id
stringlengths 24
24
| idx
int64 0
402
| paragraph
stringlengths 106
17.2k
|
|---|---|---|
65c9f0bf9138d23161ecc7b7 | 25 | We have reported an automated cavity calculation software, C3, deployed as a Python module, for the calculation of the cavity of cages, as well as aromatic contacts, solvent accessible surface area (SASA), hydrophobicity (MHP), and/or electrostatic potential (ESP) contributions. The main advantage of our method is its easy use and general applicability to a wide range of porous structures without the need for parameter adjustment. Users can select the grid spacing to achieve the desired cavity resolution and reduce calculation time by using larger grid spacing. The method was benchmarked on a wide range of cage structures. |
65c9f0bf9138d23161ecc7b7 | 26 | The cavity can be visualized with any chemical visualization software as the cavity output is stored in a PDB file containing the cavity grid points. To facilitate the use of the algorithm for nonspecialized users, a plugin for molecular viewer PyMol was developed, enabling its use without requiring computer programming knowledge. We anticipate that the developed software will streamline the characterization of molecular cages and speed up the development of novel functional designs. |
6617b71821291e5d1d9dc482 | 0 | This makes them a promising carbon-neutral energy option, especially for the longdistance and heavy-duty transportation sector, when combined with hydrogen produced via electrolysis and renewable electricity . However, to date, water management is one of the technical bottlenecks that must be resolved to ensure high performance and long durability, as the role of water is fundamentally linked to fuel cell operation. |
6617b71821291e5d1d9dc482 | 1 | Accumulation of liquid water prevents gas transport inside the gas diffusion electrode (GDE) and flow-field channels, referred to as mass transport losses, whilst too low membrane water content increases the Ohmic resistance and heating . Many studies have attempted to determine the hydration state of the fuel cell based on external signals, including cell voltage distribution , electrochemical impedance spectroscopy (EIS) , pressure drop , acoustic emission and magnetic field . |
6617b71821291e5d1d9dc482 | 2 | Mathematical models are commonly employed to estimate the water profile within a fuel cell, typically incorporating various assumptions, as experiments are costly and limited. A typical one-dimensional model presented by Hu et al. describes the water profile in all fuel cell components along the through-plane direction. The measured Ohmic resistance is compared to the membrane water content estimated by the model, revealing a consistent trend. To reduce the computational effort due to the complexity of the model, Liso et al. introduced a zero-dimensional model that is less computationally intensive to predict the water distribution, and can support operational decisions related to external humidification. Although the model matched the experimental data, its capability was subjected to the strong assumption that liquid water is indistinguishable from vapor, and, therefore, the model can only be applied in the case of small amounts of excess water. To date, modelling challenges still exist as the coupled reactions and non-uniform transportation in fuel cells are difficult to model. Some assumptions and simplifications are adopted during simulation and, consequently, deviations between experimental and modeling results are hard to avoid . |
6617b71821291e5d1d9dc482 | 3 | EIS stands out as an established tool, perhaps the most widely used characterisation method in many electrochemical devices, for investigating the various processes occurring inside fuel cells in a non-destructive way through current or voltage perturbation, which is intensively applied for characterisation, control and diagnosis of water management failures . The total cell resistance is divided into three parts with different time constants in the frequency domain: Ohmic resistance, charge transfer resistance, and mass transport resistance . Protons can be transported more easily through water-filled clusters or ionic channels after the membrane has absorbed more water, resulting in reduced Ohmic resistance, where membrane resistance is the main contributor. Similarly, the charge transfer resistance decreases with increasing relative humidity (RH), given the crucial role of water content in facilitating proton conduction at the triple-phase boundary in the catalyst layer . The mass transport resistance is also strongly correlated to water inside the cell, as flooding in the gas diffusion layer (GDL) will hinder the transport of reactants through the porous structure and catalyst sites . Toyota has successfully validated and implemented a water content control system based on AC impedance measurement , making a -30℃ cold start possible. A high resistance corresponds to reduced water content, leading to a decline in average voltage; when the membrane is humidified to saturation, the voltage becomes unstable. Keeping the resistance within a certain range helps the fuel cell system operate efficiently and stably. It is still possible to increase the accuracy of the water content control, because only Ohmic resistance is used in this study , |
6617b71821291e5d1d9dc482 | 4 | considering that beyond a certain threshold in water content, the membrane resistance will not continue to decrease. Additionally, the specific value of the water content is not directly given due to the complexity of water content measurement. Instead, the fluctuation in cell voltage is used to reflect the change in water content. Hence, there is an urgent need to study the mapping relationship between the microscopic water management state inside a fuel cell and the macroscopic impedance information. |
6617b71821291e5d1d9dc482 | 5 | neutron imaging , magnetic resonance imaging , and transparent endplate design have been reported to study water content formation, distribution and removal. However, practical limitations exist for some of these approaches, often necessitating modifications to existing designs and materials. For instance, endplates, typically made of coated steel to improve weight-specific power density, pose challenges for X-ray radiography due to a large attenuation difference between light materials and dense materials . As a result, transparent X-ray windows need to be created by removing the current collector and endplate material, which could have implications for the local performance of the cell in the area of measurement . |
6617b71821291e5d1d9dc482 | 6 | Additionally, X-ray imaging of water transport can be compromised due to the poor contrast between water and fuel cell materials such as carbon fibre. Among the various microscopy options, neutron imaging is a powerful technique that is sensitive to low density materials, which allows direct mapping of the water content in operating fuel cells and provides quantitative data. Notably, neutron imaging demonstrates a superior ability to penetrate dense structural materials without the need for significant modifications, making it versatile for applications ranging from small-scale lto full-size single cells to stacks containing multiple cells. Neutron imaging has been extensively used to investigate water management-related problems, exploring the impact of cell structure design and compression , operation conditions, materials and degradation . Lee et al. conducted neutron imaging on an operating PEFC under various humidity levels, in parallel with impedance measurements at high frequencies, to determine the membrane resistance. The results confirmed that the Ohmic resistance increases with decreasing RH levels, and that there is an inverse relationship between the Ohmic resistance and the membrane water content, in line with earlier discussions. Despite its advantages, neutron imaging is not a realistic option for online implementation of water content characterisation in vehicular or stationary fuel cell systems due to the requirement for nuclear reactors or spallation sources to produce neutrons, and it is likely to remain confined to specialist laboratories and beamlines for the foreseeable future. |
6617b71821291e5d1d9dc482 | 7 | Correlating complex characterisation methods with simpler ones that are more widely available and implementable can be a route to facilitate the monitoring of electrochemical systems. For example, Fordham et al. correlated spatially-resolved synchrotron X-ray diffraction (SXRD) and laboratory X-ray computed tomography (CT) measurements with thermal imaging and acoustic mapping of automotive battery cells to study heterogeneous effects of ageing, opening the way to the use of only acoustic signals once fully correlated with the richer but difficult to carry out X-ray techniques. |
6617b71821291e5d1d9dc482 | 8 | Machine learning (ML) approaches can mine patterns that are difficult to draw intuitively from experimental datasets . Specifically, a function can be learned from a training dataset that contains the inputs and the desired outputs. An onboard assessment of water content could become a reality if a similar approach could be taken to marry an accurate assessment of water content in fuel cells with electrical signals. |
6617b71821291e5d1d9dc482 | 9 | To this end, synchronous operando neutron imaging and EIS measurements were conducted in this study, manipulating the water content of the fuel cell by switching between dry and wet gases, which has previously been shown to provide a boost to fuel cell performance . Dual datasets are generated in which neutron measurements correspond to impedance spectra at different water content levels. Given the nonlinear relationship between impedance spectra and water content, an artificial neural network (ANN) was employed to establish this intricate relationship. Neutron imaging can provide accurate water distribution with good spatial and temporal resolution, while EIS is a universal, cost-effective tool in laboratories, a version of which could even be implemented into real-time and rapid onboard diagnostics of fuel cells . The integration of neutron imaging and EIS is intended to leverage the strengths of both techniques. By ensuring accurate correlation of the information provided by each method, we aim to develop a cheaper, more readily available, and onboard approach for real-time estimation of water content in a fuel cell stack. |
6617b71821291e5d1d9dc482 | 10 | be seen in Fig. . Following the initiation of air supply with 100% RH, the cell voltage steadily decreases from ca. 0.670 V to ca. 0.616 V, which is attributed to the accumulation of liquid water in the GDE. Subsequently, after switching to dry air, reactant transport is facilitated by the drainage effect of the dry gas, coupled with a pause in external humidification, leading to an increase in cell voltage to ca. 0.638 V. |
6617b71821291e5d1d9dc482 | 11 | Voltage decreases once again upon switching back to fully humified air. Synchronous measurements of EIS and neutron imaging were conducted and repeated throughout this process. The impact of dry-wet gas switching on the overall water content within the fuel cell is clearly visible in Fig. . Overall water content steadily increases with humidified gas supply and decreases with the introduction of dry gas. The fluctuation trend of cell voltage exhibits a reverse pattern compared to the evolution of overall water content. The changing water content can be easily realised by dry-wet gas switching, which is also faster than the RH control. thickness was calculated can be seen in the neutron imaging collection and data analysis part of the experimental procedures section. The water mapping is rotated in 3D, and the height and colour scale are proportional to the water thickness (mm). |
6617b71821291e5d1d9dc482 | 12 | The EIS measurement results at different experimental stages, namely at the beginning of the experiment (#1), at peak water content (#2), and at the lowest water content point after switching to dry gas for around 15 mins (#3), are illustrated in Fig. . The water content measured by neutron imaging and assessed by attenuation of the neutron beam at various positions is shown in Fig. (the corresponding total water distribution across the entire active area, the height of the 3D mapping is proportional to the water thickness in mm). For #1, the internal water content is approximately 0.022 g and exhibits a uniform distribution as little water is present. The Nyquist plot confirms this by having a larger intersection with the real axis (higher membrane resistance) and smaller arc (indicating low mass transfer resistance). In contrast, for #2, a significant amount of liquid water (0.213 g, an order of magnitude higher) is observed, primarily present in the channel regions. Although the intersection with the real axis shifts to the left at this stage, the impedance points in the low-frequency region, both on the real and imaginary axes, show a larger value, corresponding to a lower cell voltage (Fig. ). |
6617b71821291e5d1d9dc482 | 13 | Synchronous measurement of neutron imaging and EIS was conducted continuously and each image/EIS spectrum is referred to a sample with a number, such that increasing sample numbers represent the evolution of the experiment. Fig. illustrates the membrane resistance, anodic charge transfer resistance, and cathodic double-layer capacitance fitted through equivalent circuit modelling. Proton transport resistance is inversely proportional to water content before membrane saturation, leading to a continuous decrease in membrane resistance from 1.25 mΩ as overall water content increases. Consequently, up to sample 29, the two variables exhibit a linear inverse relationship. This relationship can be quantified using correlation coefficients ranging between 1 and -1, with 1 indicating perfect positive correlation, 0 representing no correlation and -1 denoting perfect negative correlation. Fig. shows the correlation coefficients between water content and membrane resistance for different sample sizes. When computed using the first 29 samples, the correlation coefficient is -0.823, reaching as low as -0.876 using 16 samples. However, membrane resistance ceases to decrease after saturation, achieving a minimum value of approximately 1 mΩ. |
