Abstract
Reticular chemistry is typically invoked to organize molecular building blocks, yet its ability to reprogram the intrinsic reactivity of atomically precise metal clusters remains largely unexplored. Here, we demonstrate dimensional reticulation as a decisive lever for tuning CO2 electroreduction by systematically comparing a thiolate-protected Cu7 nanocluster across molecular, one-dimensional (1D), and two-dimensional (2D) architectures. Coordination-driven assembly with linear and trigonal pyridyl linkers yields Cu7 MOF-1D and Cu7 MOF-2D, respectively, preserving cluster nuclearity while progressively transforming its geometric, electronic, and porous environment. Single-crystal analysis reveals that reticulation lifts the intrinsic metric degeneracy of the Cu7 core, imprinting directional distortions that intensify from 1D chains to 2D sheets. This evolution coincides with a transition from packing-limited solids to genuine microporous frameworks, culminating in a high-surface-area 2D material (659 m2 g–1) with enhanced CO2 uptake. These structural changes translate directly into electrocatalytic performance, where formate selectivity increases from 23.4% (Cu7 NC) to 39.0% (Cu7 MOF-2D), accompanied by a decreased Faradaic contribution of H2 despite increased overall cathodic activity. Post-reaction analyses show that the reticular frameworks retain their overall structural organization and predominantly low-valent, cluster-derived Cu coordination environments. Density functional theory reveals that reticulation strengthens CO2 binding, stabilizes *HCOO intermediates, and enriches electronic states near the Fermi level, collectively steering the reaction toward HCOOH formation. This work establishes reticular assembly as a multi-parameter strategy for programming cluster reactivity, transforming a single Cu7 motif into a hierarchy of catalytic states.
Keywords: copper cluster-organic frameworks, electrocatalytic CO2 reduction, electrochemical formate electrosynthesis, dimensionality-driven selectivity, framework-mediated catalysis


Introduction
The evolution of reticular chemistry − has been largely anchored in the use of single metal ions or small clusters as nodes, yet a transformative frontier is now emerging in which atomically precise metal nanoclusters (NCs) − are elevated to lattice points within extended frameworks. These nanocluster metal–organic frameworks (NCMOFs) redefine the foundational principles of MOF design by replacing conventional metal nodes with discrete, multi-metal entities whose nuclearity, geometry, and electronic structure are precisely encoded at the atomic level. − This “lattice-point design” paradigm enables a convergence between the molecular exactness of NCs and the long-range periodicity of crystalline frameworks, thereby unlocking bonding configurations, redox behaviors, and spatial anisotropies inaccessible to classical MOFs. Unlike ion-based secondary building units (SBUs), NC nodes inherently provide a high density of chemically addressable metal centers with delocalized electronic structures and multivalent redox flexibility, enabling more efficient charge transport and catalytic turnover. At the same time, their integration into ordered frameworks imparts site isolation, structural robustness, and architectural programmabilityfeatures typically absent in discrete NC systems. As a result, NCMOFs offer a uniquely integrated platform in which catalytic function is no longer dictated solely by composition, but can be systematically encoded through topology, connectivity, and dimensionality. Our recent studies on silver cluster frameworks have demonstrated how rigid linkers can stabilize fluxional Ag n clusters and enforce crystallographic order while preventing aggregation. − In contrast, copper NCs (Cu NCs), with their more adaptable coordination geometries and redox-active character, require a fundamentally different design logic in which flexible linkers accommodate structural distortions, facilitate self-assembly, and stabilize high-nuclearity motifs. This divergence highlights a broader principle: the integration of nanoclusters into reticular architectures is not a universal strategy, but a chemically nuanced approach that must be tailored to the intrinsic properties of each metal system. In this context, NCMOFs represent not merely an extension of MOFs but a conceptual shift toward programmable, cluster-based materials that unify precision, functionality, and structural hierarchy.
Electrocatalytic CO2 reduction (CO2RR) provides a compelling arena in which such design principles can be tested, as it demands simultaneous control over adsorption, electron transfer, and reaction selectivity in competition with the hydrogen evolution reaction (HER). , Copper-based catalysts occupy a privileged position in this field due to their unique ability to mediate multi-electron CO2 conversion pathways, including the formation of CO, formate, and hydrocarbons. − Atomically precise Cu NCs have recently emerged as model systems for probing structure–reactivity relationships, offering well-defined active sites and tunable electronic structures. − These systems can exhibit size-dependent catalytic behavior and enhanced selectivity arising from discrete electronic states; − however, their practical deployment remains hindered by intrinsic limitations. Under electrochemical conditions, Cu NCs often undergo aggregation, ligand loss, or structural reconstruction, leading to ill-defined active phases and diminished reproducibility. Furthermore, their molecular nature typically results in limited surface accessibility and inefficient integration into conductive electrodes, constraining current density and mass transport. Parallel efforts using MOF-based electrocatalysts have addressed some of these challenges by providing porous architectures and site isolation, − yet conventional ion-based MOFs frequently suffer from low electrical conductivity, limited redox flexibility, and ambiguous active-site identity, particularly under reductive bias. , These limitations underscore a central bottleneck in CO2RR catalyst design: achieving simultaneous control over structural definition, electronic structure, and catalytic stability while suppressing HER, which competes strongly for available electrons and often dominates product distributions. Thus, despite significant advances, a unifying materials platform that integrates the precision of nanoclusters with the robustness and accessibility of frameworks remains largely unrealized.
Here, we address this challenge by constructing NC-based reticular catalysts in which a thiolate-protected Cu7 cluster is embedded into one-dimensional (1D) and two-dimensional (2D) frameworks, enabling a direct and systematic interrogation of how dimensional reticulation governs catalytic behavior. By employing coordination-driven assembly with linear and trigonal pyridyl linkers, we generate Cu7 MOF-1D and Cu7 MOF-2D architectures that preserve the cluster nuclearity while progressively transforming its geometric, electronic, and porous environment. By maintaining Cu7 nuclearity across the series, this design separates nuclearity effects from the broader consequences of reticular assembly, while recognizing that dimensionality is intrinsically coupled to changes in linker identity, local coordination environment, connectivity, porosity, and electronic structure. Unlike discrete Cu7 NCs, which exhibit aggregation and HER-dominated behavior, the reticular analogues immobilize the cluster nodes within persistent coordination networks, enhance CO2 uptake through framework porosity, and introduce directional distortions that modulate adsorption and intermediate stabilization. As a result, the 2D framework achieves significantly enhanced formate selectivity and a reduced Faradaic contribution from H2 evolution, establishing a clear structure–reactivity relationship across molecular, 1D, and 2D states of the same cluster motif. Beyond the specific Cu7 system, this work introduces a broader conceptual advance: reticular assembly can serve as a programmable tool for engineering catalytic landscapes at the level of atomically precise clusters. By bridging the long-standing divide between molecular NCs and extended frameworks, NCMOFs emerge as a new class of electrocatalysts in which reactivity is not simply inherited but deliberately constructed through architecture.
Results and Discussion
From Linker-Directed Assembly to Geometric Reprogramming of Cu7 Nanocluster Nodes
Cu7 MOF-1D and Cu7 MOF-2D were obtained through a mild solution-phase cluster assembly process initiated from the thiolate-protected precursor [CuS t Bu] n . Treatment of the copper thiolate polymer with trifluoroacetic acid in dichloromethane triggers protonolysis of the Cu–S bonds, generating soluble copper–thiolate species capable of reorganizing into multinuclear clusters. Subsequent introduction of pyridyl-based organic linkers in dimethylacetamide promotes coordination-driven reticular assembly of these cluster nodes under ambient conditions. When 1,2-di(pyridin-4-yl)ethane (DPE) is employed as a linear ditopic linker, interfacial diffusion between the two solvent phases directs the formation of elongated block-like crystals of Cu7 MOF-1D (Figure S18A). In contrast, replacing the linker with the tripodal linker tris(pyridin-4-ylmethyl)amine (TPMA) leads to well-faceted truncated polyhedral crystals of Cu7 MOF-2D (Figure S18B). The distinct architectures arise from the different coordination geometries imposed by the linkers: the linear DPE linker favors propagation of cluster connectivity along a single direction, whereas the trigonal TPMA linker enables multidirectional coordination that promotes higher-dimensional framework growth. These crystallizations proceed at room temperature without solvothermal treatment, highlighting that the Cu7 NC nodes and their reticular assembly emerge spontaneously under interfacial diffusion conditions (Figure ).
1.

(A) Single-crystal X-ray structure of Cu7 MOF-1D, showing the 1D reticular architecture constructed from alternating crystallographically independent Cu7 nanocluster nodes (NC-1 and NC-2) bridged by the linear DPE linker. (B) Single-crystal X-ray structure of Cu7 MOF-2D, showing the 2D reticular architecture built from the heptanuclear Cu7 cluster node (NC-3) interconnected by the tripodal TPMA linker.
Single-crystal X-ray diffraction (SCXRD) analysis elucidates that Cu7 MOF-1D crystallizes in the monoclinic space group P21/n (Table S1) and is built from two crystallographically independent heptanuclear copper NC nodes, denoted NC-1 and NC-2 (Figure ). The asymmetric unit comprises 14 crystallographically distinct Cu sites, which assemble into these two symmetry-independent Cu7 clusters that act as the fundamental secondary building units (SBUs) of the framework. Several intermediate-layer Cu sites additionally exhibit a minor positional component (∼11% occupancy); analysis of these positions shows only modest local distortion of the Cu7 cores without altering their nuclearity or connectivity (Figure S17 and Tables S31, S32). As highlighted in Figure A, NC-1 and NC-2 are not isolated molecular entities but are incorporated into an extended coordination polymer through pyridyl ligation of the DPE linker, giving rise to a 1D reticular architecture. The framework propagates through alternating NC nodes along a single crystallographic direction, producing a zigzag chain in which the Cu7 clusters function as atomically precise SBUs and the linear ditopic DPE linker serves as the bridging strut between adjacent nodes. The presence of two non-equivalent Cu7 SBUs within the same chain indicates that the 1D framework is not generated by simple translational repetition of one unique cluster, but instead by periodic alternation of two closely related yet crystallographically distinct nanocluster environments. This alternating-node construction is structurally significant, because it shows that the DPE linker does not merely tether preformed clusters, but organizes them into a directionally anisotropic reticular motif in which discrete Cu7 cores are preserved while long-range framework order is established. The NC-1 node in Cu7 MOF-1D features a three-tiered Cu7 metal core composed of an apical Cu1 atom, a middle triangular layer defined by Cu2, Cu3, and Cu4, and a basal set of three copper atoms (Cu5–Cu7) (Figure S1). The most prominent direct metal–metal contacts are localized within the upper Cu4 crown fragment, where the Cu1–Cu2, Cu1–Cu3, and Cu1–Cu4 separations are 2.741, 2.829, and 2.785 Å, respectively (Table S3). These values fall within the range typically associated with cuprophilic Cu···Cu interactions, indicating appreciable metal–metal adjacency within the crown portion of the cluster. The non-equivalence of these distances further reveals that the Cu4 fragment adopts a distorted rather than symmetric geometry, reflecting the anisotropic coordination environment imposed by the surrounding ligand framework. This asymmetry is further reflected in the Cu–Cu–Cu angles centered at Cu1, namely 76.01° (Cu2–Cu1–Cu3), 78.59° (Cu2–Cu1–Cu4), and 70.85° (Cu4–Cu1–Cu3) (Table S4). These relatively acute angles indicate that the Cu2–Cu3–Cu4 triangle beneath Cu1 is compressed, giving rise to a capped triangular Cu4 crown geometry. These metrics show that NC-1 is best described as an anisotropic, electronically