The emerging science of electrosynbionics

Dramatic changes in electricity generation, use and storage are needed to keep pace with increasing demand while reducing carbon dioxide emissions. There is great potential for application of bioengineering in this area. We have the tools to re-engineer biological molecules and systems, and a significant amount of research and development is being carried out on technologies such as biophotovoltaics, biocapacitors, biofuel cells and biobatteries. However, there does not seem to be a satisfactory overarching term to describe this area, and I propose a new word—‘electrosynbionics’. This is to be defined as: the creation of engineered devices that use components derived from or inspired by biology to perform a useful electrical function. Here, the phrase ‘electrical function’ is taken to mean the generation, use and storage of electricity, where the primary charge carriers may be either electrons or ions. ‘Electrosynbionics’ is distinct from ‘bioelectronics’, which normally relates to applications in sensing, computing or electroceuticals. Electrosynbionic devices have the potential to solve challenges in electricity generation, use and storage by exploiting or mimicking some of the desirable attributes of biological systems, including high efficiency, benign operating conditions and intricate molecular structures.

Globally, the amount of electricity generated has increased dramatically over recent decades. Although the use of renewables has been increasing, further progress is needed in order to reduce carbon dioxide emissions sufficiently to avoid catastrophic effects due to climate change while keeping pace with increasing electricity consumption.
In 2017, in total 25 606 TWh of electricity was gen erated globally, over four times the amount produced in 1973 [1]. Despite significant increases in the use of renewable sources such as photovoltaics, the percent age of global electricity generated by the burning of coal was approximately the same in 1973 and 2017, at 38.3%-38.5% [1]. Through burning of fossil fuels and other activities, we pumped an estimated 2040 billion tonnes of CO 2 into the atmosphere between 1750 and 2011 [2]. About 40% of this remained in the air, and consequently the atmospheric concentration of CO 2 in the air increased over the same period from a little over 280 ppm to nearly 400 ppm, based on measure ments from ice cores and atmospheric air [2]. As CO 2 absorbs strongly in the infrared part of the spectrum, it contributes strongly to the retention of heat that would otherwise be reradiated into space. Between 1880 and 2012, the average global surface temper ature increased by 0.85 °C, and even with mitigation measures (i.e. reduction of CO 2 emissions) the average global surface temperature in 2100 is likely to be over 1.5 °C higher than in 1880, which could cause signifi cant loss of life and decrease in living standards due to increased frequency and severity of extreme weather events, changes to ecosystems leading to reduced food production, and geopolitical or social issues arising from increases in sea level and migration [2].
The agreement [3] reached in Paris in 2015 set the objective of keeping the global temperature rise this century to well below 2 °C and pursuing efforts to limit the rise to 1.5 °C. To address this, countries are starting to set goals of reducing carbon dioxide emissions to net zero in the coming decades. As electricity and heat production accounts for approximately 25% of green house gas emissions and transport accounts for 14% [2], further expansion of renewable electricity gen eration and use of electric vehicles will be required in order to meet such commitments. Use of intermittent renewable power sources such as photovoltaics and wind turbines implies a need for electricity storage sys tems such as battery banks, in order to match supply to demand and reduce wastage.
Improved technologies for electricity generation and energy storage would facilitate the transition to an economy with zero net carbon dioxide emissions. While progress has been made in this area, there is potential for further developments that could reshape the landscape for electricity generation and energy storage. Not only could new technologies facilitate decarbonization, but they could be more efficient than existing approaches, leading to cost savings.
Biological systems provide a rich source of ideas for a wide range of technologies, from geckoinspired adhesives to optimization algorithms, but devices inspired by or derived from biology have not become mainstream in the energy sector, despite the efforts of numerous research groups. Such technologies could harness the high efficiency of biological processes that have been finetuned over millions of years by evo lution, exploiting the intricate molecular structures of biological components. The benign conditions required for many biological processes are also advan tageous.
At present, efforts to develop biophotovoltaics, biocapacitors, biofuel cells and biobatteries are some what fragmented, and this may be partly due to the lack of an overarching term to describe such research. I pro pose the new word 'electrosynbionics', which I define as the creation of engineered devices that use components derived from or inspired by biology to perform a useful electrical function.
Here, the phrase 'electrical function' is taken to mean the generation, use and storage of electricity, where the primary charge carriers may be either elec trons or ions. The word 'electrosynbionics' is intended to capture the ELECTRical nature of the process, the presence of elements derived from or inspired by BIOl ogy, and the use of engineering methods to develop SYNthetic devices.
It is important to note that electrosynbionics is dis tinct from bioelectronics. The term 'bioelectronics' is commonly used in the context of applications in sens ing, computing or perhaps electroceuticals, none of which could be covered accurately by the definition of electrical function given above. Most 'bioelectronic' devices are used to process information, whereas elec trosynbionic devices process energy. For information processing, it is desirable to minimize power dissipa tion, but for energy applications the aim is usually to maximize power generation and/or energy storage density. For example, a living photovoltaic system that harvests light and generates current using cells like cyanobacteria would be an example of an electrosyn bionic technology [4] whereas an electrochemical bio sensor that detects a protein biomarker of sepsis would fall under the heading of bioelectronics [5].
The transformation of electrosynbionics from an emerging area to a thriving branch of mainstream sci ence will be driven by the need for new electricity gen eration and storage techniques, but for electrosynbi onic technologies to be commercially viable they must demonstrate a competitive advantage over established techniques. It is therefore essential to view electrosyn bionics in the context of the existing landscape of elec tricity generation and energy storage. To that end, I will proceed to review conventional technologies before describing the biological phenomena that could be used in their biological or bioinspired counterparts, and then summarizing some of the innovations that I suggest should be rebranded as 'electrosynbionics'. I will conclude by arguing that electrosynbionic tech nologies have great potential but much further work and investment are required to make them competi tive, and this will need to be supported by a more sys tematic approach to the testing of devices/systems and statements of their specifications.
This perspective is intended to be accessible to researchers in disciplines ranging from physical sci ences and engineering to the biological sciences. Con sequently I assume very little prior knowledge, and parts of sections 2 and 3 present material that could be regarded by specialists as elementary but will not be familiar to all readers.

Conventional technologies for electricity generation and storage
Here, 'conventional' should be taken to mean 'without biological parts'. In this discussion, I will focus solely on the technologies that could have electrosynbionic equivalents. Hence, the only electricity generation techniques to be mentioned will be photovoltaics and fuel cells.