6617b71821291e5d1d9dc482 | 14 | Following the switch to dry gas after sample 59 (#2), membrane resistance rises to 1.08 mΩ and subsequently decreases upon switching back to humidified air. Similar trends are observed for anodic charge transfer resistance and cathodic double-layer capacitance, as increased water content adversely affects proton transport in the ionomer. The correlation coefficients between these parameters and water content are depicted in Fig. and F, with 0.666 and 0.655 for anodic charge transfer resistance and cathodic double-layer capacitance, respectively, when considering all samples. This robust correlation supports impedance spectroscopy-based water content estimation, signifying a nonlinear relationship between EIS measurement and overall water content. |
6617b71821291e5d1d9dc482 | 15 | Due to the distribution of water within the cell, including in the membrane, GDE and channels, total water content is segmented based on the geometric structure of the cathode flow-field into channel regions and rib regions. The water distribution in the rib region for representative stages #1, #2, and #3 is shown in Fig. , while the water distribution in the corresponding channel regions is mapped in Fig. . Almost all water is observed in the channel region. Similar distributions are observed for other sample points (see Fig. ), with a high correlation coefficient of 0.992 between channel region and overall water content. This value for the rib region is 0.829 (rib region water content changes at least follow the overall water content). The significant presence of water in the channels is due to changes in internal water content primarily achieved by directly switching between dry and humidified gases; under 100% RH, a large amount of water accumulates at the edges of the channels (Fig. ). |
6617b71821291e5d1d9dc482 | 16 | Correlations between fitted impedance and water content for channels and rib regions are also calculated (as shown in Fig. ) to assess whether EIS can predict water content in the channels. The correlation between channels and impedance is almost identical to that with overall water content, while the correlation between rib regions' water content and impedance is weak. This may be attributed to the fact that introducing water directly or switching to dry gas initially changes the water content in the channel region, as diffusion of water to the rib region requires additional time. |
6617b71821291e5d1d9dc482 | 17 | Artificial neural networks (ANNs) are powerful tools to discover correlations in The regression results for training, validation, test and all datasets are shown in Fig. , in which the R-value is the coefficient of correlation representing how well the ANN prediction matches the real water content. R is 1 for a perfect fit, where all the data fall on a 45° line, while 0 means no correlation. R for the training data has been found to be around 0.917, indicating an excellent correlation. Although not all points fall on the 45° line, the generalisation performance is guaranteed as no overfitting happened. |
6617b71821291e5d1d9dc482 | 18 | Achieving a perfect fit with an R-value of 1 indicates that the model reproduces the training data perfectly but does not guarantee good performance on unseen data. However, the performance of ANN needs to be evaluated based on the ability to predict unseen data. When the network performance is optimal at epoch 2, the corresponding R-value of the validation data set is 0.921, and the R-value of the test set is 0.884. Due to the network's initialisation and the training data's random partition, fitting results may exhibit slight variations. |
6617b71821291e5d1d9dc482 | 19 | Even for training samples, there is a relatively large error in the predicted values for samples with higher water content. This is due to the unbalanced distribution of samples, with most samples concentrated in regions with lower water content and fewer in regions with higher water content. The typical accurately predicted points and errors, as illustrated in Fig. for the test regression curve, further confirm this observation. |
6617b71821291e5d1d9dc482 | 20 | Through the nonlinear fitting capability of ML, such as ANN, even for data that have not been previously trained, ANN can predict the overall water content based on impedance information (test samples illustrated in the Fig. and). After collecting neutron imaging and electrochemical impedance data for varying water content and training the ML model, the water content information from costly and intricate neutron imaging experiment can be linked to the EIS. In onboard systems, real-time water content estimation can thus be achieved by using simple, accessible impedance measurements when appropriate ML models have been constructed and pre-trained. This represents significant progress in the capability of simple electronic signals to perform advanced diagnostics in electrochemical systems. Estimated water content can be directly used as a feedback control parameter for water management strategies. |
6617b71821291e5d1d9dc482 | 21 | When the water content deviates from a pre-defined certain range, humidification or purge operation can be performed. With the development of localised EIS, spatially resolved water mapping can potentially be inferred from the electrochemical measurement alone, with previously correlated neutron imaging and ML providing the model to relate EIS to local water content. A promising way of revealing the fundamental mechanisms underlying local performance disparities is therefore possible, facilitating optimisation of fuel cell structure design, manufacturing processes, and operational conditions. This deviation in water content prediction arises from the coupling factors of the changing water content method and the long-time measurement in this study. When introducing a fully humidified reactant directly, the change in water content is rapid. |
6617b71821291e5d1d9dc482 | 22 | However, for synchronous EIS and neutron measurements, this process takes a longer time. The water content inside the cell undergoes certain changes during the measurement process. This limitation provides insights for future improvement. Firstly, use a more complex and precise humidity control to change the water content inside the cell and perform measurements under steady-state conditions. Secondly, synchronous measurements of multi-frequency AC voltage responses and faster neutron imaging should be considered to ensure stability. |
6617b71821291e5d1d9dc482 | 23 | In this study, the manipulation of water content within a PEM fuel cell was achieved by switching between dry and humidified air. When humidified gas was supplied, the overall water content within the cell continuously increased, and, after switching to dry air, the water content started to decrease. Communication between the potentiostat and neutron instrument enabled neutron imaging and EIS measurement synchronisation, and further data correlation and decoupling post analysis. |
6617b71821291e5d1d9dc482 | 24 | As the overall water content increased, the membrane resistance decreased; however, after saturation, the membrane resistance ceased to decline. This phenomenon is reflected in the increase of the correlation coefficient between the water content and membrane resistance from the lowest point of -0.876 to -0.612. The anodic charge transfer resistance and cathodic double-layer capacitance exhibited a proportional relationship with overall water content, with correlation coefficients of 0.666 and 0.655 when using all samples. This strong correlation suggests a nonlinear relationship between impedance spectra and overall water content. Through the nonlinear fitting capability of an ANN model using real and imaginary parts of all impedance points as the input, the correlation coefficient for 75% of the training data reached 0.917. The ANN model demonstrated a comparable correlation coefficient of around 0.884, even for untrained test samples. With the AI-assisted technology, it is possible to establish a correlative relationship between a high-end diagnostic instrument in a national facility and a low-cost tool in the laboratory that offers a viable pathway to predict the water content without expensive tools in future studies. The potential to predict water content through EIS offers a promising way for developing more effective onboard water management, and could provide crucial theoretical support for subsequent research on fuel cell performance degradation, fault diagnosis and lifespan prediction to be conducted independently by neutron imaging equipments. |
6617b71821291e5d1d9dc482 | 25 | A single cell with a 100 cm² active area (explosion diagram depicted in Fig. was customised specifically for through-plane neutron imaging experiments, with the membrane electrode assembly (MEA) oriented perpendicular to the neutron beam. The endplates, having a thickness of 25 mm, were made from aluminium to mitigate neutron absorption and anodised to enhance electrical insulation. Silicon sheets interposed between the endplate and the gold-coated current collector plate provided further electrical isolation. The 6-mm thick graphite flow-field plates were machined with multiple serpentine flows, featuring 7 channels and 6 bends. The channel dimensions were as follows: channel width -1 mm, channel depth -1 mm, and rib width -1 mm. |
6617b71821291e5d1d9dc482 | 26 | Cooling channels were engraved on the reverse side of the cathode flow-field plate, given the predominant heat generated during oxygen reduction reactions (ORR). Gas and coolant tightness was ensured by 1.2-mm thick EPDM gaskets, facilitating appropriate MEA compression. Sheathed in PVC tubing, 12 bolts were positioned along the endplate periphery. Uniform compression was achieved by incrementally increasing torque to 6 N m. Gases and coolant were both supplied from the anode endplate side to make the fuel cell position close to the neutron scintillator of the CMOS camera box. |
6617b71821291e5d1d9dc482 | 27 | The in-house fabricated MEA comprised commercial GDEs with a Pt loading of 0.4 mg cm -2 (HyPLAT, South Africa) featuring a Freudenberg H23C9 carbon paper with a 15 μm microporous layer (MPL), subjected to hydrophobic treatment. A 15 μm GORE Select polymer membrane (M820.15, GORE) was sandwiched between the GDEs. The MEA was clamped between steel plates and subjected to hot pressing between two heated plates in a hot press (Carver Inc., USA) at 150 °C for 3 minutes, applying a compression force of 2.75 MPa. Kapton film sheets were placed between the MEA and steel plates to prevent mechanical damage to GDE and membrane, and avoid GDE displacement on both sides. |
6617b71821291e5d1d9dc482 | 28 | Galvanostatic or potentiostatic modes were operated with an electronic load (PLZ664WA, Kukusui). A CompactRIO system with NI 9211 for temperature measurement and NI 9215 for voltage measurement were used to collect analogue signals. The cells operated at a temperature of 60 °C, with heavy water as the coolant (D2O barely attenuates a neutron beam compared to H2O and can only be seen faintly in radiographs), flowing through the back of the cathode flow-field plate via Huber CC-1 heating circulation thermostats. Hydrogen and air stoichiometries were maintained at 1.5 and 3, respectively. Inlet gases were humidified using the bubbling method, regulated by mass flow controllers (Bronkhorst, up to 15 L/min for air and hydrogen, up to 2 L/min for nitrogen). A 3/2-way solenoid valve (JP Fluid Control) on the air side also facilitated wet and dry gas switching. Zero-grade gases were used with the following specifications: nitrogen (99.9% purity), air (99.9% purity) and hydrogen (99.995% purity). |
6617b71821291e5d1d9dc482 | 29 | To achieve synchronous measurements of EIS and neutron imaging, communication was established between Gamry and the neutron experimental control system. Initially, Gamry remained idle and only initiated EIS measurements upon receiving an execution signal from the neutron camera control script. Immediately after an execution signal was sent to Gamry to initiate EIS measurements, the neutron CMOS detector started capturing images continuously in 10 s intervals until completion of EIS measurements, after which Gamry returned to an idle state. The experimental spectra were fitted using the RelaxIS ® software (the fitting result can be seen in Fig. ), and the frequency points above the real axis were selected. The (RC)R(RC) model was used as it has a simple structure and can be easily fitted. The first RC represents the charge transfer process on the cathode side, R in the middle represents the Ohmic impedance, and the second RC depicts the processes on the anode side. |
