João Paulo Leal

Random forests for feature selection in QSPR Models - an application for predicting standard enthalpy of formation of hydrocarbons

BackgroundOne of the main topics in the development of quantitative structure-property relationship (QSPR) predictive models is the identification of the subset of variables that represent the structure of a molecule and which are predictors for a given property. There are several automated feature selection methods, ranging from backward, forward or stepwise procedures, to further elaborated methodologies such as evolutionary programming. The problem lies in selecting the minimum subset of descriptors that can predict a certain property with a good performance, computationally efficient and in a more robust way, since the presence of irrelevant or redundant features can cause poor generalization capacity. In this paper an alternative selection method, based on Random Forests to determine the variable importance is proposed in the context of QSPR regression problems, with an application to a manually curated dataset for predicting standard enthalpy of formation. The subsequent predictive models are trained with support vector machines introducing the variables sequentially from a ranked list based on the variable importance.ResultsThe model generalizes well even with a high dimensional dataset and in the presence of highly correlated variables. The feature selection step was shown to yield lower prediction errors with RMSE values 23% lower than without feature selection, albeit using only 6% of the total number of variables (89 from the original 1485). The proposed approach further compared favourably with other feature selection methods and dimension reduction of the feature space. The predictive model was selected using a 10-fold cross validation procedure and, after selection, it was validated with an independent set to assess its performance when applied to new data and the results were similar to the ones obtained for the training set, supporting the robustness of the proposed approach.ConclusionsThe proposed methodology seemingly improves the prediction performance of standard enthalpy of formation of hydrocarbons using a limited set of molecular descriptors, providing faster and more cost-effective calculation of descriptors by reducing their numbers, and providing a better understanding of the underlying relationship between the molecular structure represented by descriptors and the property of interest.

The Nature of Protic Ionic Liquids in the Gas Phase Revisited: Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Study of 1,1,3,3-Tetramethylguanidinium Chloride

The sublimation/vaporization of the protic ionic liquid 1,1,3,3-tetramethylguanidinium chloride, [Htmg]Cl, was studied by Fourier transform ion cyclotron resonance mass spectrometry in the temperature range 298-488 K and under a reduced pressure of 3.2 x 10(-6) to 1.5 x 10(-5) Pa. The results show that no charged species are present in the vapor over the condensed phase. Furthermore, ion-molecule reaction studies found no evidence of neutral ion pairs in the gas phase. This indicates that the sublimation/vaporization of [Htmg]Cl conforms to the general mechanism postulated for the distillation of protic ionic liquids, which involves a proton transfer leading to the formation of the neutral acid and base precursors, in this case hydrogen chloride and 1,1,3,3-tetramethylguanidine.

Vaporisation of a Dicationic Ionic Liquid Revisited

The vaporization of a dicationic ionic liquid at moderate temperatures and under reduced pressures--recently studied by line-of-sight mass spectrometry--was further analyzed using an ion-cyclotron resonance mass spectroscopy technique that allows the monitoring of the different species present in the gas phase through the implementation of controlled ion-molecule reactions. The results support the view that the vapour phase of an aprotic dicationic ionic liquid is composed of neutral ion triplets (one dication attached to two anions). Molecular dynamics simulations were also performed in order to explain the magnitude of the vaporization enthalpies of dicationic ionic liquids vis-à-vis their monocationic counterparts.

Bridging the gap between ionic liquids and molten salts: group 1 metal salts of the bistriflamide anion in the gas phase.

Fourier transform ion cyclotron resonance mass spectrometry experiments showed that liquid Group 1 metal salts of the bistriflamide anion undergoing reduced-pressure distillation exhibit a remarkable behavior that is in transition between that of the vapor-liquid equilibrium characteristics of aprotic ionic liquids and that of the Group 1 metal halides: the unperturbed vapors resemble those of aprotic ionic liquids, in the sense that they are essentially composed of discrete ion pairs. However, the formation of large aggregates through a succession of ion-molecule reactions is closer to what might be expected for Group 1 metal halides. Similar experiments were also carried out with bis{(trifluoromethyl)sulfonyl}amine to investigate the effect of H(+), which despite being the smallest Group 1 cation, is generally regarded as a nonmetal species. In this case, instead of the complex ion-molecule reaction pattern found for the vapors of Group 1 metal salts, an equilibrium similar to those observed for aprotic ionic liquids was observed.