coupled Cu7 NC in which localized cuprophilic contacts and acute metal-core angles generate a preorganized multinuclear platform for subsequent ligand-supported framework construction. The NC-1 cluster is enveloped by a hexathiolate shell (S1–S6) that not only passivates the Cu7 core but also mechanically stitches together its three-tiered metal framework through a combination of μ2- and μ3-thiolate coordination modes (Figures S2 and S3). As summarized in Table S5, the Cu–S bond distances span 2.153–2.268 Å, with an average of 2.209 Å, confirming a uniformly strong Cu(I)–thiolate interaction manifold across the cluster surface. The shortest contacts, Cu2–S4 = 2.153 Å and Cu4–S3 = 2.153 Å, are closely followed by Cu2–S1 = 2.154 Å and Cu4–S6 = 2.155 Å, whereas the longest interactions, Cu7–S5 = 2.268 Å and Cu7–S6 = 2.259 Å, occur at the lower part of the cluster. This gradual expansion from the upper to the lower region indicates that the thiolate shell is not metrically uniform but adapts to the anisotropic topology of the Cu7 framework and the differing steric demands of the crown and basal domains. The three sulfur atoms in the upper part of the cluster, S1, S2, and S3, each adopt μ2-bridging coordination, linking the apical Cu1 atom to the middle-layer copper atoms Cu2, Cu3, and Cu4, respectively. Their corresponding Cu–S–Cu angles, namely Cu1–S1–Cu2 = 76.86°, Cu1–S2–Cu3 = 80.06°, and Cu1–S3–Cu4 = 78.37°, define a relatively tight set of bridging geometries that clamp the Cu4 crown into a compact arrangement (Table S6). Complementarily, the associated S–Cu–Cu angles centered at the apical and middle-layer copper atoms are also acute, with S1–Cu1–Cu2 = 49.93°, S2–Cu1–Cu3 = 48.67°, S3–Cu1–Cu4 = 49.21°, S1–Cu2–Cu1 = 53.22°, S2–Cu3–Cu1 = 51.28°, and S3–Cu4–Cu1 = 52.43°. These narrow angular metrics show that the upper thiolate bridges do not merely cap the cluster surface; rather, they compress and stabilize the apical Cu1(Cu2Cu3Cu4) substructure into a tightly fused crown-like unit. This interpretation is further reinforced by the Cu–Cu–S angles Cu1–Cu2–S4 = 133.92°, Cu1–Cu3–S5 = 139.43°, and Cu1–Cu4–S6 = 137.93°, which indicate that the sulfur donors connected to the lower shell are oriented in a way that transmits structural rigidity from the crown into the basal region. In contrast to the upper μ2-thiolate set, the lower sulfur atoms S4, S5, and S6 function as μ3-bridging ligands that interlink the middle and basal copper atoms, thereby creating a more highly cross-linked sulfur framework at the bottom of the cluster. Specifically, S4 bridges Cu2, Cu5, and Cu6; S5 bridges Cu3, Cu5, and Cu7; and S6 bridges Cu4, Cu6, and Cu7. This difference in connectivity is directly reflected in their angular distributions. For S4, the Cu–S–Cu angles are Cu2–S4–Cu5 = 113.38° and Cu2–S4–Cu6 = 106.53°, while the related Cu5–S4–Cu6 angle is 97.94°. For S5, the corresponding values are Cu3–S5–Cu5 = 87.80° and Cu3–S5–Cu7 = 121.15°, together with Cu5–S5–Cu7 = 109.52°. For S6, the analogous set comprises Cu4–S6–Cu6 = 117.72°, Cu4–S6–Cu7 = 103.44°, and Cu6–S6–Cu7 = 100.05°. The much broader span of these Cu–S–Cu angles (87.80–121.15°), relative to the upper μ2-thiolate bridges, reveals that the lower shell is geometrically more adaptive and serves as a flexible sulfur brace accommodating the expanded Cu5/Cu6/Cu7 basal platform. The associated S–Cu–Cu angles at the basal copper atoms are likewise diagnostic of this multidirectional support, including S4–Cu5–S5 = 116.72°, S4–Cu6–S6 = 112.63°, and S5–Cu7–S6 = 103.75°. Overall, the thiolate shell of NC-1 is organized in a distinctly hierarchical manner: an upper set of three μ2-thiolates (S1–S3) locks the apical Cu4 crown into a compact and angularly compressed arrangement, whereas a lower set of three μ3-thiolates (S4–S6) cross-links the middle and basal copper atoms into a more laterally expanded sulfur-supported platform. The sulfur shell therefore functions not simply as a passive protecting layer, but as a topological organizer of the NC, converting the Cu7 core into a rigid yet anisotropic SBU whose geometry is preconditioned for subsequent reticular propagation in Cu7 MOF-1D. In addition to the thiolate shell, the NC-1 cluster bears two trifluoroacetate ligands at the lower portion of the Cu7 framework, which further differentiate the coordination environments of the basal copper sites (Figures S4 and S5). Structural analysis shows that these carboxylates bind through terminal κ1-O coordination rather than Cu–O–Cu bridging, with the coordinated oxygen atoms O1 and O2 attached to Cu6 and Cu7, respectively. The corresponding Cu–O bond distances are 2.502 and 2.199 Å (Table S7), indicating that both carboxylates are bound to the cluster, while the shorter Cu7–O2 contact points to a relatively stronger Cu–O interaction at that site. As illustrated in Figure S5, these oxygen donors are localized on the basal sector of the cluster and complement the sulfur-rich coordination sphere established by the surrounding thiolate ligands. For the Cu6 site, the angles O1–Cu6–S4 and O1–Cu6–S6 are 92.14° and 104.31°, respectively, whereas at Cu7, the angles O2–Cu7–S5 and O2–Cu7–S6 are 106.85° and 108.84° (Table S8). These angular parameters demonstrate that the two trifluoroacetate ligands occupy distinct but well-defined positions within the basal coordination environment, giving rise to two related yet non-equivalent O/S-coordinated Cu sites. Accordingly, rather than functioning as bridging ligands, the trifluoroacetates act as site-selective terminal caps that subtly tune the local coordination geometry of the basal Cu centers and reinforce the intrinsic asymmetry of the Cu7 NC. The organic linker coordination in NC-1 is established through three pyridyl nitrogen donors that bind selectively to the basal copper atoms Cu5, Cu6, and Cu7, thereby defining the outward connectivity of the cluster toward the 1D framework (Figure S6). As summarized in Table S9, the Cu–N bond distances are 2.006, 2.023, and 2.025 Å for Cu5–N1, Cu6–N2, and Cu7–N3, respectively (average 2.018 Å), indicating a consistent Cu(I)–pyridyl coordination environment across the three basal sites, with the slightly shorter Cu5–N1 interaction suggesting a marginally stronger metal–linker contact. The pyridyl donors are accommodated within related yet non-equivalent mixed-donor coordination spheres defined by surrounding sulfur and oxygen atoms. At Cu5, the N–Cu–S angles are N1–Cu5–S4 = 120.25° and N1–Cu5–S5 = 123.00° (Table S10), indicating that the pyridyl donor is disposed along a well-separated coordination vector relative to the two thiolate ligands. At Cu6, the corresponding parameters are N2–Cu6–S4 = 122.11°, N2–Cu6–S6 = 120.53°, and N2–Cu6–O1 = 95.50°, whereas at Cu7, the analogous values are N3–Cu7–S5 = 123.76°, N3–Cu7–S6 = 119.83°, and N3–Cu7–O2 = 91.69°. The large N–Cu–S angles (119.83–123.76°) together with the near-orthogonal N–Cu–O angles (91.69–95.50°) indicate that pyridyl coordination projects away from the sulfur-rich face of the cluster. Consequently, the three DPE-derived pyridyl donors establish directionally organized coordination vectors on the basal face of the nanocluster, rendering NC-1 a reticularly competent SBU for propagation of the extended 1D Cu7 MOF-1D framework.
Despite sharing the same Cu7 nuclearity, hexathiolate shell, two terminal trifluoroacetates, and three pyridyl linker donors, NC-1 and NC-2 are not geometrically superposable; rather, they represent two closely related structural realizations of the same cluster formula embedded in different local framework environments. In NC-2, the apical Cu1–Cu2, Cu1–Cu3, and Cu1–Cu4 separations are 2.760, 2.793, and 2.766 Å, respectively (average 2.773 Å; Table S11), compared with 2.741, 2.829, and 2.785 Å in NC-1 (average 2.785 Å), indicating that the upper Cu4 crown of NC-2 (Figure S7) is slightly more metrically even and less distorted than that of NC-1. This trend is mirrored in the corresponding Cu–Cu–Cu angles: NC-2 exhibits 77.62° (Cu2–Cu1–Cu3), 73.69° (Cu2–Cu1–Cu4), and 74.13° (Cu4–Cu1–Cu3) (Table S12), whereas NC-1 shows a broader and more asymmetric set of 76.01°, 78.59°, and 70.85°, confirming that the crown fragment in NC-1 is more angularly compressed and anisotropic. The two nodes remain highly comparable at the level of the thiolate manifold, with both displaying the same overall average Cu–S distance of 2.209 Å; however, NC-2 spans a slightly broader range (2.135–2.271 Å; Table S13) than NC-1 (2.153–2.268 Å), and its lower μ3-thiolate network is redistributed accordingly (Figure S8). In particular, the lower-shell Cu–S–Cu angles in NC-2 extend from 93.27° to 122.94° (Table S14), compared with 87.80° to 121.15° in NC-1, showing that the sulfur-supported basal platform in NC-2 is reorganized into a somewhat less compressed but still strongly cross-linked geometry. The trifluoroacetate-capped basal copper sites (Figure S9) also remain topologically conserved but metrically distinct: NC-2 displays Cu6–O1 = 2.484 Å and Cu7–O2 = 2.175 Å (Table S15), versus 2.502 and 2.199 Å in NC-1, while the associated O–Cu–S angles (Table S16) remain closely matched (91.95–110.17° in NC-2 vs 92.14–108.84° in NC-1), indicating preservation of the same local coordination motif with subtle relaxation of the basal Cu/O/S environment. The most conspicuous difference emerges in the linker-binding (Figure S10) metrics, where the Cu–N distances in NC-2 (1.989, 1.999, and 2.031 Å; average 2.006 Å) (Table S17) are overall shorter than those in NC-1 (2.006, 2.023, and 2.025 Å; average 2.018 Å), suggesting slightly stronger pyridyl coordination in NC-2. Consistent with this, the N–Cu–S/O angular sets (Table S18) in the two clusters are very similar in pattern but not identical in value, confirming that both nodes present the same directional connectivity principle while accommodating it through subtly different local distortions. These comparisons show that NC-1 and NC-2 are compositionally equivalent but geometrically nondegenerate Cu7 SBUs: the framework preserves the same cluster archetype twice, yet allows each instance to relax into a distinct metric minimum. This duality is structurally important because it reveals that the 1D reticular chain is built not from a single rigidly repeated NC, but from two closely matched Cu7 nodes whose small but systematic differences encode the local packing and coordination anisotropy of the extended framework.
Unlike Cu7 MOF-1D, which accommodates two symmetry-independent cluster nodes within a 1D zigzag chain, Cu7 MOF-2D is built from a single heptanuclear cluster type, NC-3, whose geometry preserves the same three-tiered Cu7 motif yet redistributes the supporting ligand field in a manner that is distinctly compatible with 2D reticulation. The NC-3 core comprises an apical Cu1 atom, an intermediate triangular layer (Cu2, Cu3, Cu4), and a basal set of three copper atoms (Cu5–Cu7) (Figure S11). Within the upper Cu4 crown, the Cu1–Cu2, Cu1–Cu3, and Cu1–Cu4 separations are 2.770, 2.725, and 2.789 Å, respectively (average 2.761 Å; Table S19), while the corresponding Cu–Cu–Cu angles are 72.01° (Cu2–Cu1–Cu3), 77.46° (Cu2–Cu1–Cu4), and 72.34° (Cu4–Cu1–Cu3) (Table S20), indicating a more compressed crown than in either NC-1 or NC-2. The surrounding hexathiolate shell remains hierarchically differentiated, with S1–S3 operating as upper μ2-bridging ligands and S4–S6 as lower μ3-bridging ligands (Figure S12), but its metric distribution is subtly shifted relative to the 1D analogues: the Cu–S distances span 2.160–2.272 Å (average 2.218 Å; Table S21), and the lower-shell Cu–S–Cu angles extend from 87.73° to 126.22° (Table S22), revealing a more laterally relaxed sulfur-supported basal platform. A further distinguishing feature of NC-3 is its carboxylate manifold (Figure S13). Rather than the exclusively terminal κ1-O arrangement observed in NC-1, the trifluoroacetate environment in NC-3 becomes electronically and topologically differentiated: O1 engages both Cu5 and Cu6 with Cu–O distances of 2.751 and 2.611 Å, respectively, whereas O2 binds Cu7 at 2.335 Å (Table S23). The associated angles, including O1–Cu5–S4 = 87.83°, O1–Cu5–S5 = 111.05°, O1–Cu6–S4 = 91.72°, O1–Cu6–S6 = 92.81°, O2–Cu7–S5 = 99.94°, O2–Cu7–S6 = 100.30°, and Cu5–O1–Cu6 = 73.88° (Table S24), show that the basal Cu/O/S interface of NC-3 is more interconnected than in the 1D framework. The cluster remains reticularly competent through three pyridyl coordination vectors at Cu5, Cu6, and Cu7 (Figure S14), with Cu5–N1 = 1.992 Å, Cu6–N2 = 2.023 Å, and Cu7–N3 = 2.009 Å (average 2.008 Å; Table S25). Their associated angular setsN1–Cu5–S4 = 124.90°, N1–Cu5–S5 = 128.01°, N1–Cu5–O1 = 89.59°; N2–Cu6–S4 = 122.63°, N2–Cu6–S6 = 119.95°, N2–Cu6–O1 = 93.79°; and N3–Cu7–S5 = 123.41°, N3–Cu7–S6 = 127.45°, N3–Cu7–O2 = 91.16° (Table S26)show that the pyridyl donors project away from the sulfur-rich cluster face in three well-separated directions. In the presence of the tripodal TPMA linker, these three outwardly oriented coordination vectors are no longer confined to chain propagation, but are instead distributed across the plane to stitch neighboring NC-3 units into an extended 2D reticular sheet (Figure B). Thus, NC-3 preserves the atomically precise Cu7 cluster archetype while rebalancing its Cu–Cu, Cu–S, Cu–O, and Cu–N metrics into a coordination geometry that is intrinsically more compatible with planar, multidirectional framework growth.
