Vacuum 80(2006)1053–1065
Proton exchange membrane fuel cells
V.M.Vishnyakov Ã
Dalton Research Institute,Manchester Metropolitan University,Manchester M15GD,UK
Abstract
The desire to have compact,high power sources with low environmental impact has focused attention on proton exchange membrane fuel cells (PEMFC).The paper gives a brief overview of processes and technical challenges related to PEMFCs.The main goal is also to give a short overview of the underlying science and to show a feasible approach to implementation of these cells.The potential for
use of vacuum-based techniques in PEMFC manufacturing is also discussed.r 2006Elsevier Ltd.All rights reserved.
Keywords:Proton exchange membrane;Fuel cell;Backplate;Catalyst;Hydrogen
1.Introduction
It is said that the amount of energy used determines the technological sophistication of a society.There is a constant development of existing and a search for new energy sources.This search is always fuelled either by new energy-hungry processes,or by some energy crisis.In principle,fuel cells offer electrical and heat energies by combining fuel and oxygen without the drama of a high-temperature burning process.They were discovered a long time ago and have developed through a few phases of rises and falls of interest from academia and industry.In the past,technical problems and high production and exploita-tion costs have always managed to wipe them out of the focus of interest.It can be argued that the last rise in interest began with space exploration.This started the current stretch of development which is proving to be the longest and most intensive one.Partially,this can be explained by the potential energy density fuel cells promise to bring for critical,transportation and other portable applications and,partially,by the promise of ‘‘green’’,carbon-emission-free energy.While fuel ce
ll technology by itself is very simple in principle and huge advantages have been made in many aspects of its implementation,it still can be considered to be in its infancy for applications in the wide consumer market.There are numerous reasons for the
relatively slow development in ‘‘market-ready’’fuel cells.The more general ones are generic to all new technologies:it always takes time to develop technology for a broad consumer market.Completed implementation at all times relies on combined areas of technology and they need to be advanced all together.This takes effort and a considerable investment.
Contemporary society in developed countries enjoys,at the moment,an abundance of relatively cheap energy and this adds to a certain level of reluctance (and even resistance)to invest heavily in order to compete with developed market supplies.While in some areas fuel cells promise a revolution there is a tendency to stick with evolution up to the point of sharp crisis.At the very beginning of the current wave of interest,the cost of fuel cells was very high,but this was easily offset by their high energy/weight density as compared with other electrical energy sources.The mass market,on the other hand,needs supplies which are cheap,have high efficiency and are made using reliable technology.Considerable progress has been made to reduce costs and in the last 10–15years,the price per energy unit generated by fuel cells dropped by a factor of almost 10–20.The development is still driven by the pro
mise of new markets,security of energy supply and environmentally clean energy.
Traditionally,there are five types of fuel cells:alkaline,proton exchange membrane (PEMFC),phosphoric acid,molten carbonate and solid oxide (see accompanying article in this issue).One can argue that new types have
www.elsevier/locate/vacuum
0042-207X/$-see front matter r 2006Elsevier Ltd.All rights reserved.doi:10.1016/j.vacuum.2006.03.029
ÃTel.:+441612473385;fax:+441612473382.
E-mail address:v.vishnyakov@mmu.ac.uk.
been developed recently.Generically,all cells consume hydrogen and oxygen to produce electrical current. Different fuel cell types have their strengths and weak points and,as a result,they have their application niches.It is possible to argue that the PEMFCs are the most simple ones and,as such,are very easy to understand,implement and use.The description of processes in PEMFCs and the challenges during development can be used to some extent as an example which will form a better understanding of all fuel cells.Their basic simple implementation and expected wide penetration of PE
MFCs into the end user market have attracted much research and development effort of the PEMFC.
One selling point of fuel cell technology is its high energy-conversion efficiency.In favourable circumstances this efficiency can be almost60%.This seems not to be a high value for the‘‘green economy’’.However,car engines at low power(.y say20%of a rated maximum)can achieve only30%efficiency and the engine will only be more efficient at high powers.Fuel cells on the contrary are more efficient at low power loads.
