Various ab initio methods have been applied to the studies of the molecular structure and energetics for the low-lying electronic states of the phenoxy radical and of the mechanism of its thermal decomposition. The calculations confirm the decomposition mechanism I (Scheme ) suggested earlier from experimental consideration. It involves the formation of the bicyclic intermediate E2 followed by α-CC bond cleavage to produce E3 and the elimination of the CO group from the latter leading to the main reaction products, C5H5 + CO. The rate-determining step is E2 → E3 with transition state T2. The activation energy, 52 kcal/mol, at our best G2M(rcc,MP2) level, is 8 kcal/mol higher than the value derived from experiment. A multistep RRKM calculation has been performed for the decomposition reaction. The theoretical rate constants agree well with the experimental results at lower temperatures (T < 1200 K) but deviate from experiment above 1300 K. Plausible reasons for the deviation have been discussed. On the basis of state-specific and state-averaged CASSCF calculations, the observed electronic transitions for the phenoxy radical have been assigned.

The geometry and harmonic force field of the phenoxyl radical are determined by ab initio calculations of the electronic structure. All the π electrons are involved in strong nondynamical correlations, which are treated in a complete π active space wave function. Results from using multiconfigurational self-consistent-field orbitals in the active space are found to be closely approximated by use of the more economical unrestricted Hartree–Fock natural orbitals. Large polarized basis sets are also required to obtain qualitatively correct results. Properties of the carbonyl chromophore are particularly interesting. Compared to closed-shell analogs, the equilibrium geometry indicates the CO bond in phenoxyl to be nearly as short as a typical double bond, whereas the scaled quantum-mechanical force field shows it to have strength intermediate between those of typical single and double bonds. The calculated fundamental vibrational frequencies generally confirm current interpretations of the major features of the observed experimental resonance Raman spectra, while leading to reassignments of some minor features. Some limited theoretical characterization of valence excited states provides qualitative explanation of the observed intensity pattern. Calculations carried out at lower levels of theory consistently find the CO bond to be too long and weak, leading to misassignment of the observed vibrational spectrum.

In drug development, early assessments of pharmacokinetic and toxic properties are important stepping stones to avoid costly and unnecessary failures. Considerable progress has recently been made in the development of computer-based (in silico) models to estimate such properties. Nonetheless, such models can be further improved in terms of their ability to make predictions more rapidly, easily, and with greater reliability. To address this issue, we have used our vNN method to develop 15 absorption, distribution, metabolism, excretion, and toxicity (ADMET) prediction models. These models quickly assess some of the most important properties of potential drug candidates, including their cytotoxicity, mutagenicity, cardiotoxicity, drug-drug interactions, microsomal stability, and likelihood of causing drug-induced liver injury. Here we summarize the ability of each of these models to predict such properties and discuss their overall performance. All of these ADMET models are publically available on our website (https://vnnadmet.bhsai.org/), which also offers the capability of using the vNN method to customize and build new models.

A simple atom equivalent method for converting density functional theory (DFT) energies to enthalpies of formation is described and its performance, in conjunction with six DFT methods, was examined. For 23 stable hydrocarbons with well-established experimental data, the root mean square deviations between the calculated and experimental enthalpies of formation range from 1 to 6 kcal/mol. The smallest deviation was obtained with the B3LYP energies, and the largest deviation was obtained with the LSDA energies. The B3LYP atom equivalents of carbon and hydrogen derived from stable hydrocarbons were used without adjustment to calculate the enthalpies of formation of some free radicals and carbocations. The mean deviation between the calculated and experimental results is about 2 kcal/mol, which is of the same order as experimental uncertainties for these highly reactive species.

