1. Introduction

Machine-learning technology powers many aspects of modern society: from web searches to content filtering on social networks to recommendations on e-commerce websites, and it is increasingly present in consumer products such as cameras and smartphones. Machine-learning systems are used to identify objects in images, transcribe speech into text, match news items, posts or products with users' interests, and select relevant results of search. Increasingly, these applications make use of a class of techniques called deep learning(1). In this context, brain-inspired machine learning-based software techniques have accomplished various feats in terms of handling image, signal, and natural language-based challenges ^[1]; for example, recently, AlphaGo was the first program to achieve superhumans performance in Go. The published version, referred to as AlphaGo Fan by Silver et al ^[2], defeated the European champion Fan Hui in October 2015. AlphaGo Fan utilized two deep neural networks: a policy network that outputs move probabilities, and a value network that outputs a position evaluation. These merits of artificial intelligence to achieve superhuman proficiency are really extraordinary, but they continue to suffer in terms of many orders of high energy consumption ^[3][4][5]. One key explanation for the computing hardware's inadequacies in comparison to the brain is its architecture ^[6][7]. On the one hand, the brain uses a bottom-up strategy for construction, using biochemical ingredients in liquid and electrical activity, while existing software and hardware, on the other hand, use a top-down approach, in which all possible connections must be created before training, and their strengths are modulated during the training process, and weak connections are only precluded ^[8][9]. It results in several orders of magnitude higher energy consumption of existing hardware compared to the brain due to fully connected network training, and from a hardware standpoint, it is extremely difficult to create fully connected networks with a higher number of neurons due to space constraints for dense connectivity. Thus, the biological strategy of generating just relevant connections might be a valuable option for conserving resources, space, and energy usage in computer hardware ^{[6][7][8][9][10][11]}. With this inspiration ^[12][13][14], the materials engineering technique of electropolymerization ^{[15][16][17][18][19][20][21]} has been explored to create electrical activity-driven connections ^[12][13][14]. Monomers present in the liquid phase polymerize on the electrodes based on electrical activity in these techniques ^{[15][16][17][18][19][20][21]}. The connection between the electrodes, in this case, may be managed  by electrical activity between the electrodes, which is important for neuromorphic applications. Recently, Janzakova et al. ^[15] also demonstrated experimentally that signal parameters may be utilized to consistently generate interconnected morphologies. Koizumi et al. showed the electropolymerization of Poly(3,4-ethylenedioxythiophene) (PEDOT) derivatives and discovered that the propagation was in the form of fiber from the ends of Au bipolar electrodes (BPEs) in the parallel direction to the external electric field ^[17]. By adjusting the applied voltage, duty factor, and electrode spacing, Eickenscheidt et al. showed the formation of diverse polymeric forms ^[18]. Unlike previous neuromorphic options, the suggested technique is more closely related to biological circumstances and is created from monomers present in the liquid phase, similar to the constituents of neurotransmitters involved in synaptogenesis ^{[6][7][10][11]}. With these dendritic wires, the engineering path has opened up opportunities for neuromorphic computation, and several research groups are investigating electro-polymerization as a viable technique for neuromorphic computing. In this spirit, Akayi-Kasaya et al. demonstrated the suitability of conductive polymer wires derived by bipolar electro-polymerization for neuromorphic applications, where the morphology’s electro-polymerization development is directly related to the learning process of artificial synapses ^[19]. Hagiwara et al. demonstrated long-term potentiation and short-term potentiation using the varied frequency of continuous pulses ^[20]. Ji et al. built organic electrochemical transistors (OECTs) based on bipolar electropolymerization, and the performance of the OECTs may be modified by adjusting the electropolymerization settings ^[21]. Recently, the technique has been explored for neuromorphic functions such as Hebbian learning and pattern recognition ^[22].  We have also recently demonstrated short-term and long-term memories for neuromorphic functions with electropolymerized OECT dendrites ^[23]. The electropolymerized dendrite structures, controlled by the parameters can also be employed in several  applications needing fractal electrodes such as supercapacitors ^[24], heat transfer ^[25], fractal antennas ^[26], absorbers ^[27],  solar cells electrodes ^[28], and electronic wire connections ^[28][29], etc.  

This introductory review shows that the reported use of conventional engineering routes could have a variety of applications from neuromorphic engineering to bottom-up computing strategies.

