those of Cl–F bonds in ClF3.[50] This clearly demonstrates the covalent characteristics of these bonds, which is in accord with the conclusion from the abovementioned WBI analysis. Moreover, it is noted that Al–N bonds always exhibit larger LBO values than Al–O bonds, suggesting that the former possess more covalent component than the latter. It should be mentioned that all the LBO values are much smaller than corresponding WBI values because LBO is a definition of covalent bond order rather than total bond order, and thus, noncovalent interactions have no contribution to LBO.[50]

Localized Molecular Orbital and Electron Localization Function

To visualize the bond npan> class="Chemical">ature of the Al–N and Al–O bonds in these Al12Be–X compounds, the localized molecular orbitals (LMO) are obtained by using the Boys–Foster method[51] and are shown in Figure . The obvious localized electron density between aluminum and N/O atoms in these Al12Be–X compounds suggests the substantial covalent interaction between Al12Be and bases/BPs. Additionally, the contribution analysis reveals that the shared electron pair is mostly derived from the nitrogen or oxygen atoms of the DNA, while the Al atoms of Al12Be contribute less in these Al–N/Al–O bonds. This implies the high polarity of these linkage bonds between Al12Be and bases/BPs. Furthermore, it is noted that the differences in contribution to the shared localized electrons from O and Al atoms (61.6–70.2%) are much larger than those from N and Al atoms (50.3–53.3%), which demonstrates that the Al–O bonds possess larger polarity than Al–N bonds. In other words, the Al–O bonds have more ionic characteristics than Al–N bonds, which is in good agreement with the above topological AIM theory analysis.

Figure 2

LMO related to the Al–O and Al–N bonds in Al12Be–X. The contributions of atoms to LMO are also listed next to them.

LMO reln class="Chemical">ated to the n class="Chemical">Al–O and Al–N bonds in Al12Be–X. The contributions of atoms to LMO are also listed next to them.

n class="Chemical">Also, this can pan> class="Chemical">be further confirmed by the electron localization functions (ELF)[52] of Al12Be–X, in which the orange-red color regions between Al and N atoms are darker than those between Al and O atoms (see Figure ). Thus, the localized electron density between Al and N atoms is larger than that between Al and O atoms, verifying the more obvious covalent characteristics of Al–N bonds than those of Al–O bonds. Besides, the localized electrons between Al and O/N atoms are closer to O/N atoms, suggesting the large polarity of these covalent Al–N and Al–O bonds. Therefore, based on the abovementioned bonding nature analysis, it can be concluded that the formed Al–N and Al–O bonds can be regarded as strong polar covalent bonds, and the Al–N bonds have more covalent components than the Al–O bonds.

Figure 3

ELF pictures of Al12Be–X.

ELn class="Chemical">F pictures on class="Chemical">f n class="Chemical">Al12Be–X.

Potential Applications in O2 Activation and CO Oxidation

It is n class="Chemical">also meanipan> class="Chemical">ngful to investigate the effect of DNA bases on the electronic properties of the Al12Be superatom. As shown in Table S4, the combination of bases with Al12Be enlarged the highest occupied molecular orbital (HOMO)–lowest unoccupied molecular orbital (LUMO) gap of 2.73 eV Al12Be to those of 3.04–3.09 eV for C-1, A-1, and T-1 compounds. Besides, the charge transfer from the DNA bases to Al12Be results in the lower vertical ionization energies (VIE) and vertical electron affinities (VEA) of these Al12Be–X compounds than those of Al12Be. In particular, the VIE of G-1 is as small as 5.12 eV, which is even lower than the IE of 5.14 eV for Na, implying the high reducibility of this complex. Hence, the G-1 complex is highly expected to be capable of activating oxygen molecules and even catalyzing CO oxidation. Hence, the potential of G-1 in activating O2 and catalyzing CO oxidation has been explored in the following subsections.

