Predicting chain dimensions from an artificial neural network model

Artificial neural network models are used to investigate polymer chain dimensions. In our model, the input nodes are glass transition temperature (T_g), entanglement molecular weight (M_e), and melt density (ρ). The number of nodes in the hidden layer is eight. We found that the relative error for prediction of the characteristic ratio ranges from 0.77 to 7.5% and that the overall average error is 3.57%. Artificial neural network models may provide a new method for studying statistics properties of polymer chains. © 2000 John Wiley & Sons, Inc. J Polym Sci B: Polym Phys 38: 3163–3167, 2000     

On the use of computational neural networks for the prediction of polymer properties

A general purpose computational paradigm using neural networks is shown to be capable of efficiently predicting properties of polymeric compounds based on the structure and composition of the monomeric repeat unit. Results are discussed for the prediction of the heat capacity, glass transition temperature, melting temperature, change in the heat capacity at the glass transition temperature, degradation temperature, tensile strength and modulus, ultimate elongation, and compressive strength for 11 different families of polymers. The accuracies of the predictions range from 1–13% average absolute error. The worst results were obtained for the mechanical properties (tensile strength and modulus: 13%, 7% elongation: 12%, and compressive strength: 8%) and the best results for the thermal properties (heat capacity, glass transition temperature, and melting point: <4%). A simple modification to the overall method is devised to better take into account the fact that the mechanical properties are experimentally determined with a fairly large range (due to variability in measurement procedures and especially the sample). This modification treats the bounds on the range for the mechanical properties as complex numbers (complex, modular neural networks) and leads to more rapid optimization with a smaller average error (reduced by 3%).Dedicated to Professor Bernhard Wunderlich on the occasion of his 65th birthdayThis research was sponsored by the Division of Materials Sciences, Office of Basic Energy Sciences, U.S. Department of Energy, under Contract No. DE-AC05-84R21400 with Lockheed Martin Energy Systems, Inc. We would like to express our gratitude for the continued collaboration, support, and interest of Prof. Wunderlich in our research. We would also like to thank participants of the 1st DOE Workshop on Applications of Neural Networks in Materials Sciences for useful discussion on materials properties and neural networks.     

Artificial neural system modeling of Monte Carlo simulations of polymers

In this work, a neural network was used to learn features in potential energy surfaces and relate those features to conformational properties of a series of polymers. Specifically, we modeled Monte Carlo simulations of 20 polymers in which we calculated the characteristic ratio and the temperature coefficient of the characteristic ratio for each polymer. We first created 20 rotational potential energy surfaces using MNDO procedures and then used these energy surfaces to produce 10000 chains, each chain 100 bonds long. From these results we calculated the mean-square end-to-end distance, the characteristic ratio and its corresponding temperature coefficient. A neural network was then used to model the results of these Monte Carlo calculations. We found that artificial neural network simulations were highly accurate in predicting the outcome of the Monte Carlo calculations for polymers for which it was not trained. The overall average error for prediction of the characteristic ratio was 4,82%, and the overall average error for prediction of the temperature coefficient was 0,89%.     

Glass transition in poly(propylene glycol) networks

Difficulty in controlling and determining the structural parameters of polymer networks has hindered experimental studies on the glass transition in crosslinked polymers. A series of wellcharacterized networks of poly(propylene glycol) having narrow network chain-length distributions and average molecular weight between crosslinks M _c in the range of 425–3000 has been prepared. The glass transition temperatures T_g of these networks were found to vary linearly with M , consistent with several theoretical treatments. Both the physical crosslinking and the incorporation of crosslinking agent into the system (a “copolymer” effect) are shown to be responsible for increase in T_g upon crosslinking in this system. Varying the network chain-length distribution without changing M _c did not affect the T_g of the system. The chemical nature of the crosslinking agent, however, does affect the T_g of the network, particularly at high crosslink densities.     

Prediction of coefficients of the Langmuir adsorption isotherm using various artificial intelligence (AI) techniques

Adsorption is a process that utilizes porous solid materials to separate some solutes from gas or liquid mixtures. The extent of this separation is often determined using the adsorption isotherms, i.e., semi-empirical correlation for relating the amount of adsorbed substances by the solid medium to its associated concentration in fluid phase at constant temperature. Prior to employing an adsorption isotherm, its coefficients should be adjusted using experimental data of a considered adsorption system. In this study, the coefficients of Langmuir model have been predicted using various types of artificial neural networks (ANNs), support vector machines, and adaptive neuro fuzzy interface systems, and coupled scheme of ANN-genetic algorithm. The employed ANN types are multi-layer perceptron neural network (MLPNN), radial basis function neural network, cascade feedforward neural network, and generalized neural network. The considered coefficients tried to be modeled as functions of temperature, pH, adsorbent density, and adsorbate molecular weight. Predictive accuracies of the AI techniques have been compared utilizing different statistical indices such as correlation coefficient (R²), mean square error, and absolute average relative deviation (AARD%). The results indicated that MLPNN was the most accurate model for predicting the coefficients of Langmuir isotherm, due to its AARDs of 24.64 and 22.40% for the first and second coefficients, respectively.     