Photovoltaics
In biological and nonbiological systems, there are three key steps in light harvesting and energy conversion, namely photoexcitation, charge separation and charge transport. Typical solar cells are based on semiconductors. In the absence of any excitation, electrons in a semiconductor normally occupy a set of filled energy states known as the valence band. For charge to flow, electrons must be excited to the conduction band, which is at a higher energy than the valence band. Conduction and valence band are separated by a set of forbidden states known as the band gap. An electron can only be excited to the conduction band if it is given sufficient energy to overcome the band gap. Photoexcitation (figure 1(a)) can therefore only occur if the energy of the incident photon exceeds the band gap energy. In this case, an electron is elevated to the conduction band and a 'hole' is left behind in the valence band. The semiconductor of choice is often silicon. Conventional silicon photovoltaics are either singlecrystal (where the crystalline order extends throughout the material) or multicrystalline (where the material consists of a large number of crystalline grains).
A conventional solar cell [6] consists of a pnjunc tion, comprising two layers of semiconductor mat erial, doped with different types of impurities ( figure  1(b)). The p-type material is doped with impurities such as boron that tend to accept electrons, creating 'holes' in the valence band, and the ntype material is doped with impurities such as phosphorus that tend to donate electrons to the conduction band. Joining the n and p type materials together results in the diffusion of mobile charge carriers across the boundary, leading to the creation of a depletion zone where there is a strong electric field, which helps to separate photoexcited electron-hole pairs. Once the electrons and holes are transported to opposite electrodes, a current can flow in the external circuit. Recombination of electrons and holes prior to separation/transport tends to degrade the performance of the solar cell, as the absorbed photon energy is wasted. The efficiency of a solar cell is usually defined as the quantity of electrical energy extracted divided by the total solar energy incident on the device. The maximum efficiency achievable is fun damentally limited by thermodynamics and is referred to as the ShockleyQueisser limit after those who first calculated it [7]. For a singlejunction silicon device Shockley and Queisser originally computed the limit to be approximately 30%. Later calculations incorpo rating additional physics and utilizing a more accurate solar spectrum produced very similar values [8]. The limit depends on the material used and the value of the band gap.
Multijunction solar cells consist of several pn junctions based on materials with different band gaps. Multijunction solar cells can absorb a higher percent age of the solar spectrum and therefore have the potential to display greater efficiency. The Shockley Queisser limit for a solar cell with an infinite number of junctions has been calculated to be 68% [9]. Some research groups are seeking to circumvent the Shock leyQueisser limit by exploiting physics that was not accounted for in the original models [10].
Thinfilm solar cells [11] are made by growing lay ers of material on a substrate (as opposed to slicing up a large ingot). The layer is less than a few microns thick. Thinfilm devices use very little material and are con sequently very lightweight, but they only have a small share of the overall photovoltaic market.
In the 1990s, a completely new approach was pro posed: the dyesensitized solar cell [12]. This consists of the following components: a dye, an electrontrans porting porous material such as a colloidal film of TiO 2 nanoparticles, an electrolyte and two transparent conducting electrodes ( figure 1(c)). The dye acts like the chromophore in photosynthetic rection centres (see below), absorbing light and producing a photoex cited electron. The electron is transported through the porous material and into one of the transparent elec trodes. The electrolyte completes the circuit and con tains additives that allow regeneration of the dye after photoexcitation. The efficiency of DSSCs is typically quite low, reaching values of 12% at best [13].  [13]. ** Data from [7,9].
Over the last decade, a new technology has emerged, in the form of perovskites, materials with the chemical formula ABX 3 , where A and B are cations (A larger than B) and X is an anion. Perovskites may be used as light absorbers in DSSCs or as the basis of thinfilm style devices [14]. These devices have been developed rapidly and the efficiencies achieved have increased dramatically over the last 10 years, although some issues remain about potential environ mental issues associated with perovskitebased cells as some of them contain elements such as lead. Perovskite cells can be referred to as one of the 'emerging PV' technologies. This category also includes DSSCs and systems based on organic semiconductors, quantum dots etc.
The National Renewable Energy Laboratory (NREL) in the USA provides a muchused chart to show the highest confirmed conversion efficiencies for the different technologies as they develop over time (in a research context-rather than commercial applica tion) [15]. Tables published every six months in the journal Progress in Photovoltaics also show the best confirmed efficiencies in each category [13], some of which are shown graphically in figure 1(d), together with the thermodynamic performance limits for sin glejunction silicon cells and multijunction cells. It is essential to note that results are only included in the reference tables and graphs if they have been 'indepen dently measured by a recognized test centre' [13] and measurements are made in accordance with interna tional standards. International standards have been developed to define a reference solar spectrum and the procedure for measuring photovoltaic efficiency using artificial solar simulators. The journal Progress in Photovoltaics refers to the standard IEC609043, from the International Electrotechnical Commission. Similar standards are also provided by ASTM International (ASTM E927 and G173) [16,17], and the spectra are nearly identical. As shown in the G173 standard [17], the solar irradiance is greatest at wavelengths of 450-550 nm, dropping steeply at lower wavelengths and more gradually at higher wavelengths. The spectrum is not smooth due to atmospheric absorp tion at par ticular wavelengths, and the irradiance drops to zero in two bands centred on approximately 1380 and 1875 nm. The tail of the spectrum falls away to nearly zero in the vicinity of 2500 nm.
Of course, efficiency is not the only important factor in the success of a photovoltaic technology. Economic considerations are particularly important, and the economic competitiveness of electricity gen erating technologies is often assessed using the lev elized cost of electricity (LCOE). This includes all the costs of building and operating the technology over its lifetime, including all the infrastructure [18]. The expected service life of the systems is therefore critical, but so too is the manufacture cost, and high volume mass production methods are essential. It is also vital to consider the costs of the systems required to connect the photovoltaic modules to the grid. Other important aspects include the appearance of the cells, their mat erial content and endoflife disposal considerations.

Batteries, capacitors and fuel cells 2.2.1. Batteries
The first battery [19] was demonstrated by Alessandro Volta in 1800, and consisted of a pile of pairs of dissimilar metal discs, separated by brinesaturated cloth. Socalled 'primary' batteries like Volta's are singleuse devices, whereas 'secondary' batteries can be recharged. Inside a typical battery there are two electrodes, an electrolyte and a separator (figure 2(a)). The electrolyte is a substance through which ions can move but which is impermeable to electrons. The function of the separator is simply to prevent the two electrodes from coming into contact. In the example of the leadacid battery, when current is drawn the following reactions occur: Positive electrode: Negative electrode: For each reaction a standard electrode potential can be quoted. This is usually given as the potential difference that would be observed between the given electrode system and a socalled standard hydrogen electrode, under standard conditions. The standard voltage for the battery is the difference between the two electrode potentials.
The Nobel Prize in Chemistry 2019 was awarded for the development of lithiumion batteries [20], a revolutionary technology that has had a dramatic impact in a number of areas. In these batteries, the electrodes consist of materials in which lithium ions intercalate in the crystal structure. During discharge, lithium ions disentangle themselves from the struc ture of the negative electrode, electrons are produced and pushed into the external circuit. The ions move through the electrolyte to the other electrode, and as they intercalate in the structure an electron is pulled in from the external circuit.
There are many other possible battery chemistries.

