Deep reinforcement learning in chemistry: A review.

Reinforcement learning (RL) has been applied to various domains in computational chemistry and has found wide-spread success. In this review, we first motivate the application of RL to chemistry and list some broad application domains, for example, molecule generation, geometry optimization, and retrosynthetic pathway search. We set up some of the formalism associated with reinforcement learning that should help the reader translate their chemistry problems into a form where RL can be used to solve them. We then discuss the solution formulations and algorithms proposed in recent literature for these problems, the advantages of one over the other, together with the necessary details of the RL algorithms they employ. This article should help the reader understand the state of RL applications in chemistry, learn about some relevant actively-researched open problems, gain insight into how RL can be used to approach them and hopefully inspire innovative RL applications in Chemistry.

Induced chirality at the surface: fixation of a dynamic M/P invertible helical Co3 complex on SiO2.

Dynamic M/P invertible helicity was successfully induced at a SiO2 surface immobilized with a dynamic helical trinuclear cobalt complex, [LCo3(NHMe2)6](OTf)3, using chiral ((R) or (S))-1-phenylethylamine. Solid-state CD spectra and theoretical calculations suggested that the fixation of the M/P helical complex on the surface via coordination interactions was the key factor of the induced chirality at the surface.

Investigation of the Suzuki-Miyaura cross-coupling reaction on a palladium H-beta zeolite with DFT calculations.

Metal or metal cluster-doped zeolites catalyse a wide variety of reactions. In this work, a coupling reaction between bromobenzene and phenylboronic acid to yield biphenyl with the Pd-H-Beta zeolite catalyst was investigated with density functional theory (DFT) calculations. Utilizing a model system with tetrahedral Pd4 clusters within the H-Beta zeolite, it was demonstrated that the catalyst exhibited notable reactivity by effectively reducing the activation energy barrier for the reaction. Our investigation revealed that the zeolite framework facilitated electron transfer to the Pd cluster, thereby increasing the reaction activity. The coupling reaction was shown to be exothermic and comprise three main steps: oxidative addition of bromobenzene (C6H5Br), transmetallation with phenylboronic acid (C6H5B(OH)2), and reductive elimination of biphenyl (C12H10). Specifically, in the transmetallation step, which was the rate-determining step, the C-B bond breaking in phenylboronic acid (C6H5B(OH)2) and the phenylboronate anion (C6H5B(OH)3-) were compared under neutral and basic conditions, respectively. This comprehensive study clarifies the mechanism for the reaction with the modified Pd zeolite catalyst and highlights the essential role of the zeolite framework.

Selective synthesis of tightly- and loosely-twisted metallomacrocycle isomers towards precise control of helicity inversion motion.

Molecular twist is a characteristic component of molecular machines. Selectively synthesising isomers with different modes of twisting and controlling their motion such as helicity inversion is an essential challenge for achieving more advanced molecular systems. Here we report a strategy to control the inversion kinetics: the kinetically selective synthesis of tightly- and loosely-twisted isomers of a trinuclear PdII-macrocycle and their markedly different molecular behaviours. The loosely-twisted isomers smoothly invert between (P)- and (M)-helicity at a rate of 3.31 s-1, while the helicity inversion of the tightly-twisted isomers is undetectable but rather relaxes to the loosely-twisted isomers. This critical difference between these two isomers is explained by the presence or absence of an absolute configuration inversion of the nitrogen atoms of the macrocyclic amine ligand. Strategies to control the helicity inversion and structural loosening motions by the mode of twisting offer future possibilities for the design of molecular machines.

Element effects in endohedral metal-metal-bonding fullerenes M2@C82 (M = Sc, Y, La, Lu).

Endohedral metal-metal-bonding fullerenes have recently emerged, in which encapsulated metals form a metal-metal bond. However, the physical reasons why some metal elements prefer to form metal-metal bonds inside fullerene are still unclear. Herein, we reported first-principles calculations on electronic structures, bonding properties, dynamics, and thermodynamic stabilities of endohedral metallofullerenes M2@C82 (M = Sc, Y, La, Lu). Multiple bonding analysis approaches unambiguously reveal the existence of one two-center two-electron σ covalent metal-metal bond in M2@C82 (M = Sc, Y, Lu); however, the La-La bonding interaction in La2@C82 is weaker and could not be categorized as one metal-metal covalent bond. The energy decomposition analysis on bonding interactions between an encapsulated metal dimer and fullerene cages suggested that there exist two electron-sharing bonds between a metal dimer and fullerene cages. The reasons why La2 prefers to donate electrons to fullerene cages rather than form a standard σ covalent metal-metal bond are mainly attributed to two following facts: La2 has a lower ionization potential, while the hybridization of ns, (n - 1)d, and np atomic orbitals in La2 is higher. Ab initio molecular dynamic simulations reveal that the M-M bond length at room temperature follows the trend of Sc < Lu < Y. The statistical thermodynamics calculations at different temperatures reveal that the experimentally observed endohedral metal-metal-bonding fullerenes M2@C82 have high concentrations in the endohedral fullerene formation temperature range.

