This paper focuses on a performance analysis of single-walled-carbon-nanotube / liquid crystal classifiers produced by evolution in materio. A new confidence measure is proposed in this paper. It is different from statistical tools commonly used to evaluate the performance of classifiers in that it is based on physical quantities extracted from the composite and related to its state. Using this measure, it is confirmed that in an un-trained state, ie: before being subjected to an algorithmcontrolled evolution, the carbon-nanotube-based composites classify data at random. The training, or evolution, process brings these composites into a state where the classification is no longer random. Instead, the classifiers generalise well to unseen data and the classification accuracy remains stable across tests. The confidence measure associated with the resulting classifier’s accuracy is relatively high at the classes’ boundaries, which is consistent with the problem formulation.

This paper focuses on a performance analysis of single-walled-carbon-nanotube / liquid crystal classifiers produced by evolution in materio. A new confidence measure is proposed in this paper. It is different from statistical tools commonly used to evaluate the performance of classifiers in that it is based on physical quantities extracted from the composite and related to its state. Using this measure, it is confirmed that in an un-trained state, ie: before being subjected to an algorithmcontrolled evolution, the carbon-nanotube-based composites classify data at random. The training, or evolution, process brings these composites into a state where the classification is no longer random. Instead, the classifiers generalise well to unseen data and the classification accuracy remains stable across tests. The confidence measure associated with the resulting classifier’s accuracy is relatively high at the classes’ boundaries, which is consistent with the problem formulation.

This paper focuses on a performance analysis of single-walled-carbon-nanotube / liquid crystal classifiers produced by evolution in materio. A new confidence measure is proposed in this paper. It is different from statistical tools commonly used to evaluate the performance of classifiers in that it is based on physical quantities extracted from the composite and related to its state. Using this measure, it is confirmed that in an un-trained state, ie: before being subjected to an algorithm-controlled evolution, the carbon-nanotube-based composites classify data at random. The training, or evolution, process brings these composites into a state where the classification is no longer random. Instead, the classifiers generalise well to unseen data and the classification accuracy remains stable across tests. The confidence measure associated with the resulting classifier's accuracy is relatively high at the classes' boundaries, which is consistent with the problem formulation.

This paper presents a series of experiments demonstrating the capacity of single-walled carbon-nanotube (SWCNT)/liquid crystal (LC) mixtures to be trained by evolutionary algorithms to act as classifiers on linear and nonlinear binary datasets. The training process is formulated as an optimisation problem with hardware in the loop. The liquid SWCNT/LC samples used here are un-configured and with nonlinear current-voltage relationship, thus presenting a potential for being evolved. The nature of the problem means that derivative-free stochastic search algorithms are required. Results presented here are based on differential evolution (DE) and particle swarm optimisation (PSO). Further investigations using DE, suggest that a SWCNT/LC material is capable of being reconfigured for different binary classification problems, corroborating previous research. In addition, it is able to retain a physical memory of each of the solutions to the problems it has been trained to solve.

Evolution-in-materio is a form of unconventional computing combining materials' training and evolutionary search algorithms. In previous work, a mixture of single-walled-carbon-nanotubes (SWCNTs) dispersed in a liquid crystal (LC) was trained so that its morphology and electrical properties were gradually changed to perform a computational task. Material-based computation is treated as an optimisation problem with a hybrid search space consisting of the voltages used for creating the electrical field and the material's infinitely possible SWCNT arrangements in LC. In this paper, we study solutions using synthetic data with a non-linear separating boundary. In addition, results for two real life datasets with partly merged classes are presented. The training process is based on a differential evolution (DE) algorithm, which subjects the SWCNT/LC material to repeated electrical charging, leading to progressive morphological and electric conductivity modifications. It is shown that the material configuration the DE algorithm converges to form a non-negligible part of the solution. Furthermore, the problem's complexity is relevant to the properties of the resulting "physical solver". The material structures created when training for a problem allow the retraining for a less complex one. The result is a doubly-trained material that keeps the memory of the original more complex problem. This is not the case for doubly-trained materials where initial training is for the less complex problem. The optimal electric field found by the DE algorithm is also a necessary solution component for the material's output to be interpreted as a computation.