Capacitors
The traditional capacitor [21] consists of a dielectric sandwiched between two metal plates (figure 2(b)). When a voltage is applied across the capacitor, charge builds up on the metal plates, giving rise to an electric field in which energy is stored. A variety of capacitors exist, some of which differ in design from the 'classic' variety.
Particularly high values of capacitance are achieved in supercapacitors, also known as ultracapacitors or electrochemical capacitors. Supercapacitors [22] fall into two categories: 'pseudocapacitors', in which charge is stored via redox reactions at the electrodes, or 'electrochemical double layer capacitors'. In the lat ter, two electrodes are immersed in an electrolyte with a separator in between. When an external voltage is applied, the electrodes become charged as a result of the flow of electrons. The mobile ions in the solution move towards the oppositely charged electrodes, and a layer of ions forms at the electrode-electrolyte inter face. The socalled double layer [23] consists of the layer of ions in the electrode and the layer of counter ions in the electrolyte (Figure 2c). It is extremely thin, which means that the charge is being stored in a very narrow volume, leading to high capacitance.

Fuel cells
Fuel cells [23] produce electricity as a result of electrochemical reactions between reagents that are fed into the cell. Unlike battery electrodes, fuel cell electrodes are inert; they essentially catalyse the reaction. Fuel cells are generally classified by electrolyte, and some systems run at very high temperatures (a few hundred degrees C). The energy conversion can be very efficient because these systems are not subject to the thermodynamic limit for heat engines. Fuels include hydrogen and hydrocarbons, and oxygen/air is usually required.

Overview
The energy storage devices described here are suited to different applications. In general, fuel cells are high energy systems, and supercapacitors are highpower systems, while batteries are intermediate in both domains.
Key performance indicators for energy storage devices include energy density (either volumetric or gravimetric i.e. by mass), power density, lifetime (and number of chargerecharge cycles), efficiency, and safety.

Ions as biological charge carriers
Many biological processes take place in an aqueous environment, in the presence of various ions such as sodium, potassium and chlorine. The fundamental building block of biology is the cell, which is bounded by a membrane consisting of a phospholipid bilayer (figure 3(a)). Various biological molecules are embedded in the bilayer, including larger molecules such as proteins. The membrane itself is insulating but there are several mechanisms by which substances can be transported from the exterior of a cell to the interior or vice versa. For instance, ions can move through ion channels, protein pores embedded in the membrane.
The conditions inside and outside the cell may dif fer, in terms of electrical potential (i.e. voltage) and the concentration of ions. For a given type of ion, the tendency of an ion to cross the membrane (outwards) through an open ion channel is quantified by the ion motive force, which has units of volts. This is given by: where R is the molar gas constant, z is the charge on the ion in multiples of the electronic charge e, and F is the Faraday constant [24]. The first term in the IMF equation is the mem brane voltage, where a negative sign for this quantity indicates that the inside is negative with respect to the outside. The second term in the IMF equation is the chemical potential, where the square brackets denote the concentration of the indicated species. In the case of H + ions, the term 'proton motive force' is used for the IMF.
If only one ionic species is relevant in a particular situation, and the ion is in equilibrium, the ion motive force is zero and the membrane voltage is equal to the chemical potential. In this case, the IMF expression reduces to the form known as the Nernst equation, which defines the membrane voltage at equilibrium. Where multiple ionic species are involved, the more complex Goldman-Hodgkin-Katz (GHK) equa tions can be used. The GHK voltage equation applies in equilibrium but the GHK current equation applies out of equilibrium.
If the motive force on an ionic species is non zero, the ions can cross the membrane through ion channels. Such channels can be gated, such that they only open in the presence of a stimulus, in the form of a transmembrane potential, chemical ligand, or a mechanical force. Ion pumps transport ions against the prevailing ion motive force and require a source of energy to operate. Ion channels and pumps are often highly specific to particular types of ion, because they possess a structural feature known as a selectivity fil ter, in which the amino acids of the protein structure are arranged such that only an ion of the correct size and charge can pass through. In humans, malfunc tions of ion channels are associated with a range of diseases known as channelopathies, including a num ber of neurological, cardiovascular and autoimmune diseases.
The operation of ion channels underpins the transmission of nerve impulses along neurons [25]. When a neuron is resting, and there is no impulse, the membrane potential is approximately −60 mV. When an action potential is initiated, sodium ion channels begin to open and sodium ions flood into the cell, increasing the membrane potential ( figure 3(b)). This causes the opening of additional sodium ion chan nels that are voltagegated. The voltage becomes posi tive, and when it reaches a critical value, the sodium ion channels close, and potassium ion channels open. Potassium ions flow out of the cell, causing the mem brane potential to decrease. The original configuration is restored by ion pumps. The action potential prop agates along the cell because depolarization of one area of the membrane triggers opening of sodium channels in the next section of the membrane. The classic mathematical model for nerve impulse transmission was derived by Hodgkin and Huxley in 1952, and they were awarded the Nobel Prize in Physi ology and Medicine in 1963, together with Eccles. Hodgkin and Huxley modelled the neuron membrane as a capacitor, the ion channels as a set of resistors in parallel, and the ion motive force as a batterylike entity [26].
The mechanism by which an electric eel gener ates electricity (figure 3(c)) is similar to that of the action potential propagation. The cells responsible for generating the eel's shock are known as electro cytes, and they are arranged in columns along the eel. At rest, the transmembrane potential of an electro cyte is around −85 mV [27]. When an electric shock is triggered, sodium ion channels open on one side of the electrocyte, and the potassium channels in the same area close. The membrane on that side of the electrocyte depolarizes, creating a potential difference between one side of the electrocyte and the other. The total voltage of all the stacked electrocytes can be as much as 600 V [28].
Ion channels also underpin the mechanism of sight. In the eye, light is absorbed by the molecule reti nal, which is bound to the opsin protein, forming the structure 'rhodopsin'. Rhodopsin is embedded in the membrane that bounds subcellular disks in rod cells in the eye. The absorption spectrum of retinal depends on the structure of the opsin to which it is bound [29]. Light absorption triggers a conformational change in retinal ('photoisomerization'), which causes the pro tein to change shape. This transformation initiates a cascade of events that ultimately leads to the transmis sion of a nerve impulse to the brain. Socalled chan nelrhodopsins, found in organisms such as algae, are lightdriven ion pumps, which have been exploited in applications such as optogenetics, the science of con trolling gene expression using light. Bacteriorhodopsin (figure 3(d)) is a specific exam ple of a lightdriven ion pump, and it is found in the purple membranes of halobacteria, an extremely halo philic type of Archaea (not actually Bacteria, despite the name) [30]. The purple membrane can be isolated and is approximately 25% lipid and 75% protein by weight. Bacteriorhodopsin itself is a membranespan ning trimeric protein, and each monomer consists of 7 transmembrane helices surrounding a central pore.
As with the rhodopsin protein in the human eye, the chromophore is retinal. Photoisomerization of retinal disrupts a water molecule that is situated in a criti cal position, and a series of conformational changes ensues, resulting in the transfer of protons between various amino acid residues in the protein. Ulti mately, the net result is the transfer of a proton across the membrane, and the protein returns to the ground state. The timescales for the key processes have been measured spectroscopically. Bacteriorhodopsin has potential applications in a range of devices, including photovoltaics (to be discussed below), and can even be acquired commercially. It can form lattices on surfaces, which can be imaged by AFM [31].