Recent progress in controlling the photoluminescence properties of single-walled carbon nanotubes by oxidation and alkylation.

The functionalization of single-walled carbon nanotubes (SWCNTs) has received considerable attention in the last decade since highly efficient near-infrared photoluminescence (PL) has been observed to be red-shifted compared with the intrinsic PL peak of pristine SWCNTs. The PL wavelength has been manipulated using arylation reactions with aryldiazonium salts and aryl halides. Additionally, simple oxidation and alkylation reactions have proven effective in extensively adjusting the PL wavelength, with the resulting PL efficiency varying based on the chosen reaction techniques and molecular structures. This review discusses the latest developments in tailoring the PL attributes of SWCNTs by oxidation and alkylation processes. (6,5) SWCNTs exhibit intrinsic emission at 980 nm, and the PL wavelength can be controlled in the range of 1100-1320 nm by chemical modification. In addition, recent developments in chiral separation techniques have increased our understanding of the control of the PL wavelength, extending to the selection of excitation and emission wavelengths, by chemical modification of SWCNTs with different chiral indices.

Acceleration and deceleration of chirality inversion speeds in a dynamic helical metallocryptand by alkali metal ion binding.

We report that the chirality inversion kinetics of a trinickel(II) cryptand can be controlled by guest recognition in the cryptand cavity. When the guest was absent, the nickel(II) cryptand underwent a dynamic interconversion between the P and M forms in solution, preferring the M form, with a half-life of t1/2 = 4.99 min. The P/M equilibrium is reversed to P-favored by binding with an alkali metal ion in the cryptand cavity. The timescale of this M→P inversion kinetics was both notably accelerated and decelerated by the guest binding (t1/2 = 0.182 min for K+ complex; 186 min for Cs+ complex); thus, the equilibration rate constants differed by up to 1000-fold depending on the guest metal ions. This acceleration/deceleration can be explained in terms of the virtual binding constants at the transition state of the P/M chirality inversion; K+ binding more stabilizes the transition state rather than the P and M forms to result in the acceleration.

Mechanistic Studies of Regiocontrolled Bisaddition of Fullerenes Driven by Oriented External Electric Fields.

The challenge of achieving regioselective multifunctionalization on highly symmetric C60 and C70 fullerenes persists as a significant hurdle. In this study, we present a novel approach involving the participation of an oriented external electric field (OEEF) to facilitate the regioselective formation of bisadducts in C60/C70 fullerenes. These products are obtained through consecutive Diels-Alder cycloaddition reactions. We constructed the field strength-barrier relationship and elucidated the OEEF-driven modulation mechanisms quantitatively. Leveraging the interplay between molecular dipoles and electric fields, the diverse reactions at distinct sites exhibit varying degrees of sensitivity to the applied electric fields, thereby leading to a pronounced regioselectivity in the bisaddition process. Our proposition suggests that the angle formed between the bonding direction (referred to as the reaction axis) and the external field can conveniently function as a predictive descriptor for the reactivity of different sites on the fullerene surface when subjected to electric fields.

Clarifying the Electronic Structure of Anion-Templated Silver Nanoclusters by Optical Absorption Spectroscopy and Theoretical Calculation.

Electronic structures of anion-templated silver nanoclusters (Ag NCs) are not well understood compared to conventional, template-free Ag NCs. In this study, we synthesized three new anion-templated Ag NCs, namely [S@Ag17(S-4CBM)15(PPh3)5]0, [S@Ag18(S-4CBM)16(PPh3)8]0, and [Cl@Ag18(S-4CBM)16(PPh3)8][PPh4], where S-4CBM = 4-chlorobenzene methanethiolate, and single-crystal X-ray crystallography revealed that they have S@Ag6, S@Ag10, and Cl@Ag10 cores, respectively. Investigation of their electronic structures by optical spectroscopy and theoretical calculations elucidated the following unique features: (1) their electronic structures are different from those of template-free Ag NCs described by the superatomic concept; (2) optical absorption in the range of 550-400 nm for S2--templated Ag NCs is attributed to the charge transitions from S2--templated Ag-cage orbitals to the s-shaped orbital in the S2- moiety; (3) the Cl--templated Ag NCs can be viewed as [Cl@Ag18(S-4CBM)16(PPh3)8]0[PPh4]0 rather than the ion pair [Cl@Ag18(S-4CBM)16(PPh3)8]-[PPh4]+; and (4) singlet-coupled singly occupied orbitals are involved in the optical absorption of the Cl--templated Ag NC.