65039773b338ec988a7e0c2f | 0 | The selective functionalization of unactivated C-H bonds remains one of the greatest challenges in catalysis. Inspiration can be taken from nature, where selective enzymatic C-H halogenation plays a critical role in numerous biochemical processes. In addition to their incorporation in natural products and related biological and medicinal applications, halogenated compounds find a wide variety of uses in therapeutic drugs and agrochemicals. The enzymes that carry out C-H activation are typically a-ketoglutarate (aKG)-dependent enzymes that operate under ambient conditions with strong regioselectivity and stereoselectivity resulting in the catalysis of a wide variety of reactions , including halogenation. For non-heme iron halogenases in particular , e.g., SyrB2 , CytC3 , WelO5 , and BesD, the active sites are similar to other non-heme iron enzymes with only one major change. In comparison to more well understood non-heme iron hydroxylases, the halogenases have a halide coordinating the metal center, i.e., native chloride or non-native bromide. This halide is bound in place of a carboxylate of the 2-His-1-carboxylate facial triad of the hydroxylases because in the halogenases there is no carboxylate-bearing aspartate or glutamate residue present and instead a non-coordinating alanine residue is typically present to create a hydrophobic pocket for halide binding. Despite this understanding of halogenases, it is extremely hard to develop selective processes for C-H halogenation using synthetic methods due to the inertness of C-H bonds and their high bond dissociation energy. |
65039773b338ec988a7e0c2f | 1 | Non-heme iron halogenases natively carry out C-H chlorination , and C-H bromination has been observed in vitro in SyrB2 but with lower efficiency than chlorination. This is corroborated by the abundance of chlorinated and brominated natural products. On the contrary, C-H fluorination has never been observed in any of the non-heme iron rationalize selectivity of some albeit not all non-heme iron halogenases, we also study isomer energetics and isomerization reaction coordinates (RCs) to understand the isomer energetic favorability of key intermediates such as Fe(IV)=O and Fe(III)-OH formed during the catalytic cycle with all three halides. |
65039773b338ec988a7e0c2f | 2 | The active site of WelO5 (PDBID: 5IQS), a representative, carrier-protein-free nonheme iron halogenase, was extracted from the enzyme's crystal structure, following the protocol from prior work. The native halide, i.e., chloride, was replaced with fluoride and bromide to generate three active site models corresponding to the three halides (Supporting Information Figure ). Active site intermediates and corresponding isomers of these active site models containing water (i.e., as in the crystal structure), O2, oxo, hydroxo, and succinate (monodentate and bidentate) ligands were generated by modifying the crystal structure with molSimplify, which uses OpenBabel as a backend. Consistent with prior work, the metal-distal carboxylate oxygen of aKG and succinate ligands along with Nd atoms of His ligands were protonated using Avogadro v1.2.0 to ensure an overall neutral charge for the final active site models (Supporting Information Figure ). All hydrogen atoms added to the active site models were force-field optimized in Avogadro with the UFF force field while holding the heavy atoms fixed. |
65039773b338ec988a7e0c2f | 3 | All geometry optimizations were carried out in ORCA 83 v.4.0.1.2 and v.4.2.1 using density functional theory (DFT) with the generalized gradient approximation (GGA) global hybrid PBE0 and the def2-TZVP basis set and employing semi-empirical D3 dispersion with Becke-Johnson damping. The resolution of the identity (RI) approximation was used, and the auxiliary basis set for def2-TZVP was generated automatically by ORCA to accelerate calculations. Consistent with prior work , the five heavy atoms of succinate and the methyl carbon atoms of His ligands were held fixed to mimic the ligand positions in the enzyme (Supporting Information Figure ). Optimizations were performed in the gas phase and in solution using the BFGS algorithm in redundant internal coordinates. The default thresholds of 5 x 10 -6 hartree for the change in self-consistent field (SCF) energy between steps and 3 x 10 -4 hartree/bohr for the maximum gradient were used. The solution calculations were carried out using the conductor-like polarizable continuum model 94 (C-PCM) solvation energies along with the conductor-like screening solvent model (COSMO) epsilon function type. A solvent dielectric value of e = 10 was used, approximately mimicking the protein environment. As observed in prior work , we find that the inclusion of environment effects through implicit solvent alters most gas-phase geometries very little (i.e., most are within 0.01-0.03 Å while a few outliers have larger changes), thus we carry out further analyses only in the gas phase (Supporting Information Table ). Low-spin (LS) singlet states were simulated in a spin-restricted formalism while non-singlet LS doublet, intermediate-spin (IS), and high-spin (HS) states were simulated in an unrestricted formalism (Supporting Information Table ). All initial and optimized structures for gas-phase and implicitly solvated optimizations are provided in the Supporting Information. |
65039773b338ec988a7e0c2f | 4 | Reaction coordinates (RCs) for isomerization of axial to equatorial (i.e., of oxo/hydroxo/halide ligands) were obtained using the protocol from prior work at the PBE0-D3/def2-TZVP level of theory. These RCs were sampled by rotating the oxo/hydroxo/halide ligand with respect to the axial His in 1° increments of the angle formed by the ligand, the Fe center, and nitrogen of the axial His from 90° to 180° (Supporting Information Figure ). Reaction coordinates for hydrogen atom transfer (HAT) and radical rebound were obtained at the B3LYP-D3/6-31G* level of theory. The B3LYP/6-31G* method/basis set combination was chosen to reduce the computational cost associated with the larger def2-TZVP basis set due to increased system size from the presence of substrate atoms in these calculations and for consistency with prior work. Hydrogen atom transfer (HAT) RCs were obtained using Fe(IV)=O and lysine, which is the substrate of a representative non-heme iron halogenase, BesD . The radical rebound RCs (i.e., with both hydroxo and halogen F, Cl, or Br) were modeled following prior work as the rebound of a 2-methylbutane radical to an Fe(III)-OH intermediate, where 2-methylbutane was chosen as the model substrate because it closely resembles most substrates of Fe(II)/aKGdependent enzymes (Supporting Information Figure ). |
65039773b338ec988a7e0c2f | 5 | Additionally, the relevant RC angles were constrained in isomerization RCs (Supporting Information Figure ). Heavy atoms of the lysine substrate and the distance between the oxo ligand and the relevant hydrogen atom of lysine were also constrained and varied in increments of 0.05 Å for HAT RCs (Supporting Information Figure ). For rebound RCs, the distance between the carbon radical of the model substrate and oxygen of OH or halide was constrained and varied in increments of 0.1 Å (Supporting Information Figure ). High-energy structures along all RCs were used to obtain vibrational frequencies using gas-phase numerical Hessian calculations carried out at the corresponding levels of theory. The Hessian was computed using the central difference approach after 6N atomic displacements. The presence of an imaginary frequency along the RC confirmed characterization of high-energy structures along the RCs as transition states. |
65039773b338ec988a7e0c2f | 6 | Single-point energy calculations on the optimized geometries of active-site intermediates and their isomers were performed using domain-localized pair natural orbital coupled cluster with singles, doubles and perturbative triples (i.e., DLPNO-CCSD(T) ) in ORCA v4.0.1.2. Two-point extrapolation to the complete basis set (CBS) limit was carried out using Dunning-style correlation-consistent double-z and triple-z (i.e., aug-cc-pVDZ and aug-cc-pVTZ) basis sets. All reported DLPNO-CCSD(T)/CBS energies were obtained using tight PNO thresholds, consistent with prior work (Supporting Information Table ). Mayer bond order analysis to quantify the strength of binding of aKG, monodentate, and bidentate succinate ligands to iron was performed using Multiwfn 100 . |
65039773b338ec988a7e0c2f | 7 | Prior work based on the crystal structures and spectroscopic studies of nonheme iron halogenases along with the related non-heme iron dioxygenases and hydroxylases has led to the proposal of the following mechanism in the catalytic cycle for non-heme iron halogenases. In its resting state (1), the hexa-coordinated iron active site consists of an Fe(II) coordinated to two His ligands, a bidentate aKG ligand, a halide ligand, and a water molecule, which is loosely bound to the iron center (Figure ). When the substrate enters the binding pocket in the active site, the water molecule is displaced, followed by the binding of molecular oxygen (2) to the iron center (Figure ). The bound molecular oxygen immediately attacks the carbonyl carbon of the bidentate aKG ligand, leading to O-O bond cleavage and oxidative decarboxylation of aKG (Figure ). During this catalytic step, the decarboxylation of an initially bidentate coordinating aKG leads to release of a carbon dioxide molecule, leaving behind the succinate substrate that has a carboxylate group that can bind iron in a monodentate or bidentate fashion (Figure ). This conversion leads to the appearance of an active site consisting of a highly reactive terminal Fe(IV)-oxo intermediate (3), which abstracts a hydrogen atom from the substrate, forming a radical substrate and an Fe(III)-OH intermediate (4) (Figure ). This rate-determining HAT step is followed by rebound halogenation of the substrate radical by the halide ligand, which returns the active site to its resting state (1) (Figure ). We assume that the Fe(IV)=O intermediate can form after oxidative decarboxylation based on our prior study of an engineered SadX enzyme with fluoride and chloride bound, where both could form hydroxylated products, and therefore the Fe-oxo intermediate must have formed, but only the chloride bound active site led to chlorinated products. Thus, we focus on the HAT and rebound steps (i.e., 3 and 4), and we study this mechanism for three halides: fluoride, chloride, and bromide. Hydrogen, carbon, nitrogen, oxygen, fluorine, chlorine, iron, and bromine atoms are shown in white, gray, blue, red, cyan, green, brown, and maroon, respectively. |
65039773b338ec988a7e0c2f | 8 | All the intermediates formed during the catalytic cycle can potentially exist in two or more isomers. For example, in intermediates ( ) and ( ), water and molecular oxygen are represented in axial positions corresponding to their respective axial configurational isomers due to the frequency with which water is found in the axial position in crystal structures of halogenases (Figure ). However, these intermediates can also have water and molecular oxygen present in equatorial positions, corresponding to equatorial configurational isomers (Figure and Supporting Information Figure ). In the Fe(IV)=O and Fe(III)-OH intermediates, the active site can be hexacoordinated or penta-coordinated depending on whether succinate is coordinated to iron in a bidentate or monodentate configuration (Figure and Supporting Information Figure ). In the case of a hexa-coordinated iron active site with bidentate succinate, oxo or hydroxo moieties and the halide ligand can be present in axial or equatorial positions, resulting in three configurational isomers: axial oxo/hydroxo and equatorial halide, equatorial oxo/hydroxo and axial halide, and equatorial oxo/hydroxo and halide (Figure and Supporting Information Figure ). These three configurational isomers can also be observed when we have a penta-coordinated iron active site with monodentate succinate (Figure and Supporting Information Figure ). Combining the monodentate and bidentate succinate coordination cases, we potentially have a total of six configurational isomers for intermediates (3) and (4) (Figure and Supporting Information Figure ). Along with computing reaction coordinates, we will evaluate the properties and energetics for interconversion of all of these isomers to identify which ones are most likely present during catalysis. |
65039773b338ec988a7e0c2f | 9 | It is known that fluoride-bound halogenase variants of the engineered enzyme SadX solely hydroxylate, whereas chloride-bound variants can both chlorinate and hydroxylate substrates , but the lack of strong differences in energetics we previously identified and further expand upon this work motivates investigation into alternative explanations for this difference in reactivity. For example, isomerization has been invoked to rationalize selectivity of some albeit not all non-heme iron halogenases. In the cases where it has been motivated, such as cases like WelO5 where the substrate in the active site would be distant from the donating halogen in the absence of isomerization , isomerization of the active site after high-valent metal-oxo formation has been suggested as a key factor in the selective C-H halogenation carried out by non-heme Fe(II) enzymes . In particular, prior experimental and computational studies suggested that the key intermediates formed during the catalytic cycle, i.e., Fe(IV)=O and Fe(III)-OH, could isomerize to favor selective halogenation over hydroxylation by positioning the rebounding substrate radical closer to the halogen. While the relative favorability of Fe(IV)=O and Fe(III)-OH isomers in chloride intermediates has been investigated in prior studies, it is not known whether other halogen species exhibit different preferences. Therefore, to identify the ground spin states and to understand the relative energetics and stabilities of isomers of Fe(IV)=O and Fe(III)-OH intermediates, we study active site models of these intermediates and their isomers across all three halides, i.e., fluoride, chloride, and bromide, in all possible spin states (Supporting Information Table and Figure ). Across all halides, we find that all isomers of both studied intermediates have HS ground states (Supporting Information Table ). Furthermore, the HS state is strongly preferred over IS and LS states for all species, and the HS ground states have longer bonds in most intermediates in comparison to the LS states (Supporting Information Tables ). |