Bioelectrochemistry
In solid state devices current is usually carried by electrons and holes moving through conductive materials, but in biology we usually think of electrons in the context of electrochemistry and electron transport chains. Here, electrons are passed between different types of molecule. This is captured in the classic concepts of reduction (gain of electrons) and oxidation (loss of electrons), collectively referred to as 'redox' reactions. The reduction/oxidation potential quantifies the energy required for a species to gain/lose electrons.

Photosynthesis
Oxygenic photosynthesis is the process by which green plants, algae and cyanobacteria extract energy from sunlight. The net result of photosynthesis is the production of a sugar and oxygen from carbon dioxide and water. The details of the biomolecular mechanisms of photosynthesis depend on the type of organism, and the description that follows refers to oxygenic photosynthesis in plants.
Photosynthesis [25] involves light reactions and dark reactions. Both reactions take place in chloro plasts, which contain a number of stacked membrane structures called thylakoids, in which photosystems are embedded. Each photosystem consists of antennae (also known as light harvesting complexes) and reac tion centres. In the light reactions, photons are cap tured by the pigments comprising the antennae, arrays of chromophores in a protein structure. The anten nae funnel the photons' energy to the reaction centre, which is a complex comprising proteins, a chromo phore (a chlorophyll) and other components (figure 3(e)). Photoexcitation occurs, elevating an electron to a higher energy state. Charge separation occurs rapidly, and the electron passes along a transport chain, losing energy as it does so. The transport chain consists of a series of molecules that can accept or donate electrons, changing oxidation state as they do so. The energy of the electron is used to pump protons that then drive ATP synthesis. The electron is then transferred to another photosystem, and it is photoexcited again. It moves on once more, and is ultimately used to produce NADPH. The ATP and NADPH can donate electrons for the dark reactions, which take place in the stroma, the medium surrounding the thylakoids, and produce a sugar (glucose). The holes created in the first pho toexcitation step are used to split water, which means that water acts as the original source of electrons.

Electromicrobiology
Various microbes have evolved the ability to transport electrons, along distances ranging from nm to cm. This enables them to use materials in their environment (e.g. ironbased minerals) as electron donors or acceptors, to supply or remove electrons for metabolic reactions [32].
The term exoelectrogen has been applied to organ isms that transfer electrons in and out of the cell. There are three main mechanisms for electron transfer in electrically active microbes (figure 3(f)): the use of soluble electron 'shuttle' molecules that are produced by the cell, short range electron transfer from mol ecules embedded on the outer surface of the microbe, and long range transfer through extended structures. Although these mechanisms have been studied inten sively, many questions still remain.
Examples of electron shuttles include the flavins produced by Shewanella onedensis, which are secreted into the extracellular matrix. When in an appropriate state, flavins can reduce insoluble Fe (III) [32].
Short range direct electron transfer to electrodes or substrates usually involves cytochromes. These are proteins that contain heme complexes, small struc tures consisting of ringlike molecules with iron at the centre. Shewanella requires certain cytochromes in addition to flavins. Pumpprobe spectroscopy has been used to measure the rate of charge transfer between heme groups in cytochromes [33].
In the electrically active species Geobacter sulfurreducens, longrange electron transfer occurs through nanowirelike structures called pili, which extend from the cells into the surroundings. Experiments have demonstrated that each pilus is conductive, and when the gene for the pilus protein PilA is knocked out the cells can attach to Fe(III) oxide surfaces but are unable to grow [34]. Sitespecific modification of amino acids in PilA can increase the conductivity of the pilus dra matically and reduce its diameter [35]. Shewanella also possesses external nanowires, but some of these struc tures appear to differ significantly from the pili of Geobacter. For instance, some Shewanella nanowires are significantly less conductive than those of Geobacter, and it is believed that this is because the conduction mechanism in Geobacter is almost metallike, whereas in Shewanella electrons are thought to move in a step wise manner along a chain of redoxactive molecules [32].
Extremely long range transfer has been seen in 'cable bacteria' [36]. These remarkable filamentous microbes grow in marine sediments, forming centi metrescale chains that contain unbranched chains of as many as 10 000 cells or even more [37]. At the bot tom of the filament, there is no oxygen but there is a source of sulphide. At the top of the filament there is oxygen but limited sulphide. Bacteria in different positions in the filament therefore perform differ ent redox reactions. At the top of the filament oxygen is reduced to form water, using electrons generated at the bottom of the filament by oxidation of the sul phide source. The outer membranes of the individual cells merge, to cover the complete filament, and elec trons are transported through the space between the outer membrane and the inner membranes. nAscale currents have been measured directly in filaments of intact cells, and similar size currents were seen from the isolated fibre sheath (the part of the filament out side the inner membranes) [38]. Although the mech anism of electron transport is not yet fully known, a recent study suggested that a protein known for form ing pilitype structures was the most abundant pro tein in one particular species of cable bacteria [39], and Raman spectr oscopy results reported in another study indicated a lack of cytochromes in the isolated fibre sheath [38]. One hypothesis is that the electrons are transported through an assembly of pilitype pro teins located in the periplasm (between the two mem branes).

Electrosynbionics: technologies
The phenomena described above can be used as the basis for a wide variety of technologies, some of which I will describe here. Technologies that use chemical sources of energy will be discussed alongside closely related technologies that harvest energy from sunlight.