Considering the relevance of several topics that we have addressed, manually biological and artificial neural networks (BioNNs and ANNs), including neurons and synapses along with the associated gated ions channels, electropolymerized dendrite structures, OECTs, supercapacitors, heat transfer, fractal antennas, fractal absorbers, solar cell electrodes, and electronic wire connections, a few of them are briefly discussed below.

2. Biological and Artificial Neural Systems

The nervous system in vertebrates (e.g. human beings) is where intelligence resides. It is composed of the brain, the spinal cord, and peripheral nerves all over the body ^[30][31]. In this system, computations that support intelligent functions such as memorizing and forgetting, learning, and decision making are carried out in various neural circuitries in the brain. As a side-by-side comparison, Figure 1 provides a high-level overview of the human nervous system and artificial neural system built with emerging neuromorphic devices. Both systems span multiple levels of organization. Three basic levels can be distinguished. On the top level, the human nervous system is a vast mesh of different types of neural networks organized hierarchically to support different computational functions such as vision, audition, emotion, etc. On the medium level, the basic unit is a neuron, which is composed of a soma, with many dendrites to receive inputs, and a single axon (usually with many branches) to send out outputs. These neurons connect to one another via synapses. On the bottom level, different types of ion channels form the molecular basis for electrical activities in neurons and support the transmission and processing of information. The number and properties of ion channels in a neuron are regulated by other cellular signaling machinery. Similarly, a typical Al chip is composed of different types of ANNS that can be mapped onto cross-point arrays of resistive switching elements, where the artificial synapses and neurons can be operated through conduction channels (filaments) that are associated with ion movements driven by electric field, Joule heating or electrochemical potential.

Figure 1 - A high-level comparison of the human nervous system and an artificial neural system built with emerging neuromorphic devices. The human nervous system (left panel) has different types of neural networks whose basic functional elements are neurons and synapses, in which different types of ion channels underlie electrical neuronal activity. Similarly, an Al chip (right panel) is composed of different types of ANNs that can be mapped with cross-point arrays of artificial synapses and neurons, whose operation mechanism could be conductive filaments associated with electrically induced lon movements as in the case of RRAM(resistive random-access memory).

From the top level of BioNNs down to the cell level, the basic units in the nervous system where signal processing happens are mainly neurons and synapses. They are also the mostly mimicked biological components using emerging neuromorphic devices so far. It is thus worthwhile to carefully review and compare the structures and mechanisms of biological and artificial neurons and synapses.

Figure 2 - Schematics of a typical neuron and chemical synapse. A typical neuron is composed of a soma, an axon, and dendrites. Synapse connects the axon terminal of presynaptic neuron and the dendrite of postsynaptic neuron. The space at the chemical synapses is called synaptic cleft, and the critical feature is the presence of small, spherical, membrane-bonded organelles called synaptic vesicles within the presynaptic neuron. Calcium ion influx through ion channels causes the release of neurotransmitters from these vesicles, and when they reach the synaptic cleft, they bind with the receptors in the postsynaptic neuron, leading to the ion channels open or close, causing the potential change in the postsynaptic neuron. ^[30].

In biological nervous systems, a typical neuron, as illustrated in Figure 2 consists of several structural and functional parts: a soma, an axon, and dendrites. The junction that connects an axon terminal of one (pre-synaptic) neuron and a dendrite of another (postsynaptic) neuron is the synapse. Each of these parts plays distinct roles in the generation and communication of signals between cells ^[30][31]. Since biological neurons are delimited by lipid bilayer membranes which act as capacitors and are also punctuated by small holes on the membrane (ion channels) acting as reconfigurable resistors. A membrane potential gradient is also set up by the difference in ion concentrations across the membrane (which is maintained by  pumps that consume energy in the form of adenosine triphosphate (ATP), as shown in Figure 3b) and acts as a battery. Typically, the membrane resting potential is around -70 mV with the inside of the cell being at a lower potential. The most basic electric operation of a biological neuron can be understood as a RC circuit with a battery. The membrane potential (voltage) can be changed when currents are injected mainly through the ion channels. The capacitance and electric potential are normally impossible to change over a short time scale, but the resistance can be changed by opening and closing of ion channels, which give rise to current flows and underlie diverse electric operations in the neurons.