65039773b338ec988a7e0c2f | 10 | Examination of Fe(IV)=O isomers in the HS state reveals that all bidentate succinate isomers are comparable in energy and more favorable than the monodentate succinate isomers by ca. 3 kcal/mol for intermediates containing fluoride and chloride (Figure and Supporting Information Table and Figure ). The monodentate succinate isomer with fluoride in the active site was obtained using additional constraints on Fe and oxygen atoms of succinate to ensure monodentate coordination. In the absence of these constraints, this monodentate isomer optimizes to a bidentate succinate isomer, indicating that for fluoride intermediates, interactions with the greater protein environment, such as a hydrogen bond to either the proximal or distal succinate carboxylate groups that would pull the co-substrate away from iron, are likely necessary to facilitate monodentate succinate coordination. In contrast, for bromide intermediates, the monodentate succinate isomer is more favorable than two bidentate succinate isomers by ca. 1-2 kcal/mol (Figure ). We find that the most stable Fe(IV)=O isomer is different for each halogen element (Figure ). However, the range of relative isomer energetics is consistent, i.e., for a given halide, all isomers differ in energy by less than 3.5 kcal/mol, suggesting that isomerization of Fe(IV)=O is feasible for all halides (Figure and Supporting Information Table and Figure ). ) oxo, axial (ax.) X, bidentate (bident.) succinate (succ.); axial oxo and equatorial X, bidentate succinate; axial oxo, equatorial X, monodentate (monodent.) succinate; equatorial oxo and X, bidentate succinate. (Top to bottom, right) Isomers are equatorial OH, axial X, monodentate succinate; axial OH, equatorial X, monodentate succinate; equatorial OH and X, bidentate succinate; axial OH, equatorial X, bidentate succinate; equatorial OH, axial X, bidentate succinate. The dashed blue line corresponds to an estimated DLPNO/CBS energy. Hydrogen, carbon, nitrogen, oxygen, iron, and bromine are shown in white, gray, blue, red, brown, and maroon, respectively. |
65039773b338ec988a7e0c2f | 11 | Next, we study the relative energetics of Fe(III)-OH isomers, because it has been suggested that isomerization of the Fe(III)-OH intermediate could selectively favor substrate halogenation over hydroxylation during the substrate rebound reaction, meaning that a species that was limited in its ability to isomerize would lead primarily to hydroxylated products in the absence of other selectivity-determining mechanisms. We find that the relative energetic ordering of most Fe(III)-OH isomers is consistent across all halides with small energetic differences of ca. 1 kcal/mol for some isomers as a result of slightly different optimized geometries (Figure and Supporting Information Table and Figure ). The most stable bidentate succinate isomer with axial halide and equatorial OH is strongly preferred over monodentate isomers across all halides by ca. 5-9 kcal/mol (Figure ). This suggests a potential difficulty in isomerization (see calculation of barriers in Sec. 4.3) of the most stable Fe(III)-OH isomer, implying that selective halogenation succ., ax. oxo, eq. X could be expected to take place only when the substrate is positioned more proximal to the halide by second-sphere interactions or the isomerization takes place prior to HAT. |
65039773b338ec988a7e0c2f | 12 | Examination of optimized geometries of halide intermediates in their ground spin states shows that iron-halide bond lengths increase on average with halide size from 1.81 Å (F) to 2.25 Å (Cl) to 2.41 Å (Br) in Fe(IV)=O and Fe(III)-OH intermediates (Supporting Information Table ). To assess whether this increase in bond lengths is solely due to differences in the size of the halides, we considered bond lengths scaled by the covalent radii of participating atoms (Supporting Information Table ). The scaled Fe-F bond is shorter than Fe-Cl (0.84 vs 0.90), while scaled Fe-Cl and Fe-Br bond lengths are comparable. The shorter scaled and unscaled Fe-F bonds suggest that it would be harder for the substrate to approach the halide in the oxo or hydroxo intermediates, which could explain the challenges with C-H fluorination. All other metal-ligand bond lengths are mostly comparable across fluoride, chloride, and bromide intermediates (Supporting Information Table ). The Fe-Cl and Fe-Br distances observed in optimized geometries of Fe(II)-H2O intermediates, i.e., 2.38 Å and 2.54 Å, respectively, are roughly comparable to those observed in the crystal structures of BesD and SyrB2, suggesting that the shortened Fe-F bond lengths are treated reliably by DFT (Supporting Information Table ). |
65039773b338ec988a7e0c2f | 13 | We next examine the binding strength of ligands to Fe as quantified through the Mayer bond order to understand how the differences in binding of monodentate and bidentate succinate to Fe relate to their energetic stabilities in oxo and hydroxo intermediates. Consistent with prior work on intermediates with chloride, bidentate succinate binds more strongly to Fe than monodentate succinate but not as strongly as the bidentate aKG for intermediates with fluoride and bromide (Figure ). In general, binding strengths of succinate and aKG to Fe are slightly weaker in intermediates with fluoride compared to the larger chloride and bromide halides, both of which exhibit similar bond orders (Figure ). However, for Fe(III)-O2 intermediates, aKG binds more strongly to Fe when bromide/fluoride is in the active site compared to chloride, i.e., bond order of 1.40 vs 1.23 (Figure ). For Fe(II)-H2O intermediates with fluoride in the active site, the water molecule in the axial position can form a hydrogen bond (HB) with fluoride (Supporting Information Figure ). We find that aKG binds more strongly to Fe in the isomer with a HB between water and fluoride than it does in the isomer without the HB, i.e., bond order of 0.57 vs 0.47 (Figure ). Overall, binding strengths of ligands to Fe demonstrate limited dependence on the active site halide, with succinate and bidentate aKG displaying slightly stronger binding strengths for increasingly heavy halides (Figure ). Thus, the overall reactivity and bonding should be qualitatively similar regardless of halide. The stronger binding of aKG to Fe in Fe(III)-O2 |
65039773b338ec988a7e0c2f | 14 | intermediates of fluoride and bromide could suggest that oxidative decarboxylation to form Fe(IV)=O with succinate ligand might be slightly more difficult for these halides compared to the native chloride species. Nevertheless, prior experimental results demonstrate that the oxidative decarboxylation step occurs in a fluoride-bound version of the engineered enzyme SadX as proficiently as in the case where chloride or bromide are bound . Thus, we focused our reaction coordinate analysis on only the HAT and rebound steps. |
65039773b338ec988a7e0c2f | 15 | We study two reactions that are crucial for C-H halogenation for fluoride, chloride, and bromide intermediates: the rate-determining HAT reaction and the competing rebound hydroxylation or halogenation reactions. We investigate HAT and rebound reactions to understand if differences across three halides for these reactions could explain why C-H chlorination or bromination are observed but fluorination is not observed. We study rebound of the substrate radical to both We model the HAT reactions using a bidentate succinate isomer with axial oxo and equatorial halide, which is the most stable isomer for the chloride Fe(IV)=O intermediate. While this is not the most stable isomer for fluoride or bromide intermediates, it is used for HAT reactions across all halides to enable consistent comparisons across halides and also because all bidentate succinate isomers are comparable in energy for the three halides (Figure ). We find that the HAT reactions are accompanied by a significant reaction barrier (ca. 25 kcal/mol) for all halides suggesting that the nature of HAT is largely independent of the halide in the active site (Figure ). |
65039773b338ec988a7e0c2f | 16 | This barrier height is in the middle range of those reported in prior work. Specifically, it is slightly lower than the HAT reaction barriers of ca. 30 kcal/mol observed for BesD in prior work owing to additional constraints used in that study. In comparison to another study, we find that the HAT barriers we observe are higher by ca. 5 kcal/mol, likely because we evaluate HAT reaction using first coordination sphere residues and a model substrate whereas first and second coordination sphere residues along with the complete substrate are used to compute HAT barrier in this other work. We also find that the hydroxo intermediate along with the substrate radical formed after HAT are higher in energy by ca. 10-13 kcal/mol relative to the reactants, i.e., the oxo intermediate and substrate, across all three halides (Figure ). The high-energy structures along the RCs observed at an O•••H distance of 1.18 Å are confirmed to be transition states through Hessian calculations and the presence of a single imaginary frequency along the HAT RC (Figure and Supporting Information Table ). Furthermore, we obtained optimized geometries of these transition states along with the corresponding Hessian calculations with a single imaginary frequency along the HAT RC (Supporting Information Table and Figure ). Next, we study rebound RCs in which the substrate radical reacts with hydroxo or halide ligands (Figure ). Rebound reactions with hydroxo/halide are carried out using the most stable bidentate succinate isomers with axial hydroxo/halide, which are comparable in energy (Figure ). |
65039773b338ec988a7e0c2f | 17 | Rebound reactions where the substrate radical rebounds to the hydroxo ligand are barrierless for larger halides, i.e., chloride and bromide but have a barrier of ca. 3 kcal/mol for the smaller fluoride intermediates (Figure ). Rebound reactions where the substrate radical rebounds to the halide are barrierless for bromide and exhibit a minor barrier of ca. 1 kcal/mol for chloride (Figure ). However, a significant barrier of 6.5 kcal/mol is observed for fluoride intermediates, suggesting that selective C-H fluorination could be challenging (Figure ). Consistent with the Hammond-Leffler postulate 105, 106 , the strong exothermicity corresponds to transition states that are much more similar to reactants than to the product rebound intermediate (Figure ). The moderate to significant barriers for fluoride intermediates suggest that the rebound reaction could make C-H fluorination difficult in comparison to hydroxylation, consistent with prior work on SadX. Further examination of rebound RCs reveals that the OH-rebound intermediates are energetically more favorable than halide-rebound intermediates for fluoride, chloride, and bromide intermediates, highlighting a greater thermodynamic driving force for hydroxylation (Figure ). |
65039773b338ec988a7e0c2f | 18 | Thus, to avoid the hydroxylated product which could serve as a thermodynamic trap, enzymes must have chemical or structural strategies to favor halogenation over hydroxylation. For larger halides chloride and bromide, the OH-rebound intermediates are much more favorable than the halide-rebound intermediates by almost 35 kcal/mol, despite the fact that it is known that several enzymes do selectively brominate or chlorinate (Figure ). Surprisingly, we observe that the OHrebound intermediate is more stable than fluoride-rebound intermediate by a much smaller amount, ca. 18 kcal/mol (Figure ). Despite the smaller energetic difference between OH-rebound and halide-rebound intermediates for fluoride relative to chloride/bromide, the higher barrier for rebound to fluoride could make C-H fluorination challenging compared to C-H chlorination or bromination (Figure and Supporting Information Table ). |
65039773b338ec988a7e0c2f | 19 | We next studied isomerization energy landscapes of Fe(IV)=O intermediates for all three halides to assess their likelihood of interconversion. We compare the active site isomerization across fluoride, chloride, and bromide intermediates to understand any differences in isomerization barriers and preferred orientation of oxo/halide ligands in intermediates of the catalytic cycle. We study two isomerization reaction coordinates (RCs) corresponding to the isomerization of a) halide and b) oxo ligands from axial to equatorial positions with both monodentate and bidentate succinate isomers. As previously discussed, we expect isomerization of Fe(IV)=O intermediates to be favored when the active site has a monodentate succinate configuration to ensure the needed coordination flexibility within the active site to enable isomerization. Moreover, monodentate metal-oxo isomers are only slightly less favorable than the bidentate isomers by ca. 3 kcal/mol (Figure ). |