To further interrogate how reticular assembly perturbs the intrinsic geometry of the Cu7 cluster scaffold, the inorganic skeletons of NC-1, NC-2, and NC-3 were compared using a common set of structurally equivalent Cu and S reference positions (Figure A). As illustrated in Figure A, the analysis defines a homologous local motif comprising an apical copper site (Cuα), a middle-layer copper (Cuβ), a basal copper (Cuγ), and the corresponding upper and lower sulfur sites (Sα and Sβ). This representation makes it possible to track how the same cluster archetype responds to different reticular constraints in the 1D and 2D frameworks. The most revealing trend is seen in the Cu–Cu metrics (Figure B). The shorter Cuα–Cuβ distances (bond indices 1–3), which correspond to the apical-to-middle-layer crown contacts, remain comparatively similar among NC-1, NC-2, and NC-3, indicating that the immediate Cu4 crown is largely preserved irrespective of framework dimensionality. In contrast, the longer Cuα···Cuγ separations (bond indices 4–6), which span from the apical site to the basal layer, display substantially larger divergence. This anisotropic expansion is most pronounced for NC-3, showing that incorporation into the 2D TPMA-linked network imposes a stronger directional distortion on the Cu7 scaffold than the DPE-linked 1D framework. In other words, while the apical crown remains relatively conserved, the apical-to-basal projection of the cluster is selectively elongated along specific directions once the node is embedded in an extended reticular environment. By comparison, an idealized discrete molecular Cu7 nanocluster would be expected to show essentially no such directional splitting. A complementary picture emerges from the Cu–S analysis (Figure C). Here, bond indices 1–3 correspond to Cuα–Sα interactions, whereas 4–6 represent the longer Cuα···Sβ separations. Relative to the Cu–Cu data set, the Cu–S distances exhibit much smaller variation across all three clusters, and the spread within each cluster is correspondingly lower. This indicates that the Cu–S framework is appreciably more resistant to reticular distortion than the Cu–Cu framework, and that the modest variations in Cu–S metrics are most plausibly a secondary consequence of metal-core deformation rather than a primary driver of it. On this basis, Figure shows that the structural plasticity of these Cu7 nodes is expressed predominantly through anisotropic Cu–Cu reorganization, while the thiolate scaffold remains comparatively robust and acts as a geometrically buffering shell around the deformable copper core. A related directional effect is evident in the linker-coordination metrics shown in Figure . Figure A compares the three pyridyl coordination vectors of the DPE linker in NC-1 and NC-2, labeled as Cuγ–Nα, Cuγ–Nβ, and Cuγ–Nγ, and shows that although the absolute Cu–N distances differ between the two crystallographically independent nodes, their ordering is preserved in a point-symmetry-related manner. This trend is quantified in Figure B: in both NC-1 and NC-2, the Cuγ–Nγ bond is the longest, whereas Cuγ–Nα is the shortest and Cuγ–Nβ is intermediate. Thus, the same directional hierarchy is retained in both nodes despite their nonequivalent local environments. Structurally, this is highly significant because the longest Cu–N contact always coincides with the coordination vector oriented along the framework expansion direction. The pyridyl bond aligned with the reticular growth axis is therefore consistently the most elongated, indicating that linker binding is not isotropically accommodated by the Cu7 node but instead carries a directional structural signature of framework propagation. This finding reinforces the broader conclusion from Figure : reticular assembly does not simply connect preformed Cu7 NCs but imprints a measurable and direction-dependent distortion onto both the metal core and its linker-facing coordination sphere. In the 1D framework, that distortion is encoded symmetrically across the alternating NC-1/NC-2 pair, whereas in the 2D framework, it is amplified into a more spatially distributed deformation that supports multidirectional extension.
2.

(A) Definition of the common structural motif used for comparison of the inorganic skeletons of NC-1, NC-2, and NC-3, highlighting the equivalent reference sites Cuα, Cuβ, Cuγ, Sα, and Sβ. (B) Comparison of the Cu–Cu distances within the three Cu7 nanocluster nodes, showing that the shorter apical-to-middle-layer contacts remain relatively conserved, whereas the longer apical-to-basal separations exhibit pronounced directional divergence, particularly in NC-3. (C) Comparison of the Cu–S distances within the same cluster nodes, showing that the thiolate shell remains comparatively robust and less sensitive to reticular distortion than the Cu–Cu framework.
3.

(A) Definition of the three pyridyl coordination vectors in NC-1 and NC-2, denoted Cuγ–Nα, Cuγ–Nβ, and Cuγ–Nγ, showing the orientation of the DPE-derived linker coordination relative to the Cu7 cluster core. (B) Comparison of the corresponding Cu–N bond distances for NC-1 and NC-2. In both nodes, the same directional hierarchy is preserved, with Cuγ–Nγ being the longest Cu–N bond, while Cuγ–Nα is the shortest and Cuγ–Nβ is intermediate.
A useful reference point for interpreting the framework-embedded clusters is the discrete Cu7 NC, which exposes the extent to which reticular assembly breaks the intrinsic metric degeneracy of the parent metal–thiolate scaffold. In the isolated cluster, the Cu7 core (Figure S15) adopts a highly regular arrangement in which the three apical crown contacts, Cu1–Cu2, Cu1–Cu3, and Cu1–Cu4, are all identical at 2.712 Å (Table S27), and the three corresponding Cu–Cu–Cu angles, Cu2–Cu1–Cu3, Cu2–Cu1–Cu4, and Cu4–Cu1–Cu3, are likewise degenerate at 60.82° (Table S29). Even the longer center-to-center separations from the apical Cu atom to the basal Cu atoms, Cu1–Cu5, Cu1–Cu6, and Cu1–Cu7, are equivalent at 5.324 Å (Table S28). Thus, unlike NC-1, NC-2, and NC-3, which all display measurable directional differentiation in both Cu–Cu distances and Cu–Cu–Cu angles, the discrete nanocluster represents a metrically undistorted Cu7 reference state with essentially zero dispersion within each equivalent bond set. The same contrast is observed in the sulfur coordination environment (Figure S16). In the discrete cluster, the three upper Cu1–S1/2/3 contacts are all 2.315 Å, while the longer Cu1···S4/5/6 separations are each 3.973 Å (Table S30), again reflecting a fully symmetric ligand field. By comparison, the framework-embedded clusters show clear splitting of both short and long Cu–S metrics: NC-1 spans 2.153–2.268 Å, NC-2 spans 2.135–2.271 Å, and NC-3 spans 2.160–2.272 Å, with the lower sulfur manifold especially responsive to reticular constraints. In other words, whereas the isolated nanocluster preserves an idealized and internally equivalent Cu7/S6 geometry, incorporation into Cu7 MOF-1D and Cu7 MOF-2D lifts that degeneracy and resolves the cluster into a set of directionally inequivalent metal–metal and metal–sulfur vectors. This comparison is structurally important because it shows that the framework does not merely immobilize a preformed Cu7 core; rather, it mechanically imprints anisotropy onto an originally more symmetric NC scaffold. The effect is already evident in the 1D architecture, where the discrete archetype splits into two nondegenerate variants (NC-1 and NC-2), and becomes still more pronounced in the 2D network, where NC-3 exhibits the strongest directional redistribution of Cu–Cu, Cu–S, Cu–O, and Cu–N metrics. From this perspective, the discrete Cu7 NC may be regarded as the symmetry-preserving parent structure, whereas the reticular frameworks represent symmetry-lowered, mechanically biased descendants in which linker-driven assembly selectively stretches, compresses, and reorients the cluster without destroying its atomically precise identity.
From Atomic Architecture to Bulk Function: Structural Fidelity, Porosity, and Electronic Identity
To verify that the crystallographically resolved cluster frameworks persist in the bulk state, we first examined the powdered samples by powder X-ray diffraction (PXRD). The experimental patterns of Cu7 MOF-1D and Cu7 MOF-2D reproduce the simulated traces generated from the SCXRD models with close fidelity over the full measured range, confirming that both materials are obtained as phase-pure, long-range ordered cluster-assembled solids rather than as poorly correlated cluster aggregates (Figure A). In Cu7 MOF-1D, the unusually intense and sharp reflection at 24.5° can be indexed to the (270) plane on the basis of the monoclinic cell of the SCXRD structure. Its enhanced prominence in the experimental trace is most reasonably attributed to preferred orientation of the elongated block-like crystallites during sample mounting, together with strong periodic registry along the corresponding lattice planes, rather than to any secondary phase. The absence of extra reflections or diffuse features further indicates that the heptanuclear Cu7 nodes propagate regularly through the solid, yielding a periodic 1D chain lattice in Cu7 MOF-1D and an extended 2D sheet lattice in Cu7 MOF-2D. Scanning electron microscopy (SEM) images support this crystallographic picture: Cu7 MOF-1D appears as elongated block-like crystals, whereas Cu7 MOF-2D forms well-faceted truncated polyhedra, and energy-dispersive X-ray spectroscopy (EDX) mapping confirms homogeneous distributions of Cu, S, N, O, and F throughout each crystal, excluding obvious phase segregation or elemental inhomogeneity (Figure S19).
4.

(A) Overlay of the simulated and experimental PXRD patterns of Cu7 MOF-1D and Cu7 MOF-2D, evidencing close correspondence with the crystallographically resolved structures. (B) N2 adsorption–desorption isotherms of Cu7 NC, Cu7 MOF-1D, and Cu7 MOF-2D measured at 77 K. (C) Pore-size distribution of Cu7 MOF-2D obtained from NLDFT analysis. Inset: calculated pore size of ∼0.77 nm. (D) CO2 adsorption–desorption isotherms of Cu7 MOF-1D and Cu7 MOF-2D measured at 298 K. (E) High-resolution Cu 2p XPS spectra of Cu7 MOF-1D and Cu7 MOF-2D. (F) Cu K-edge XANES spectra of Cu7 MOF-1D and Cu7 MOF-2D, benchmarked against reference materials (Cu foil, Cu2O, CuO, and CuS t Bu).
Gas-sorption measurements further reveal how dimensional reticulation transforms the accessible free volume of the Cu7 cluster platform. All three materialsCu7 NC, Cu7 MOF-1D, and Cu7 MOF-2Ddisplay steep uptake at very low relative pressure, consistent with micropore-dominated adsorption behavior (Figure B). However, their accessible surface areas differ dramatically. The discrete Cu7 NC exhibits only a small BET area (40.9 m2 g–1; Figure S20A), which is not indicative of a true permanently porous molecular framework; rather, it most likely originates from extrinsic voids created by inefficient packing of ligand-coated clusters, minor interparticle free volume, and limited accessibility to shallow surface recesses created by the thiolate shell. In other words, the low surface area of the discrete cluster should not be interpreted as intrinsic molecular porosity, but as the adsorption signature of a densely packed molecular solid that still contains a small amount of accessible free space. A similarly low value is observed for Cu7 MOF-1D (35.6 m2 g–1; Figure S20B), despite its reticular nature. In this case, the cluster nodes are extended only along one dimension, and the bulky StBu– shell together with the DPE struts prevents efficient close packing of neighboring chains, leaving behind interchain free volume rather than well-developed intrinsic framework channels. By contrast, Cu7 MOF-2D exhibits a markedly higher BET area (659.0 m2 g–1; Figure S20C), more than 18-fold greater than that of Cu7 MOF-1D. The BET fitting ranges were selected individually from the appropriate linear regions of the BET plots, considering the distinct adsorption regimes of the three materials and the established consistency criteria for porous materials (Figure S20). This large increase arises because reticulation in two dimensions creates an extended and periodically accessible void network: the TPMA-linked sheets resist collapse, maintain intersheet galleries, and at the same time preserve intralayer free volume around the cluster nodes. Thus, the progression from discrete Cu7 NC to 1D Cu7 MOF-1D to 2D Cu7 MOF-2D can be viewed as a transition from packing-derived extrinsic free space, to weakly accessible interchain voids, to genuine framework-accessible microporosity. NLDFT analysis is fully consistent with this trend: Cu7 MOF-2D shows a clear micropore-size distribution centered at 0.72 nm, in excellent agreement with the crystallographically estimated pore width of ∼0.77 nm (Figure C), whereas Cu7 NC and Cu7 MOF-1D display negligible pore-size distributions (Figure S21). The same structural hierarchy is echoed in low-pressure CO2 sorption at 298 K. Cu7 MOF-2D adsorbs 11.5 cm3 g–1 of CO2, clearly exceeding Cu7 MOF-1D (8.1 cm3 g–1) (Figure D). This higher uptake is not simply a consequence of larger surface area, although that is an important factor; rather, it reflects the combined effect of greater accessible micropore volume, better diffusion pathways, and a more favorable polar environment within the 2D framework. The narrow microporous galleries in Cu7 MOF-2D place CO2 in closer contact with the sulfur-rich cluster surface, pyridyl nitrogens, and fluorinated carboxylate-derived environment, all of which can enhance interaction with the quadrupolar CO2 molecule. In Cu7 MOF-1D, by contrast, adsorption is limited by the fact that most of the available free volume is interchain and less uniformly distributed, so the gas experiences a more discontinuous adsorption landscape. From a catalytic viewpoint, this distinction is important because higher CO2 uptake in Cu7 MOF-2D suggests a stronger ability to locally enrich the reactant near the Cu active motif.
The surface and local electronic structures of the two framework materials were then probed by X-ray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS). Survey spectra confirm the simultaneous presence of Cu, S, N, O, C, and F, in agreement with the cluster core, thiolate shell, pyridyl linker, and trifluoroacetate ligands (Figure S22). In the high-resolution Cu 2p spectra, Cu7 MOF-1D shows Cu 2p3/2 and Cu 2p1/2 components at 932.0 and 951.9 eV, while Cu7 MOF-2D exhibits corresponding peaks at 931.6 and 951.5 eV (Figure E). These relatively low binding energies, together with the absence of intense shake-up satellite features in the 940–945 eV region, are diagnostic of Cu(I) rather than Cu(II). This is a critical distinction: Cu(II) species typically display pronounced satellite structure arising from final-state effects associated with an open-shell 3d9 configuration, whereas Cu(I), with its closed-shell 3d10 electronic structure, does not. The XPS data therefore show that the cluster nodes in both frameworks retain a predominantly Cu(I)-centered electronic state, fully consistent with the thiolate-rich coordination environment resolved crystallographically. This assignment is reinforced by Cu K-edge X-ray absorption near-edge structure (XANES) and Fourier transform extended X-ray absorption fine structure (FT-EXAFS) analyses. The Cu K-edge XANES spectra of both Cu7 MOF-1D and Cu7 MOF-2D lie much closer to the edge position of Cu2O than to those of CuO or Cu foil (Figure F), corroborating the predominance of Cu(I) in both frameworks. In other words, the Cu centers are neither metallic nor strongly oxidized, but remain in a low-valent, cluster-stabilized coordination state. The FT-EXAFS profiles provide a complementary real-space picture of the local coordination environment. For Cu7 MOF-1D, a dominant feature appears at ∼2.4 Å (phase-uncorrected), attributable primarily to Cu–Cu scattering, while additional features at ∼1.8 Å and ∼1.4 Å are assigned to Cu–S and Cu–O contributions, respectively (Figure S32B). Cu7 MOF-2D shows an analogous pattern, with a major Cu–Cu contribution at ∼2.3 Å and lower-R components at ∼1.8 Å and ∼1.3 Å arising from Cu–S and Cu–O coordination (Figure S32C). The persistence of a strong Cu–Cu shell in both materials is especially significant, because it demonstrates that the multinuclear copper framework of the cluster node remains intact after reticulation. XANES and EXAFS converge on the conclusion that Cu7 MOF-1D and Cu7 MOF-2D preserve an atomically precise, Cu(I)-rich, metal–metal-coupled cluster environment rather than collapsing into disordered mononuclear or oxide-like Cu species.
