Polymer nanocomposite data set for prediction and modeling of transmitted light intensity via machine learning models

Light transmission from polymers doped with carbon-based nanofillers such as multi-walled carbon nanotubes (MWCNTs) is one of the key challenges in optimal design of conductive nanocomposite materials. Incorporation of MWCNT nanofillers into polymers at high concentrations remarkably improves electrical conductivity of neat, insulating polymers, however; it also leads to lower transmission of photons from heterogeneous medium of resultant nanocomposite since MWCNT nanofillers act as light scatters. In this respect, it is experimentally important to determine the critical concentration of MWCNTs in preserving the optical transparency and electrical conductivity of the material, simultaneously. In our data set, we show transmitted light (photon) intensities of polystyrene (PS) latexes doped with MWCNT nanofillers at different concentrations (i.e mass fractions). Emulsion polymerization experiments were first conducted to align particle size and molecular weight values of latex particles based on initiator and surfactant concentrations used during polymerization reaction. Three different sets of PS latexes, each having different molecular weight and particle diameter, were then mixed with nanofillers at different mass fractions up to 20 wt%. Afterward, each prepared nanocomposite solution was deposited on transparent glass plates by the same amounts of solution droplets to be annealed in oven. Transmitted light intensity of the samples was recorded using UV-Vis spectrophotometer after each annealing step in a broad temperature range varying between 100 and 250 Celsius degree. Each measurement parameter was listed in a separate column and 512 different measurement data (where each row represents one measurement) were collected in total. The first four columns (initiator concentration, surfactant concentration, molecular weight of PS latex and mean particle diameter of PS latex particles) belong to PS latex polymers. The fifth column (MWCNT concentration) represents how much MWCNT nanofillers were added into PS latexes. The last two columns stand for annealing temperature and measured transmitted light intensity values. Our experimental data clearly show that transmitted light intensity of PS/MWCNT nanocomposite films is significantly low in the presence of high MWCNT content and at low annealing temperatures. Our optical data set collected from photon transmission measurements is suitable for studying polymer film physics applying percolation and other statistical theories, material characterization, mathematical modeling and machine learning. Our experimental data set can be of great interest for researchers and computational scientists in developing neural network topologies, support vector regression (SVR), adaptive neuro fuzzy interference system (ANFIS) and other machine learning models combined with many advanced optimization algorithms such as genetic algorithm (GA), particle swarm optimization (PSO) and artificial bee colony algorithm (ABC).

掺杂多壁碳纳米管（multi-walled carbon nanotubes, MWCNTs）这类碳基纳米填料的聚合物的光传输性能，是导电纳米复合材料优化设计中的核心挑战之一。将多壁碳纳米管纳米填料以高浓度掺入聚合物中，可显著提升纯绝缘聚合物的导电性能，但同时也会导致所得纳米复合材料非均质介质的光子传输率下降，因为多壁碳纳米管纳米填料会充当光散射体。就此而言，实验上确定同时维持材料光学透明度与导电性能的多壁碳纳米管临界浓度，具有重要意义。本数据集收录了掺杂不同浓度（即质量分数）多壁碳纳米管纳米填料的聚苯乙烯（polystyrene, PS）胶乳的透射光（光子）强度数据。首先开展乳液聚合实验，依据聚合反应过程中使用的引发剂与表面活性剂浓度，调控胶乳颗粒的粒径与分子量参数。随后将三组分别具有不同分子量与粒径的聚苯乙烯胶乳，与浓度最高达20wt%的不同质量分数的纳米填料混合。之后，将每份制备好的纳米复合溶液以等量液滴的形式涂覆于透明玻璃板上，随后置于烘箱中进行退火处理。在100至250摄氏度的宽温度区间内，每完成一次退火步骤后，均使用紫外-可见分光光度计（UV-Vis spectrophotometer）记录样品的透射光强度。每项测量参数均单独列为一列，最终共收集到512组测量数据（每一行代表一次测量结果）。前四列分别为引发剂浓度、表面活性剂浓度、聚苯乙烯胶乳的分子量以及聚苯乙烯胶乳颗粒的平均粒径，对应聚苯乙烯胶乳聚合物的相关参数。第五列为多壁碳纳米管浓度，代表掺入聚苯乙烯胶乳中的多壁碳纳米管纳米填料的用量。最后两列则分别为退火温度与实测透射光强度值。本实验数据清晰表明，当多壁碳纳米管含量较高且退火温度较低时，聚苯乙烯/多壁碳纳米管纳米复合薄膜的透射光强度会显著降低。本数据集通过光子传输测量获得的光学数据，适用于基于渗流理论与其他统计理论开展聚合物薄膜物理研究、材料表征、数学建模以及机器学习相关工作。本实验数据集对于开发神经网络拓扑结构、支持向量回归（support vector regression, SVR）、自适应神经模糊干扰系统（adaptive neuro fuzzy interference system, ANFIS）以及结合遗传算法（genetic algorithm, GA）、粒子群优化（particle swarm optimization, PSO）、人工蜂群算法（artificial bee colony algorithm, ABC）等多种先进优化算法的其他机器学习模型的研究人员与计算科学家而言，具有重要的参考价值。