This paper presents a series of experiments demonstrating the capacity of single-walled carbon-nanotube (SWCNT)/liquid crystal (LC) mixtures to be trained by evolutionary algorithms to act as classifiers on linear and nonlinear binary datasets. The training process is formulated as an optimisation problem with hardware in the loop. The liquid SWCNT/LC samples used here are un-configured and with nonlinear currentvoltage relationship, thus presenting a potential for being evolved. The nature of the problem means that derivative-free stochastic search algorithms are required. Results presented here are based on differential evolution (DE) and particle swarm optimisation (PSO). Further investigations using DE, suggest that a SWCNT/LC material is capable of being reconfigured for different binary classification problems, corroborating previous research. In addition, it is able to retain a physical memory of each of the solutions to the problems it has been trained to solve.

The potential of Evolution in Materio (EiM) for machine learning problems is explored here. This technique makes use of evolutionary algorithms (EAs) to influence the processing abilities of an un-configured physically rich medium, via exploitation of its physical properties. The EiM results reported are obtained using particle swarm optimisation (PSO) and differential evolution (DE) to exploit the complex voltage/current relationship of a mixture of single walled carbon nanotubes (SWCNTs) and liquid crystals (LCs). The computational problem considered is simple binary data classification. Results presented are consistent and reproducible. The evolutionary process based on EAs has the capacity to evolve the material to a state where data classification can be performed. Finally, it appears that through the use of smooth signal inputs, PSO produces classifiers out of the SWCNT/LC substrate which generalise better than those evolved with DE.

Evolution In Materio (EIM) is concerned with solving computational problems by exploiting the physical properties of materials. This paper presents the results of using a particle swarm optimisation (PSO) algorithm for evolving logic circuits in single-walled carbon nanotubes (SWCNT) based composites on a special custom made platform. The material used is a composite of SWCNT dispersed randomly in a polymer forming a complex conductive network. Following the EIM methodology the conductance of the material is manipulated for evolving threshold based logic circuits. The problem is formulated as a constrained, mixed integer optimisation problem. It is solved using PSO in conjunction with the shortest position value rule. The results showed that the conductive properties of SWCNT can be used to configure these materials to evolve multiple input/output logic circuits.

Evolution-in-materio concerns the computer controlled manipulation of material systems using external stimuli to train or evolve the material to perform a useful function. In this paper we demonstrate the evolution of a disordered composite material, using voltages as the external stimuli, into a form where a simple computational problem can be solved. The material consists of single-walled carbon nanotubes suspended in liquid crystal; the nanotubes act as a conductive network, with the liquid crystal providing a host medium to allow the conductive network to reorganise when voltages are applied. We show that the application of electric fields under computer control results in a significant change in the material morphology, favouring the solution to a classification task.

We report on the use of a liquid crystalline host medium to align single-walled carbon nanotubes in an electric field using an in-plane electrode configuration. Electron microscopy reveals that the nanotubes orient in the field with a resulting increase in the DC conductivity in the field direction. Current versus voltage measurements on the composite show a nonlinear behavior, which was modelled by using single-carrier space-charge injection. The possibility of manipulating the conductivity pathways in the same sample by applying the electrical field in different (in-plane) directions has also been demonstrated. Raman spectroscopy indicates that there is an interaction between the nanotubes and the host liquid crystal molecules that goes beyond that of simple physical mixing.