Synthesis and Characterization of Novel Methacrylates Derived from Morpholine and Pyrrolidine: The Determination of Kinetic Parameters with Thermogravimetric Analysis

Abstract

The synthesis of two new methacrylate esters containing morpholine and pyrrolidine group are described. The monomers produced from the reaction of corresponding morpholino chloroacetamide and pyrrolidino chloroacetamide with sodium methacrylate were polymerized in DMSO solution at 65°C using AIBN as an initiator. The monomers and their polymers were characterized by Fourier transform infrared (FTIR), ¹H‐ and ¹³C‐NMR spectroscopy. The glass transition temperature of the polymers were investigated by DSC and thermal decomposition activation energies were calculated by the Ozawa method using the SETARAM Labsys thermogravimetric analysis (TGA) thermobalance, respectively. By using gel permeation chromatography, weight average (M¯_w) and number average (M¯_n) molecular weights and polidispersity indices of the polymers were determined.     

Threshold tear strength of elastomers

Tear strengths have been measured for a wide variety of molecular networks under threshold conditions; i.e., at high temperatures, low rates of tearing, and with swollen samples. For all of the polymers examined, the threshold tear strength was found to be proportional to the square root of the average molecular weight M_c of network strands, in agreement with theory. However, for the same M_c, and hence for similar values of elastic modulus, different polymers showed major differences in threshold tear strength. The tear strength of polydimethylsiloxane networks was only about one-third that for networks of polybutadiene and cis-polyisoprene, and the values obtained for polyphosphazene networks were only about one-fifth as large, at the same M_c. These striking differences are attributed to differences in network strand length and extensibility for the same molecular weight. The threshold tear strengths are shown to be in satisfactory quantitative agreement with theoretically predicted values on this basis.     

盐湖水化学类型的人工神经网络判别方法

研究了作为典型径向基函数网络之一的概率神经网络在盐湖水化学类型分类预测中的应用，验证了该方法的可靠性，得到了满意的分类预测结果。实验结果和网络结构分析表明，概率神经网络方法比熟知的反向传播算法(BP)网络要好。概率神经网络的研究应用为化学模式识别提供了一个新工具。     

Effect of prepolymer molecular weight on solid state polymerization of poly(bisphenol a carbonate) with nitrogen as a sweep fluid

The effect of prepolymer molecular weight on the solid‐state polymerization (SSP) of poly(bisphenol A carbonate) was investigated using nitrogen (N₂) as a sweep fluid. Prepolymers with different number–average molecular weights, 3800 and 2400 g/mol, were synthesized using melt transesterification. SSP of the two prepolymers then was carried out at reaction temperatures in the range 120–190 °C, with a prepolymer particle size in the range 20–45 μm and a N₂ flow rate of 1600 mL/min. The glass transition temperature (T_g), number–average molecular weight (M_n), and percent crystallinity were measured at various times during each SSP. The phenyl‐to‐phenolic end‐group ratio of the prepolymers and the solid‐state synthesized polymers was determined using 125.76 MHz ¹³C and 500.13 MHz ¹H nuclear magnetic resonance (NMR) spectroscopy. At each reaction temperature, SSP of the higher‐molecular‐weight prepolymer (M_n = 3800 g/mol) always resulted in higher‐molecular‐weight polymers, compared with the polymers synthesized using the lower molecular weight prepolymer (M_n = 2400 g/mol). Both the crystallinity and the lamellar thickness of the polymers synthesized from the lower‐molecular‐weight prepolymer were significantly higher than for those synthesized from the higher‐molecular‐weight prepolymer. Higher crystallinity and lamellar thickness may lower the reaction rate by reducing chain‐end mobility, effectively reducing the rate constant for the reaction of end groups. © 2008 Wiley Periodicals, Inc. J Polym Sci Part A: Polym Chem 46: 4959–4969, 2008     

Extension of the chain‐end,free‐volume theory for predicting glass temperature as a function of conversion in hyperbranched polymers obtained through one‐pot approaches