Europium(III) tetrakis(β-diketonate) complex as an ionic liquid: a calorimetric and spectroscopic study.

An intrinsic photoluminescent ionic liquid based on europium(III) tetrakis(β-diketonate) complex with a tetraalkylphosphonium as counterion was synthesized. Calorimetric measurements showed a melting point at 63 °C, which allows the ionic liquid classification. When cooling the material from the liquid state, metastable supercooled ionic liquid is obtained, as seen from NMR spectroscopy as well. Eu(III) photoluminescence is clearly observed while the absorption spectra of the ligand is dominant, showing the antenna effect. This was confirmed with submicrosecond time scale luminescence spectroscopy, where a rise of Eu(III) emission is observed with the correspondent decay of the ligand excited state. Temperature effects in the photoluminescence are also shown, being prominent above the melting point where the intensity decreases with Arrhenius behavior. Eu(III) luminescence decays also show features characteristic of energy migration between homologue Eu(III) species. Solvent effects were also studied by NMR and Luminescence spectroscopies, highlighting that the nucleophilicity of organic solvents such as n-alcohols leads to a coordination with Eu(III), which ultimately compromises the stability of the complex.

Gas phase actinide ion chemistry: Activation of alkanes and alkenes by thorium cations

Abstract The activation of methane and of several small alkanes and alkenes by thorium cations, as studied by Fourier transform ion cyclotron resonance/mass spectrometry, is described. Thermalized Th + ions dehydrogenate methane with a rather low efficiency ( k/k L = 0.02) to form ThCH 2 + . Th + ions react exothermically with the studied alkanes (ethane, propane, n -butane, isobutane and cyclopropane) and alkenes (ethene, propene and 1-butene): single and/or double dehydrogenation is observed for all the substrates studied and, in the case of cyclopropane, propene and 1-butene, loss of hydrocarbons is also observed. In secondary reactions of the primary products of cyclopropane and the alkenes C-C coupling processes may occur. This study indicates that Th + ions are more reactive than U + ions. A tentative preview of 5f metal ion reactivity is also presented, based on comparisons of the data reported herein with available data on lanthanide and uranium ions.

Additive Methods for Prediction of Thermochemical Properties. The Laidler Method Revisited. 1. Hydrocarbons

A new parameterization of the Laidler method for estimation of atomization enthalpies and standard enthalpies of formation at 298.15 K for several families of hydrocarbons alkanes, alkenes, alkynes, polyenes, poly-ynes, alkyl radicals, cycloalkanes, cycloalkenes, benzene derivatives, and polyaromatics is presented. A total of 200 compounds 164 for liquid phase are used for the calculation of the parameters. Comparison between the experimental values and those calculated using the group additive scheme led to an

Standard enthalpies of formation of sodium alkoxides

Abstract The standard enthalpies of formation of several crystalline sodium alkoxides, Δ H o f (NaOR, cr), have been determined by reaction-solution calorimetry. A linear correlation has been found between Δ H O f (NaOR, cr) and Δ H o f (ROH, 1/cr) for R = n-alkyl, enabling the prediction of data for other sodium alkoxides. The results were also used to derive the lattice energies and the thermochemical radii of the anions OR − .