65039773b338ec988a7e0c2f | 20 | We first study the isomerization RC that connects equatorial and axial halide isomers with equatorial oxo and monodentate succinate via an RC described by the NHis-Fe-X angle formed with the His ligand that is trans to the axial moiety (Figure ). Comparison of this isomerization RC across three halides reveals differences in the stabilities of the equatorial halide configuration, which is not a minimum on the RC, but which we refer to as the equatorial isomer (Figure ). This equatorial fluoride isomer is more favorable than the equatorial chloride and bromide isomers by ca. 2 kcal/mol and 3 kcal/mol, respectively (Figure ). While the equatorial halide isomer is energetically more stable than the axial halide isomer for all three halides, their energies are more comparable for larger halides, i.e., the equatorial isomer is strongly preferred over the axial isomer by 4.5 kcal/mol for fluoride but only by 2.5 kcal/mol for bromide (Figure ). We also find that the lowest-energy structure positions the halide tilted out of the equatorial plane and is observed at increasingly obtuse RC angles with increasing halide size, i.e., the lowest-energy NHis-Fe-X angle is 137°, 140°, and 147° in fluoride, chloride, and bromide intermediates, respectively (Figure ). |
65039773b338ec988a7e0c2f | 21 | However, axial halide isomers are comparable in energy across all halides with axial fluoride and bromide isomers being slightly more stable than axial chloride isomer by ca. 1 kcal/mol (Figure ). Overall, we expect that the differences in equatorial halide stabilities and halide-dependent NHis-Fe-X RC angle during isomerization could play a role in enabling C-H halogenation. |
65039773b338ec988a7e0c2f | 22 | The second RC corresponds to isomerization between the axial and equatorial positions for the oxo while the halide remains equatorial and succinate remains monodentate (Figure ). We describe this RC as a function of NHis-Fe=O angle formed with the His ligand that is trans to the axial moiety (Figure ). We find that the energies of both minima corresponding to axial oxo and equatorial oxo isomers differ by less than 1 kcal/mol between the smaller fluoride and larger chloride/bromide intermediates (Figure ). Furthermore, the angular RCs are broadly consistent across all halides with a barrier of ca. 8 kcal/mol for oxo isomerization (Figure ). The high-energy structures along these RCs are confirmed to be transition states with a single imaginary frequency along the angular RC mode (Supporting Information Table ). |
65039773b338ec988a7e0c2f | 23 | Although isomerization may occur in the Fe(IV)=O intermediate, positioning the reactive moiety away from the substrate at the cost of more sluggish but selective halogenation, it could alternately occur after hydrogen abstraction but before the radical rebound step. We thus also explore a similar isomerization RC that connects axial OH and equatorial OH where halide remains equatorial and succinate remains monodentate (Supporting Information Figure ). This RC is captured by the change in NHis-Fe-OH angle with the His trans to the axial hydroxo (Supporting Information Figure ). Consistent with oxo isomerization RCs, the hydroxo isomerization RCs are also comparable across all halides (Supporting Information Figure ). However, while an axial oxo isomer is favored along the oxo isomerization RC, here we observe that a hydroxo intermediate where OH tilts away from the axial position (NHis-Fe-OH = 154°) is favored for all halides (Figure and Supporting Information Figure ). Comparable oxo and hydroxo isomerization energy landscapes suggest that the preferred isomers before and after hydrogen atom abstraction are consistent for all halides. This indicates that the difficulty with C-H fluorination could stem from differences in halide isomerization that distinguish the behavior of fluoride and chloride/bromide intermediates. |
65039773b338ec988a7e0c2f | 24 | We next obtained isomerization RCs where succinate is forced to remain bidentate along the RC to isolate the restrictive effect of bidentate coordination on isomerization and compare it to the RCs obtained with monodentate succinate (Figure ). Consistent with prior work , while bidentate succinate is more stable than its monodentate counterpart, we find that isomerization may not be feasible with fully bidentate succinate geometries for any of the three halides (Figure ). Comparison of isomerization RCs obtained with bidentate and monodentate isomers reveals that the structures on the bidentate RC for fluoride are energetically more favorable than those for chloride/bromide by ca. 2-3 kcal/mol (Figure ). isomers for fluoride which could explain why C-H fluorination is not observed in non-heme Fe(II) enzymes (Figure ). On the contrary, isomerization RCs of fluoride intermediates obtained as single-point calculations at chloride/bromide geometries exhibit comparable or even slightly reduced barriers when the Fe-F bond is lengthened (Supporting Information Figure ). These observations suggest that targeted modifications to residues in the enzyme active site to introduce residues that form halogen bonds with fluoride or create local electric fields that elongate Fe-F bond lengths could make selective C-H fluorination more favorable. This longer Fe-F bond would then be expected to make the isomerization and rearrangement of the fluoride proximal to the substrate radical more favorable. |
65039773b338ec988a7e0c2f | 25 | While non-heme iron halogenases are known to carry out C-H chlorination and bromination, C-H fluorination activity has never been observed in these enzymes. We showed through a combination of DFT and WFT that there are significant differences between the smaller fluoride and the larger chloride/bromide intermediates in terms of their structural and energetic preferences. While HAT reactions were found to be comparable for all three halides, we observed crucial differences in radical rebound reaction barriers for fluoride relative to chloride/bromide intermediates. The larger halide-rebound reaction barriers for fluoride intermediates in comparison to chloride/bromide intermediates could explain the challenges of C-H fluorination relative to C-H chlorination and bromination. |
65039773b338ec988a7e0c2f | 26 | Isomerization is expected to play a role in reaction selectivity in non-heme iron halogenases. We found that while chloride and bromide Fe(IV)=O intermediates have readily stabilized monodentate succinate isomers, fluoride monodentate isomers collapse back to bidentate structures, suggesting additional interactions with the greater protein environment might be necessary to enable the formation of this isomer for fluoride intermediates. Furthermore, we find that the much shorter Fe-F bonds relative to Fe-Cl/Fe-Br bonds suggest difficulty in the substrate approaching the oxo or hydroxo intermediates which could explain why C-H fluorination is challenging. |
65039773b338ec988a7e0c2f | 27 | To further distinguish across all three halide intermediates, we studied active site isomerization of monodentate Fe(IV)=O between equatorial and axial halide isomers and observed differences in equatorial halide stabilities and the halide-dependent NHis-Fe-X RC angle during isomerization, which could play a role in C-H halogenation. We also found that bidentate isomers have additional stability for fluoride relative to chloride/bromide intermediates, suggesting difficulty in isomerization between bidentate and monodentate isomers for fluoride, which could explain why C-H fluorination is not observed in non-heme iron halogenases. To understand the effect of size differences, we obtained isomerization RCs of chloride/bromide intermediates with Fe-Cl and Fe-Br bonds constrained to shorter bond lengths that revealed increased reaction barriers. Conversely, fluoride RCs where Fe-F bond was elongated resulted in slightly reduced barriers suggesting that targeted modifications in the enzyme active site to position residues nearby that can form halogen bonds with fluoride and elongate Fe-F bonds could make selective C-H fluorination possible. |
65039773b338ec988a7e0c2f | 28 | Overall, our study highlights the differences between the smaller fluoride and the larger chloride/bromide intermediates that could explain why selective C-H fluorination is not observed in non-heme iron halogenases. The differences in Fe-halide bond lengths and radical rebound reactions described in this work set the stage for further studies about targeted modifications of the enzyme active site to enable favorable C-H fluorination in non-heme iron halogenases. |
65039773b338ec988a7e0c2f | 29 | ASSOCIATED CONTENT Supporting Information. Extracted active site of WelO5 with three halides, i.e., F, Cl, Br; active site isomers and constrained atoms, angles, and distances; differences between implicit solvent and gas-phase metal-ligand bond lengths; potential spin states of active site intermediates with three halides (F, Cl, Br); tight PNO thresholds used in DLPNO-CCSD(T)/CBS calculations; active site isomers of intermediates formed during catalytic cycle; DLPNO-CCSD(T)/CBS spin-splitting energies of oxo and OH intermediates; differences between high-spin and low-spin metal-ligand bond lengths; relative energies of isomers obtained with DLPNO-CCSD(T)/CBS; optimized geometries of Fe(IV)=O isomers with three halides (F, Cl, Br); optimized geometries of Fe(III)-OH isomers with three halides (F, Cl, Br); optimized metal-ligand bond lengths of all active site isomers; covalent radii of fluorine, chlorine, iron, and bromine atoms; Fe-Cl and Fe-Br bond lengths in optimized geometries and crystal structures; optimized geometries of Fe(II)-H2O and Fe(III)-O2 isomers; NHis-Fe=O angles in transition state geometries along isomerization RC; RC for isomerization between axial and equatorial hydroxo; relaxed RCs and RCs obtained as singlepoints along relaxed RCs; O•••H distances in high-energy structures along HAT RCs; O•••H distances in optimized transition state geometries along HAT RCs; optimized geometries of transition states along HAT RCs; relative energetics of halide-rebound and hydroxo-rebound intermediates (PDF) |
65039773b338ec988a7e0c2f | 30 | Initial geometries, gas-phase optimized geometries, and solvent-phase optimized geometries of isomers of active site intermediates for all three halides, i.e., fluoride, chloride, and bromide; initial and optimized geometries along isomerization reaction coordinates for axial and equatorial oxo isomerization as well as axial and equatorial halide isomerization for fluoride, chloride, and bromide intermediates (ZIP) |
60c757ef337d6c4944e29104 | 0 | The diatomic species C2 has a long and literally colourful history, one nicely summarised recently. Much of the known chemistry and particularly the spectroscopy relates to the gaseous species at high temperatures, but the very high reactivity has also meant that ambient temperature generation in the condensed phase and the mechanism of subsequent reactions has hitherto been less studied. Interest in this species has been particularly sparked in the last decade by renewed discussion of its chemical bonding in the form of a proposal that it sustains a quadruple bond between the carbons as in C⩸C, a claim that generated much stimulating debate and continues to do so unabated. Most recently, the attention has extended to species with quadruple bond patterns between carbon and other elements such as C⩸Fe or related main group elements with similar four-fold bond characteristics as in B⩸Rh or Si⩸Rh. Such diverse reports raise the distinct possibility that room temperature/solution studies of species containing quadruple bonds to carbon will become possible and perhaps even routine. In this context then, the recent proposal of a room temperature chemical synthesis in the condensed phase on a relatively rapid timescale (~minutes) of C2 itself was particularly significant, since this would open an avenue for exploring the reactivity of carbon in this unusual bonding state in new media and a new temperature range. The reaction was thought to proceed at ambient or low temperatures from the transient zwitterionic intermediate 2, formed by treating precursor 1 with a source of fluoride anion. Unimolecular fragmentation of 2 would then produce iodobenzene and free singlet state C2 (Figure ). It should be noted that other reagents that deliver a dicarbon fragment have been reported, although there the mechanism was not thought to involve any free C2. Likewise dicarbon stabilized by a single phosphine ligand has been characterised which can undergo, inter alia, an interesting intermolecular C-H activation upon thermolysis. Experiments trapping 2 in solution with either 9,10dihydroanthracene or galvinoxyl radical (Figure ) enabled isolation of products from which participation of free C2 was inferred. The same conclusion also following from detection of polymeric carbon products such as amorphous carbon and even C60. An experiment carried out using solid reagents putatively produced "C2 gas", as inferred by using argon to flush any volatile products out of the reagent flask and into a second flask, where they were again observed to be trapped using solid galvinoxyl. Further pertinent experimental observations are the reported results of isotopic substitution in 2. In dichloromethane solutions, it was asserted that 13 C⩸C 12 , as apparently formed from labelled 2 and trapped using galvinoxyl, results in a 71:29 product ratio in favour of a C label |
60c757ef337d6c4944e29104 | 1 | in the a position of the product rather than b (see Figure ). For the experiment conducted without solvent, the isotope distribution was found to be almost equal (52:48). The former result was attributed to a fast radical pairing between C2 and galvinoxyl in solution, prior to ejection of iodobenzene from the solvent cage. Changes in the isotope patterns with solvent were thought to arise from differing solvent viscosities. |