Biobatteries, enzymatic fuel cells, biosupercapacitors and sub-cellular biophotovoltaics
The principle of enzymatic fuel cells and biobatteries is simply that enzymes catalyse reactions that release electrons, which can be harvested and used to drive a current in an external circuit. The enzymes may be in solution, immobilized on a surface or trapped in a gel. Electron transfer from enzymes to electrodes may be either direct or facilitated by a mediator. Mediators must be chosen such that their redox properties match the process to be mediated. One possible application for an enzymatic fuel cell or biobattery [40] is as an implantable power source for a devices such as a pacemaker. Key factors to consider in the assessment of the technologies include stability, power density and costs. Enzymes can potentially be made comparatively cheaply at large scales in a fermenter [40].
Technically, the term fuel cell implies that the fuel is supplied from a source external to the device, whereas the term battery implies an internal fuel source, but in practice some papers use the terms interchangeably, in part because some devices can be used in either way. However, it is important to draw a distinction between enzymatic and microbial fuel cells. Enzymatic fuel cells use subcellular components in the form of enzymes extracted from cells or made synthetically, whereas microbial fuel cells are based on living cells and will be discussed below.
For one biobattery, a synthetic reaction pathway involving 13 enzymes was designed de novo [41]. In the anode compartment, maltodextrin was broken down and electrons were transferred to the electrode. The reaction pathway was cyclic, meaning that the enzymes were regenerated. In the cathode compartment, oxygen was consumed and water was produced. The maximum power seen was 0.8 mW cm −2 at 50 °C but the enzyme activity decreased after a few days.
When enzymes are immobilized on the surface of electrodes ( figure 4(a)) it is common to use high sur face area electrodes, with some degree of micro or nanostructuring, in order to achieve efficient electron transfer [42] and, for immobilized enzymes, a high level of enzyme loading. For example, enzymes such as laccase can be chemically conjugated to structures such as carbon nanotubes [43]. The carbon nano tube aggregate known as buckypaper can also be used to make electrodes for biobatteries [44]. In systems based on immobilized enzymes, the orientation of the enzyme must be optimized for effective electron transfer, as must the behaviour of the redox mediators. It has been shown that orientation and mediation can be controlled using a synthetic molecule that performs multiple functions, anchoring itself to the electrode, binding to the enzyme in an appropriate way and also acting as a good mediator [45].
Some architectures have extended lifetimes, such as the 'enzymatic nanomembrane' system, in which nanomembranes of titanium were rolled up and the enzymes were trapped between the leaves of the roll [46]. In this device the enzymes used were bilirubin oxidase (on the cathode), glucose dehydrogenase and diaphorase (on the anode). The biobattery discharged over 452 h. The operational lifetime of biobatteries/ enzymatic fuel cells can be limited by availability of fuel, in addition to enzyme degradation, and this can be addressed by the use of flowthrough systems, in which fuel is continually replenished [47].
Energy is stored in biobatteries and biofuel cells in the form of chemical fuel that is then metabolized. How ever, a new dimension can be introduced in the form of capacitive storage. It has been shown that cytochromes are responsible for the enhanced capacitance seen in Geobacter biofilms [48] and this has inspired the con struction of synthetic biosupercapacitors, including one device based on biophilized graphene oxide and myoglobin [49]. Myoglobin stores charge by means of protonation or deprotonation of charged residues, but the mechanism appears to persist even if the secondary structure of the protein is disrupted.
In selfcharging biosupercapacitors (also known as supercapacitor/biofuel cell hybrids), enzyme reac tions charge the inbuilt capacitance. As in conven tional supercapacitors, the charge can be stored either through electrochemical double layer capacitance or pseudocapacitance, created by redox reactions in the vicinity of the electrode surfaces. In one example, the electrodes consisted of enzymes integrated with car bon nanotubes [50], and after the system discharged itself it was observed to recharge. It performed nume rous charge/discharge cycles but when the glucose fuel was removed the open circuit voltage fell sharply because the enzyme substrate was no longer present. In a later case, graphite electrodes were coated with an osmiumbased redox polymer to create a charge stor age layer. This was then overcoated with a layer of enzymes trapped in the same polymer. The bioanode was based on a glucose dehydrogenase, while the bio cathode was based on bilirubin oxidase. During opera tion, the action of the enzymes changed the redox state of the polymers, leading to capacitive charge storage. This system was referred to as a Nernstian biosuper capacitor [51] because the same polymer was used for both electrodes and the Nernst equation governed the open circuit voltage of the complete device.
It has been shown that supercapacitor/biofuel cell hybrid devices can be used to generate pulses similar to those needed for the operation of a pacemaker [52]. Devices can be made from flexible materials, such that the power output was not changed significantly when the cell was bent, even by an angle of 60° [53].
The biobattery, biosupercapacitor and enzymatic fuel cell systems described so far all charge biochemi cally, i.e. through metabolism of a fuel. However, in biology many organisms harness solar energy through photosynthesis, and the machinery they use for this can be isolated and incorporated into subcellular biophotovoltaics. This term refers to devices in which electricity is generated from sunlight using comp onents such as chloroplasts and thylakoids that have been extracted from cells, in contrast to the systems I call 'living photovoltaics', which are based on living cells and will be discussed below.
An early example of a 'photoelectrochemical cell' (i.e. a subcellular biophotovoltaic system) employed thylakoid membranes extracted from spinach [54]. Photocurrents of several µA were observed, and this effect was suppressed by the addition of particular herbicides, which were known to inhibit photosyn thesis. Current could be increased to tens of µA by the addition of artificial electron acceptors and the immobilization of thylakoids in a matrix. In general, in enzymatic fuel cells, biosupercapacitors and bio photovoltaics, immobilization can either enhance or detract from performance, depending on the details of the device under test.
In a more recent study [55], the anode comprised spinach thylakoid membranes chemically tethered to a matrix of multiwall carbon nanotubes depos ited on a gold electode. The cathode was similar, apart from the use of the enzyme laccase in place of thylakoids ( figure 4(b)). The purpose of the laccase was to regenerate water from oxygen. The maximum power observed was 5.3 µW cm −2 , the system being characterized under a lamp. Another setup [56] was also based on thylakoid membranes, but utilized the enzyme bilirubin oxidase and electrodes made of ITO glass-the same material that is standard in dyesen sitized solar cells. At peak power output, 0.6 µW cm −2 was delivered. This system was characterized in ambient natural light and was referred to as a solar biosuper capacitor.
In subcellular bioPV systems, it is not necessary to use intact thylakoid membranes, and it is possible to build solar cells using isolated photosystems or reac tion centres. In one study, an architecture similar to that of a dyesensitized solar cell was used [57], with photosystem1 from thylakoid membranes of cyano bacteria as the 'dye'. The natural redox mediators were replaced with an electrolyte and a nanostruc tured metal oxide, in the form of TiO2 paste or ZnO nanowires attached to a conducting glass electrode. The system was tested with AM1.5 standard simulated sunlight, and a power of 81 µW cm −2 was achieved, with an external efficiency of 0.08%. In this study, the photosystems were stabilized using designer peptides, which have been shown to protect proteins like this from structural degradation for an extended period [58].
Interestingly, in DSSCstyle devices based on pho tosystems, performance can be improved by incor poration of an artificial lightharvesting antenna-a molecule that absorbs photons and passes their energy to the photosystem. In a recent study, this yielded a power conversion efficiency of 0.13%, measured with a standard solar simulator (AM1.5) [59].
In an alternative approach, the entire Zscheme of photosynthesis was mimicked using an anode based on photosystem 2 and a cathode based on photosys tem 1 [60]. In the anode compartment, water acted as the electron donor, while methyl viologen acted as the electron acceptor in the cathode compartment. Methyl viologen was ultimately reoxidized by molecular oxy gen. In this case, the photosystems were attached to the electrodes via a redox hydrogel, which acted as a bridge for the electrons that flowed through the exter nal circuit during illumination. The power obtained was 23 nW cm −2 but this was increased subsequently to almost 2 µW cm −2 , with corresponding efficiency of 0.0045%, by changing the redox hydrogel used for the anode [61]. However, these power figures are not exactly comparable as illumination conditions were not identical-although LEDs were used in both cases, the stated light intensity was different.
Photocurrent can be generated from isolated Rhodobacter sphaeroides reaction centres adsorbed on Biophotovoltaic system based on immobilized thylakoid membranes and enzymes (geometry not depicted accurately), similar to the system described in [55]. (c) Microbial fuel cell operating in a marine environment, as described in [78]. (d) Plant microbial fuel cell, as described in text and multiple references therein. (c) Bacteriorhodopsinbased photovoltaic system, as described in [96]. (d) Schematic of electric eel inspired hydrogel battery, as described in [27]. silver electrodes with cytochrome c [62]. When the electrodes were nanostructured using electrochemical roughening, the number of reaction centres loaded on the surface increased. The rate of electron transfer per reaction centre also increased, and this was attributed to plasmonenhanced lightharvesting. In another experiment, prior to dropcasting of reaction centres and cytochromes, electrodes were coated with self assembled monolayers (SAMs) of 11mercaptound ecanol or 11mercaptoundecanoic acid, to facilitate measurements of electrical properties of the proteins [63]. For the cytochromes to act as an effective bridge between the electrode and the reaction centres, it was found necessary for them to have some mobility; when they were chemically immobilized the photocurrent decreased dramatically.
One major issue with the use of subcellular parts of the photosynthetic machinery in vitro is their lim ited lifetime. It has been noted that isolated photosys tems will only survive for 30-40 min of illumination in vitro, with whole chloroplasts being able to last a little longer (normally a few hours at most) [64]. This may be acceptable for some niche applications [65], but it will not be suitable for mainstream scenarios, where electrical infrastructure is expected to last for years. As mentioned above, additives such as designer peptides may stabilize the structure of photosynthetic proteins to some extent [57,58], but it is not clear whether this will be sufficient to prevent longterm functional dete rioration under continuous illumination. As an alter native, regeneration may be possible. For instance, in one study it was demonstrated that a photosystem Iloaded gold electrode could be replenished with fresh photosystem from solution [66]. In this case, the other electrode was made from eutectic galliumindium, which is flexible. When illuminated under standard AM1.5 conditions, the response was too small to be measured, but a photocurrent was observed under laser illumination. The efficiency was 0.0025%.