HYDROCARBYL DERIVATIVES OF UCL2HB(PZ)32 : SYNTHESIS, CHARACTERIZATION AND REACTIVITY STUDIES TOWARDS PROTIC SUBSTRATES AND KETONES

Abstract The reaction of [UCl 2 HB(pz) 3 2 ] ( 1 ) with lithium alkyls LiR (R = Me, CH 2 SiMe 3 , C 6 H 4 - o -CH 2 NMe 2 ) in the 1:1 or 1:2 molar ratio affords the compounds [UCIRHB(pz) 3 2 ] (pz = C 3 H 3 N 2 , R = Me ( 2 ), CH 2 SiMe 3 ( 3 ), C 6 H 4 - o -CH 2 NMe 2 ( 4 )) and [UR 2 HB(pz) 3 2 ] (R = Me ( 5 ), CH 2 SiMe 3 ( 6 )) respectively in 60–80% yield. Complex 2 can also be obtained (60% yield) by redistribution at room temperature between the complexes 1 and 5 . Compounds 2, 3 and 4 react with pzH providing [UCl(pz)HB(pz) 3 2 ] ( 7 ) in almost quantitative yield and 5 and 6 react also with pzH leading to [U(pz) 2 HB(pz) 3 2 ] ( 8 ). The alkoxide [U(OC 6 H 4 - o -OMe) 2 HB(pz) 3 2 ] ( 9 ) was synthesized by reacting 5 or 6 with guaiacol. By reacting the chlorohydrocarbyls 2, 3 or 4 with excess of acetone the aldolate [UCl(OCMe 2 CH 2 (C=O)Me)HB(pz) 3 2 ] ( 11 ) was obtained, due to the activation of α -CH bond of acetone; however, for 3 the reaction is not clean and a mixture of 11 and [UCl(OCMe 2 CH 2 SiMe 3 )HB(pz) 3 2 ] ( 12 ) is always obtained. Stoichiometric amounts of acetone insert into the metal-carbon bonds of 2 and 5 yielding [UCl(O t Bu)HB(pz) 3 2 ] and [U(O t Bu) 2 HB(pz) 3 2 ] respectively, while the insertion product [U(OCMe 2 CH 2 SiMe 3 ) 2 HB(pz) 3 2 ] ( 10 ) can only be obtained when 6 reacts with excess of this substrate. 9 crystallizes from toluene/hexane in the triclinic space group P 1¯ with unit cell dimensions a = 12.295(2) A, b = 12.640(2) A, c = 13.994(2) A, α = 76.10(1)°, β = 72.50(1)°, γ = 80.71(1)°, V = 2004(2) A 3 and Z = 2. Recrystallization of a mixture containing [UCl(O t Bu)HB(pz) 32 ] and 11 led to a decomposition product which has been characterized by X-ray structural analysis as [UCl(Hpz)(O t Bu)(μ-O)B(μ-pz)(pz) 2 ] 2 ( 13 ): monoclinic P 2 1 / n , a = 13.701(3) A, b = 11.337(2) A, c = 14.857(4) A, β = 104.65(2)°, V = 2233(1) A 3 and Z = 2. Extended Huckel molecular orbital calculations provided some information on the bonding capabilities of the [UHB(pz) 3 2 ] fragment compared to [U(C 5 Me 5 ) 2 ] and on the vulnerability of the poly(pyrazolyl)borate ligands to nucleophilic attack.

Radiolytic degradation of gallic acid and its derivatives in aqueous solution

Polyphenols, like gallic acid (GA) released in the environment in larger amount, by inducing some unwanted oxidations, may constitute environmental hazard: their concentration in wastewater should be controlled. Radiolytic degradation of GA was investigated by pulse radiolysis and final product techniques in dilute aqueous solution. Subsidiary measurements were made with 3,4,5-trimethoxybenzoic acid (TMBA) and 3,4,5-trihydroxy methylbenzoate (MGA). The hydroxyl radical and hydrogen atom intermediates of water radiolysis react with the solute molecules yielding cyclohexadienyl radicals. The radicals formed in GA and MGA solutions in acid/base catalyzed water elimination decay to phenoxyl radicals. This reaction is not observed in TMBA solution. The hydrated electron intermediate of water decomposition adds to the carbonyl oxygen, the anion thus formed protonates on the ring forming cyclohexadienyl radical or on the carbonyl group forming carbonyl centred radical. The GA intermediates formed during reaction with primary water radicals in presence of oxygen transform to non-aromatic molecules, e.g., to aliphatic carboxylic acids.