UV–vis diffuse reflectance spectroscopy (DRS) was carried out not to evaluate photoactivity, but to probe the electronic structure and metal–ligand electronic communication relevant to electrocatalytic CO2RR. In electrocatalysis, such measurements help identify the ligand-centered and metal-perturbed transitions that shape the frontier electronic landscape, which in turn influences charge delocalization, redox accessibility, and the ability of the cluster node to mediate electron transfer during catalysis. For Cu7 MOF-1D, the sharp absorption at 259 nm mirrors that of the DPE linker and is most reasonably assigned to a ligand-centered π→π* transition of the pyridyl framework (Figure S23A). The broader feature centered near 304 nm is likewise inherited largely from the linker but is expected to be perturbed by coordination to the Cu7 node, giving it mixed ligand-centered/metal-perturbed charge-transfer character. A closely related situation is observed for Cu7 MOF-2D and TPMA (Figure S23B), which both show a sharp band at 259 nm and a broader band at 322 nm. The red-shifted broad absorption in the TPMA-derived system suggests a somewhat more extended or more easily polarized electronic manifold, consistent with the more multidirectional coordination environment of the 2D framework. Thus, the DRS data indicate that the optical response of both Cu7 MOFs is dominated by linker-based transitions, yet these transitions are clearly modulated by coordination to the Cu7 cluster and therefore provide a window into the electronic coupling environment that underpins CO2RR. Finally, thermogravimetric analysis (TGA) was employed to evaluate framework robustness, identify the release of guest molecules, and delineate the sequence of ligand loss. Under N2, the discrete Cu7 NC remains intact up to ∼170 °C, while Cu7 MOF-1D and Cu7 MOF-2D remain stable to ∼145 °C and ∼180 °C, respectively (Figure S24), indicating that both cluster-assembled materials possess meaningful thermal robustness despite their relatively soft, ligand-rich architectures. For Cu7 MOF-1D, the initial 7.1% mass loss up to ∼145 °C is attributable to desorption of trapped solvent molecules, followed by a 10.7% loss between ∼145 and 160 °C assigned to trifluoroacetate removal, a further 27.5% loss between ∼160 and 200 °C associated with decomposition of the (S t Bu)6 shell, and an additional 22.8% loss between ∼200 and 230 °C corresponding to cleavage of the DPE linker. For Cu7 MOF-2D, the first 4.1% loss up to ∼125 °C likewise reflects guest removal, followed by a much larger 36.0% loss between ∼125 and 210 °C arising from combined loss of CF3COO– and (S t Bu)6, and a final 20.1% loss between ∼210 and 350 °C associated with decomposition of the TPMA linker. These stepwise profiles are consistent with the crystallographically established ligand hierarchy and show that the two Cu7 MOFs preserve their cluster-based skeletons over a practically useful temperature range before undergoing sequential ligand dissociation and framework collapse.
From Electrode Integration to Formate Selectivity: Reticular Control of Cu7 Electrocatalysis
For electrochemical evaluation, particular care was taken to ensure that differences in electrode preparation and Cu loading did not obscure the intrinsic comparison among the three Cu7-based catalysts. All three crystalline materials were processed using the same electrode-fabrication protocol, without the addition of carbon black or other conductive powders. After thorough grinding, Cu7 NC, Cu7 MOF-1D, and Cu7 MOF-2D were dispersed in 2-propanol containing Nafion, ultrasonicated to form homogeneous catalyst inks, and drop-cast onto identical carbon-paper substrates. Because the three materials possess different ligand/framework compositions and therefore different Cu weight fractions, their total catalyst masses were adjusted to provide equivalent Cu loading: 17.4 mg of Cu7 NC, 12.3 mg of Cu7 MOF-1D, and 12.0 mg of Cu7 MOF-2D were employed. This Cu-normalized electrode preparation provides a consistent basis for comparing their electrochemical responses while avoiding differences arising simply from unequal amounts of Cu or from the presence of different conductive additives. The initial electrochemical signature of this architectural effect is already evident in the LSV profiles recorded in CO2-saturated 0.1 M KHCO3 (Figure A). Across the cathodic scan, Cu7 MOF-2D delivers the largest reduction current, followed by Cu7 MOF-1D, whereas the discrete Cu7 NC shows the lowest response. This ordering indicates that the 2D reticular material provides the most favorable combination of electrochemically addressable Cu sites, interfacial mass transport, and charge-transfer connectivity under CO2RR conditions. The LSVs therefore do more than rank activity: they imply that dimensional reticulation progressively transforms the Cu7 platform from a ligand-crowded molecular object into a more open and electronically connected electrocatalyst. In the discrete nanocluster, the Cu core is embedded in a comparatively compact ligand envelope and is prone to aggregation within the catalyst layer, which can limit the effective density of electrochemically accessible catalytic motifs. In the 1D framework, reticulation begins to organize these nodes into a periodic chain, but the accessible void space and transport pathways remain modest. In the 2D framework, by contrast, the more extended porous architecture, higher CO2 uptake, and multidirectional linker-supported connectivity create a more favorable local environment for reductive turnover. The higher reduction current of Cu7 MOF-2D is therefore consistent with a catalyst that not only admits CO2 more effectively, but also sustains greater charge flow through a more percolated cluster network.
5.

(A) LSV profiles of Cu7 NC, Cu7 MOF-1D, and Cu7 MOF-2D recorded in 0.1 M KHCO3 aqueous electrolyte under CO2 flow. (B) FE of the products formed on Cu7 NC, Cu7 MOF-1D, and Cu7 MOF-2D at different applied potentials. (C) Partial current densities of the products on Cu7 NC, Cu7 MOF-1D, and Cu7 MOF-2D at −1.4 V vs RHE in 0.1 M KHCO3 aqueous electrolyte under CO2 flow. Error bars represent mean ± SD (n = 3). (D) Total current density for Cu7 NC, Cu7 MOF-1D, and Cu7 MOF-2D at different applied potentials.
Controlled-potential electrolysis in a flow-cell shows that these structural differences translate directly into product-selective CO2 reduction. GC and 1H NMR analyses identify H2, CO, and HCOOH as the major products, but their distributions vary strongly with catalyst architecture. Control experiments under Ar are especially important here: only H2 was detected in the absence of CO2, establishing that the observed CO and HCOOH originate exclusively from reduction of the flowing CO2 feed rather than from adventitious carbon sources. Within this verified CO2RR regime, all three catalysts produce HCOOH, yet the efficiency with which they channel current into this pathway changes systematically across the structural series. At −1.4 V vs RHE, the maximum FEHCOOH values are 23.4% for Cu7 NC, 28.9% for Cu7 MOF-1D, and 39.0% for Cu7 MOF-2D. The same trend is already apparent for Cu7 MOF-2D at less negative potentials, where FEHCOOH reaches 26.1% at −1.0 V, 32.0% at −1.2 V, and 39.0% at −1.4 V, showing that the 2D reticular catalyst promotes formate production over a broad applied-potential window. By contrast, the discrete cluster remains strongly biased toward the competing HER, delivering FEH2 = 72.5% at −1.2 V and 71.2% at −1.4 V. The fractional HER contribution decreases in the reticular analogues: Cu7 MOF-1D shows FEH2 = 62.4% at −1.2 V and 61.9% at −1.4 V, while Cu7 MOF-2D exhibits the lowest FEH2, reaching 57.6% at −1.2 V and 56.2% at −1.4 V. Importantly, this decrease in FEH2 should not be interpreted as a decrease in absolute H2 production. As shown by the H2 partial current densities in Figure C, the framework catalysts sustain higher absolute H2-forming currents because their total cathodic activities are also substantially greater. Rather, reticulation redistributes the fractional current contribution: although HER remains active, a larger proportion of the enhanced cathodic current is directed toward CO2-derived products, particularly HCOOH. This evolution in selectivity is particularly striking because the three catalysts contain the same number of Cu atoms per cluster unit. The divergence in performance can therefore be distinguished from a change in Cu nuclearity and instead reflects the coupled effects of ligand and coordination environment, cluster geometry, framework connectivity, porosity, and interfacial accessibility that accompany reticular organization of the Cu7 motif. In the discrete cluster, the Cu centers are protected by a compact ligand shell and are embedded in a relatively nonporous molecular solid, so proton reduction remains kinetically competitive and often dominant. In Cu7 MOF-1D, one-dimensional reticulation partially reorganizes the ligand environment and begins to expose the Cu7 node to a more extended coordination landscape, lowering the relative weight of HER and improving HCOOH production. In Cu7 MOF-2D, this effect becomes much more pronounced: the 2D framework combines higher porosity, greater CO2 adsorption capacity, and a more directionally open coordination environment around the cluster node. Together, these factors should increase the probability that adsorbed intermediates involved in the two-electron formate pathway are stabilized relative to proton discharge. Said differently, the 2D framework appears to reshape the catalytic microenvironment so that carbon fixation becomes more competitive than hydrogen evolution. This provides a compelling structural lesson: reticulation is not a passive support strategy, but an active means of rewriting the reaction landscape around an atomically precise Cu7 core.
The partial current densities reinforce this interpretation. At −1.4 V vs RHE, Cu7 MOF-2D exhibits the highest partial current density for HCOOH, followed by Cu7 MOF-1D and then Cu7 NC (Figure C). The same ordering is maintained at −1.0 V and −1.2 V (Figure S25), demonstrating that the superior formate productivity of the 2D catalyst is not confined to a single bias point but is persistent across the potential range examined. This is an important result, because FE alone can sometimes improve simply by suppressing one side reaction; here, however, the increase in FEHCOOH is accompanied by a corresponding increase in the absolute current directed to HCOOH formation. Thus, the 2D material is not merely more selectiveit is also intrinsically more productive toward formate. Structurally, this behavior is fully consistent with the combination of features already established above: higher accessible surface area, genuine framework microporosity, stronger CO2 uptake, and a more spatially distributed distortion of the Cu7 node. Those characteristics likely cooperate to enhance CO2 enrichment near the cluster, improve transport of reactants and products through the catalyst layer, and bias the local Cu environment toward carbon-bound reduction intermediates rather than H adsorption. In this sense, Cu7 MOF-2D can be regarded as the most effective architectural translation of the Cu7 motif into an electrocatalyst, because it turns a compositionally fixed cluster into a structurally optimized platform for HCOOH generation. The total current density traces further support this structure–reactivity relationship. As shown in Figure D, the overall cathodic current follows the same order as the LSV and product-current trends, namely Cu7 MOF-2D > Cu7 MOF-1D > Cu7 NC across the applied potential range. This consistency implies that the enhanced formate production of the reticular materials, especially the 2D framework, is not an isolated selectivity anomaly but part of a broader improvement in electrocatalytic turnover. The electrochemical fingerprints traced by the FE profiles, product current densities, and total current densities reveal that the stepwise evolution from molecular cluster to 1D chain to 2D sheet tunes the Cu7 catalyst toward lower fractional HER contribution, stronger HCOOH-forming current, and higher total cathodic throughput. The fact that all of these changes occur while the Cu7 nuclearity remains constant underscores the central mechanistic point of this work: it is the geometric structure, ligand-supported microenvironment, and dimensionality of reticulation, rather than Cu atom count alone, that govern how the Cu7 unit behaves under CO2RR conditions.
Stability measurements show that the reticular catalysts maintain these advantages under sustained bias. The current densities at fixed applied potentials remain well-defined over the 1 h measurements, and longer chronoamperometric experiments at −1.4 V vs RHE reveal stable responses for both Cu7 MOF-1D and Cu7 MOF-2D over 8 h (Figure S26). Equally important, the total FE remains essentially unchanged during this period, and the individual contributions from HCOOH, CO, and H2 are likewise retained without obvious drift. This behavior indicates that the cluster-based active motifs remain functionally intact during electrolysis and that the reticular frameworks provide sufficient structural resilience to prevent rapid deactivation or major selectivity drift. In practical terms, the 1D and 2D cluster-assembled materials are therefore not only more selective than the discrete cluster precursor, but also sufficiently robust to sustain their catalytic identity over extended electrolysis. The 2D framework, in particular, emerges as the most compelling design in this series: it combines the highest current response, the strongest preference for HCOOH formation, the largest framework-accessible surface area, and stable catalytic operation, all while preserving the atomically precise Cu7 cluster as the operative structural motif.