Ag6 Cu8 (C=CAr)14 (DPPB)2 : A Rigid Ligand Co-Protected Bimetallic Silver(I)-Copper(I) Cluster with Room-Temperature Luminescence.

Metal clusters have become increasingly important in various applications, with ligands playing a crucial role in their construction. In this study, we synthesized a bimetallic cluster, Ag6 Cu8 (C=CAr)14 (DPPB)2 (Ag6 Cu8 ), using a rigid acetylene ligand, 3,5-bis(trifluoromethyl)phenylacetylide. Through single-crystal structure characterization, we discovered that the butterfly-shaped Ag2 Cu2 motifs were subject to distortion due to steric hindrance imposed by the rigid ligand. These motifs assembled together through shared vertices and edges. Mass spectrometry analysis revealed that the primary fragments detected during electrospray ionization (ESI) testing corresponded to the Ag2 Cu2 motifs. Furthermore, we conducted a comprehensive investigation of the cluster's solution properties employing 31 P NMR, UV-vis absorption, and photoluminescent measurements. In contrast to previously reported Ag/Cu bimetallic clusters protected by flexible ligands, Ag6 Cu8 protected by rigid ligands exhibited intriguing room temperature fluorescence properties alongside excellent thermal stability. DFT calculations on Ag6 Cu8 and Ag6 Cu8 with the rigid aromatic ring removed revealed that the presence of the rigid aromatic ring can lower the electronic energy levels of the cluster, and reduce the energy gap from 4.05 eV to 3.45 eV. Moreover, the rigid ligand further suppressed the non-radiative transition process, leading to room temperature fluorescence emission.

Control of functionalized single-walled carbon nanotube photoluminescence via competition between thermal rearrangement and elimination.

We conducted the chiral separation of functionalized single-walled carbon nanotubes (SWNTs) with dibromopropane derivatives. Depending on their chirality and diameter, the thermal treatment of functionalized SWNTs leads to a shift in the emission radiation to longer wavelengths owing to rearrangement reaction in competition with elimination reaction.

Selective emergence of photoluminescence at telecommunication wavelengths from cyclic perfluoroalkylated carbon nanotubes.

Chemical functionalisation of semiconducting single-walled carbon nanotubes (SWNTs) can tune their local band gaps to induce near-infrared (NIR) photoluminescence (PL). However, tuning the PL to telecommunication wavelengths (>1300 nm) remains challenging. The selective emergence of NIR PL at the longest emission wavelength of 1320 nm was successfully achieved in (6,5) SWNTs via cyclic perfluoroalkylation. Chiral separation of the functionalised SWNTs showed that this functionalisation was also effective in SWNTs with five different chiral angles. The local band gap modulation mechanism was also studied using density functional theory calculations, which suggested the effects of the addenda and addition positions on the emergence of the longest-wavelength PL. These findings increase our understanding of the functionalised SWNT structure and methods for controlling the local band gap, which will contribute to the development and application of NIR light-emitting materials with widely extended emission and excitation wavelengths.

Photoluminescence control by atomically precise surface metallization of C-centered hexagold(i) clusters using N-heterocyclic carbenes.