Microbial fuel cells and living photovoltaics
An obvious way to circumvent the instability of isolated biological parts is to avoid extracting them in the first place, and work with living systems that can refresh proteins as they wear out. These systems include microbial fuel cells (MFCs) and living photovoltaics. MFCs produce electricity by exploiting the metabolic processes of microorganisms and require a chemical fuel to operate. Some living photovoltaics are described as photoMFCs, and may or may not require additional chemical fuel. As will be seen, there is some overlap between the categories. Terminology is not applied in the same way by all researchers.

MFCs (non-photosynthetic)
In an MFC, microbes oxidize a substrate and pass electrons to an anode [67]. The electrons then flow through an external load to the cathode. The substrate will ultimately be used up and must be replenished.
The typical design consists of two chambers separated by a cation exchange membrane such as Nafion. Each chamber contains an electrode. The anode must be conductive, biocompatible and stable, and carbon is often chosen. Cathode materials depend on the application.
For large MFCs, the high internal resistance can limit the power density [68]. In any case, the power output is usually quite low, which means that stacks of MFCs may be needed to generate sufficient power (depending on the application). Some of the comp onents can be expensive, so lower cost options would be useful.
An emerging trend in the area of microbial fuel cells is the use of this technology to simultaneously generate electricity and remove contaminants from wastewater at treatment plants, and pilot projects have taken place [68]. For practical application of this technique it is necessary to operate microbial fuel cells at scale, using large volume reaction chambers. When scaling up from laboratory scale, it may be neces sary to introduce design modifications to achieve the necessary performance. For instance, in a 1.5 lscale system, it was shown that increasing the number of electrodes (both anode and cathode) improved per formance significantly [69]. For much bigger systems, at a scale of tens of litres, recirculation of the wastewa ter can increase the maximum power density by up to 17%, but this depends on the recirculation time and flow path [70].
In one example of a study of largescale micro bial fuel cells for simultaneous electricity produc tion and wastewater treatment, 48 fuel cell modules were connected in a series stack, the total volume of the anode compartments being around 100 l [71]. For power extraction, a commercial energy harvest ing system was used, with the capacity for 'maxi mum power point tracking', a means of optimizing output by matching the load impedance in real time. The power output from the system was 84-130 mW and it removed over 75% of the organic matter as well as 80% of the solid content. It was noted that if one weaker source was connected in series with stronger sources, the overall performance could be degraded. Wastewaterbased MFCs can function for extended periods of time, but their behaviour is not very repro ducible [72].
At the opposite end of the size scale, microbial fuel cells could potentially be used to provide longterm power for autonomous sensors that form part of a network. In order for this to be practical, it is neces sary for the technology to be miniaturized. One early example of a miniature MFC consisted of two cham bers made from nonconducting plastic, each having a volume of only 1.2 cm 3 and housing a carbonbased electrode [73]. The two chambers were separated by a Nafion proton exchange membrane. The microbe used was Shewanella oneidensis, metabolising lactate. Measurements were made over a period of 7 d, with the addition of lactate where necessary. The maximum output power obtained was 0.3 mW cm −2 .
Subsequently, even smaller MFCs were developed, reaching microlitrescale. For one device, based on a single chamber with a volume of 25 µl, gold anodes were fabricated on glass using standard photolithog raphy techniques [74], and a biofilm was formed on the gold surface. The reaction chamber was made from PMMA, and this was placed between the glass and a proton exchange membrane. This device was also based on Shewanella oneidensis and produced a maxi mum power of 2.9 µW cm −2 (power per geometric area of electrode).
In standard labonchip microfluidic fabrication, it is common to use PDMS instead of PMMA, but PDMS can cause problems in MFCs because it allows oxygen to enter the reaction chamber. It has been shown that coating PDMS with parylene C significantly reduces the oxygen permeability and this was used in a minia ture MFC based on Geobacteraceae, in a microfluidic system with a channel 55 µm deep. The peak power den sity observed was approximately 72 µW cm −2 , where this was achieved when micropillars had been made on the electrode [75]. It has been shown that electrodes made from conducting nanofibres of the polymer PEDOT worked effectively in a microscale MFC based on Shewanella oneidensis [76], a device which pro duced a peak power density of around 2.25 µW cm −2 , with an anode chamber volume of 12 µl. Electrodes with micro or nanostructuring can produce per formance enhancements due to the increased area for electron transfer and biofilm formation, but the sur face must be wellsuited for biofilm growth.
At an intermediate size scale, microbial fuel cells can be used to generate power in marine environ ments. An anode buried in sediment under seawater is colonized by microbes, and if connected to a cathode immersed in the water above through a resistive load, power is generated [77] (figure 4(c)). Such a device can run for an extended period, and in 2001 was observed to give an average power of 2.8 µW cm −2 at an output of 0.27 V over four and a half months. Later, similar systems were used to power a meteorological buoy, and it has been noted that they can operate for long peri ods without maintenance, potentially for years [78]. Recently a patent was filed for a device of this type that could be deployed remotely [79].
Microbial fuel cells utilize microorganisms to generate electricity using a chemical fuel source. It is also possible to use microbes to generate chemical fuel using electricity, and the use of bioelectrochemical sys tems for electrosynthesis has been attracting increas ing attention [80].