Reticular Confinement Preserves Cu7 Identity under Catalytic Turnover
To directly correlate catalytic performance with structural integrity under operating conditions, we next examined the fate of the Cu7 motifs after CO2RR. Particular attention was paid to distinguishing non-destructive aggregation, framework robustness, and electronic-state invariance, as these factors collectively determine whether the observed catalytic behavior originates from the designed cluster architectures or from unintended restructuring. For the discrete Cu7 NC, the pre-catalytic dispersion state already reveals an important limitation. Dynamic light scattering (DLS) analysis of the catalyst slurry shows a bimodal distribution dominated by micron-scale aggregates, centered at ∼1.9 μm (95.4%) with a minor population at ∼5.4 μm, yielding an average size of ∼2.1 μm and notably no detectable population below 1 μm (Figure S27). Given that the intrinsic molecular dimension of the Cu7 cluster is ∼1.4 nm, this disparity demonstrates that the clusters are not present as individually dispersed catalytic entities, but rather exist as colloidally assembled superstructures. Importantly, such aggregation is not inherently destructive: the thiolate-protected Cu7 cores are known to retain their internal structure upon clustering, implying that the system initially consists of structurally intact but spatially congested catalytic domains. This aggregation state is already consistent with the inferior electrochemical accessibility and HER-dominated behavior observed for the discrete NC. The situation evolves further under catalytic conditions. TEM analysis reveals that, after CO2RR, both the average particle size and the breadth of the size distribution increase (Figure S28), indicating that aggregation is not static but dynamically amplified during electrolysis. The magnitude of this growth exceeds what would be expected for purely reversible, non-destructive clustering, suggesting that partial irreversible coalescence or restructuring occurs alongside benign aggregation. However, the persistence of identifiable cluster-derived features implies that this process is not a complete transformation into bulk metallic particles; rather, the system likely undergoes hybrid aggregation, in which a fraction of clusters remains intact while others partially fuse or reorganize. This behavior provides a mechanistic explanation for the instability of catalytic selectivity in molecular systems: the active-site definition becomes progressively blurred as the reaction proceeds. In sharp contrast, the reticular analogues exhibit a fundamentally different post-reaction behavior. Because the Cu7 nodes are embedded within extended coordination frameworks, non-destructive aggregation is structurally suppressed; instead, the relevant question becomes whether the framework itself remains intact. PXRD comparisons before and after CO2RR show that both Cu7 MOF-1D and Cu7 MOF-2D retain their crystallographic signatures with high fidelity (Figure S29), confirming that the long-range periodic arrangement of cluster nodes is largely preserved during electrolysis. For Cu7 MOF-1D, the diffraction pattern is essentially unchanged, indicating that the 1D chain architecture is structurally rigid under the applied conditions. For Cu7 MOF-2D, a subtle but informative deviation is observed: the lowest-angle reflection, indexed to the (011) plane, becomes attenuated after catalysis. This reflection corresponds to the interlayer periodicity of the stacked 2D sheets, and its selective disappearance indicates a degree of interlayer slippage or partial delamination. Crucially, all higher-angle reflectionswhich encode the intralayer ordering and the integrity of the Cu7 cluster nodesremain intact. Thus, the catalytic process induces only a weak perturbation along the stacking direction, while the fundamental 2D framework and its atomically precise cluster SBUs are preserved. In practical terms, this represents a minor structural relaxation rather than framework degradation and may even be beneficial by slightly increasing interlayer accessibility without compromising the catalytic motif.
The electronic stability of the Cu centers further corroborates this structural resilience. High-resolution Cu 2p XPS spectra collected before and after CO2RR show no detectable shift in binding energies or emergence of satellite features for either Cu7 MOF-1D or Cu7 MOF-2D (Figure S30), consistent with retention of a predominantly low-valent Cu environment throughout catalysis. Because Cu 2p XPS alone cannot unambiguously distinguish Cu(I) from Cu(0), this result is interpreted together with the XANES and EXAFS data rather than as independent proof of an unchanged Cu(I) state. X-ray absorption spectroscopy provides a more sensitive probe of both oxidation state and local coordination. The Cu K-edge XANES spectra of Cu7 NC and Cu7 MOF-1D before and after reaction are essentially indistinguishable (Figure S31A,B), indicating that no major change in the average Cu electronic environment occurs during catalysis. For Cu7 MOF-2D, a slight shift of the absorption edge toward higher energy is observed after CO2RR (Figure S31C), but the edge position remains much closer to that of Cu2O than to CuO, indicating that the Cu centers are still predominantly low-valent and not oxidized to Cu(II). This subtle shift is best interpreted as a minor redistribution of electron density within the cluster–ligand manifold, possibly associated with interlayer relaxation or adsorbate-induced electronic perturbation, rather than a change in formal oxidation state. The FT-EXAFS spectra reinforce this conclusion at the level of local structure. After catalysis, both Cu7 MOF-1D and Cu7 MOF-2D retain a dominant Cu–Cu scattering feature, accompanied by secondary contributions from Cu–S and Cu–O coordination shells (Figure S32). The persistence of the Cu–Cu peak is particularly diagnostic: it demonstrates that the multinuclear Cu7 core remains intact, rather than fragmenting into mononuclear species or restructuring into bulk metallic copper. At the same time, the retention of Cu–S and Cu–O contributions confirms that the ligand framework surrounding the cluster is preserved, maintaining the characteristic mixed-donor environment established crystallographically. Any minor variations in peak intensity or position can be attributed to subtle changes in coordination geometry or thermal/disorder effects under operating conditions, rather than to a fundamental structural transformation. These post-reaction analyses establish a clear structural dichotomy between the molecular and reticular systems. The discrete Cu7 NC operates as a dynamically evolving ensemble, in which aggregation and partial restructuring progressively modify the catalytic landscape. In contrast, Cu7 MOF-1D and Cu7 MOF-2D function as structurally encoded catalysts, where the cluster-derived Cu environments are immobilized within persistent frameworks that substantially retain their local structural and electronic characteristics during CO2RR. The minor interlayer relaxation observed for the 2D material represents a benign structural adjustment rather than degradation and does not compromise the integrity of the active Cu7 motif. From a broader perspective, these findings underscore a central advantage of reticular cluster assembly: it transforms atomically precise but inherently mobile nanoclusters into topologically stabilized catalytic units, in which both structural identity and electronic configuration are retained under reaction conditions. This stabilization not only prevents aggregation-induced loss of definition but also ensures that the catalytic behavior can be directly traced to the designed Cu7 architecture, thereby closing the structure–property loop established in the preceding sections.
Mechanistic Reprogramming of CO2RR at Cu7 Nodes: Pathways, Intermediates, and Formate Selectivity
To mechanistically rationalize how dimensional reticulation reprograms catalytic behavior at the Cu7 node, we carried out density functional theory (DFT) calculations using structurally faithful cluster fragments that preserve the local coordination environment while eliminating artificial edge states. In each case, the smallest chemically meaningful repeating unit was extracted and appropriately terminated to maintain charge balance and coordination saturation (Figure A–C). This approach enables direct comparison of the intrinsic reactivity of the Cu7 motif across the molecular, 1D, and 2D architectures without convolution from long-range periodic artifacts. For each catalyst, three distinct adsorption sites-denoted as Site 1, Site 2, and Site 3-were identified and systematically examined for CO2 and H adsorption (Figure S33). For the framework catalysts, the possible Cu centers corresponding to Site 3 show essentially equivalent three-dimensional coordination environments and closely similar DOS characteristics near the Fermi level; accordingly, one representative Site 3 environment was employed in the adsorption calculations (Figure S34). Across all three catalysts, the considered sites exhibit stronger CO2 binding than H binding, indicating a thermodynamic preference for CO2 adsorption at these sites. This adsorption preference favors competitive access of CO2 to the catalytic surface, but should not be interpreted as suppression of the overall HER rate, because HER remains thermodynamically accessible at other reaction coordinates and active sites. However, a complete CO2RR pathway analysis at every site is computationally demanding. Therefore, representative active sites were selected based on the largest energy difference between E b(CO2) and E b(H). Accordingly, Site 1 and Site 3 were chosen for Cu7 NC, while Site 1 and Site 2 were selected for both Cu7 MOF-1D and Cu7 MOF-2D for subsequent CO2RR analysis. Before discussing the CO2RR pathways in detail, it is essential to analyze the HER activity, as hydrogen is found to be the dominant product across all three catalysts. The HER free-energy profiles for all sites are presented in Figure A–C. The HER overpotential (ηHER) for each catalyst is defined as the minimum overpotential among the three adsorption sites, i.e., ηHER = min [ηHER (Site 1), ηHER (Site 2), ηHER (Site 3)], where ηHER (Site i) represents the overpotential for HER at Site i. For example, in the case of Cu7 NC (Figure A), ηHER values of 1.17, 0.76, and 1.21 V are obtained for Site 1, Site 2, and Site 3, respectively, yielding an overall ηHER of 0.76 V. Similarly, the ηHER values for Cu7 MOF-1D and Cu7 MOF-2D are 0.45 and 0.08 V, respectively. This analysis provides insight into the thermodynamic feasibility of the HER across the studied catalysts. However, it does not fully explain the overall activity trend, as the observed activity is confined to a single active site. To develop a more complete understanding of the activity trend, it is important to examine the full reaction free-energy profiles at other possible active sites as well. The CO2RR pathways considered in this study include CO and HCOOH formation routes. CO production proceeds via the following steps:
6.

Structures and the analyzed sites for (A) Cu7 NC, (B) Cu7 MOF-1D, and (C) Cu7 MOF-2D. The cumulative density of states (DOS) of all the Cu atoms in (D) Cu7 NC, (E) Cu7 MOF-1D, and (F) Cu7 MOF-2D.
7.

Free-energy diagrams for the electrocatalytic reactions on Cu7-based catalysts. (A–C) HER free-energy profiles for Cu7 NC, Cu7 MOF-1D, and Cu7 MOF-2D, respectively, evaluated at the representative adsorption sites. (D–F) CO2RR free-energy profiles for Cu7 NC, Cu7 MOF-1D, and Cu7 MOF-2D, respectively, illustrating the competing pathways leading to CO and HCOOH formation. Here, * denotes the pristine catalyst, and *molecule denotes an adsorbed reaction intermediate.
Pathway 1 (CO formation):
| 1 |
| 2 |
| 3 |
HCOOH production occurs through two possible pathways:
Pathway 2 (HCOOH via *HCOO):
| 4 |
| 5 |
| 6 |
Pathway 3 (HCOOH via *COOH):
| 7 |
| 8 |
| 9 |
Analysis of the CO2RR free-energy diagrams (Figure D–F) reveals that, at 0 V, the HCOOH formation pathways are energetically more favorable than the CO formation pathway for all three catalysts and all selected sites. For Cu7 NC, the overpotentials for HCOOH formation (ηHCOOH) are 0.79 V for both Site 1 and Site 3, which are comparable to the overpotentials for CO formation (ηCO) of 0.78 and 0.74 V, respectively. At these applied potentials, HER is also activated (ηHER = 0.76 V), resulting in the simultaneous production of H2, CO, and HCOOH, with HCOOH being favored over CO. For Cu7 MOF-1D, the ηHCOOH and ηCO values for Site 1 and Site 2 are [0.84 V, 1.09 V] and [0.78 V, 0.61 V], respectively. Similarly, for Cu7 MOF-2D, the corresponding ηHCOOH and ηCO values are [0.86 V, 0.86 V] for Site 1 and [0.80 V, 1.06 V] for Site 2. Therefore, applying potentials of approximately 0.84 V for Cu7 MOF-1D and 0.86 V for Cu7 MOF-2D induces the formation of both HCOOH and CO while simultaneously increasing the likelihood of HER across all sites. Although the calculated overpotentials confirm the concurrent formation of H2, CO, and HCOOH-with H2 being the dominant product-they do not fully explain the progressive suppression of CO formation and enhancement of HCOOH selectivity from Cu7 NC to Cu7 MOF-1D and further to Cu7 MOF-2D. This behavior can be understood by closely examining the thermodynamics of the key surface-bound intermediates, namely *COOH and *HCOO, which govern the branching between CO and HCOOH formation pathways. For Cu7 NC (Figure D), the formation of both *COOH and *HCOO intermediates is thermodynamically uphill. However, the energy penalty associated with *HCOO formation is smaller than that for *COOH. As a result, the reaction preferentially proceeds through Pathway 2 rather than Pathway 1, while the unfavorable formation of *COOH effectively limits the activation of Pathway 3. This energetic landscape explains why HCOOH formation is favored over CO production on Cu7 NC. In contrast, the energetics change significantly for Cu7 MOF-1D (Figure E). At Site 1, *COOH formation is nearly thermoneutral (∼0.01 eV), whereas at Site 2 it becomes exothermic (−0.26 eV). Meanwhile, *HCOO formation is exothermic at both sites, further strengthening the preference for Pathway 2 relative to Cu7 NC. Although the exothermic stabilization of *COOH at Site 2 opens the possibility for Pathway 3, this site favors CO formation through Pathway 1, which proceeds with a substantially lower overpotential (0.61 V) compared to HCOOH formation via Pathway 3 (1.09 V). Consequently, Pathway 1 is kinetically preferred over Pathway 3 at Site 2, resulting in higher HCOOH production and reduced CO formation compared to Cu7 NC. For Cu7 MOF-2D (Figure F), *COOH formation remains nearly thermoneutral at Site 1 (∼0.01 eV) but becomes strongly exothermic at Site 2 (−0.68 eV), indicating a much stronger stabilization of this intermediate relative to Cu7 MOF-1D. At Site 2, CO formation via Pathway 1 requires a higher overpotential (1.06 V) than HCOOH formation via Pathway 3 (0.86 V), making Pathway 3 the more favorable route. This shift leads to a further increase in HCOOH selectivity and a more pronounced suppression of CO formation compared to Cu7 MOF-1D. Overall, these energetic trends result in a clear increase in HCOOH formation in the order Cu7 MOF-2D > Cu7 MOF-1D > Cu7 NC, with the opposite trend observed for CO production. To gain deeper insight into the origin of this enhanced catalytic behavior, we examined the cumulative density of states (DOS) of all Cu atoms in the three catalysts (Figure D–F). Both Cu7 MOF-1D and Cu7 MOF-2D display a noticeably higher density of electronic states near the Fermi level than Cu7 NC, reflecting increased electronic activity. The greater availability of states close to the Fermi level strengthens the interaction between the catalyst surface and the oxygen atoms of CO2-derived intermediates, particularly *HCOO, thereby stabilizing this intermediate and facilitating HCOOH formation more effectively in Cu7 MOF-1D and Cu7 MOF-2D than in Cu7 NC. The DFT results reveal that the catalytic evolution across the Cu7 series is not governed by a single factor, but by the interplay of three tightly coupled effects: (i) adsorption thermodynamics, which favor CO2 over H at all sites; (ii) intermediate stabilization, which progressively lowers the energetic cost of key CO2RR steps; and (iii) electronic delocalization, which enhances charge-transfer capability and reshapes the reaction landscape. Reticular assembly therefore emerges not as a passive structural modification, but as a multifunctional tuning strategy that simultaneously redefines adsorption, reaction energetics, and electronic structure. In doing so, it converts an atomically precise but electronically constrained Cu7 nanocluster into a hierarchically optimized catalytic platform for selective CO2-to-HCOOH conversion.