O2 Activation

To find the pan> class="Chemical">combining site of O2 on the surface of G-1, the electrostatic potential (ESP) of G-1 has been obtained and is shown in Figure S11. By combining O2 with the points with a relatively large negative potential on Al12Be, several initial structures of the G-1–O2 complex have been constructed. Figure shows the most stable structure of G-1–O2 after optimization. Note that the O2 molecule was adsorbed on G-1 by forming three Al–O bonds with a large adsorption energy of −5.03 eV, and meanwhile, the Al–N bond between Al12Be and guanine in isolated G-1 was split. Moreover, the length of the O–O bond is elongated from 1.189 to 1.459 Å in this G-1–O2 complex. The obviously prolonged O–O bond indicates that O2 is activated on G-1. Numerous studies reported that the activation of oxygen is strongly dependent on the electron transfer from the substrate to the 2π* antibonding orbital of O2,[53−55] leading to the elongated O–O bond as that in the peroxo or superoxo state. The same case is also found in this study. As shown in Table , the NPA charges on O2 indicate that the O2 molecule obtains 1.41 electrons from G-1 in the resulting G-1–O2 complex, which is consistent with the charge transfer phenomenon as described in previous studies.[53−55] For example, Zhang et al.[54] found that O2 can obtain about 1.3 electrons from the MgO-supported Au20 cluster, resulting in a peroxo state of O2 with an O–O bond length of 1.52 Å.

Figure 4

Most stable configuration of G-1-O2 with the corresponding bond lengths and absorption energies Ead (Ead = Egas-complex – Egas – Ecomplex).

Table 3

Atomic Spin Population on Oxygen Atoms, Bond Lengths (dO–O, in Å) and WBI Values of O–O Bonds, and the NPA Charges on O2 (Q′, in |e|) for G-1–O2, and Several Isolated Oxygen Species

species	α population	β population	dO–O	WBI	Q′
3O2	S1.96p2.53d0.01	S1.93p1.55d0.02	1.189	1.516	0.00
1O2	S1.95p2.04d0.02	S1.95p2.04d0.02	1.187	2.031	0.00
2O2–	S1.97p2.52d0.01	S1.96p2.03d0.01	1.322	0.510	–1.00
1O22–	S1.99p2.51d0.00	S1.99p2.51d0.00	1.522	1.082	–2.00
G-1–O2	Oa: S1.69p2.23d0.01	S1.69p2.23d0.01	1.459	0.988	–1.41
	Ob: S1.91p2.21d0.01	S1.91p2.21d0.01

Most stable n class="Chemical">conpan> class="Chemical">figuration of G-1-O2 with the corresponding bond lengths and absorption energies Ead (Ead = Egas-complex – Egas – Ecomplex).

To deeply reveal the activation degree of O2 in G-1–O2, the atomic spin population on oxygen atoms, bond lengths, and WBI values of O–O bonds, and charges on the O2 units of G-1–O2 and several oxygen species, including triplet 3O2 and some reactive oxygen species (singlet oxygen 1O2, superoxide anion 2O2–, and peroxide dianion 1O22–)[56] are also shown in Table . It is observed that the O2 subunit in G-1–O2 is similar to the peroxide dianion 1O22–, considering they have a similar atomic spin population on oxygen atoms, bond lengths, and WBI values of O–O bonds. For example, a bond length of 1.459 Å and WBI of 0.988 for the O–O bond in G-1–O2 are nearly equal to those of 1.522 Å and 1.082 for 1O22–, respectively. It is reported that the 1O22– and peroxides are able to oxidize CO into CO2,[57−60] and therefore, the adsorbed O2 on G-1 is highly expected to promote the CO oxidation reaction.