60c757ef337d6c4944e29104 | 2 | A fundamental aspect of any ambient temperature reaction occurring on a relatively short time scale is its energetics. These were addressed as matter arising from the original synthetic report, in which analysis of the computed thermodynamics of this reaction led to the conclusion that the production of free C2 and iodobenzene was likely to be highly endoenergic. The energetics of the equilibrium (Eqn. 1, R=Me,Ph) were in the range of +(43-53) kcal mol -1 using three different estimates, anchored by a calibrated CCSD(T)/Def2-TZVPPD/SCRF=dichloromethane calculation for a simplified model (Eqn. 1, R=Me) for which DG298 +47.1 kcal mol -1 . Eyring theory tells us that at 298K, unimolecular reactions with a half-life of respectively 1 minute and 1 hour correspond to free energy barriers of 20.0 or 22.5 kcal mol -1 , significantly lower than the energy range predicted above. |
60c757ef337d6c4944e29104 | 3 | same reagents (Figure ) prior to any release of C2. Since generation of C2 by reactions such as these has the potential for much exploitation, it is important to try to establish not only the energetics but also the mechanisms by which the observed products might be forming. Following on from the previous energetic study (Figure ), the present article reports the results of a computational exploration of the bimolecular reaction mechanism between 2 and the trapping reagents used in the original experimental study (Figure ), including the selfreaction of 2 and further similar steps which result in polymerisation giving linear carbon chains. |
60c757ef337d6c4944e29104 | 4 | To study the energetics and mechanism of these reactions, the ωB97XD 16 /Def2-SVPD 17 /SCRF 18 =dichloromethane solvent density functional procedure was selected 19 as computationally more feasible than coupled-cluster methods such as CCSD(T) for computing large species such as galvinoxyl and for evaluating intrinsic reaction coordinates (IRCs). This DFT method was first calibrated against both the CCSD(T) 20 /Def2-TZVPPD model and experiment (Table ). This revealed that the relative energy of free C⩸C itself is predicted to be too high by ~28 kcal mol -1 using the ωB97XD/Def2-SVPD/SCRF=DCM method. Transition states were verified using IRC pathways. A model 1,2-substitution reaction which is similar to the reactions shown in Figure and which allows the CCSD(T) level transition state to be located using symmetry alone (C2h) was selected for calibration (Eqn. 2). |
60c757ef337d6c4944e29104 | 5 | A bimolecular mechanistic model avoiding the formation of free C2 is here proposed. This involves concerted 1,1-substitution directly on 2 by nucleophilic attack from e.g. the galvinoxyl oxygen atom and with iodobenzene acting as a nucleofuge, to form a b-labelled product if isotopic substitution is present in 2 and a-labelled product for 1,2-substitution (Figure ). The transition state structures for these two alternatives are shown in Figure . Classical nucleophilic substitution at trigonal 23 and digonal carbon is suggested to proceed via a 1,1- |
60c757ef337d6c4944e29104 | 6 | mode involving either addition/elimination or a direct SN2 like structure, but 1,2-modes have apparently never been previously proposed. A similar 1,1-substitution mechanism which avoids liberating the free cation C⩸N + (a species isoelectronic with C⩸C) may apply when cyanogen chloride or bromide (Cl-C≡N) reacts with benzene in the presence of e.g., aluminium chloride to produce benzonitrile, with benzene as nucleophile displacing the chloride nucleofuge directly at carbon. . The issue now is whether either of these alternative mechanisms have overall lower activation free energies than the previously mooted pathway generating C2 itself, whether unbound or as a "solvent-cage trapped" species. The ωB97XD/Def2-SVPD/SCRF=DCM model predicts the barriers for this bimolecular reaction (Eqn. 2 and Table , column 8) to be close to those obtained at the CCSD(T) levels and also that the differences between the more accurate Def2-TZVPPD and the computationally faster Def2-SVPD basis sets are acceptably small (<1 kcal mol -1 ). It was also possible to compare the free energy of the 1,2-substitution transition states (Eqn. 2) with that of free C2 + two Me-I molecules (Table , 28 column 7). If a correction of ~+28 kcal mol -1 noted above is applied to DG298 using the ωB97XD/Def2-SVPD/SCRF=DCM model, the 1,2-substitution reaction of Me-I + -C≡C -by Me-I as nucleophile still emerges as ~10 kcal mol -1 lower in energy than the pathway involving free C2. Replacing Me-I by phenoxyl radical as a better substituting nucleophile suggests that this 1,2-reaction is now ~21 kcal mol -1 lower in free energy than generation of free C2, and that the activation free energy itself (~21 kcal mol -1 ) is compatible with a facile room temperature reaction. |
60c757ef337d6c4944e29104 | 7 | The results for increasingly complete models of the galvinoxyl trap are shown in Tables 1 for both the 1,1-and 1,2-substitution mechanisms. The reaction between 2 and phenoxyl radical has a slightly lower barrier (ΔG ‡ ~16.9-18.8 kcal mol -1 ) in the gas phase than in dichloromethane solution (~19.4-19.6), due to solvent stabilization of the ionic 2. The free energies of the 1,1and 1,2-substitutions tend to be similar but not identical, which would account for the small variations in isotopic ratios of the final product. Such a model no longer requires stipulating fast radical pairing in a solvent cage to account for unequal isotope ratios in solutions. For the full galvinoxyl model, 1,2-substitution resulting in a-labelled C-product is computed as slightly lower in free energy, in accord with observation. These transition states also have slightly different dipole moments (1,2 isomer 12.2D, vs 1,1-isomer 11.0D), which suggests that such differences may explain the changes in isotope ratios as a function of solvent observed in the original experiments (cf Figure ). The height of the dichloromethane solution free energy barrier (25.9 kcal mol -1 ) is now perhaps 3-4 kcal/mol higher than expected for a facile room temperature reaction, but the size of the system has thus far precluded full conformational optimisation to identify any conformers with lower energy barriers. The energies of both the 1,1-and 1,2-substitution transition states are lower than the computed combined free energies of the trapping species + free C2 + iodobenzene by ~12.3 and 14.3 kcal mol -1 respectively. Bimolecular reaction between free C2 and any trap would augment that free energy difference because of an additional free energy barrier induced by loss of entropy (see e.g. the entry in Table for the reaction between free C2 and dichloromethane), reinforcing the conclusion that the route involving bound rather than free C2 is the more probable mechanism. The transition state for reaction of 2 with 9,10-dihydroanthracene shows much greater discrimination between 1,1-and 1,2-substitution, with the latter being clearly favoured (Figure and Table ). The former has a small degree of biradicaloid character (<S 2 >= 0.4129) and is highly asynchronous, tending towards formation of HC≡C• and 9,10-dihydroanthracen-9-yl radical as a "hidden intermediate", but which eventually results in hydrogen abstraction from the latter by the former to give the final trapped products. The more stable 1,2-isomer has no biradicaloid character at the equally asynchronous transition state and at this point approximately corresponds to hydride abstraction to form to a 9,10-dihydroanthracen-9-ylium cation and a HC≡C -hidden-intermediate ion-pair instead, which then collapses to final observed products. Importantly, a thermally accessible barrier is computed for this reaction (ΔG ‡ ~23.8 kcal mol -1 ), which again is lower than the combined (corrected) energies of the species involved in unbound C2 by ~19.4 kcal mol -1 plus any additional entropic barrier for bimolecular reaction of C2 (see above). The next mechanism to be addressed here relates to the observation 10,11 that along with trapping of assumed unbound C2 itself, other major products are clearly carbon oligomers, including the formation of C60. Can these too arise without the intermediacy of free/unbound C2? The reaction of 2 with itself to form a new C-C bond provides an obvious route for such a process (Figures and). A 1,2-transition state is clearly lower than the 1,1-mode and hence provides a facile thermal route to formation of a bound C4 species (ΔG ‡ 15.4 kcal mol -1 ). The geometry of the former has a novel aspect in having two-fold (C2h) symmetry, with each molecule of 2 acting as the nucleophile attacking the other and both iodobenzene units apparently acting as the nucleofuge. An IRC (Figure ) reveals that this symmetry is initially maintained following the transition state, with apparent elimination of a free C4 unit, but eventually the energy landscape breaks symmetry to bifurcate and the unit of C4 is "frustrated" by recombining with one PhI only to form PhIC4. Such a bifurcating potential energy surface is reminiscent of the dimerization of cyclopentadiene. Further low barrier reactions between this product and more of 2 extends the carbon chain to six, this time favouring 1,1-substitution. The process can be repeated to form longer linear or even branched carbon chains (Table , footnotes d and e). |
60c757ef337d6c4944e29104 | 8 | Eventually these chains will undertake further complex reactions to result in e.g., polymers such as amorphous carbon and C60, the energetics of which will be investigated in future work. The final transition states investigated are those involving reaction of 2 with a solvent such as dichloromethane, in which the original solution experiments were performed. Predominant trapping of 2 by other species such as galvinoxyl would require the barrier for reaction with solvent to be significantly higher than with the trap, especially since the concentration of the solvent is much greater (~12M) than that reported for the trapping species (~0.02-0.033M). The bimolecular reaction between 2 and dichloromethane involves hydride abstraction to give an acetylide anion and a 1,1-dichloromethylium cation (no biradical character was detected). The transition state (Figure ) for 1,2-substitution corresponds to a reaction free energy barrier of 33.3 kcal mol -1 , which is high enough to preclude facile reaction of 2 with 12M solvent, as observed. Contrast this with the free energy barrier calculated for reaction of singlet free C2 itself with dichloromethane (Figure ), which is very much lower (DG298 ‡ 10.7 kcal mol -1 or 15.6 kcal mol -1 for the lowest triplet state of C2, Table ). If unbound C2 were indeed to be generated in dichloromethane solutions at ambient temperatures, a free energy barrier this low would certainly mean its rapid trapping by the solvent. |
60c757ef337d6c4944e29104 | 9 | A reaction that can produce the simple diatomic species C2 under mild conditions would open up a new landscape for the synthesis of carbon-rich species, including C60 itself and generate potential access routes to species in which carbon sustains the hypothesized quadruple bonding pattern. Given the recent report of exactly such a reaction, it appeared desirable to apply quantitative computational techniques as a reality check, not only regarding the energetics of such a process but also to the mechanisms by which it may proceed. These computed mechanisms reveal that ambient temperature routes for reaction of the zwitterion 2 ("bound" C2) as the active species are lower in energy than those involving unbound or free C2. Compounds such as 2 serve as potentially useful and potentially selective precursors or synthons for C2, reacting readily with species such as galvinoxyl or 9,10-dihydroanthracene and less readily with solvents such as dichloromethane, whereas the very high energy free dicarbon species 3 is likely to be less selective. |
60c757ef337d6c4944e29104 | 10 | These low energy bimolecular mechanisms bring into question whether free C2 as generated by unimolecular fragmentation of 2 actually participates in the solution-phase reactions. It also raises the issue of what is happening in the reported experiment where a flask containing solid-state reactants is flushed by argon gas into a second flask containing galvinoxyl. The assumption was that the only species sufficiently volatile to be transferred between flasks would be "C2 gas", which would then be trapped and crucially that 2 itself was too involatile to be so transferred. An experiment whereby 2, via the aryl group, be covalently anchored to a solid-phase support and placed in the first flask would eliminate any possibility that it is 2 and not C2 that is being transferred and trapped in the second flask. |
617e93c2f9f05b6743e75dca | 0 | The organometallic chemistry of iron has been dominated by strong-field supporting ligands such as CO, CN, and phosphines, and by macrocyclic N ligands like porphyrins. The active sites of hydrogenase enzymes incorporate S-based ligands, and these are low-spin due to the influence of carbonyl and cyanide donors. However, the interesting reactions of nitrogenase enzymes, and a new generation of low-valent iron catalysts, have predominantly weak-field ligands and have led to increasing interest in organometallic iron complexes with higher spin states. We focus here on iron coordination environments that result from a mixture of C and S donors -choices that are particularly compelling since the six "belt" iron atoms in the ironmolybdenum cofactor (FeMoco) of nitrogenase have a mixed C/S coordination sphere. This unusual combination of potentially strong-field C donors and weak-field S donors could lead to changes in spin states during catalysis, which has been linked to changes in barriers and selectivity. Thus, the study of C-and S-ligated iron has relevance for both fundamental coordination chemistry and bioinorganic mechanisms. However, CO-free iron complexes with supporting ligands that coordinate through only carbon and sulfur donors are rare. Of these examples, only one multidentate C/S ligand is known to support N2 binding. In addition, Qu has provided important studies on Cp*-supported iron dimers bridged by dithiolates, and this C/S ligand sphere can bind and facilitate the reduction of nitrogenase-relevant NxHy substrates. Our group's recent work on SCS systems began with a dithiolate ligand having a central arene that can interact with iron through backbonding into its π-system in low oxidation states. While this hemilabile interaction stabilizes the reduced complexes, complete dissociation of the SCS ligand occurred in the presence of Brønsted acids, and backbonding into N2 competed with backbonding into the arene. In order to address this instability, we then moved to tridentate pincer 4 Chart 1. Typical palladium SCS pincer complexes and closely-related iron complexes with SNS and SCS pincers. In the bottom pictures, the square represents a coordination site with various ligands. |