Conclusions
In this study, we demonstrate that the catalytic behavior of an atomically precise Cu7 nanocluster can be systematically redefined through dimensional reticular assembly, without altering its nuclearity. By comparing Cu7-based motifs across molecular, 1D, and 2D architectures while maintaining Cu nuclearity, we establish a structure–function relationship in which reticular organizationand the accompanying changes in ligand environment, connectivity, porosity, and electronic structuregoverns catalytic performance in CO2 electroreduction.
The key findings are summarized as follows:
Reticulation-induced structural reconfiguration: Incorporation of Cu7 clusters into extended frameworks lifts the intrinsic symmetry of the discrete nanocluster, introducing direction-dependent distortions that become more pronounced from 1D chains to 2D sheets.
Emergence of true framework porosity: While the discrete Cu7 NC and 1D analogue exhibit primarily packing-derived voids, the 2D framework develops genuine microporosity with a high surface area (659 m2 g–1), enabling enhanced CO2 uptake and diffusion.
Architecture-governed catalytic selectivity: A clear evolution in product distribution is observed, with HCOOH selectivity progressively increasing while the fractional contribution of H2 to the total Faradaic efficiency decreases in the order Cu7 NC > Cu7 MOF-1D > Cu7 MOF-2D.
Stabilization of active motifs under operation: Unlike the discrete cluster, which undergoes aggregation and partial restructuring, the reticular frameworks immobilize Cu7 nodes within robust coordination networks, preserving both structural integrity and Cu(I) electronic states during CO2RR.
Mechanistic origin of selectivity control: DFT analysis reveals that reticulation simultaneously modulates adsorption thermodynamics, stabilizes key *HCOO intermediates, and enhances electronic delocalization near the Fermi level, collectively steering the reaction pathway toward HCOOH formation.
These results show that reticular assembly is not merely a structural extension strategy, but a powerful means of encoding catalytic function into the architecture itself. By transforming a nuclearity-conserved Cu7 motif into distinct reticular catalytic states, this work introduces a design paradigm in which reactivity can be programmed through the coupled influence of ligand coordination, topology, connectivity, porosity, and framework dimensionality. Looking forward, this concept opens broad opportunities for the rational design of cluster-based catalysts, where reticular engineering can be used to precisely tune adsorption environments, reaction pathways, and electronic structures. Extending this strategy to other metal NCs, mixed-metal systems, and multielectron transformations could enable the development of next-generation electrocatalysts for sustainable chemical synthesis and energy conversion.
Methods
Synthesis of Copper tert-Butylthiolate [CuS t Bu] n
CuS t Bu was synthesized according to a previously reported procedure. Copper(II) nitrate trihydrate (3.0 mmol) was dissolved in acetonitrile (10.0 mL), followed by the addition of triethylamine (10.0 mL), which resulted in the formation of a blue precipitate. While stirring the suspension, tert-butyl mercaptan (9.0 mmol) was added, which caused the precipitate to dissolve and a yellow solution to form. The reaction mixture was stirred briefly, and the solvent was subsequently removed under reduced pressure. The resulting solid was washed with methanol and dried to afford copper tert-butylthiolate [CuS t Bu] n as a yellow powder in approximately 95% yield.
Synthesis of Cu7 MOF-1D
First, the [CuS t Bu] n complex (10.0 mg) was dispersed in DCM (1 mL) to form a homogeneous suspension. CF3COOH (0.09 mmol, 6.68 μL) was then added, and the mixture was stirred continuously at room temperature until a clear solution (solution 1) was obtained. Separately, the DPE linker (0.065 mmol, 12.0 mg) was dissolved in DMAc (1.0 mL) to give a clear solution (solution 2). An aliquot of solution 2 (1.0 mL) was transferred to a cent tube, followed by the careful addition of solution 1 (1.0 mL). The tube was left undisturbed in the dark for 2 days. Yellow elongated block-like crystals formed at the liquid–liquid interface and were isolated in 19.5% yield based on Cu.
Synthesis of Cu7 MOF-2D
First, the [CuS t Bu] n complex (10.0 mg) was dispersed in DCM (1 mL) to form a homogeneous suspension. CF3COOH (0.09 mmol, 6.68 μL) was then added, and the mixture was stirred continuously at room temperature until a clear solution (solution 1) was obtained. Separately, the TPMA linker (0.065 mmol, 18.9 mg) was dissolved in DMAc (1.0 mL) to give a clear solution (solution 2). An aliquot of solution 2 (1.0 mL) was transferred to a cent tube, followed by the careful addition of solution 1 (1.0 mL). The tube was left undisturbed in the dark for 2 days. Yellow well-faceted truncated polyhedral crystals formed at the liquid–liquid interface and were isolated in 20.1% yield based on Cu.
Synthesis of Cu7 Nanocluster
A Cu7 nanocluster was synthesized following the method previously described in the literature.
Preparation of Electrocatalysts for Electrochemical CO2 Reduction
The crystalline Cu7 NC, Cu7 MOF-1D, and Cu7 MOF-2D samples were each thoroughly ground in an agate mortar for 30 min to obtain fine powders and minimize differences arising from crystal size. To ensure an equivalent Cu loading across the three catalysts, the catalyst masses were adjusted according to their calculated Cu contents, using Cu7 MOF-2D (12.0 mg) as the reference. Accordingly, 17.4 mg of Cu7 NC, 12.3 mg of Cu7 MOF-1D, and 12.0 mg of Cu7 MOF-2D were used for electrode preparation. Each catalyst was dispersed in 1050 μL of 2-propanol containing 10 μL of 5 wt % Nafion dispersion and ultrasonicated for 30 min to obtain a homogeneous catalyst ink. No carbon black or other conductive additive was incorporated into any of the catalyst inks. The resulting inks were drop-cast onto carbon-paper substrates (SIGRACET GDL 36BB, diameter: 2 cm), followed by vacuum drying overnight to remove residual solvent and obtain the working electrodes for electrochemical CO2 reduction.
Electrochemical CO2 Reduction Measurements
All electrochemical measurements were performed using an ECstat-302 potentiostat (EC FRONTIER) in a flow-cell configuration (EC FRONTIER). The catalyst-coated carbon paper was employed as the working electrode, while an Ag/AgCl electrode and a platinum mesh were used as the reference and counter electrodes, respectively. Electrochemical experiments were conducted in 0.1 M KHCO3 aqueous electrolyte. Prior to measurements, the electrolyte was purged with CO2 at a flow rate of 25 mL min–1 for 15 min to ensure saturation. To activate the electrode surface and remove residual impurities, cyclic voltammetry (CV) was performed in the potential range of 0.0 to 1.2 V vs RHE. Subsequently, chronoamperometry (CA) measurements were carried out at constant potentials for 1 h under continuous CO2 flow (10 mL min–1). After each CA run, the electrolyte was replaced and resaturated with CO2 prior to subsequent measurements. CA experiments were conducted at applied potentials ranging from −1.0 to −1.4 V vs RHE. Linear sweep voltammetry (LSV) was performed under continuous flow of CO2 (10 mL min–1) at a scan rate of 20 mV s–1 over the potential range of 0.0 to −1.5 V vs RHE. Long-term stability tests were conducted by chronoamperometry at −1.4 V vs RHE for 8 h. The geometric current density was calculated based on an electrode area of π cm2.
Product Analysis and Faradaic Efficiency Determination
Gas-Phase Product Analysis
Gaseous products were analyzed using an online gas chromatograph (Shimadzu GC-8A) equipped with a thermal conductivity detector (TCD) and a flame ionization detector (FID). Ar or N2 was used as the carrier gas depending on the detector configuration. For analysis, a 5 mL gas sample collected at the end of each CA measurement was injected into the sampling loop. Quantification of gaseous products was performed using calibration curves constructed from standard gases.
Liquid-Phase Product Analysis
Liquid products were quantified by 1H NMR spectroscopy using a Bruker AVANCE III 400 MHz spectrometer. After electrolysis, 590 μL of electrolyte was collected and mixed with 10 μL of DMSO (internal standard) and 65.6 μL of D2O. The concentration of liquid products was determined from the integral ratios of the product peaks relative to the DMSO reference signal.
Faradaic Efficiency Calculation
Faradaic efficiencies (FEs) for each product were calculated by comparing the charge required for product formation with the total charge passed during electrolysis, according to
| 10 |
where n is the number of electrons transferred, F is the Faraday constant, N is the amount of product formed, and Q is the total charge passed.
Computational Details
All catalyst structures and reaction intermediates were fully optimized, and their ground-state energies were calculated using the open-source Quantum ESPRESSO package. − The exchange–correlation interactions were described within the generalized gradient approximation (GGA) using the Perdew–Burke–Ernzerhof (PBE) functional. The convergence thresholds for electronic and ionic energies were set to 7.35 × 10–7 Ry (∼10–5 eV) and 1 × 10–4 Ry (∼0.001 eV), respectively, with a force convergence criterion of 7.78 × 10–4 Ry/Bohr (0.02 eV/Å) applied in all geometry optimizations. Electronic and ionic convergence thresholds were set to 7.35 × 10–7 Ry (∼10–5 eV) and 1 × 10–4 Ry (∼0.001 eV), respectively, while a force convergence criterion of 7.78 × 10–4 Ry/Bohr (0.02 eV/Å) was applied for all geometry optimizations. Brillouin-zone sampling was performed using a 1 × 1 × 1 Monkhorst–Pack k-point grid. Gaussian smearing with a smearing width of 0.01 Ry was employed to improve electronic convergence. Long-range dispersion interactions were accounted for using the DFT-D3 van der Waals correction with damping. , To avoid spurious interactions between periodic images, a vacuum layer of 20 Å was introduced along the z-direction.
The Gibbs free-energy change (ΔG) for each elementary reaction step was calculated according to
| 11 |
where ΔE denotes the DFT-calculated energy difference between the adsorbed intermediate and the pristine surface, ΔZPE represents the zero-point energy correction, and ΔS corresponds to the entropy change of gas-phase and adsorbed species evaluated at 298.15 K.
Supplementary Material
Acknowledgments
We gratefully acknowledge Prof. Y. Idemoto and Dr. T. Ichihashi (Tokyo University of Science) for the TEM measurements. This study was supported by the Scientific Research on Innovative Areas “Aquatic Functional Materials” (grant no. 22H04562), JSPS KAKENHI (grant nos. 23H00289, 23KK0098, and 26KJ0596), the Yazaki Memorial Foundation for Science and Technology, and the FUSO Innovative Technology Fund. R.T. thanks the DST–FIST (SR/FST/PS-I/2024/306) and the National Supercomputing Mission (NSM) for providing access to PARAM Rudra for conducting DFT-based computations.
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmaterialsau.6c00199.
Crystallographic data; atomically resolved structural analysis of Cu7 cores and ligand coordination environments; physicochemical characterization (PXRD, SEM–EDX, TEM, XPS, DRS, TGA, and gas sorption); electrochemical CO2RR measurements and product analysis; post-reaction structural and electronic characterization (XANES and EXAFS); and DFT computational methods and adsorption energetics (PDF)
#.
R.N., J.S., and S.G. contributed equally. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.
The authors declare no competing financial interest.