Figure 3 - Schematics of an action potential (AP). a) Illustration of the change of the membrane potential during an AP. Diagrams b-e) illustrate the ion channels that act in each phase: b) In the resting state, Na⁺/K⁺ pumps transport Na⁺ to outside of the membrane and K⁺ to inside of the membrane. The ion concentration gradient drives K⁺ to flow outside of the membrane, which results in approximately -70 mV membrane potential (the potential inside minus the potential outside the membrane). c) When the membrane potential increases up to a threshold, the voltage-gated Na⁺ channels are activated and Na⁺ flows from outside of the membrane to inside, thus the membrane potential goes up rapidly. This process is called depolarization. d) With the increase of membrane potential, voltage-gated K⁺ channels are activated and the voltage goes down again, which is called repolarization. e) The voltage usually "undershoot" below resting potentials due to the outflow of K⁺. This is called hyperpolarization.

We review the main structural and functions of each part of the neuron below.

Soma 

The soma, or the cell body of a neuron, contains the cell nucleus and cellular organelles. From the soma grows two types of cytoplasmic protrusions, namely the axon and the dendrites. A specialized region of the neuron, the axon hillock, is where the axon originates. Within the axon hillock, an AP (Figure 3a) is generated by integrating the synaptic inputs on dendrites that cause sufficient excitation over a certain threshold, and then pass to the axon (output) as a traveling impulse. The generation of APs is generally termed as neuronal "firing."

Axon

The axon is considered as the output element of the neuron according to the description of "dynamic polarization by Cajal ^[32]. APs generated at the axon hillock propagate along the axon to the nerve terminals, where synapses are formed on the postsynaptic dendrites (and sometimes soma). The most important function of AP is probably to ensure reliable transmission of signals across a long distance while it might also have important computational functions. Some recent research has reported that despite the faithful conduction in the axons, the complex time and voltage dependences of APs can lead to activity-dependent changes in spike shapes and the resting potential, affecting the temporal fidelity of spike conduction ^[33]. With the exception of SNNs, “axons” in most ANNs typically transmit a real number firing rate rather than all-or-none spike events.

Dendrites

Dendrites are the input elements of a neuron that branch off from the soma. Dendrites play a critical role in filtering and integrating the synaptic inputs to determine whether to fire at signal or not. Traditionally the dendrite is considered to be a passive element and does not amplify the input signals. Different from the single axon derived from the soma, there are many dendrites and thus, the morphology such as branch density and grouping patterns are highly relevant to the function of the neuron. Recent studies however have emphasized the generation of active electric events in the dendrites akin to those happening at the axon hillock ^[34]. Therefore, dendrites themselves may be capable of complicated processing of incoming stimuli. Current ANN model only describes the passive integration part of the dendrite function in a very simplified manner, such as summations.

Synapses

Synapses are the connecting structure where the axon terminal of the presynaptic neuron transmits signals to the postsynaptic neuron. The branches of a single axon may form synapses with as many as thousands of other neurons. There are two funda- mentally different types of synapses: electrical and chemical synapses. Electrical synapses pass ionic current to induce the voltage changes in the postsynaptic cell directly. While for the chemical synapse, which is the majority of synapses in the human brain, electric activity in the presynaptic cell is converted to chemicals (neurotransmitters) and chemical signals are converted back to electrical activity in the postsynaptic cell. This arrangement isolates the two cells electrically and can give rise to many complex signal transformations at the synapse in both amplitude and time domain. Depending on the direction (inward or outward) of the synaptic current, the membrane potential is increased (depolarized) or decreased (hyperpolarized), the current and potential are termed excitatory/inhibitory postsynaptic current and potential (EPSC/IPSC and EPSP/ IPSP), respectively. In this way, the connection between the two communicating neurons is modulated, i.e., the weight of the synapse is changed via synaptic plasticity (will be discussed in Section 4). The 3D structures of two common types of chemical synapses, excitatory and inhibitory synapses, have been revealed by the latest cryo electron microscopy technique ^[35]. In biological systems, it is generally believed that synapses are either excitatory or inhibitory, and furthermore each neuron only sends out either excitatory or inhibitory synapses but can receive neurotransmitters from both types of synapses. This is termed as Dale's law. In ANNs, the synapse is normally simply modeled as a number (synaptic weight) to be multiplied with and can be of both positive and negative signs. Change in synaptic properties, most importantly synaptic weight, is the main manner that learning is realized in both ANNs and BioNNs.