CO Oxidation on G-1

It is well known that pan> class="Chemical">CO oxidation is not only a very important reaction dealing with the poisonous CO constituent of the automobile exhaust systems but also serves as a prototype reaction for the heterogeneous catalysis community. There are two kinds of mechanisms for CO oxidation, that is, Langmuir–Hinshelwood (LH) and Eley–Rideal (ER) mechanisms. In the first one, both CO and O2 molecules are first co-adsorbed on the surface and then form a −OCOO– transition state (TS), while the CO molecule tends to insert into the O–O bond of the preadsorbed O2 molecule to form a carbonate-type TS in the second one.[61] Thus, these two mechanisms for CO oxidation on G-1 have been considered and are shown in Figure , while corresponding structures and critical geometric parameters of initial states (IS), intermediate products (IM), TS, and final states (FS) are given in Figures S12 and S13.

Figure 5

Reaction pathway profiles of CO oxidation on G-1. All energies are given respect to the reference energy of reactants, that is, the sum of electronic energies of G-1, two CO, and one O2 molecules.

Reaction pn class="Chemical">athway pron class="Chemical">files of CO oxidation on G-1. All energies are given respect to the reference energy of reactants, that is, the sum of electronic energies of G-1, two CO, and one O2 molecules.

As shown in Figure S12, the LH mechanpan>ism begins with the coadsorption of CO and O2 on G-1 as shown in IS1, in which the distance between two O atoms in O2 molecule is 1.317 Å. Then, the adsorbed CO slowly approaches O2 until an −OOCO– intermediate in IM1 is formed, in which the O–O bond length is elongated to 1.450 Å. Afterward, IM1 goes through the TS1 with an energy barrier of 20.39 kcal/mol to release a CO2 molecule and leave a single activated oxygen atom on G-1 in FS1. This energy barrier is a bit smaller than that of 24.1 kcal/mol for gold-cluster Au3,[62] demonstrating the better performance of G-1 in CO oxidation than Au3. Furthermore, the second step of CO + O* → CO2 has been also considered to reveal whether a second CO molecule could be oxidized by the O* remained on G-1 after the first step. The barrier energy of this step is computed to be 17.54 kcal/mol, which is a bit smaller than that of the first step.

Unlike the LH mechanism, as shown in n class="Chemical">Figure S13, the oxidation of CO via the ER mechanism starts from the chemical adsorption of the O2 molecule on G-1, in which the O2 molecule is first activated into 1O22– as mentioned above. Then, a free CO molecule approaches the activated O2 and then inserts to the O–O bond, resulting in a carbonate-type CO3 intermediate in IM1. Then, IM1 needs to undergo the first TS (TS1) with a small energy barrier of 7.78 kcal/mol. It is noted that this energy barrier is much lower than that of 20.39 kcal/mol for the LH mechanism, suggesting the preference of the ER pathway over the LH mechanism for the CO oxidation catalyzed by G-1. In particular, it is observed that this energy barrier is greatly smaller than those of 36.20–38.74 kcal/mol for the Au3–X (X = DNA bases) complex,[31] indicating that the guanine-modified G-1 compound has better performance than these DNA base-decorated gold clusters in CO oxidation. As for the second step, it only needs to overcome a smaller energy barrier of 5.02 kcal/mol, which further confirms that the first step is the rate-limiting step for the CO oxidation on G-1.

Conclusions

The interaction n class="Chemical">betweenpan> supern class="Chemical">atom Al12Be and DNA nucleobases/BPs has been investigated by exploring the most stable structures of Al12Be–X (X = DNA nucleobases and BPs) complexes. The calculated Eb and ΔG values of the resulting compounds demonstrate that Al12Be tends to combine with guanine by forming two Al–O and Al–N bonds via the bidentate mode among the four DNA nucleobases (G, C, A, and T). The WBI, NPA charges, AIM theory, LMO, and ELF analyses reveal that the formed Al–N and Al–O bonds can be regarded as strong polar covalent bonds, and the Al–N bonds have more covalent components than the Al–O bonds. In particular, it is interesting to find that the electron transfer from the DNA bases to Al12Be results in the lower VIE and VEA values of Al12Be–X compounds as compared with the isolated Al12Be. Therefore, by taking G-1 as an example, the potential applications of these Al12Be–X complexes in oxygen activation and CO oxidation have been verified in the present work. It is highly hoped that this work could pave the way to extend the application of superatoms in nanobiotechnology or chemical catalysis.