References
- Yaghi, O. M. ; Kalmutzki, M. J. ; Diercks, C. S. . Introduction to Reticular Chemistry: metal-Organic Frameworks and Covalent Organic Frameworks; Wiley–VCH: Weinheim, Germany, 2019. [Google Scholar]
- Gropp C., Canossa S., Wuttke S., Gándara F., Li Q., Gagliardi L., Yaghi O. M.. Standard Practices of Reticular Chemistry. ACS Cent. Sci. 2020;6(8):1255–1273. doi: 10.1021/acscentsci.0c00592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yaghi O. M.. Reticular Chemistry in All Dimensions. ACS Cent. Sci. 2019;5(8):1295–1300. doi: 10.1021/acscentsci.9b00750. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yaghi O. M.. Reticular Chemistry: Molecular Precision in Infinite 2D and 3D. Mol. Front. J. 2019;3(01):66–83. doi: 10.1142/S2529732519400054. [DOI] [Google Scholar]
- Lin H., Song X., Chai O. J. H., Yao Q., Yang H., Xie J.. Photoluminescent Characterization of Metal Nanoclusters: Basic Parameters, Methods, and Applications. Adv. Mater. 2024;36(25):e2401002. doi: 10.1002/adma.202401002. [DOI] [PubMed] [Google Scholar]
- Yang Z., Wang Y., Zhang R., Chen T., Xie J.. Precise Surface Engineering of Metal Nanoclusters: Ligand Programming for Functionality Design. Adv. Mater. 2025;37(50):e08578. doi: 10.1002/adma.202508578. [DOI] [PubMed] [Google Scholar]
- Li S., Li N.-N., Dong X.-Y., Zang S.-Q., Mak T. C. W.. Chemical Flexibility of Atomically Precise Metal Clusters. Chem. Rev. 2024;124(11):7262–7378. doi: 10.1021/acs.chemrev.3c00896. [DOI] [PubMed] [Google Scholar]
- Kang X., Zhu M.. Tailoring the Photoluminescence of Atomically Precise Nanoclusters. Chem. Soc. Rev. 2019;48:2422–2457. doi: 10.1039/C8CS00800K. [DOI] [PubMed] [Google Scholar]
- Du Y., Sheng H., Astruc D., Zhu M.. Atomically Precise Noble Metal Nanoclusters as Efficient Catalysts: A Bridge between Structure and Properties. Chem. Rev. 2020;120(2):526–622. doi: 10.1021/acs.chemrev.8b00726. [DOI] [PubMed] [Google Scholar]
- Jin R., Zeng C., Zhou M., Chen Y.. Atomically Precise Colloidal Metal Nanoclusters and Nanoparticles: Fundamentals and Opportunities. Chem. Rev. 2016;116(18):10346–10413. doi: 10.1021/acs.chemrev.5b00703. [DOI] [PubMed] [Google Scholar]
- Jin R., Li G., Sharma S., Li Y., Du X.. Toward Active-Site Tailoring in Heterogeneous Catalysis by Atomically Precise Metal Nanoclusters with Crystallographic Structures. Chem. Rev. 2021;121(2):567–648. doi: 10.1021/acs.chemrev.0c00495. [DOI] [PubMed] [Google Scholar]
- Weerawardene K. L. D. M., Häkkinen H., Aikens C. M.. Connections Between Theory and Experiment for Gold and Silver Nanoclusters. Annu. Rev. Phys. Chem. 2018;69:205–229. doi: 10.1146/annurev-physchem-052516-050932. [DOI] [PubMed] [Google Scholar]
- Gupta R. K., Wang Z., Mohan B., Tung C.-H., Sun D.. Atomically Precise Coinage Metal Nanoclusters with Exposed Metal Sites: Unlocking Reactivity and Functionality. Adv. Funct. Mater. 2025;35(48):2507047. doi: 10.1002/adfm.202507047. [DOI] [Google Scholar]
- Seong H., Lee D.. Atomically Precise Metal Nanoclusters for Energy Conversion. Bull. Korean Chem. Soc. 2024;45(5):435–450. doi: 10.1002/bkcs.12842. [DOI] [Google Scholar]
- Hirai H., Ito S., Takano S., Koyasu K., Tsukuda T.. Ligand-Protected Gold/Silver Superatoms: Current Status and Emerging Trends. Chem. Sci. 2020;11:12233–12248. doi: 10.1039/D0SC04100A. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chakraborty I., Pradeep T.. Atomically Precise Clusters of Noble Metals: Emerging Link between Atoms and Nanoparticles. Chem. Rev. 2017;117(12):8208–8271. doi: 10.1021/acs.chemrev.6b00769. [DOI] [PubMed] [Google Scholar]
- Maity S., Kolay S., Chakraborty S., Devi A., Rashi Patra R., Patra A.. A Comprehensive Review of Atomically Precise Metal Nanoclusters with Emergent Photophysical Properties towards Diverse Applications. Chem. Soc. Rev. 2025;54(4):1785–1844. doi: 10.1039/D4CS00962B. [DOI] [PubMed] [Google Scholar]
- Jadzinsky P. D., Calero G., Ackerson C. J., Bushnell D. A., Kornberg R. D.. Structure of a Thiol Monolayer-Protected Gold Nanoparticle at 1.1 Å Resolution. Science. 2007;318(5849):430–433. doi: 10.1126/science.1148624. [DOI] [PubMed] [Google Scholar]
- Lu Y., Chen W.. Sub-Nanometre Sized Metal Clusters: From Synthetic Challenges to the Unique Property Discoveries. Chem. Soc. Rev. 2012;41(9):3594–3623. doi: 10.1039/c2cs15325d. [DOI] [PubMed] [Google Scholar]
- McKenzie L. C., Zaikova T. O., Hutchison J. E.. Structurally Similar Triphenylphosphine-Stabilized Undecagolds, Au11(PPh3)7Cl3 and [Au11(PPh3)8Cl2]Cl, Exhibit Distinct Ligand Exchange Pathways with Glutathione. J. Am. Chem. Soc. 2014;136(38):13426–13435. doi: 10.1021/ja5075689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gan Z., Chen J., Wang J., Wang C., Li M.-B., Yao C., Zhuang S., Xu A., Li L., Wu Z.. The Fourth Crystallographic Closest Packing Unveiled in the Gold Nanocluster Crystal. Nat. Commun. 2017;8(1):14739. doi: 10.1038/ncomms14739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Brust M., Walker M., Bethell D., Schiffrin D. J., Whyman R.. Synthesis of Thiol-Derivatised Gold Nanoparticles in a Two-Phase Liquid–Liquid System. J. Chem. Soc., Chem. Commun. 1994;30:801–802. doi: 10.1039/C39940000801. [DOI] [Google Scholar]
- Zhang L., Wang E.. Metal Nanoclusters: New Fluorescent Probes for Sensors and Bioimaging. Nano Today. 2014;9(1):132–157. doi: 10.1016/j.nantod.2014.02.010. [DOI] [Google Scholar]
- Xiao Y., Wu Z., Yao Q., Xie J.. Luminescent Metal Nanoclusters: Biosensing Strategies and Bioimaging Applications. Aggregate. 2021;2(1):114–132. doi: 10.1002/agt2.11. [DOI] [Google Scholar]
- Sahoo K., Gazi T. R., Roy S., Chakraborty I.. Nanohybrids of Atomically Precise Metal Nanoclusters. Commun. Chem. 2023;6(1):157. doi: 10.1038/s42004-023-00958-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kawawaki T., Imai Y., Suzuki D., Kato S., Kobayashi I., Suzuki T., Kaneko R., Hossain S., Negishi Y.. Atomically Precise Alloy Nanoclusters. Chem.Eur. J. 2020;26(69):16150–16193. doi: 10.1002/chem.202001877. [DOI] [PubMed] [Google Scholar]
- Das S., Sakai J., Nakatani R., Kondo A., Tomioka R., Das S., Takahashi S., Kawawaki T., Biswas S., Negishi Y.. Tunable Structural Rearrangement in Cu Cluster Assemblies through Linker and Solvent Alterations. Chem. Sci. 2025;16(6):2600–2608. doi: 10.1039/D4SC07730J. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Irie T., Nakatani R., Sreelakshmi K. S., Sasaki K., Nozaki M., Kondo A., Kawawaki T., Das S., Negishi Y.. Reticular assembly of 1D and 2D Cu10 cluster-assembled materials as a recyclable luminescent probe for highly sensitive creatinine detection. Nanoscale. 2026;18(14):7789–7801. doi: 10.1039/D5NR05249A. [DOI] [PubMed] [Google Scholar]
- Xu Y., Dong Q.-G., Dong J.-P., Zhang H., Li B., Wang R., Zang S.-Q.. Assembly of Copper-Clusters into a Framework: Enhancing the Structural Stability and Photocatalytic HER Performance. Chem. Commun. 2023;59:3067–3070. doi: 10.1039/D3CC00275F. [DOI] [PubMed] [Google Scholar]
- Wang Q.-Y., Zhang L., He W.-M., Yang L., Zhang C., Wang Z.-Y., Zhang R., Chen J.-H., Wang S., Zang S.-Q., Mak T. C. W.. High-Performance Primary Explosives Derived from Copper Thiolate Cluster-Assembled Materials for Micro-Initiating Device. Chem. Eng. J. 2020;389:124455. doi: 10.1016/j.cej.2020.124455. [DOI] [Google Scholar]
- Liu Z., Fang J.-J., Wang Z.-Y., Xie Y.-P., Lu X.. Assembly of Copper Alkynyl Clusters into Dimensionally Variable and Luminescent Frameworks. Inorg. Chem. 2024;63(24):11146–11154. doi: 10.1021/acs.inorgchem.4c00822. [DOI] [PubMed] [Google Scholar]
- Sakai J., Biswas S., Irie T., Mabuchi H., Sekine T., Niihori Y., Das S., Negishi Y.. Synthesis and Luminescence Properties of Two Silver Cluster-Assembled Materials for Selective Fe3+ Sensing. Nanoscale. 2023;15:12227–12234. doi: 10.1039/D3NR01920A. [DOI] [PubMed] [Google Scholar]
- Nakatani R., Biswas S., Irie T., Sakai J., Hirayama D., Kawawaki T., Niihori Y., Das S., Negishi Y.. A New Two-Dimensional Luminescent Ag12 Cluster-Assembled Material and Its Catalytic Activity for Reduction of Hexacyanoferrate(III) Nanoscale. 2023;15:16299–16306. doi: 10.1039/D3NR03343K. [DOI] [PubMed] [Google Scholar]
- Nakatani R., Biswas S., Irie T., Niihori Y., Das S., Negishi Y.. Unlocking the Potential of an Atom-Precise Ag12 Cluster Assembled Material as a Highly Efficient SERS Sensor for the Detection of Hg2+ Ions. ACS Mater. Lett. 2024;6(2):438–445. doi: 10.1021/acsmaterialslett.3c01198. [DOI] [Google Scholar]
- Sakai J., Sasaki K., Nakatani R., Das S., Negishi Y.. A Silver Cluster-Assembled Material as a Matrix for Enzyme Immobilization towards a Highly Efficient Biocatalyst. Nanoscale. 2024;16:21767–21775. doi: 10.1039/D4NR02506G. [DOI] [PubMed] [Google Scholar]
- Nakatani R., Sakai J., Saha A., Kondo A., Tomioka R., Kawawaki T., Das S., Negishi Y.. Designed construction of two new atom-precise three-dimensional and two-dimensional Ag12 cluster-assembled materials. Nanoscale. 2025;17:813–822. doi: 10.1039/D4NR03992K. [DOI] [PubMed] [Google Scholar]
- Irie T., Sasaki K., Nozaki M., Kawawaki T., Takahashi S., Das S., Negishi Y.. Reticular Construction of a Two-Dimensional Ag12 Cluster-Assembled Material for Rapid and Selective Luminescent Sensing of Antibiotics in Aqueous Media. Nanoscale. 2025;17:25495–25504. doi: 10.1039/D5NR03351A. [DOI] [PubMed] [Google Scholar]
- Irie T., Nakatani R., Sreelakshmi K. S., Kondo A., Kawawaki T., Takahashi S., Das S., Negishi Y.. Precision Ladder-Structured Silver Cluster Assembly Bridged by Tetrazine Linkers for Highly Sensitive and Selective Luminescence Recognition of Amino Acids. Nanoscale. 2026;18:8789–8800. doi: 10.1039/D6NR00302H. [DOI] [PubMed] [Google Scholar]
- Han J., Bai X., Xu X., Bai X., Husile A., Zhang S., Qia L., Guan J.. Advances and challenges in the electrochemical reduction of carbon dioxide. Chem. Sci. 2024;15(21):7870–7907. doi: 10.1039/D4SC01931H. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dattila F., Seemakurthi R. R., Zhou Y., López N.. Modeling Operando Electrochemical CO2 Reduction. Chem. Rev. 2022;122(12):11085–11130. doi: 10.1021/acs.chemrev.1c00690. [DOI] [PubMed] [Google Scholar]
- Nitopi S., Bertheussen E., Scott S. B., Liu X., Engstfeld A. K., Horch S., Seger B., Stephens I. E. L., Chan K., Hahn C., Nørskov J. K., Jaramillo T. F., Chorkendorff I.. Progress and Perspectives of Electrochemical CO2 Reduction on Copper in Aqueous Electrolyte. Nanoscale. 2019;119(12):7610–7672. doi: 10.1021/acs.chemrev.8b00705. [DOI] [PubMed] [Google Scholar]
- Xiao C., Zhang J.. Architectural Design for Enhanced C2 Product Selectivity in Electrochemical CO2 Reduction Using Cu-Based Catalysts: A Review. ACS Nano. 2021;15(5):7975–8000. doi: 10.1021/acsnano.0c10697. [DOI] [PubMed] [Google Scholar]
- Li T., Sun R., Yu M.. Strategies for Improving Product Selectivity in Electrocatalytic Carbon Dioxide Reduction Using Copper-Based Catalysts. Adv. Funct. Mater. 2024;34(51):2410186. doi: 10.1002/adfm.202410186. [DOI] [Google Scholar]