Living photovoltaics
Living photovoltaics contain cells (such as cyanobacteria) that harvest energy from sunlight and export electrons, generating a voltage/current. As these devices possess the ability to renew damaged components, they should be more durable than those counterparts based on isolated subcellular parts. As an example of a living photovoltaic system, when cyanobacteria are grown on the anode of a microbial fuel cell, the electrode acts as an extracellular electron acceptor, and a rapid increase in voltage is seen when the cell is exposed to light [4]. The voltage decreases dramatically when the light is switched off. The response to light of different wavelengths is consistent with the known properties of cyanobacterial light harvesting systems. A photovoltage is also seen from a naturally occurring consortium of organisms. Non photosynthetic bacteria do produce a voltage in such an MFC, but the voltage produced is independent of illumination. It is important to note that various species of photosynthetic microorganisms can be used in living photovoltaics including types of algae [81].
It was recently shown that the use of a flow cell could enhance the performance of cyanobacteria based living photovoltaics [82]. Here, the cyanobac teria were exposed to light in one chamber and the electron transfer occurred in a second chamber. Care ful design of the second chamber and the flow param eters also enabled elimination of the membrane that is usually used as a separator between the electrodes. This simplified fabrication and avoided one of the causes of biofouling.
It has also been demonstrated that it is possible to 'print' living photovoltaic systems using a modified HP inkjet printer, where cyanobacteria were incorpo rated into a 'bioink' and electrodes were printed using a conductive ink [83]. The solar cell was able to power small electronic devices, namely an LED or a simple digital clock. It also produced some current in the dark due to usage of internal reserves produced in the light. The combination of energy storage and electricity gen eration in the same device has the potential to reduce cost and complexity in comparison with systems based on a photovoltaic panel connected to a battery.
Despite considerable efforts by a number of groups, the efficiency and power output realized experimentally for living photovoltaics is thought to be considerably below the theoretical maximum [64,84]. Results obtained up to 2015 were summarized in a review of that year [64], in a form not dissimilar to that of tables for conventional solar cells. The caveat here is that measurement conditions (e.g. illumina tion) for living photovoltaics differ between studies, whereas measurement techniques are standardized for nonbiological technologies. A major challenge in the development of living photovoltaics is that the organ isms typically being used have not evolved efficient mechanisms for extracellular electron transfer [64,84]. Effort is being made accordingly to use synth etic biol ogy to genetically reengineer some of these organ isms, for example by expressing elements of the Shewanella electron transport system in other organisms such as E. coli and cyanobacteria [84].
In an alternative approach, a system was recently presented in which two engineered microorganisms worked together to harvest light and produce current [85]. The cyanobacterium Synechococcus elongatus used solar energy to produce Dlactate, which was used as a metabolite by Shewanella oneidensis, the organism responsible for exporting electrons. When lactate pro duction and electron export were separated both spa tially and temporally, it was possible to obtain power for tens of days at levels up to about 20 µW cm -2 .

Plant microbial fuel cells
Electrical power can also be extracted from plant microbial fuel cells ( figure 4(d)). Green plants undergo photosynthesis as described in section 3.2.1, and ultimately use energy from light to produce carbon containing compounds such as glucose from carbon dioxide and water. A significant quantity of carbon is released by the plants into the medium in which they grow, as a result of secretion, cell death and other phenomena. This carbon can be metabolized by bacteria, which can export electrons that can be used to produce current in an external circuit. In one such plant microbial fuel cell, rice plants grew in a matrix in which anodes had been embedded [86]. The cathode was placed in the water layer above the growth matrix. The matrix had been inoculated with bacteria, and the reactors were kept for a period of months in a greenhouse, under lamps that provided a 16 hour day and an 8 hour night. The temperature was variable. Different growth matrices were tested. The power extracted was a factor of 7 higher in the presence of plants than in the control measurements with no plants present. For the plant MFC, the average power produced over representative stable periods was 3.3 µW cm −2 (anode geometric area).
Another plant MFC was based on Spartina anglica [87], which grows in salt marshes. In this case, the anode consisted of a bed of biofilmcovered graphite granules, in which the plant grew. The cathode was made from graphite felt and light was provided for 14 hours per day at a fixed illumination level. Duplicate plantMFCs were tested, alongside blanks (control systems without plants). After an incubation period of 3-4 months, the plantMFCs started to produce current. During the period of current generation, the average power was about 1-2 µW cm −2 while no sig nificant current was generated from the blanks. Power increased when ferricyanide was added. When the bac teria in a plantMFC based on Spartina anglica were analysed, it was found that the most abundant types were Proteobacteria [88]. It is important to note that the use of a salt marsh plant in a plantMFC would enable deployment of the technology on land that is not cultivated for agricultural purposes [87].
The term 'vascular plant biovoltaics' has also been applied to the plantMFC technology [89]. Different plants give different results; for instance, when plant MFCs were made using either the rice species Oryza sativa or the weed Echinochloa glabrescens, which com monly invades rice paddies, it was found that the rice based system produced around ten times the power of the weedbased version [89]. In this case, the plant MFCs were not innoculated with bacteria, and relied on selfforming communities.
For practical use of plant microbial fuel cells out side the laboratory, it will be necessary to ensure long term operation, and the ability to withstand changing weather conditions. In one study a plantMFC was deployed on a roof, but the plants did not survive the winter [90]. Indoor plants can also be used [91]. As a demonstration of how a plantMFC could be used, a concept design was produced for a table incorporat ing such a system based on moss (the 'Moss Table') [92]. Although the lamp in the table was not in fact powered by electricity generated biologically, the Moss Table was intended to demonstrate how technology like this could be used in the future.
As an aside, in connection with the use of plants in electrosynbionic devices, it is interesting to note that conductive materials have been synthesized in living plants, to make wires and capacitors [93].