Computational Methods

The geometric structures on class="Chemical">f n class="Chemical">Al12Be–X complexes have been obtained by two methods. In the first one, a large number of initial geometries were constructed artificially by combining the Al3 face, Al–Al edge, and Al-point of Al12Be with the most possible sites of DNA bases and BPs, such as N1, N3, and N7 atoms of adenine, O2 and O4 of thymine, N3, N7, and O6 of guanine, and O2 and N3 of cytosine because these sites with lone pairs can easily contribute electrons to the LUMO of Al12Be. The second one is to use the stochastic search procedure, that is, the genmer tool of the Molclus program,[63] in which the initial configurations of Al12Be and base/BPs are respectively frozen and then combined in a random way to further confirm that all the global minima are found in the first procedure. Subsequently, the obtained initial configurations of Al12Be–X complexes as well as all geometries of IS, IM, TS, and FS involved in O2 activation and CO oxidation were optimized by using the Minnesota density functional (M06-2X)[64,65] with the all-electron 6-311+G(d) basis set.[66] The analytical frequency calculations were computed to identify the nature of stationary points (minima and TSs possess zero and one imaginary frequency, respectively) at the same computational level. The M06-2X developed by Zhao and Truhlar has been found to be reliable for studying the weak interactions and barrier heights of noncovalently interacting systems and catalytic reactions along with Pople’s basis sets.[67−70] Note that the polarizable continuum model[72] was employed to take the effect of solvent (water) into account during studying the interaction between Al12Be and DNA bases/BPs but was not used in the calculation about the application of G-1 in O2 activation and CO oxidation.

The sin class="Chemical">ngle-point enpan>ergies were cpan> class="Chemical">alculated at the M06-2X/6-311++G(d,p) level to obtain the binding energies Eb and Gibbs free energy differences (ΔG) of these Al12Be–X complexes defined as followswhere E(Al12Be)/G(Al12Be), E(X)/G(X), and E(Al12Be–X)/G(Al12Be–X) are the zero-point energy-corrected electronic energies/thermal free energies of Al12Be, nucleobases/BP, and Al12Be–X, respectively. Herein, the Gibbs free energies are the sum of the calculated single-point energies and the thermal corrections to Gibbs free energy at a temperature of 298 K and pressure of 1 atm obtained from frequency calculations.

To deeply understand the nn class="Chemical">ature opan> class="Chemical">f bonds formed between Al12Be and bases/BPs, the AIM theory,[39−41] LMO,[51] and ELF[52] analyses were performed in this work. To clearly analyze the charge transfer between Al12Be and nucleobases/BPs in Al12Be–X, the fragment EDD for each compound was determined by eq .where ρ(Al12Be–X) is the electron density of Al12Be–X, while ρ(Al12Be) and ρ(X) are the unperturbed electron densities of Al12Be and nucleobases/BP, respectively.

When studying the catalytic performance of G-1, the intrinsic reaction coordinate[71] calculations were performed at the M06-2X/6-311+G(d) level to confirm that a given TS connects a particular pair of consecutive minima. When the reaction energy profiles are constructed, the total energies of the initial reactive systems are set as zero, and the used electronic energies are also obtained at the same level.

All the abovementionpan>ed cpan> class="Chemical">alculations in this work were performed using the Gaussian 16 software.[72] The WBI and natural population analysis[73] were conducted using the Natural Bond Orbital 3.1 program implemented in the Gaussian 16 package. The output files produced at the M06-2X/6-311++G(d, p) level by Gaussian16 were used as inputs into Multiwfn 3.3.7 software[74] to carry out the AIM, ELF, ESP, LMO, and WBI decomposition analyses. Dimensional plots of molecular configurations and orbitals were generated with the GaussView program.[75]