- Dinh K. H., Menisa L. T., Warkentin H., Nguyen T. N., Dinh C.-T.. Metal Cluster Catalysts for Electrochemical CO2 Reduction. ACS Catal. 2025;15(7):5731–5759. doi: 10.1021/acscatal.4c07952. [DOI] [Google Scholar]
- Yang D., Wang J., Wang Q., Yuan Z., Dai Y., Zhou C., Wan X., Zhang Q., Yang Y.. Electrocatalytic CO2 Reduction over Atomically Precise Metal Nanoclusters Protected by Organic Ligands. ACS Nano. 2022;16(10):15681–15704. doi: 10.1021/acsnano.2c06059. [DOI] [PubMed] [Google Scholar]
- Kawawaki T., Okada T., Hirayama D., Negishi Y.. Atomically precise metal nanoclusters as catalysts for electrocatalytic CO2 reduction. Green Chem. 2024;26:122–163. doi: 10.1039/D3GC02281A. [DOI] [Google Scholar]
- Kawawaki T., Okada T., Takemae K., Tomihari S., Negishi Y.. Size-Dependent Carbon Dioxide Reduction Activity of Copper Nanoparticle and Nanocluster Electrocatalysts. ChemNanomater. 2024;10(5):e202300575. doi: 10.1002/cnma.202300575. [DOI] [Google Scholar]
- Wu Q.-J., Si D.-H., Sun P.-P., Dong Y.-L., Zheng S., Chen Q., Ye S.-H., Sun D., Cao R., Huang Y.-B.. Atomically Precise Copper Nanoclusters for Highly Efficient Electroreduction of CO2 towards Hydrocarbons via Breaking the Coordination Symmetry of Cu Site. Angew. Chem., Int. Ed. 2023;62(36):e202306822. doi: 10.1002/anie.202306822. [DOI] [PubMed] [Google Scholar]
- Li J.-K., Dong J.-P., Liu S.-S., Hua Y., Zhao X.-L., Li Z., Zhao S.-N., Zang S.-Q., Wang R.. Promoting CO2 Electroreduction to Hydrocarbon Products via Sulfur-Enhanced Proton Feeding in Atomically Precise Thiolate-Protected Cu Clusters. Angew. Chem., Int. Ed. 2024;63(48):e202412144. doi: 10.1002/anie.202412144. [DOI] [PubMed] [Google Scholar]
- Biswas S., Shingyouchi Y., Kamiyama M., Ogami M., Song H., Li B., Wang S., Kawawaki T., Jiang D.-E., Negishi Y.. Atomically Precise [Cu23H4(SC7H7)18(PPh3)6] Nanocluster: Structural Integration of Johnson Solids through a Cu(0) Center and Electrocatalytic Functionality. J. Am. Chem. Soc. 2025;147(27):23733–23742. doi: 10.1021/jacs.5c05665. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Liu L.-J., Wang Z.-Y., Wang Z.-Y., Wang R., Zang S.-Q., Mak T. C. W.. Mediating CO2 Electroreduction Activity and Selectivity over Atomically Precise Copper Clusters. Angew. Chem., Int. Ed. 2022;61(35):e202205626. doi: 10.1002/anie.202205626. [DOI] [PubMed] [Google Scholar]
- Shingyouchi Y., Ogami M., Biswas S., Tanaka T., Kamiyama M., Ikeda K., Hossain S., Yoshigoe Y., Osborn D. J., Metha G. F., Kawawaki T., Negishi Y.. Ligand-Dependent Intracluster Interactions in Electrochemical CO2 Reduction Using Cu14 Nanoclusters. Small. 2025;21(16):2409910. doi: 10.1002/smll.202409910. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Han B.-L., Zhao L.-C., Yuan Z.-R., Wang Z., Yu Q., Luo G.-G., Wang L.-K., Tung C.-H., Sun D.. Dipropyne-Modified N-Heterocyclic Carbene Stabilized Atomically Precise Copper(I) Nanocluster Catalysts for CO2 Electroreduction. Adv. Funct. Mater. 2025;35(29):2500149. doi: 10.1002/adfm.202500149. [DOI] [Google Scholar]
- Zhu J., Chen J.-L., Qi X.-Z., Zhao J.-W., Jiang X.-C., Li W.-X., Liu J.-X.. Electron-Rich Subnanometer Cu Clusters Facilitate CO–CO Coupling in CO2 Electroreduction. J. Am. Chem. Soc. 2026;148(4):4008–4019. doi: 10.1021/jacs.5c12495. [DOI] [PubMed] [Google Scholar]
- Ding M., Tang L., Ma X., Song C., Wang S.. Effects of ligand tuning and core doping of atomically precise copper nanoclusters on CO2 electroreduction selectivity. Commun. Chem. 2022;5(1):172. doi: 10.1038/s42004-022-00779-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Biswas S., Tanaka T., Song H., Ogami M., Shingyouchi Y., Hossian S., Kamiyama M., Kosaka T., Nakatani R., Niihori Y., Das S., Kawawaki T., Jiang D.-E., Negishi Y.. Highly Selective Methanol Synthesis Using Electrochemical CO 2 Reduction with Defect-Engineered Cu 58 Nanoclusters. Small Sci. 2025;5(2):2400465. doi: 10.1002/smsc.202400465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ge T., Chen R., Liu X., Jia Z., Liu C., Huang J., Sun Y.. Steering CO2 Electroreduction to Methane with an Atomically Precise Copper Nanocluster. ACS Catal. 2026;16(4):3758–3767. doi: 10.1021/acscatal.5c08305. [DOI] [Google Scholar]
- Biswas S., Shingyouchi Y., Kamiyama M., Jena M. K., Ogami M., Kawawaki T., Pathak B., Negishi Y.. Deciphering Electrocatalytic Activity in Cu Nanoclusters: Interplay Between Structural Confinement and Ligands Environment. Small. 2025;21(25):2500302. doi: 10.1002/smll.202500302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang B., Tan X., Zhang Z.-H., Liu C.-Y., Hu F., Hu H.-S., Chen C., Wang Q.-M.. Atomically Precise Copper Nanoclusters with Cu-N Interfaces Toward Efficient CO2-to-CH4 Electrocatalysis. CCS Chem. 2026;8:1122–1129. doi: 10.31635/ccschem.025.202505606. [DOI] [Google Scholar]
- Patel S., McKelvey K., Liu L.. Metal–Organic Frameworks for Electrocatalytic CO2 Reduction: Developments and Prospects. Chem. Mater. 2024;36(20):10054–10087. doi: 10.1021/acs.chemmater.4c01137. [DOI] [Google Scholar]
- Zhang C., Lin Z., Jiao L., Jiang H.-L.. Metal–Organic Frameworks for Electrocatalytic CO2 Reduction: From Catalytic Site Design to Microenvironment Modulation. Angew. Chem., Int. Ed. 2024;63(50):e202414506. doi: 10.1002/anie.202414506. [DOI] [PubMed] [Google Scholar]
- Shah S. S. A., Najam T., Wen M., Zang S.-Q., Waseem A., Jiang H.-L.. Metal–Organic Framework-Based Electrocatalysts for CO 2 Reduction. Small Struct. 2022;3(5):2100090. doi: 10.1002/sstr.202100090. [DOI] [Google Scholar]
- Tang M.-H., Yang Z.-W., Xu H., Zhao B.. Strategies to Optimize Metal–Organic Frameworks for Efficient Electrocatalytic CO2 Reduction. Adv. Funct. Mater. 2026;36(1):e13412. doi: 10.1002/adfm.202513412. [DOI] [Google Scholar]
- Kornienko N., Zhao Y., Kley C. S., Zhu C., Kim D., Lin S., Chang C. J., Yaghi O. M., Yang P.. Metal–Organic Frameworks for Electrocatalytic Reduction of Carbon Dioxide. J. Am. Chem. Soc. 2015;137(44):14129–14135. doi: 10.1021/jacs.5b08212. [DOI] [PubMed] [Google Scholar]
- Hod I., Sampson M. D., Deria P., Kubiak C. P., Farha O. K., Hupp J. T.. Fe-Porphyrin-Based Metal–Organic Framework Films as High-Surface Concentration, Heterogeneous Catalysts for Electrochemical Reduction of CO2 . ACS Catal. 2015;5(11):6302–6309. doi: 10.1021/acscatal.5b01767. [DOI] [Google Scholar]
- D’alessandro D. M.. Exploiting redox activity in metal–organic frameworks: concepts, trends and perspectives. Chem. Commun. 2016;52:8957–8971. doi: 10.1039/C6CC00805D. [DOI] [PubMed] [Google Scholar]
- Huang Z., Eder D.. Harnessing the structural evolution of metal–organic frameworks under electrocatalytic conditions. Commun. Chem. 2025;8:359. doi: 10.1038/s42004-025-01747-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Prabhu P., Jose V., Lee J.-M.. Heterostructured Catalysts for Electrocatalytic and Photocatalytic Carbon Dioxide Reduction. Adv. Funct. Mater. 2020;30(24):1910768. doi: 10.1002/adfm.201910768. [DOI] [Google Scholar]
- Burgers I., Pérez-Gallent E., Goetheer E., Kortlever R.. Electrochemical CO2 Reduction on Copper in Propylene Carbonate: Influence of Water Content and Temperature on the Product Distribution. Energy Technol. 2023;11(8):2201465. doi: 10.1002/ente.202201465. [DOI] [Google Scholar]
- Song R.-B., Zhu W., Fu J., Chen Y., Liu L., Zhang J.-R., Lin Y., Zhu J.-J.. Electrode Materials Engineering in Electrocatalytic CO2 Reduction: Energy Input and Conversion Efficiency. Adv. Mater. 2020;32(27):1903796. doi: 10.1002/adma.201903796. [DOI] [PubMed] [Google Scholar]
- Biswas S., Pal A., Jena M. K., Hossain S., Sakai J., Das S., Sahoo B., Pathak B., Negishi Y.. Luminescent Hydride-Free [Cu7(SC5H9)7(PPh3)3] Nanocluster: Facilitating Highly Selective C–C Bond Formation. J. Am. Chem. Soc. 2024;146(30):20937–20944. doi: 10.1021/jacs.4c05678. [DOI] [PubMed] [Google Scholar]
- Giannozzi P., Baseggio O., Bonfà P., Brunato D., Car R., Carnimeo I., Cavazzoni C., de Gironcoli S., Delugas P., Ferrari Ruffino F., Ferretti A., Marzari N., Timrov I., Urru A., Baroni S.. Quantum ESPRESSO toward the Exascale. J. Chem. Phys. 2020;152(15):154105. doi: 10.1063/5.0005082. [DOI] [PubMed] [Google Scholar]
- Giannozzi P., Baroni S., Bonini N., Calandra M., Car R., Cavazzoni C., Ceresoli D., Chiarotti G. L., Cococcioni M., Dabo I., Dal Corso A., de Gironcoli S., Fabris S., Fratesi G., Gebauer R., Gerstmann U., Gougoussis C., Kokalj A., Lazzeri M., Martin-Samos L., Marzari N., Mauri F., Mazzarello R., Paolini S., Pasquarello A., Paulatto L., Sbraccia C., Scandolo S., Sclauzero G., Seitsonen A. P., Smogunov A., Umari P., Wentzcovitch R. M.. QUANTUM ESPRESSO: A Modular and Open-Source Software Project for Quantum Simulations of Materials. J. Phys.: Condens. Matter. 2009;21:395502. doi: 10.1088/0953-8984/21/39/395502. [DOI] [PubMed] [Google Scholar]
- Giannozzi P., Andreussi O., Brumme T., Bunau O., Buongiorno Nardelli M., Calandra M., Car R., Cavazzoni C., Ceresoli D., Cococcioni M., Colonna N., Carnimeo I., Dal Corso A., de Gironcoli S., Delugas P., DiStasio R. A. Jr, Ferretti A., Floris A., Fratesi G., Fugallo G., Gebauer R., Gerstmann U., Giustino F., Gorni T., Jia J., Kawamura M., Ko H.-Y., Kokalj A., Küçükbenli E., Lazzeri M., Marsili M., Marzari N., Mauri F., Nguyen N. L., Nguyen H.-V., Otero-de-la-Roza A., Paulatto L., Poncé S., Rocca D., Sabatini R., Santra B., Schlipf M., Seitsonen A. P., Smogunov A., Timrov I., Thonhauser T., Umari P., Vast N., Wu X., Baroni S.. Advanced Capabilities for Materials Modelling with Quantum ESPRESSO. J. Phys.: Condens. Matter. 2017;29:465901. doi: 10.1088/1361-648X/aa8f79. [DOI] [PubMed] [Google Scholar]
- Perdew J. P., Burke K., Ernzerhof M.. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996;77:3865–3868. doi: 10.1103/PhysRevLett.77.3865. [DOI] [PubMed] [Google Scholar]
- Monkhorst H. J., Pack J. D.. Special Points for Brillouin-Zone Integrations. Phys. Rev. B. 1976;13:5188–5192. doi: 10.1103/PhysRevB.13.5188. [DOI] [Google Scholar]
- Grimme S., Antony J., Ehrlich S., Krieg H.. A consistent and accurate ab Initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 2010;132(15):154104. doi: 10.1063/1.3382344. [DOI] [PubMed] [Google Scholar]
- Grimme S., Ehrlich S., Goerigk L.. Effect of the Damping Function in Dispersion Corrected Density Functional Theory. J. Comput. Chem. 2011;32(7):1456–1465. doi: 10.1002/jcc.21759. [DOI] [PubMed] [Google Scholar]
- Kohlrausch E. C., Centurion H. A., Lodge R. W., Luo X., Slater T., Santos M. J. L., Ling S., Mastelaro V. R., Cliffe M. J., Goncalves R. V., Alves Fernandes J. A.. A High-Throughput, Solvent-Free Method for Dispersing Metal Atoms Directly onto Supports. J. Mater. Chem. A. 2021;9(47):26676–26679. doi: 10.1039/D1TA08372D. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