We demonstrate the manipulation of single-walled carbon nanotube/liquid crystal composites using in-plane electric fields. The conductivity of the materials is shown to be dependant on the application of a DC bias across the electrodes. When the materials are subjected to this in-plane field, it is suggested that the liquid crystals orientate, thereby forcing the SWCNTs to follow in alignment. This process occurs over many seconds, since the SWCNTs are significantly larger in size than the liquid crystals. The opportunity for applying this material to unconventional computing problems is suggested.

This paper explores the use of single-walled carbon nanotube (SWCNT)/poly(butyl methacrylate) composites as a material for use in unconventional computing. The mechanical and electrical properties of the materials are investigated. The resulting data reveal a correlation between the SWCNT concentration/viscosity/conductivity and the computational capability of the composite. The viscosity increases significantly with the addition of SWCNTs to the polymer, mechanically reinforcing the host material and changing the electrical properties of the composite. The electrical conduction is found to depend strongly on the nanotube concentration; Poole-Frenkel conduction appears to dominate the conductivity at very low concentrations (0.11% by weight). The viscosity and conductivity both show a threshold point around 1% SWCNT concentration; this value is shown to be related to the computational performance of the material. A simple optimization of threshold logic gates shows that satisfactory computation is only achieved above a SWCNT concentration of 1%. In addition, there is some evidence that further above this threshold the computational efficiency begins to decrease.

This paper presents results of computations based on threshold logic performed by a thin solid film, following the general principle of evolution in materio. The electrical conductivity is used as the physical property manipulated for evolving Boolean functions. The material used consists of a composite of single-wall carbon nanotubes (SWCNTs) and the polymer poly(methyl methacrylate). The SWCNTs are randomly dispersed in the polymer forming a complex conductive network at the nano-scale. The training is formulated as an optimisation problem with continuous and binary constraints and is subsequently solved by two derivative-free algorithms, the Nelder-Mead (NM) and the Differential Evolution (DE) algorithms. This approach has been used to evolve gates and circuits. The NM fails to converge for all computational tasks, whereas the DE is always successful. The computation tasks considered are simple threshold logic gates and more complicated circuits. The thin film composite is very stable and its behavior remains the same after the optimal solution has been achieved.

Electroless growth of zinc oxide has been accomplished using palladium catalysts coordinated to pulsed plasma deposited poly(4-vinylpyridine) nanolayers. Four different and concurrent functional properties have been identified, which encompass photoconductivity, superhydrophobicity, photo-switchable wetting, and bacterial killing.

The effects of processing conditions on the properties of organic light emitting devices (LEDs) based on rubrene-doped poly[2-methoxy-5-(2-ethylhexyloxy)-1, 4-phenylene vinylene] and a new electron transporting material, 2, 5-bis[2-(4-tert-butylphenyl)-1, 3, 4-oxadiazol-5-yl]pyridine, are reported. These dual-layer LEDs exhibited a higher quantum efficiency than observed for structures incorporating the more widely used electron transport compound 1, 3-bis[2-(4-tert-butylphenyl)-1, 3, 4-oxadiazol-5-yl]benzene (OXD-7). However, the as-prepared devices degraded relatively rapidly on storage (10-1 mbar, no applied bias). Thermal annealing of the degraded devices at 160 °C for 30 minutes restored the currents and light outputs close to those measured for fresh devices. The annealed LEDs exhibited a significant increase in their operating lifetime. Lifetime improvements could also be achieved by increasing the deposition rate and thickness of the thermally evaporated aluminium top electrode. These effects are attributed to better adhesion between the aluminium top electrode and the underlying electron transport layer.

Thin films containing single-wall carbon nanotubes (SWCNTs) have been prepared using the Langmuir–Blodgett (LB) technique. Atomic force microscopy has been used to investigate the morphology of these films. Films of pure SWCNTs were in the form of a network of interconnected bundles of tubes. The in-plane electrical characteristics of the films were measured at room temperature. Approximately ohmic electrical conductivity was observed at low applied voltages for LB films of pure SWCNTs. High power dissipation was found to lead to a reorganisation of the nanotubes in these ultra-thin films. In contrast, multilayer architectures built up from mixtures of SWCNTs and cadmium arachidate exhibited a current I versus voltage V dependence of the form ln(I/V)∝V1/2, indicative of Poole–Frenkel conductivity.