The artificial leaf
As an alternative to the direct use of biological organisms and subcellular components for harvesting solar energy, it is possible to adopt a biomimetic approach, using entirely inorganic solid state materials. This led to the creation of the 'artificial leaf', a device based on a silicon photovoltaic cell coated with catalysts that evolve hydrogen or oxygen [94,95]. This system produces hydrogen and oxygen by using solar energy to split water. It can use water from a variety of sources, does not require wires, and operates under benign conditions. It has been suggested that this type of system could be particularly advantageous in the developing world.

Borrowing bacteriorhodopsin
Bacteriorhodopsin and its relatives have attracted considerable interest for the purpose of novel photovoltaics and other optoelectronic systems. Bacteriorhodopsin is usually described as very stable i.e. more stable than chloroplasts. If this is correct, using such a molecule could provide a way to use a biological light harvesting agent without introducing the complexities of a living system and with fewer difficulties with longterm stability than for isolated chloroplasts, thylakoids and photosystems. Bacteriorhodopsin can be reconstituted in a lipid membrane for photoelectric measurements, and a photocurrent is seen [96]. Fragments of bacteriorhodopsinloaded purple membrane can also be embedded in synthetic materials, for which the photovoltaic properties can be measured. In an early study (in the 1980s), it was stated that the efficiency of energy conversion from one such membrane was approximately 0.5% but this could be improved by the use of better techniques for ensuring the correct orientation of bacteriorhodopsin [97].
In 2009, a patent application was filed for 'protein based photovoltaics and methods of use' [98]. In this invention, one or more oriented layers of bacteriorho dopsin molecules pump protons from the vicinity of one electrode to the vicinity of the other, creating a gradient ( figure 4(e)). Hydrogen is generated at the electrode that has a surplus of protons, drawing elec trons through the external circuit from the other elec trode, where oxygen is generated. The hydrogen and oxygen can be used to produce electricity. The patent also covers variants based on other types of rhodopsin, including mutants that pump chloride ions instead of proto ns, and notes that it is possible to prepare mutants with an enhanced ability to bind to substrates. The patent was granted in 2015.
As a variant on this theme, some groups have developed a bacteriorhodopsinbased version of the dyesensitized solar cell described above, where bacte riorhodopsin replaces the dye [99].

Electric eel batteries
Studies of electric eels/rays and other examples of physiological electricity provided inspiration for the development of the earliest batteries [100]. It was in 1775 that Henry Cavendish gave to the Royal Society 'An Account of Some Attempts to imitate the Effects of the Torpedo [electric ray] by Electricity' [101]. Electric fish have continued to inspire us into the twenty first century. For example, it has been suggested that a biobattery could be made using synthetic cells that act somewhat like electrocytes, consisting of self assembled structures enclosed by phospholipid bilayers or equivalent, with embedded ion channels [102].
Recently, a hydrogelbased analogue of an electric organ was constructed [27], in which gel droplets were used to mimic electrocytes (figure 4(f)). The gels were modified chemically to make them selective for differ ent types of ion. In the initial configuration, the drop lets were isolated and the concentration of salt varied between droplets. When the droplets were placed in contact, ions could move between them, down the concentration gradient, moderated by the selectiv ity of the gels. A potential difference was established between the terminal gels, which was then measured as the opencircuit voltage.

Discussion
As described above, there are many instances of biological molecules and systems being reengineered to form the basis of new technologies with electrical applications, or biological phenomena being used as inspiration for new devices. However no overarching term has previously been suggested to cover these endeavours, the term bioelectronics normally being reserved for biosensors etc. I propose that the community should rebrand the technologies described in section 4 (and similar ones yet to be developed) as 'electrosynbionics'. This term captures the key elements of this emerging area, and the synthetic nature of the devices is central.
In addition to rebranding, a more systematic approach may be needed, with the adoption of testing and measurement methods similar to those developed in engineering and physics laboratories for conven tional devices. For example, electrosynbionic solar cells should be tested using solar simulators, using controlled conditions, as is standard for nonbiological devices. While this approach has been adopted in some papers, it is far from universal. Properties of electrosynbionic devices should be measured and specified in a manner that enables reliable comparison to existing technolo gies, to enable competitiveness to be determined.
At present the comparison is likely to be unfavour able to electrosynbionic technologies, which are mostly at low technology readiness levels. At the moment, most devices have short lifetimes and low energy output, and major unanswered questions remain over their long term competitiveness. However, further investment would be likely to lead to major developments in the field, and could even drive a paradigm shift as the full potential of semibiological and bioinspired systems is realized, particularly as they offer the opportunity to generate and store electricity in the same device, with the advantages of cost and simplicity.
The critical challenges to be addressed are the efficiency of generating or storing electricity, and the lifetime or stability of biological components. The first of these two challenges could be addressed by sys tematic design and exhaustive experimentation, while the latter is likely to be more difficult to overcome. For some technologies the answer may lie in living systems that selfrenew, replacing elements that wear out. However, in other cases, this will introduce more difficulties than it solves, and it will be necessary to replace some biological components with solid state analogues, to produce hybrid devices and capture the best of both worlds. There will be significant questions to be answered about how to construct the interface between the biological and the nonbiological parts, and insights in this domain may also advance areas other than electrosynbionics [54].
If the challenges can be addressed successfully, it will be possible to exploit the unique advantages of bio logical and bioinspired systems. Many biological pro cesses are characterized by excellent energy efficiency, operating near to the limits set by thermodynamics, and where there is scope for improvement this can be addressed by exploiting the capabilities of synthetic biology and genetic engineering, potentially mak ing electrosynbionic technologies more competitive [84,103]. The functionality of biosystems is usually underpinned by intricate molecular structures, which can be utilized to organize functional groups at the nanoscale. The specificity of biomolecular interactions allows precise targeting mechanisms to be designed, while the function of biological parts can be altered by deliberate changes to their structure. The operat ing and storage conditions for biological systems are usually benign, which facilitates handling. In contrast, the process of cleaning silicon wafers involves the use of some of the most dangerous chemicals ever used in laboratories, such as hydrofluoric acid.
There are great opportunities in the energy sector as new and better technologies are sought to make elec tricity generation and storage cleaner and cheaper. The market for electricity generation, storage and supply is huge [1] and it is possible that in the future electrosyn bionic systems will be part of the energy landscape, but a great deal of work remains before these approaches will be commercially viable.