3. Organic Electrochemical Transistors

Organic electrochemical transistors (OECT) have emerged as a platform for wearable and point-of-care testing devices owing to their advantages of low operation voltage, high sensitivity, and excellent compatibility with flexible substrates. They are widely used in biosensing, such as  ions, dopamine, epinephrine, and pH detection. Based on the physicochemical doping/dedoping analyses, ionic mass-transfer, electronic transport, and ionic-electronic coupling have been extensively studied, revealing the device physics and its sensing mechanism in increasing detail. Further, the development of the structure-property relationship led to a significant advance in the synthetic design of materials and manufacturing methods. With regard to active materials, due to its commercial availability, poly(3,4-ethylenedioxythiophene):poly(4-styrenesulfonate) (PEDOT:PSS) is by far the semiconductor most commonly used to construct OECTs. The PEDOT phase exhibits high electron mobility supporting charge transportation along conjugated backbones, while the PSS phase is intrinsic hygroscopic, which enables fast transport of the solvated ions. Other polymers, such as the complexes of alkoxysulfonate modified PEDOT and dioctylammonium (DOA), copolymers of 3,4-(15-crown-5) thiophene and 3,4-ehtylenedioxythiophene (EDOT), conjugated polymer electrolyte of poly(2-(3,3’-bis(2-(2-(2-methoxyethoxy)ethoxy)ethoxy)-[2,2’-bithiophen]5-yl)thieno [3,2-b]thiophene, have also shown great potential in applications.

In terms of manufacturing methods, OECTS are usually fabricated by either photolithography patterning or by a printing technique. Several research efforts are contributing to the photolithographic patterning approach. For instance, additive photolithographic patterning, widely used since its inception, involves the procedures of photo-etching of perylene-C double layers and spin-coating of semiconductor layers. Concerning subtraction photolithographic patterning, silver has proven to be the ideal material for etching masks to protect the semi conducting films. Based on OECTS fabricated through this method, the artificial synapse, the emulsions sensor, and the cytokine detector have been proposed recently. The jet printing method, which shows the advantages of low cost and high throughput, is also popular. Ji et al ^[36] are the first to utilize the alternating current (AC) electrodeposition method to fabricate OECTs in situ onto a microfluid chip. By using this method, biosensors with integrated long-term stable OECTs ^[37] were realized. AC electrodeposition, however, produces only one OECT device each time. Thus, low-cost and massive production is still challenging.

In the bipolar electrochemical system, however, arrays of the bipolar electrodes (BPE) could be operated simultaneously. For example, the BPE array was used to detect glucose, lactate, and choline. The pH profile could be displayed by the inter digitated BPE array when hydroquinone is oxidized to benzoquinone. High-throughput screening of catalysts and genotyping of single nucleotide polymorphisms using the BPE array have also been reported. Recently, the synthesis of PEDOT:PSS fibers ^[17] and a fiber array were realized by bipolar electrodepositions under AC voltage. Inspired by these developments, Shida et al. ^[38] utilizes the AC bipolar electrodeposition (AC-BED) to realize the mass production of PEDOT:PSS films for OECT construction. This method paves the way for batch fabrication of low-cost OECT arrays for use in biosensor applications. A schematic illustration of the AC-BDE is shown in Figure 4.

Figure 4- Schematic illustrations of AC bipolar electrodepositions (AC-BED) as well as the electrochemical impedance spectroscopy (EIS) measurement. (a) Scheme of the experimental setup for AC-BED. (b) The equivalent circuit of AC-BED. (c) Scheme of the experimental setup for EIS measurements. The grey color indicates the portion of the electrode covered with glue to expose the right pad (Pad) and the right tip (Tip). (d) The equivalent circuit used for EIS spectra fitting. (e) The scheme of the AC-BED principle. (0) The transient distribution of the electric potential across a pair of bipolar electrodes (BPEs).

4. Fractal Tree Antennas

Figure 5 -Different forms of fractal tree antenna

Early, scholars mainly studied the miniaturization of the dipole antennas using fractal tree structures as end loads.

In 2000 and 2002, the miniaturization of the two dimensional and three dimensional fractal tree dipole antennas was researched. In ^[26] the fractal tree structures are similar to Figure 5 (a,b). The dipole antennas using fractal tree structures as end loads have lower resonant frequency than a standard dipole with equal length, namely fractal tree structures achieve miniaturization effectively. In 2002, Joshua S. Petko and D. H. Werner improved the miniaturization of the 3-D fractal tree antennas by increasing the density of branches It is found that the density of fractal tree antenna using fractal tree structures as end loads plays an important role in the designing of miniature antennas effectively. And increasing the density of branches can improve the miniaturization prominently. In 2004, Joshua S. Petko and D.H. Werner improved the antenna miniaturization by varying the elevation angle or the density of the branches in ^[26].