617e93c2f9f05b6743e75dca | 1 | Thioamides have two tautomers, which form the same anion upon deprotonation (Figure , right). Most often, the thioamide coordinates as an iminothiolate through the S donor rather than the potential N donor as a result of the relative weakness of the C=S π bond and the better overlap of the S atom with metal orbitals. After coordination, protonation at the N atom gives a formally neutral thione donor that maintains the M-S interaction (Figure , brackets). Even as a formally neutral ligand, the thione C-S bond is significantly polarized toward a C δ+ -S δ-form that affords some anionic character at S. The ability to tune the donor properties of iron-coordinated sulfur via protonation/deprotonation of the supporting ligand could be used to probe how protonation state can affect the redox potential, spectroscopic characteristics, and reactivity of the iron center. The potential for proton responsiveness is also relevant to nitrogenase mechanisms. The Thorneley-Lowe kinetic scheme for nitrogenases proposes that one proton is transferred to the FeMoco with each reduction step, but the location of these protons has been controversial. |
617e93c2f9f05b6743e75dca | 2 | Computational studies have showed that there are many potential sites of protonation on the FeMoco, including the sulfides, the Mo-coordinated homocitrate, and even the carbide. Studies on multi-site proton-coupled electron transfer (PCET) have demonstrated that various protonation sites can influence the redox potentials and bond dissociation free energies (BDFE). These inspirations motivated us to pursue thermochemical studies on well-characterized (SCS)Fe complexes with proton-responsive ligands to examine the effects of distant protonation events on potentially biologically-relevant iron sites. |
617e93c2f9f05b6743e75dca | 3 | The ligand is easy to prepare on a useful scale, and the stability of the iron complexes is highlighted by the ability of an iron(III) complex to be handled in air and water. Although no iron complex of the new SCS ligand was observed to bind N2, we use EPR and Mössbauer spectroscopy alongside magnetometry measurements to elucidate the unusual electronic structures of complexes with different exogenous donors, including biologically-relevant amide, thiolate, ammonia, and CO. In the amide complex, the electronic structure may be influenced to the presence of nearby cations. |
617e93c2f9f05b6743e75dca | 4 | Ligand Synthesis and Metalation. Using a procedure modified from the literature, isophthalic acid was treated with thionyl chloride followed by 2,6-diisopropylaniline, which provided the diamide (Scheme 1). Reaction of this compound with P2S5 in toluene at 100 °C yielded the bis(thioamide) 1, in an overall yield of 69% in two steps from commercial starting materials. Treating 1 with a slight excess of Fe(PMe3)4 in Et2O at room temperature led to effervescence and a color change from yellow-brown to dark green, and iron complex 2 was isolated in 93% yield (Scheme 1). Scheme 1. Synthesis and metalation of SCS pincer ligand 1. Dipp = 2,6-diisopropylphenyl. |
617e93c2f9f05b6743e75dca | 5 | Crystallization of 2 gave green blocks suitable for X-ray diffraction (Figure , top). The diffraction data showed an octahedral iron site with a meridional pincer ligand as expected. The Fe-C bond length is 1.9502(18) Å, and the average Fe-S length is 2.268(9) Å. The three Fe-P bonds are only slightly different, with a length of 2.2484(3) Å for the phosphine trans to C and lengths of 2.2553(4) Å and 2.2608(4) Å for the phosphines cis to C. There was disorder at the thioamide N-H site, which was satisfactorily modeled with each thioamide having 50% hydrogen atom occupancy, consistent with single protonation of the supporting ligand in 2. The Mössbauer spectrum of 2 has a doublet with δ = 0.21 mm/s and |ΔEQ| = 1.25 mm/s (Figure ). Proton NMR spectra in either C6D6 or THF-d8 show signals in the range expected for a diamagnetic complex of low-spin iron(II). The most downfield resonance integrates 1:1 with proton groups from each side of the molecule (Figure ). Adding D2O made this signal disappear, and therefore it is assigned as the thioamide proton. Based on the NMR spectra, 2 is unsymmetric in solution, and it is best described has having one neutral thioamide S donor and one anionic iminothiolate donor. The solid-state IR spectrum of 2 shows a broad peak centered at 3336 cm -1 (Figure , red trace), which is distinct from the analogous signal in free ligand 1 at 3136 cm -1 (Figure , gray trace), consistent with an N-H stretching vibration. We hypothesize that the thioamide arm is deprotonated during the synthesis of 2 by a transient iron hydride to form H2, which explains the effervescence during metalation. Electron and Proton Transfer in Phosphine Complexes. We serendipitously discovered that neutral, fully-deprotonated iron(III) compound 4 could also be prepared in one pot from 1 and Fe(PMe3)4 by simply exposing the crude product 2 to air after metalation was complete. Proton NMR and UV-visible spectroscopy of air-exposed 2 in situ demonstrates rapid and complete conversion to 4, with no monoprotonated iron(III) transient species observed. This one-pot procedure afforded 4 in 88% yield from 2, and this method was used to prepare 4 for subsequent experiments. Additional proton and electron transfer reactions were carried out on 2 and 4. |
617e93c2f9f05b6743e75dca | 6 | Treating the neutral iron(II) compound 2 with stoichiometric sodium bis(trimethylsilyl)amide provided anionic complex 3-Na. Oxidation of 3-Na using AgPF6 afforded neutral iron(III) complex 4. We generated the potassium analogue, 3-K, by treatment of neutral iron(III) compound 4 with stoichiometric KC8. Both have sharp peaks in their 1 H NMR spectra suggesting low-spin iron(II), and their structures were inferred based on these spectra (Figures and). Their yields (88% for 3-Na and 91% for 3-K) were determined by integration relative to an internal standard. |
617e93c2f9f05b6743e75dca | 7 | shows an Fe-C bond length of 1.9413(12) Å, which is indistinguishable from the analogous distance in the monoprotonated iron(II) complex 2. The average Fe-S length is 2.23(2) Å in 4, which is slightly shorter than in 2. Again, all three Fe-P bond distances are different, with the phosphine trans to C having the shortest bond of 2.2692(6) Å while the cis phosphines have lengths of 2.3007(6) Å and 2.3050(5) Å. The average Fe-P bond in 4 is shorter than in 2 (2.23(1) Å vs. |
617e93c2f9f05b6743e75dca | 8 | As expected for complexes lacking protonated thioamides, 3-Na and 4 do not have any IR bands above 3050 cm -1 that correspond to N-H stretching vibrations (Figure , blue and green traces). Proton NMR spectra of ferric 4 contain very broad peaks between -20 and +12 ppm (Figure ), and the 31 P NMR spectrum is featureless, as expected for nuclei directly bound to a paramagnetic metal. The relative integrations of the evident peaks in the 1 H NMR spectra of 3-M in THF-d8 are consistent with C2v symmetry. Taken together, the IR and NMR data support double deprotonation of the thioamide groups of the supporting ligand in 3-Na and 4. Scheme 2. Conversion of 2 to 4 using stepwise and concerted H atom removal. Yields were determined by 1 H NMR using an internal standard. Cyclic voltammetry (CV) studies of 4 showed a reversible Fe 3+/2+ couple at E1/2 = -1.42 V vs. Fc + /Fc in THF (Figure ). The peak currents displayed a linear dependence on the square root of the scan rate, indicating that the process is diffusion controlled (Figure ). Next, we tested the ability of different organic bases to deprotonate 2, using changes in the 31 P NMR chemical shifts to determine the ratio of 2 and deprotonated 2 since the conjugate base and acid are in rapid equilibrium. We assumed that the chemical shifts from the 31 P NMR spectrum of 3-K represent the shifts of fully-deprotonated 2. Addition of neither Et3N (pKaH = 12.5 in THF) nor DBU (pKaH = 16.9 in THF) gave changes in the NMR spectra of 2. An initial 1.10 equiv addition of triazabicyclodecene (TBD, pKaH = 21.0 in THF) equilibrated with 2, on the other hand, gave a 31 P NMR spectrum with peaks at 12.1 ppm and 10.8 ppm. These shifts are 4.8 ppm upfield from the 31 P resonances of 3-K but 1.4 ppm downfield of 31 P resonances of 2 (Figure ). This result indicates that the pKa of 2 is 22 ± 1 in neat THF-d8. An equivalent NMR experiment in 0.3 M [N n Bu4][PF6] in THF-d8 (the electrolytic solution used for CV) indicated a pKa of 21 ± 1. (Figure ) These data enable us to define the bond dissociation free energy (BDFE) of the N-H bond using the Bordwell equation, with CG(THF) = 60.4 ± 2 kcal/mol. This analysis gave a BDFE(N-H) of 56 ± 2 kcal/mol (Scheme 2). |
617e93c2f9f05b6743e75dca | 9 | Other redox and protonation tests were used as well. Electrochemical oxidation of 2 is irreversible under the same CV conditions used for 4 (Figure ). To test whether 4 could be protonated, it was treated with one equiv of [H(OEt2)2][BF4] in THF. A dark brown solid precipitated from the reaction mixture, and the IR spectrum of the solid showed a single N-H stretch at 3250 cm -1 , which is much lower than the N-H stretching frequency of 2 at 3336 cm -1 . |
617e93c2f9f05b6743e75dca | 10 | Spectroscopy of Phosphine Complexes. The Mössbauer signal for 2 (δ = 0.21 mm/s and |ΔEQ| = 1.25 mm/s, Figure ) is similar to those for 3-Na (δ = 0.24 mm/s and |ΔEQ| = 0.94 mm/s, Figure ) and 3-K (δ = 0.25 mm/s and |ΔEQ| = 1.05 mm/s, Figure ), consistent with low-spin iron(II) complexes. Solid 4 has a lower isomer shift of 0.16 mm/s and much larger quadrupole splitting of 3.45 mm/s, and its doublets are highly asymmetric, with ΓL = 0.57 mm/s, ΓR = 0.32 mm/s. To determine the spin state of 4, solid state and C6D6 solution magnetic moments were obtained. Both measurements gave µeff = 1.6 µB at 298 K, which indicates a ground-state spin of |
617e93c2f9f05b6743e75dca | 11 | Spectroscopy and DFT calculations using B3LYP/ZORA-def2-TZVP were further employed to reveal the electronic structure of 4. The EPR spectrum of 4 in 2methyltetrahydrofuran (MeTHF) at 77 K was modeled as an S = 1/2 spin system with g = [2.200, 2.120, 1.998] and coupling to three 31 P nuclei: two with |A| = [71, 83, 71] MHz and one with |A| = [0, 37, 37] MHz (Figure ). Mulliken spin population analysis showed spin populations of -0.036 on the axial phosphorus atoms and -0.019 on the equatorial phosphorus, in agreement with the fit to two equivalent 31 P nuclei having larger A than a third 31 P nucleus. The spin population analysis also showed an averaged spin density of 0.031 on the two ligand nitrogen atoms, though the smaller nuclear magnetic moment of N relative to P likely contributed to smaller, unresolvable N hyperfine splitting, which was modeled by anisotropic H-strain of [66, 27, 0] MHz in the simulation. |
617e93c2f9f05b6743e75dca | 12 | We next investigated the feasibility of removing the phosphine ligands. When iron(III) species 4 was treated with a slight excess of triphenylborane in THF, the color of the reaction mixture changed from green to orange, and precipitation of Me3P-BPh3 was observed (Scheme 3). Crystallization from Et2O gave 5-Et2O in 58% yield (Figure ). Dissolving 5-Et2O in THF and removing solvent under vacuum several times led to the formation of 5-THF (Figure ). We refer to these compounds collectively as 5-Solv because of their similarity and ability to interconvert as a function of solvent. ). Mössbauer spectra of 5-THF differ, however, between the solid state and a frozen THF solution. Specifically, there is a higher isomer shift of 0.45 mm/s for the frozensolution spectrum, which may suggest that the dimer is broken up with THF replacing the bridging sulfur (Figure ). The frozen solution also shows an increased quadrupole splitting and marked doublet asymmetry (|ΔEQ| = 4.26 mm/s, ΓL = 0.73 mm/s, ΓR = 0.49 mm/s). The 1 H NMR spectrum of 5-Et2O in THF-d8 shows six broad signals between 12 and -81 ppm (Figure ). The number and integration of the peaks is consistent with C2v symmetry; a seventh expected peak integrating to 4H was not resolved, likely due to broadening. This is also consistent with the dimer breaking up, although it is also possible that there is some other dynamic phenomenon. The 1 H NMR spectra of 5-Solv in C6D6 have at least 12 peaks with considerable broadness, which prevented further analysis but points to a lowered symmetry in the dimer. |
