Nanoparticles and polyaniline electrical conductivity

Polyaniline is a conductive polymer that attracts the attention of many researchers around the world. The history of this polymer begins in 1862 when Letheby first reported this material. Since then, a myriad of studies has been conducted on this material, and new works continue to investigate the potential of this material. Polyaniline has been improved with the help of Nanotechnology. The use of nanofillers has been seen as a quick and economical way to modify materials, driving innovations based on new physical and chemical properties from the conductive polymer materials and nanoparticles joining. Several works address the use of different nanoparticles, which leads to the practical impossibility of sifting through all this information. Thus, this work proposes to systematically collect data in the literature and investigate which nanoparticles can increase the electrical conductivity of Polyaniline (PAni). The results obtained demonstrate that among the possible nanofillers, graphene and carbon nanotubes have great prominence. Furthermore, the results of the meta-analysis prove that PAni's conductivity increases when this polymer is modified with the aforementioned nanofillers.

The data in Figure 1 follow a polynomial of order 2. The coefficient of determination (R 2 ) found was equal to 0.989. The model and the associated R 2 numerically demonstrate that interest in the PAni topic is accelerating since the 1980s. With the advent of nanotechnology, it is evident that these numbers will continue to grow over the next years.
Then, focusing on the key question of this research, 392 documents were found using the search key TITLE-ABS-KEY ( nano* AND polyaniline AND improv* AND "electrical conductiv*" OR "electrical resistivit*" ) AND ( LIMIT-TO ( DOCTYPE , "ar")). The titles and abstracts of these documents were saved in RIS format and analyzed using VOSviewer software. The obtained results are shown in Figure 2. Besides, the *MAP.txt and *NET.txt files generated by VOSviewer are available at https://github.com/ftir-mc/Nanoparticlesimproving-PAni-conductivity.git. Thus, two new keywords, "carbon nanotubes" and "graphene", gained relevance. So these two new words were added to the search key (TITLE-ABS-KEY ( nano* AND polyaniline AND improv* AND "electrical conductiv*" OR "electrical resistivit*" AND graphene OR "carbon nanotub*") AND ( LIMIT-TO ( DOCTYPE , "ar " ) )), in a new refinement, which returned 177 documents (132-308).
All of these 177 documents were downloaded and their data were scratched looking for statistically relevant information for constructing the analysis here proposed. Table 1 shows the 1 st author of the study and the year of publication, the used nanofiller as well as the number of experimental conditions tested, the total number of replicates, and the Adjusted Correlation (adjR).
From the 177 selected documents, 15 presented useful data for the proposed meta-analysis. However, among the documents evaluated, some had more than one useful case. So, a total of k=20 studies were included in the analysis. The observed Fisher r-to-z transformed correlation coefficients ranged from 0.3713 to 3.1320, with the majority of estimates being positive (100%  Funnel plot is shown in Figure 3.  Fisher's transformation is commonly used to eliminate a possible bias in the untransformed correlation coefficient (309). As the population value of correlation becomes further from zero, the sampling distribution of correlation coefficients becomes skewed, and Fisher's transformation normalizes this sampling distribution. Empirical evidence suggests that transforming the correlation coefficient can be beneficial (310). Mainly because many meta-analytic methods assume that the sampling distribution of observed results is normal. When the correlation in a particular study is far from zero, and the sample size is small, then the gross sample distribution of the correlation becomes skewed, no longer being closely approximated by a normal distribution. In this context, Fisher's r-to-z transformation is an effective normalization transformation, which makes the statistical analysis of correlations independent of unknown quantities (311 (white), 95% (gray), 99% (dark gray), and the rest are beyond 99% (red) probability.
Forest plots are a key-graphical method used in meta-analysis. The forest plot is the graphical representation resulting from quantitative systematic reviews. This representation is designed to compare the effects of treatments in quantitative studies. The term "forest" comes from the idea that the graph resembles a forest of lines. Originally the forest plot was designed to compare randomized clinical trials that addressed a common theme. Currently, however, this representation is quite prevalent in observational studies, to visually present the mathematical significance of the joint conclusions of several works analyzed as a block (313)(314)(315).
The Forest Plot, shown in Figure 4, lists all selected studies. The relevance of the studies is presented in percentage form. All studies had similar percentage relevance. The effects and their 95% probability confidence limits are shown to the right side of the percent relevance. The modeled effect is equal to 1.64, with the lower bound equal to 1.33 and the upper bound equal to 1.96.
Thus, the modeled effect is positive and nonzero. So, the meta-analysis proved that the nanofillers in this study increase the conductivity of PAni. Therefore, graphene and CNT should be prioritized in future works involving PAni until new evidence suggests the use of different nanofillers.