The special Fibonacci number sequence which leads to variant branch length ratios can be utilized to design a novel fractal tree antenna in ^[39]. The novel fractal tree antenna gains better performance in the miniaturization, compared with the conventional designs. The methods mentioned above are just theoretical research. As the number of iterations increases, the complexity of the antennas grows rapidly. So these antennas occupying a certain space are not easy to be manufactured.

A fractal tree structure is utilized to design a dipole radio frequency identification (RFID) tag antenna on papery substrate. The length of the fractal tree arm decreases approximately 18.4% compared to the half-wave dipole. The fractal tree antenna with simple structure is easier to process compared to the Manderlbrot structure. Other authors propose a novel tree-shaped fractal patch antenna using arc branch of fractal tree structures as end loads. Geometry of the tree-shaped fractal dipole is similar to Figure 5 (c). The resonant frequency of the fractal antenna reduces approximately 37.5% compared to the zero iteration. It is found that the tree-shaped fractal structure achieves good performance in the miniaturization. A fractal tree antenna is designed using Rectangular, triangular and wired structure based on fractal tree geometry. Other workers have designed a compact RFID tag antenna using fractal tree structure similar to Fig. 5 (d). element of the antenna consists of two radiating arms and one rectangular loop. The two radiating arms with fractal tree shape achieve the character of miniaturization and dual band, and the rectangular loop realizes an inductively coupled feeding. The electrical size of the antenna is. It can be seen that the fractal tree structure can realize size reduction efficiently. In addition, the structure of the antenna is simple and easy to be manufactured.

Fractal tree structures are also used to design reconfigurable antennas. In 2004, Petko and Werner ^[26] introduced several design examples using reactive LC traps or RF switches in the tree structure to design miniature reconfigurable fractal tree antennas. Because of self-similarity property, fractal tree structure is widely used to design multiband fractal antennas in ^[40] et al. The multiband electromagnetic of the ternary tree-like fractal structure studied in ^[40] is similar to Figure 5(e). Aggarwal and Kartikeyan ^[41] designed a fractal patch antenna by exploiting the geometry of  Pythagoras tree, which was similar to Figure 5(f).

The structure of fractal tree can also be exploited to design ultra-wideband antennas. For example, a tree-like UWB antenna has been designed for the UWB applications in the MB-DFDM system. In 2012, Naser-Moghadasi et al. ^[42] designed a novel UWB fractal tree-like antenna by using unit-cells of the fractal tree in ^[42], the tree structure being similar to Figure 5 (g). The third iteration antenna has a bandwidth of 2.1-11.52 55 GHz (138%) with a compact dimension of 24x24x1mm³.

In summary, the fractal tree antenna can promote the development of fractal antenna engineering and accelerate the development of wireless communication, but further research is needed to attain these objectives.

5. Dye-sensitized Solar Cell Electrodes

Dye-sensitized solar cells (DSCs) are potential candidates for cost-effective and clean energy conversion devices due to their low cost, relatively high energy conversion efficiency, and environmentally friendly fabrication process. In a typically DSC, the counter electrode (CE) is a crucial DSC component to facilitate the electron translocation from the external circuit back to the redox electrolyte, and speed up the the reduction of triiodide to iodide. Therefore, CE materials with high electrical conductivity and excellent electrocatalytic activity are in demand. Up to now, the standard CE has still been made of platinum (Pt) due to its high electrical conductivity, electrocatalytic activity, and chemical stability. For commercialization, it is imperative to reduce or eliminate the dependence on noble Pt due to its high cost and limited reserve on this plant. Many groups have reported an appreciable performance from various carbon-based materials such as Carbon Black, activated carbon, graphene and carbon nanotubes (CNTs).

Among them, CNTs reveal some of the promising potential for use as alternative CEs due to their features of large surface area, superior electronic conductivity and excellent mechanical strength. Nevertheless, the efficiency of DSCs using CNT based CE_Ss has been found to be lower as compared to Pt ones. To overcome this shortcoming, there have been several attempts to combine the highly conductive CNTs with excellent electrocatalytic materials. Huang et al. ^[43] reported that an enhanced photovoltaic performance can be achieved by reducing Pt nanoparticles on the multi-wall carbon nanotube (MWCNT) owing to its increased surface area. MWCNTs with TiN nanoparticles as CE, which demonstrated a comparable performance to that of Pt CE, and an enhanced performance of CoS/MWCNT composite catalytic film by electrophoresis of MWCNTs onto a conducting glass substrate followed by CoS electrodeposition, were also tested. Additionally, conducting polymers are promising candidates for CE materials in DSCs due to their unique features, such as low cost and superior redox properties. In particular, poly(3,4 ethylenedioxythiophene) (PEDOT) has received ever-accelerating interest from scientists throughout the world due to its high conductivity, electrochemical stability, transparency, and catalytic ability. However, PEDOT was still inferior in conductivity as compared to Pt metal or carbon-based materials.