The electronic and optoelectronic behavior of white organic light-emitting devices (OLEDs) based on blue (FIrpic) and red [Ir(piq)2(acac)] phosphorescent dyes doped into the same layer of a polyvinylcarbazole (PVK) host are reported. The conductivity of all the OLEDs studied appeared to be dominated by space-charge injection effects, exhibiting a current I versus voltage V dependence of the form I∝Vn, with n≈7 at applied voltages at which electroluminescence was observed. Systematic studies of the current versus voltage and light-emitting behavior of the OLEDs have identified different excitation processes for the two dyes. It is suggested that electroluminescence from the FIrpic molecules originates by direct transfer of the exciton energy from the PVK to the dye molecules, while the process of light emission from the Ir(piq)2(acac) molecules involves carrier trapping. The efficiency of the devices can be tuned, to some extent, by varying the thickness of the organic film. Luminous efficiencies and luminous power efficiencies of 8 cd A−1 and 3 lm W−1 were measured for these blended-layer OLEDs, with Commission Internationale de l’Eclairage coordinates of 0.35, 0.35.

There is a current upsurge in research on devices with nanoparticles embedded in dielectrics. Such structures can operate as memories with high speed, high density, low voltage and low cost. Here, we report on hybrid gold nanoparticle-based metal-insulator-semiconductor (MIS) memory devices combining silicon technology and organic thin film deposition. The nanoparticles are deposited using a self-assembly technique at room temperature onto a 4.5 nm thermal silicon oxide layer. A 40 nm thin film of pentacene (deposited by flash thermal evaporation), polymethylmethacrylate (spin coated) and cadmium arachidate (deposited using the Langmuir- Blodgett technique) are used as insulators. Distinct capacitance-voltage (C-V) hysteresis is observed with a memory window that increases linearly with increasing voltage programming range. Clockwise and anticlockwise hysteresis in devices based on p-type and n-type silicon, respectively are observed, indicating that charges are injected from the top electrode to the nanoparticles rather than tunnelling through the thin SiO2 layer. However, thermal growth of SiO2 at a temperature below 800 °C resulted in a hysteresis in the opposite direction. The detailed electrical behaviour of the MIS devices will be discussed.

We report on the electrical behaviour of metal–insulator–semiconductor (MIS) structures fabricated on p-type silicon substrates and using polymethylmethacrylate (PMMA) as the dielectric. Gold nanoparticles, single-wall carbon nanotubes and C60, deposited at room temperature, were used as charge-storage elements. In all cases, the MIS devices containing the nanoparticles exhibited hysteresis in their capacitance versus voltage characteristics, with a memory window depending on the range of the voltage sweep. This hysteresis was attributed to the charging and discharging of the nanoparticles from the gate electrode. A relatively large memory window of about 2.2 V was achieved by scanning the applied voltage of an Al/PMMA/C60/SiO2/Si structure between 4 and −4 V. Gold nanoparticle-based memory devices produced the best charge retention behaviour compared to the other MIS structures investigated.

A bootstrapped inverter incorporating pentacene organic thin-film transistors (OTFTs), with poly(methyl methacrylate) as the gate dielectric, has been designed, fabricated and tested. The inverter uses capacitive coupling and bootstrapping effects, and exhibits superior performance to the normal diode-connected load inverter. The pentacene OTFTs used for the inverter possess a field-effect mobility of 0.32 cm(2)/V/s, a threshold voltage of -10.0 V, a subthreshold slope of 1.5 V per decade and an on/off current ratio of 2.2 x 10(6). The inverter has a 30 mu s rise time and a 450 ms fall time, at an operating frequency of 1 kHz and 30 V drive voltage.