Compared with linear polymers, more factors may affect the glass‐transition temperature (T_g) of a hyperbranched structure, for instance, the contents of end groups, the chemical properties of end groups, branching junctions, and the compactness of a hyperbranched structure. T_g's decrease with increasing content of end‐group free volumes, whereas they increase with increasing polarity of end groups, junction density, or compactness of a hyperbranched structure. However, end‐group free volumes are often a prevailing factor according to the literature. In this work, chain‐end, free‐volume theory was extended for predicting the relations of T_g to conversion (X) and molecular weight (M) in hyperbranched polymers obtained through one‐pot approaches of either polycondensation or self‐condensing vinyl polymerization. The theoretical relations of polymerization degrees to monomer conversions in developing processes of hyperbranched structures reported in the literature were applied in the extended model, and some interesting results were obtained. T_g's of hyperbranched polymers showed a nonlinear relation to reciprocal molecular weight, which differed from the linear relation observed in linear polymers. T_g values decreased with increasing molecular weight in the low‐molecular‐weight range; however, they increased with increasing molecular weight in the high‐molecular‐weight range. T_g values decreased with increasing log M and then turned to a constant value in the high‐molecular‐weight range. The plot of T_g versus 1/M or log M for hyperbranched polymers may exhibit intersecting straight‐line behaviors. The intersection or transition does not result from entanglements that account for such intersections in linear polymers but from a nonlinear feature in hyperbranched polymers according to chain‐end, free‐volume theory. However, the conclusions obtained in this work cannot be extended to dendrimers because after the third generation, the end‐group extents of a dendrimer decrease with molecular weight. Thus, it is very possible for a dendrimer that T_g increases with 1/M before the third generation; however, it decreases with 1/M after the third generation. © 2004 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 42: 1235–1242, 2004     

A bicyclic conjugated diene monomer,tetrahydroindene, for cationic polymerization and a novel alicyclic hydrocarbon polymer with heat resistance

This study was directed toward the cationic polymerization of tetrahydroindene (i.e., bicyclo[4.3.0]‐2,9‐nonadiene), a bicyclic conjugated diene monomer, with a series of Lewis acids, especially focusing on the synthesis of high‐molecular‐weight polymers and subsequent hydrogenation for novel cycloolefin polymers with high service temperatures. EtAlCl₂ or SnCl₄ induced an efficient and quantitative cationic polymerization of tetrahydroindene to afford polymers with relatively high molecular weights (number‐average molecular weight > 20,000) and 1,4‐enchainment bicyclic main‐chain structures. The subsequent hydrogenation of the obtained poly(tetrahydroindene) with p‐toluenesulfonyl hydrazide resulted in a saturated alicyclic hydrocarbon polymer with a relatively high glass transition (glass‐transition temperature = 220 °C) and improved pyrolysis temperature (10% weight loss at 480 °C). The new diene monomer was randomly copolymerized with cyclopentadiene at various feed ratios in the presence of EtAlCl₂ to give novel cycloolefin copolymers, which were subsequently hydrogenated into alicyclic copolymers with variable glass‐transition temperatures (70–220 °C). © 2006 Wiley Periodicals, Inc. J Polym Sci Part A: Polym Chem 44: 6214–6225, 2006     

弹性反传神经网络法预测烷基苯的疏水性常数

 介绍了应用人工神经网络预测烷基苯分子疏水性常数的方法。该法同传统方法相比 ,具有操作简便 ,适用范围广的特点。基于误差反传神经网络 ,建立了分子连接性指数 (χ)、范德华表面积 (Aw)和疏水性常数 (logP)之间的数学模型。应用该模型对烷基苯分子的疏水性常数进行预测 ,其平均相对偏差为 0 6 7%。并且通过与标准误差反传算法和自适应学习算法相比较 ,发现弹性反传算法具有训练速度快 ,参数选择简单的特点。     

Characteristics of liquid-crystalline cholesteryl and cyanobiphenyl side chains pendent to trans-polypentenamer

trans-Polypentenamers with thermotropic liquid-crystalline side chains cholesteryl and cyanobiphenyl were prepared by ring-opening polymerization of vinylcyclopropane monomers with proper substituents. Molecular weights of the polymers were in the range of 25000 to 80000 and the ratios of weight- to number-average molecular weights M _w/M _n were between 3.3 and 3.8. The glass transition temperature values of the polymers were 35°C ( 4a ) and 39°C ( 4b ). Monomers 3a and 3b present cholesteric and smectic mesomorphism, respectively. On the other hand, polymers 4a and 4b present only a smectic mesophase.     