The properties of metal clusters are highly dependent on their molecular surface structure. The aim of this study is to precisely metallize and rationally control the photoluminescence properties of a carbon(C)-centered hexagold(i) cluster (CAuI6) using N-heterocyclic carbene (NHC) ligands with one pyridyl, or one or two picolyl pendants and a specific number of silver(i) ions at the cluster surface. The results suggest that the photoluminescence of the clusters depends highly on both the rigidity and coverage of the surface structure. In other words, the loss of structural rigidity significantly reduces the quantum yield (QY). The QY in CH2Cl2 is 0.04 for [(C)(AuI-BIPc)6AgI3(CH3CN)3](BF4)5 (BIPc = N-isopropyl-N'-2-picolylbenzimidazolylidene), a significant decrease from 0.86 for [(C)(AuI-BIPy)6AgI2](BF4)4 (BIPy = N-isopropyl-N'-2-pyridylbenzimidazolylidene). This is due to the lower structural rigidity of the ligand BIPc because it contains a methylene linker. Increasing the number of capping AgI ions, i.e., the coverage of the surface structure, increases the phosphorescence efficiency. The QY for [(C)(AuI-BIPc2)6AgI4(CH3CN)2](BF4)6 (BIPc2 = N,N'-di(2-pyridyl)benzimidazolylidene) recovers to 0.40, 10-times that of the cluster with BIPc. Further theoretical calculations confirm the roles of AgI and NHC in the electronic structures. This study reveals the atomic-level surface structure-property relationships of heterometallic clusters.

Theoretical insights into the support effect on the NO activation over platinum-group metal catalysts.

We systematically explored NO activation at metal/oxide interfaces by the combination of Sr3Ti2O7, Sr3Fe2O7, CeO2, anatase-TiO2, ZrO2, and γ-Al2O3 supports and the platinum-group metal cluster (Pd4, Pt4, and Rh4) using slab-model density functional theory calculations. These metal clusters can be strongly adsorbed at these metal oxide surfaces. The Pt4 and Rh4 clusters show larger adsorption energies than the Pd4 cluster, yet the γ-Al2O3(100) surface shows smaller adsorption energies than other metal oxide surfaces. One oxygen vacancy close to the metal cluster was constructed to evaluate the NO activation at those metal/oxide interfaces. The O atom of NO refills the oxygen vacancy after NO dissociation, while the N adatom is left on the metal cluster. The exothermic process was identified for the NO activation except for the Sr3Fe2O7 case, indicating the significant role of the interplay between the metal cluster and oxygen vacancy.

Photoluminescence Properties of Single-Walled Carbon Nanotubes Influenced by the Tether Length of Reagents with Two Reactive Sites.

The functionalization of single-walled carbon nanotubes (SWNTs) is an effective method for controlling a local band gap, resulting in photoluminescence (PL) in the near-infrared region. Herein, SWNTs were functionalized using a series of bromoalkanes and dibromoalkanes to evaluate the effects of their length on the nanotube PL properties. When bromoalkanes (Cn H2n+1 Br) or dibromoalkanes (Cn H2n Br2 ) with tether lengths of six or more were utilized for six different semiconducting SWNTs, the obtained SWNT adducts exhibited two new PL peaks, whereas dibromoalkanes with tether lengths of 3-5 (Cn H2n Br2 : n=3-5) produced single peaks. Combined with theoretical calculations, the results suggested that the tether length of reagents changes the formation mechanism of functionalized adducts, that is, Cn H2n Br2 (n=3-5) tends to result in kinetic products.

Synergistic Surface Modification of Tin-Lead Perovskite Solar Cells.

Interfaces in thin-film photovoltaics play a pivotal role in determining device efficiency and longevity. In this work, the top surface treatment of mixed tin-lead (≈1.26 eV) halide perovskite films for p-i-n solar cells is studied. Charge extraction is promoted by treating the perovskite surface with piperazine. This compound reacts with the organic cations at the perovskite surface, modifying the surface structure and tuning the interfacial energy level alignment. In addition, the combined treatment with C60 pyrrolidine tris-acid (CPTA) reduces hysteresis and leads to efficiencies up to 22.7%, with open-circuit voltage values reaching 0.90 V, ≈92% of the radiative limit for the bandgap of this material. The modified cells also show superior stability, with unencapsulated cells retaining 96% of their initial efficiency after >2000 h of storage in N2 and encapsulated cells retaining 90% efficiency after >450 h of storage in air. Intriguingly, CPTA preferentially binds to Sn2+ sites at film surface over Pb2+ due to the energetically favored exposure of the former, according to first-principles calculations. This work provides new insights into the surface chemistry of perovskite films in terms of their structural, electronic, and defect characteristics and this knowledge is used to fabricate state-of-the-art solar cells.

Understanding electronic structures, chemical bonding, and fluxional behavior of Lu2@C2n (2n = 76-88) by a theoretical study.