[1] We collected and processed a large amount of high-quality broadband teleseismic waveform data recorded by the 48 Chinese National Digital Seismic Network stations to estimate large-scale lateral variations of crustal thickness and V p /V s ratio (hence Poisson's ratio) beneath China. A statistical method was used to select mutually coherent receiver functions at each station, which yielded over 200 traces for most of the stations. With the conventional HÀk (the crustal thickness and V p /V s ratio) approach, there is a large trade-off between H and k. Consequently, multiple maxima are frequently observed in the HÀk domain. We introduced a weight function that measures the coherence between the P-to-S conversion and the reverberation phases at each HÀk grid to reduce the trade-off. A 4th-root stacking method was further applied to reduce uncorrelated noise relative to the linear stack. These modifications turned out to be very effective in reducing the HÀk trade-off and yielded reliable estimates of crustal thickness and V p /V s ratio. The crust beneath eastern China is as thin as 31-33 km and the underlying Moho is relatively flat and sharp. In the western part of China, the crust is considerably thicker and shows large variations. The Moho is observed at about 51 km depth along the Tian Shan fold system and about 84 km deep beneath the central part of the Tibetan Plateau. The transition occurs at the so-called N-S belt between about 100°and 110°E, which is featured by unusually high seismicity and large gravity anomalies. The average V p /Vs ratio over the mainland China crust is about 1.730 (s = 0.249), significantly lower than the global average 1.78 (s = 0.27) of the continental crust. This lower V p /V s ratio may suggest a general absence of mafic lowermost crustal layer beneath China.

Density functional theory B3LYP/6-31G** and ab initio MP2/6-31G** and MP4(SDQ)/6−311++G** calculations were carried out to study the structures and isomerization and decomposition mechanisms of oxalic acid. The B3LYP structures and relative energies of the rotational isomers of oxalic acid are found very similar to MP2 results, confirming that the most stable isomer is the doubly intramolecular hydrogen-bonded C 2 h structure E1, with four other planar isomers within 6 kcal/mol. It is predicted that unimolecular formation of carbon dioxide and dihydroxycarbene (DHC) from oxalic acid has an activation barrier of 31 kcal/mol and that unimolecular formation of HCOOH from DHC has an activation barrier about 31 kcal/mol higher. The unimolecular formation of CO2, CO and H2O from oxalic acid via a concerted transition state has an activation barrier of only 42 kcal/mol, indicating it is a more favorable unimolecular decomposition channel. On the other hand, hydrogen migration from oxygen to carbon of DHC to produce HCOOH can be accomplished through a hydrogen exchange with H2O (a model for oxalic acid) with an activation barrier of less than 37 kcal/mol. Transition state theory calculations indicate that this bimolecular channel might be responsible for the rapid formation of CO2 and HCOOH in gas phase oxalic acid thermal decomposition, thus confirming the proposal of Bock and Redington. With increasing temperature the unimolecular channel to produce CO2, CO, and H2O might become significant.

A new molecular lipoaffinity descriptor was introduced in this paper to account for the effect of molecular hydrophobicity on blood-brain barrier penetration. The descriptor was defined based on Kier and Hall's atom-type electrotopological state indices. Its evaluation requires 2-D molecular bonding information only. A multiple linear regression equation using this descriptor and molecular weight reproduces the experimental logBB values of 55 training set compounds and 11 test set compounds satisfactorily with statistical parameters nearly identical to the best models based on polar surface area and ClogP. The results indicate that the lipoaffinity descriptor defined in this paper may be a significant descriptor for molecular transport properties across lipid bilayers.

A simple QSPR model, based on seven 1D and 2D descriptors and artificial neural network, was developed for fast evaluation of aqueous solubility. The model was able to predict the molar solubility of a diverse set of 1312 organic compounds with an overall correlation coefficient of 0.92 and a standard deviation of 0.72 log unit between the calculated and experimental data. Considering the fact that the estimated uncertainty of the experimental data is no less than 0.5 log unit, the results demonstrate that carefully chosen physically meaningful 1D and 2D descriptors encode sufficient molecular information for fast and reasonably reliable prediction of aqueous solubility with a simple neural network. As a comparison, we calculated the solubility of a test set of 258 compounds, ranging from simple hydrocarbons to more complex multifunctional organic molecules, with a commercial program (QMPR+ version 2.0.1 of SimulationPlus Inc.) and compared the results with predictions from our model. Statistical parameters indicate that for small and simple organic compounds, QMPR+ outperforms our model. However for more complex multifunctional molecules, our model is superior.