Usage of Artificial Neural Network (Back Propagation) in Optimising Salicylic Acid Determination with Ferric(III) Nitrate

Abstract

The use of artificial neural networks (ANN) in optimizing salicylic acid (SA) determination is presented in this paper. A simple and rapid spectrophotometric method for salicylic acid (SA) determination was carried out based on the complexation of salicylic acid–ferric(III) nitrate, SAFe(III). The SA forms a stable purple complex with ferric(III) nitrate at pH 2.45. The useful dynamic linear range is 0.01–0.35 g/L. It has a maximum absorption at 524 nm and the stability is more than 50 hours. The results were used for artificial neural networks (ANNs) training to optimize data. For training and validation purposes, a back‐propagation (BP) artificial neural network (ANN) was used. The results showed that ANN technique was very effective and useful in broadening the limited dynamic linear response range mentioned to an extensive calibration response (0.01–0.70 g/L). It was found that a network with 22 hidden neurons was highly accurate in predicting the determination of SA. This network scores a summation of squared error (SSE) skill and low average predicted error of 0.0078 and 0.00427 g/L, respectively.     

The effect of network chain-length distribution,specifically bimodality,on strain-induced crystallization

Polyurethane elastomers were prepared from a series of poly(ethylene oxide) samples by end-linking the chains into “model” trifunctional networks. The molecular weight M_c between crosslinks in such networks is simply the number-average molecular weight M_n of the precursor polymer. End-linking samples separately gave networks with unimodal distributions of network chain lengths, whereas end-linking mixtures of two samples having very different values of M_n gave bimodal distributions with average values of M_c equal to the average value of M_n for the two samples. Stress-strain isotherms in elongation were obtained for these networks, both unswollen and swollen to various extents. Strain-induced crystallization was manifested in elastic properties that changed significantly with changes in temperature. Swelling has more complicated effects, since it causes deformation of the network chains as well as melting of some of the crystallites. Comparisons among stress-strain isotherms at constant M_c indicate that bimodality facilitates strain-induced crystallization.     

Influence of molecular weight on the phase behaviors of comb‐like polymers with long n‐alkyl side chains based on mesogen‐jacketed liquid crystalline polymers

A series of comb‐like polymers, poly{2,5‐bis[(4‐octadecyloxyphenyl)oxycarbonyl]‐styrenes{ (P‐OC18s) with different molecular weights (M_n) and low molecular weight distributions have been successfully synthesized via atom transfer radical polymerization. The phase behaviors have been investigated by a combination of techniques including differential scanning calorimetry, polarized optical microscopy, wide‐angle X‐ray diffraction, and temperature‐variable FTIR spectroscopy. One hand, phase behaviors of the alkyl tails were strongly influenced by the mesogens of polymers, leading to the poor packing of the alkyl tails and the low melting. The other hand, the liquid crystalline phase structures of polymers were found to be strongly M_n dependent. The samples with M_n ≤ 4.6 × 10⁴ formed a smectic phase in low temperature and an isotropic phase in high temperature. The samples with M_n ≥ 5.2 × 10⁴ displayed a reentrant isotropic phase, which was separating the smectic phase and columnar nematic phase. Meantime, the experiment results showed that the glass temperature and the transition temperature from smectic phase to isotropic phase both slightly increased with the increase of M_nS; however, the transition temperature from isotropic phase to columnar phase sharply decreased with the M_nS improved. The reappearance of isotropic phase is due to the competing between the driving force of the enthalpy and the driving force of the entropy. © 2012 Wiley Periodicals, Inc. J Polym Sci Part A: Polym Chem, 2013     

Dendritic analogues of engineering plastics - a general one-step synthesis of dendritic polyaryl ethers

The first general single-step route to dendritic or cascade polyaryl ethers analogous to common linear polyaryl ethers is described. The sodium salts of four AB₂ monomers each containing a single phenolic hydroxyl group and two aryl fluorides activated toward nucleophilic substitution by carbonyl, sulphonyl, or tetrafluorophenyl moieties are shown to polymerize in hot N, N-dimethylacetamide. The products are high molecular weight polymers (7000 < M_n < 36000), have narrow polydispersities (1.50 < M_w/M_n < 4.50), and are highly soluble in organic solvents. The molecular weights of two of the polymers increase with monomer concentration. The polymers are thermally stable (500 °C under N₂) and have glass transition temperatures ranging from 135 to 231 °C.     

MC-determination in elastomers by 1H-NMR relaxation and 2H-NMR spectroscopy

We show how ¹H–NMR transversal relaxation and ²H–NMR spectroscopy can be used for the determination of the number average molecular mass M_C in typical elastomers at temperatures well above the glass transition temperature. M_C-results of the different NMR methods are compared among one another and with M_C-results of other common independent methods. Moreover, the NMR measurements provide a number of additional useful parameters: correlation times, portions of dangling chain ends and of sol molecules, molecular order.