Endohedral metal-metal-bonding fullerenes, in which encapsulated metals form covalent metal-metal bonds inside, are an emerging class of endohedral metallofullerenes. Herein, we reported quantum-chemical studies on the electronic structures, chemical bonding, and dynamic fluxionality behavior of endohedral metal-metal-bonding fullerenes Lu2@C2n (2n = 76-88). Multiple bonding analysis approaches, including molecular orbital analysis, the natural bond orbital analysis, electron localization function, adaptive natural density partitioning analysis, and quantum theory of atoms in molecules, have unambiguously revealed one two-center two-electron σ covalent bond between two Lu ions in fullerenes. Energy decomposition analysis with the natural orbitals for chemical valence method on the bonding nature between the encapsulated metal dimer and the fullerene cage suggested the existence of two covalent bonds between the metal dimer and fullerenes, giving rise to a covalent bonding nature between the metal dimer and fullerene cage and a formal charge model of [Lu2]2+@[C2n]2-. For Lu2@C76, the dynamic fluxionality behavior of the metal dimer Lu2 inside fullerene C76 has been revealed via locating the transition state with an energy barrier of 5 kcal/mol. Further energy decomposition analysis calculations indicate that the energy barrier is controlled by a series of terms, including the geometric deformation energy, electrostatic interaction, and orbital interactions.

Inverse design of molecule-metal nanoparticle systems interacting with light for desired photophysical properties.

Molecules close to a metal nanoparticle (NP) have significantly different photophysical properties from those of the isolated one. In order to harness the potential of the molecule-NP system, appropriate design guidelines are required. Here, we propose an inverse design method of the optimal molecule-NP systems and incident electric field for desired photophysical properties. It is based on a gradient-based optimization search within the time-dependent quantum chemical description for the molecule and the continuum model for the metal NP. We designed the optimal molecule, relative molecule-NP spatial conformation, and incident electric field of a molecule-NP system to maximize the population transfer to the target electronic state of the molecule. The design results were presented and discussed. The present method is promising as the basis for designing molecule-NP systems and incident fields and accelerates discoveries of efficient molecular plasmonics systems.

N-Heterocyclic carbene-based C-centered Au(I)-Ag(I) clusters with intense phosphorescence and organelle-selective translocation in cells.

Photoluminescent gold clusters are functionally variable chemical modules by ligand design. Chemical modification of protective ligands and introduction of different metals into the gold clusters lead to discover unique chemical and physical properties based on their significantly perturbed electronic structures. Here we report the synthesis of carbon-centered Au(I)-Ag(I) clusters with high phosphorescence quantum yields using N-heterocyclic carbene ligands. Specifically, a heterometallic cluster [(C)(AuI-L)6AgI2]4+, where L denotes benzimidazolylidene-based carbene ligands featuring N-pyridyl substituents, shows a significantly high phosphorescence quantum yield (Φ = 0.88). Theoretical calculations suggest that the carbene ligands accelerate the radiative decay by affecting the spin-orbit coupling, and the benzimidazolylidene ligands further suppress the non-radiative pathway. Furthermore, these clusters with carbene ligands are taken up into cells, emit phosphorescence and translocate to a particular organelle. Such well-defined, highly phosphorescent C-centered Au(I)-Ag(I) clusters will enable ligand-specific, organelle-selective phosphorescence imaging and dynamic analysis of molecular distribution and translocation pathways in cells.

Cluster-Geometry-Associated Metal-Metal Bonding in Trimetallic Carbide Clusterfullerenes.

Geometry configurations of the metallic clusters play a significant role in the involved bonding nature. Herein, we report the crystallographic characterization of unprecedented erbium-based trimetallic clusterfullerenes, namely, Er3C2@Ih(7)-C80, in which the inner Er3C2 cluster presents a lifted bat ray configuration with the C2 unit elevated by ∼1.62 Å above the Er3 plane. Within the plane, the Er···Er distances for Er1···Er2, Er1···Er2A, and Er2···Er2A are 3.4051(15), 3.4051(15), and 3.3178(15) Å, respectively, falling into the range of the metal-metal bonding. Density functional theory calculations unveil the three-center-one-electron Er-Er-Er bond in Er3C2@Ih(7)-C80 with one electron shared by three metals, and thus, its exceptional electronic structure can be expressed as (Er3)8+(C2)2-@C806-. Interestingly, with the further observation on the geometry configurations of the encapsulated clusters in M3C2@C2n (M = Sc, Y, and Lu) series, we find that the lifted bat ray configuration of the inner cluster is explicitly associated with the formation of the bonding interactions between the inner metals. This finding provides insights into the nature of metal-metal bonding and gives guidelines for the design of the single-molecule magnet.