Until now, several studies have employed PEDOT: poly (styrene sulfonate) (PSS)/CNT blends as the hole collecting electrode in organic photovoltaic devices and the thermoelectrical material due to the enhanced conductivity of PEDOT by incorporation of CNT electrical network. In view of both the electrical conductivity and electrocatalytic activity Xiao et al. ^[28] successfully synthesized for the first time nano-meadows PEDOT-coated MWCNT CE for Pt-free DSCs. The DSC assembled with the PEDOT/MWCNT CE exhibited a superior photovoltaic conversion of 7.03% to those of DSCs using the Pt, MWCNT, and PEDOT CEs.

For this system, EIS measurements were conducted to elucidate the electrocatalytic activity of the CEs for the reduction of  by measuring the charge transfer resistance (Rct), which is an index representing the electrocatalytic performance of CEs.

Fig 6 illustrates the Nyquist plots of the Pt, MWCNT, PEDOT, and PEDOT/MWCNT CESs. The intercept of the real axis at high frequency is the ohmic series resistance (Rs) including the sheet resistance of two identical CESs and the electrolytic resistance. The semicircle at high frequency refers to the Rct for the I₃^-, reduction at the electrolyte/CE interface, while the semicircle at low frequency represents the Nernst diffusion impedance (W) corresponding to the diffusion resistance of the iodide  redox species. The constant phase element (CPE) is frequently used as a substitute for a capacitor in an equivalent circuit to fit the impedance behavior of the electrical double layer more accurately, while the double layer does not behave as an ideal capacitor. It should be noted that the Warburg impedance of the polymer-based CE originating from the charge transport the catalyst cannot be ignored due to its low electrical conductivity, compared with that of the Pt catalyst. Therefore, an element (W_cat) is introduced to the equivalent circuit to represent the Warburg impedance from the charge transport resistance in the PEDOT film.

Figura 6- Nyquist plots of symmetrical (a) Pt, (b) MWCNT, (c) PEDOT, and (d) PEDOT/MWCNT CEs, respectively.

It was concluded that the PEDOT/ MWCNT CE is a promising candidate as highly efficient and low cost CE material for Pt-free DSCs.

6. Conclusion

In conclusion, we are reviewing studies and models to understand and optimize the electropolymerization technique as a bottom-up strategy for neuromorphic computation and other applications. 

Since one can employ different conditions of chemical material type,  experimental conditions, electrical parameters, and electrode geometry configurations, it is important to understand the phenomena, generic dependence of parameters, affecting their growth process to well tune their morphologies for desired connection types. In all of the preceding situations, the experimental investigations primarily highlight the growth and final pattern of dendritic structures under various electrical activities and experimental settings. Very recently, the potential generated in the process has been attempted to be mapped by electrochemiluminescence ^[44]. However, since the growth occurs from one specific composition phase, it is beyond the experimental and optical microscopy scope to see the motion and deposition of such ingredients forming a particular morphology and systematically predict it under various chemical compositions. There are currently no modeling methodologies available to describe the evolution of electro-polymerized structures under diverse electrical,  chemical and geometrical settings. Ab-initio simulations based on atomic and molecular interactions are not capable of modelling structures with orders of magnitude larger than molecular sizes. Furthermore, the time-varying signal and evolving electrode (due to electro-polymerization on the electrodes) make such simulations challenging to run on a commercial tool. Taking these considerations into account, we discuss a mesoscale simulations study, that incorporates particle interaction and mobility, based on the assumption that mesoscale dendritic morphologies are yielded by charged particle aggregation of oligomers preformed and dispersed in the electrolyte prior to seeding on the electrode. The modeling technique can be useful in signals and dendrite-morphology. The review provided here reflects the many forms of neurons and synapses that are regulated by a few factors such as growth area, pruning time scale, and spatial distribution of neurotrophic particles. Apart from the neuromorphic engineering community, the investigation can be an ^[45] important contribution to different domains of science interested in pattern formation. As an example, uniform to fractal-like atomic growth with growth conditions of Molecular Beam Epitaxy ^[46], coffee ring effects ^[47], based on liquid evaporation conditions and investigation of different morphologies with dc-electrodeposition ^[48][49].