617e93c2f9f05b6743e75dca | 13 | Cyclic voltammetry of 5 showed three irreversible reduction events (Figure ). An attempt to reference the reduction potentials to a ferrocene internal standard was unsuccessful due to reactivity with Fc and Fc + . The lack of electrochemical reversibility may be due to large structural changes occurring different oxidation states. Undeterred, we proceeded to stoichiometrically reduce 5 with 1 equiv of KC8 per iron at -78 °C in THF, which resulted in a deep purple solution. Crystallization from diethyl ether at -40 °C give iron(II) complex 6 in 46% yield. Its X-ray crystal structure, shown in Figure , shows a tetrameric assembly in the solid state, with K + bridging S atoms and Dipp groups. There are six (SCS)Fe units in each asymmetric unit, and the average Fe-C and Fe-SPincer bond lengths are 1.947(4) Å and 2.23(9) Å, respectively. The average distance between S and K in the central bridge is 3.19(1) Å. As in 5-Solv, the Fe-S bond lengths are ~0.05 Å longer on the side of the pincer where the S is bridging two iron centers. |
617e93c2f9f05b6743e75dca | 14 | Complex 6 is extremely air sensitive and decomposes in solution over a few days. ppm, reflecting C1 symmetry in solution. We were unable to assign resonances to specific proton environments due to peak broadness and overlapping in the 0 to 8 ppm region (Figure ). Despite the complexity of the 1 H NMR spectrum and crystal structure, the solid-state Mössbauer spectrum of 6 (Figure ) has a single quadrupole doublet with an isomer shift similar to that of the starting material (0.34 mm/s for 6 vs. 0.35 mm/s for 5-Et2O), but a smaller quadrupole splitting (1.87 mm/s for 6 vs. 3.91 mm/s for 5-Et2O). The Mössbauer spectrum suggests that all iron sites are equivalent. |
617e93c2f9f05b6743e75dca | 15 | Next, we sought to test 6 for the ability to bind N2 at low temperatures using variabletemperature UV-visible spectroscopy. Negligible spectral changes were observed between 25 and -100 °C in THF, Et2O, or toluene (Figure ). Addition of stoichiometric amounts of 18crown-6 slightly shifted the absorption maxima, but the resulting species also did not exhibit notable temperature-dependent spectral changes (Figure ). Thus, it does not appear that 6 undergoes speciation changes or N2 binding at these concentrations and temperatures. Further reduction of the iron(II) tetrameter 6 led to a silent NMR spectrum and a broad, featureless UV-Vis spectrum. The products formed in this reaction are unknown, and attempts to isolate or identify the species present were unsuccessful. |
617e93c2f9f05b6743e75dca | 16 | To study monomeric complexes of our SCS iron framework in the absence of strong phosphine donors, we next treated 5-Et2O with ligands containing S and N donors, which gave the products in Scheme 4 within 1 hour at room temperature. Their syntheses, NMR spectra, and IR spectra are described here, and their magnetism, EPR spectra and electronic structures are described in following sections. Addition of a bulky thiolate (potassium 2,6-dimesitylphenylthiolate) to 5-Et2O in THF generated complex 7 in 97% yield. Addition of 18-crown-6 followed by crystallization furnished 7-crown (Figure ). 7-Crown displays an axially coordinated THF molecule and a nearly square pyramidal geometry with τ5 = 0.08 (Figure ). The Fe-C, average Fe-SPincer, and Fe-SAr, bond lengths are 2.0006(17) Å, 2.244(6) Å, and 2.2758(5) Å, respectively. Addition of Et2O, arenes, or alkanes to solid 7 prior to the addition of 18-crown-6 led to a color change from orange to green, indicating the formation of the THF-free analogue (8) in non-coordinating solvents. |
617e93c2f9f05b6743e75dca | 17 | Compound 8 was crystallized from toluene in the presence of 18-crown-6 as 8-crown (Figure , top, and Figure ). The geometry is distorted from square planar, with τ4 = 0.20. The Fe-SAr bond distance is 2.2448(8) Å, which is shorter than the Fe-SAr bond in 7-crown (Table ). Unlike the structure of the THF adduct 7-crown, one mesityl group of THF-free 8-crown is twisted to cover an axial iron site; however, the long 3.06 Å distance between iron and the arene centroid is inconsistent with a direct electronic interaction. We previously observed a similar orientation of this thiolate ligand in another SCS iron complex and ascribed this phenomenon to crystal packing effects. We were unable to crystallize four-coordinate thiolate 8 without crown ether. The 1 H NMR spectrum of 8 in THF-d8 has ten broad peaks between 15 and -63 ppm (Figure ). Upon dissolution in THF-d8, the color of 8 changes from dark brown to red, indicating that the THFadduct 7 is likely formed. The number and integration of the signals suggests C2v symmetry, which implies that the crystallographically observed lack of symmetry caused by the bulky aryl thiolate is not preserved in THF-d8 solution. |
617e93c2f9f05b6743e75dca | 18 | Amide complex 9 was synthesized in 87% yield by treating 5 with KN(TMS)2, and it was crystallized with and without 18-crown-6 to give structures of 9-crown (Figure ) and 9 (Scheme 4, middle, and Figure ), respectively. Crown-free 9 is unstable as a solid at room temperature and in solution even at -40 °C, as evidenced by the appearance of multiple new peaks between 11 and -43 ppm in its NMR spectrum after a few days (Figure ). Once this unknown impurity formed, we have been unable to remove it by recrystallization or washing. 9-Crown, however, is more stable. The crystallographic structures of 9 and 9-crown lack axial solvent coordination (despite 9 being crystallized from THF), and their color is the same in coordinating and non-coordinating solvents. These observations suggest that solvent coordination to the N(TMS)2 adduct is unfavorable, possibly because the TMS groups block the axial sites. The crystallographic structure of 9 displays a square planar molecular geometry at iron with τ4 = 0.09-0.11. Fe-N bond lengths in 9 are Fe1-N14 = 1.915(3) and Fe2-N18 = 1.899(3), which are distinguishable. Interestingly, the structure of 9 shows a potassium cation lying in the pincer plane, with a K1-S11 distance of 3.133(1) Å and K2-S15 distance of 3.126(1) Å. The Fe-S bond lengths on the side with potassium are 2.2459(9) Å (Fe1) and 2.2463(9) Å (Fe2), which are indistinguishable. The Fe-S bond length is, however, significantly shorter on the side of the molecule without the close S-K contact, with a difference of 0.0065(12) Å at Fe1 and 0.0080 (12) Å at Fe2. Metrical parameters for 9-crown are shown in Table . The 1 H NMR spectra of 9 and 9crown in THF-d8 have eight and nine, respectively, paramagnetically-shifted peaks between 19 and -68 ppm (Figures and), consistent with C2v symmetry. Their chemical shifts are nearly the same. |
617e93c2f9f05b6743e75dca | 19 | The molecule containing Fe2 shows a hydrogen bond between one NH3 proton and one uncoordinated THF. Hence, the shorter Fe-N length in Fe2 may be due to the hydrogen bond, which increases the basicity of the NH3. While the Fe-S bond lengths of 10 are similar to those in 8-crown and 9-crown, the Fe-C bond length is significantly shorter in 10 (Table ). This difference may arise because the thiolate and amido ligands exert a larger trans influence than the ammine. The solid-state IR spectrum of 10 shows four weak bands at 3352, 3299, 3236, and 3159 cm -1 that are assigned as N-H stretches (Figure ). Their frequencies are not suggestive of any significant coordination-induced N-H bond weakening. The 1 H NMR spectrum of 10 in THF-d8 has nine resonances between 175 ppm and -78 ppm, consistent with C2v symmetry (Figure ). Since the complex is Cs symmetric with THF coordination on one face, the spectrum in THF-d8 could indicate fast THF exchange on both sides of the complex. Its spectrum in C6D6 has severe broadening compared to THF-d8, and there are at least 15 resonances (Figure ). Assuming that coordinated THF has two different proton environments and that NH3 rotation on the NMR timescale renders its protons equivalent, a Cs symmetric geometry (with the mirror plane perpendicular to the pincer) would predict 14 peaks. Thus, the number of peaks in the C6D6solvated NMR spectrum suggests 10 has no symmetry in C6D6, though the reason for the low symmetry is not obvious. ). Low isomer shifts such as that of 11 are frequently observed in iron carbonyl complexes, reflecting the withdrawal of d-electron density from the iron nucleus by backbonding. The 1 H and C NMR spectra of 11 show 7 and 14 narrow peaks, respectively (Figures and). These spectra are consistent with a C2v-symmetric diamagnetic species. |
617e93c2f9f05b6743e75dca | 20 | The two most-downfield In an effort to produce complexes with fewer CO ligands, we treated 6 with substoichiometric CO. This treatment gave solutions with 1 H NMR spectra showing a large number of peaks with chemical shifts indicative of multiple paramagnetic species. Though we were unable to isolate any of these species, the mixture converted to diamagnetic 11 upon addition of greater than three equiv of CO (Figure ). |
617e93c2f9f05b6743e75dca | 21 | To gain insight into the Fe-S bonding in 8-crown, which has iron(III) in the most biomimetic S3C coordination sphere, Fe K-edge X-ray absorption spectra were measured. The spectrum showed two pre-edge features at ca. 7112 and 7114 eV. TD-DFT calculations reproduced the pre-edge features of the experimental spectrum (Figure , bottom). Frontier quasi-restricted orbital (QRO) analysis showed three singly-occupied orbitals of primarily Fe 3d character with < 10% S 3p mixing and a LUMO composed of 39.3% Fe 3d character and 21.3% S 3p character (Figure , top). The greater covalency between the Fe and S ligands in the LUMO is responsible for its increased energy. Based on the QRO analysis, the first pre-edge feature at 7112 eV is assigned to excitations of Fe 1s electrons into the three nearly-degenerate singly-occupied orbitals, while the second feature at 7114 eV is assigned to an Fe 1s to LUMO excitation. As expected for iron in the +3 oxidation state, the large contributions of ligand S orbitals to the frontier QROs indicate a large degree of Fe-S bond covalency. These moments are consistent with a well-isolated S = 3/2 ground state. Note that in THF solution, thiolate complex 8 is in equilibrium with its THF adduct 7. At 298 K, the solid-state magnetic susceptibilities of 8, 9, and 10 show cMT of 1.4, 2.3, and 1.9 cm 3 K mol -1 , respectively, under a 5000 Oe applied field. These values are also consistent with an S = 3/2 ground state and little mixing of excited states (Figures ). To further evaluate the zero-field splitting (ZFS) parameters, we measured the low-temperature solution X-band EPR spectra in MeTHF and magnetization curves at ≤ 10 K under field strengths of 1 T to 7 T. The data were fit within the constraints of the usual spin Hamiltonian for an S = 3/2 spin system. |
617e93c2f9f05b6743e75dca | 22 | Variable-temperature (3 K to 10 K) magnetization data of thiolate complex 8 showed nesting of curves obtained at different fields, with saturation around 1.6 µB (Figure ]. An acceptable simulation of the remainder of the EPR spectrum was obtained using S = 3/2, g and D values derived from the magnetization data, and E/D of 0.20. D strain of 28 cm -1 (100% of D) was required to account for the broad spectrum. The necessity of large D strain to capture the overall shape of the spectrum indicates that a wide distribution of ZFS parameters were sampled in the spectrum. EPR spectra with similarly straindominated line shapes have been observed for other S = 3/2 iron complexes. In our system, the crystallographically-observed flexibility of the thiolate ligand may result in a mixture of conformers with varying ZFS. Based on the range of observed g values, the range of E/D for different conformers of 8 present in MeTHF is estimated to be between 0.1 and 0.2. |
617e93c2f9f05b6743e75dca | 23 | and D = -30(2) cm -1 (Figure , middle) No reasonable fit was found using D > 0. The 5 K EPR spectrum shows a sharp, intense resonance at geff = 6.35 with broader major signals at geff = 2.55 and 1.57 (Figure , middle). These features were simulated as an S = 3/2 spin system with the g and D values from the magnetization, also introducing E/D = 0.333. Similar to the spectrum of 8, D strain of 5 cm -1 (17% of D) was necessary to account for line shapes. Unresolved hyperfine coupling to the ammine N nucleus may also contribute to broadening. There are small signals at |
617e93c2f9f05b6743e75dca | 24 | The magnetization curves for ammonia complex 10 show a pronounced field dependence with saturation occurring around 2.7 µB. They were fit to g = [2.01, 2.01, 2.07] (giso = 2.03) and D = -3.5(1) cm -1 (Figure , bottom). Fitting with D = +3.5 cm -1 also gave acceptable agreement with experimental data. At 10 K, the EPR spectrum of 10 shows features at geff = 6.15, 4.75, and 3.40, while signals at 2.36, 2.07, 2.05, and 1.99 are assigned to impurities (Figure , bottom). The spectrum was simulated as S = 3/2 with E/D = 0.106 using the g and negative D from the magnetization fit. An S = 1/2 impurity with g = [2.07, 2.05, 1.99] was included in the simulation in 1% relative abundance to the major S = 3/2 signal. D strain of 2.4 cm -1 (57% of D) was applied to account for the broad line shapes, though like 9, broadening could be compounded by unresolved hyperfine coupling to the ammine 14 N nucleus. A spectrum similar to that of 10 was reported for a square pyramidal ferric intermediate spin complex with E/D = 0.107. Simulations of the EPR spectrum with D = +3.5 cm -1 had a much smaller g = 6.15 component, which is inconsistent with the relatively intense signal seen experimentally. The g = 6.15 signal is most likely associated with transitions within the |±3/2⟩ doublet, and its high intensity at 10 K suggests that the ground state is |±3/2⟩, i.e. D < 0. |
617e93c2f9f05b6743e75dca | 25 | (top, green) at ~10 K, amide complex 9 (middle, purple) at ~5 K, and ammonia complex 10 (bottom, red) at ~10 K as 1 mM solutions in MeTHF. Each spectrum was collected using a 19 G modulation amplitude and a microwave power of either 0.064 mW (8) or 0.020 mW (9 and 10). |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.