Thinfilm technologies were always used to improve performance and to reduce cost of devices.Vacuum-based processes are known for creation of high-qualityfilms. Some very encouraging work has been published,mostly in the last decade on the application of vacuum techniques for various PEMFC components.
2.General considerations
The principal construction of a contemporary fuel cell is shown in Fig.1.In a very basic form it has only two catalyst layers and a proton exchange membrane.In this basic form,the proton exchange membrane is just a piece of thin special plastic and a catalyst layer would comprise around100g of catalyst per square metre(10mg/cm2).The side supplied with hydrogen(or other hydrogen-containing f
uel)is named the anode,the oxygen(or air)side is named the cathode.Products of the reaction(electrical current and water)must be taken away.From this practical point of view one needs immediately two rigid and rather compli-cated constructions named backplates.The backplates would also help to dissipate heat in a high-power cell. Fuel cell surface areas range from a few square centimetres for a very small cells to0.1–1square metres for large power cells.The backplates in all cases should make very good and close contact with the catalyst.It is very expensive to make backplates with high surface tolerances which can be maintained over a wide temperature range.The solution is to use a carbon cloth placed between a rigid backplate and the catalyst.The cloth allows gases and water to access the reaction area and it is a good electron current conductor. Being made of soft and pliable material it also helps to deal with imperfections in backplate size,lower the required tolerances for machining and,thus,to minimise costs of backplate production.
For demonstration purposes only the components of the fuel cell are shown separately in Fig.1,but in a working cell all parts should be pressed tightly together.
In all fuel cells there are two contributors to currentflow. The usual,external to fuel cell,electrical current consists of moving negative electrons.The current of charged ions is confined to the fuel cell itself.In the acid type of fuel cell the proton H+moves between two catalyst layers.
At the anode the hydrogen molecule is split into two protons and two electrons
H2!2Hþþ2eÀ.
At the cathode oxygen reacts with protons and consumes electrons to form water
O2þ4Hþþ4eÀ!2H2O.
The electrolyte,separating electrodes and reactions,should only allow protons to pass which means that electron currentflows outside the fuel cell to produce electrical energy.Some polymers can be prepared in such a way that they conduct protons and thus they act as solid electrolytes. The thinfilm of polymer works in this case as an electrochemical membrane and PEMFCs are devices using such a polymer.
Both reactions at the cathode and anode have potential barriers.The barriers are high enough to reduce reactions proceeding at low temperatures and pressures to unusable low rates.This can be overcome by the use of catalysts and/ or high temperatures.There is a limited number of
Fig1.Proton exchange membrane fuel cell essential components.
V.M.Vishnyakov/Vacuum80(2006)1053–1065 1054
materials which can catalyse reactions and survive for a long time without degradation in the very stringent environment of the fuel cell.The choice of catalysts is essentially narrowed to platinum and some platinum-containing alloys.
The volt–ampere characteristic of fuel cells can be,in principle,understood from basic thermodynamics and electrochemistry.If one is to imagine and analyse a fully reversible process then the maximum voltage generated (denoted as E )should correspond to the electromotive force (EMF)generated at the output.The process is fully reversible if the process combining hydrogen (or any suitable fuel)and oxygen into water in the fuel cell does not make unrecoverable losses (for example does not produce heat).In the reversible case one can roughly assume that the energy change in the system (Gibbs free energy of formation)is converted into electrical energy.Then D G T ;P ;f n ÀÁ
¼nNeE ,Or
E ¼D G T ;P ;f n ÀÁ
=nNe ,
Where D G is a change of Gibbs free energy of formation in the system,n is the number of electrons transferred during the reaction (2in the case of hydrogen),N is Avogadro’s number,and e is the charge of an electron.Minus signs were omitted on both sides of the equation.
The Gibbs free energy of formation depends on the temperature (T ),pressure (P )and the phase state of the reactants and product (liquid or gas,f n ).For all materials in a fuel cell in the gaseous state,at atmospheric pressure and temperature 353K (around 801C,a quite typical temperature of operation for a PEMFC)one can easily calculate that E ¼1.17V.This derived voltage can be treated as the maximum one can get under no-loss conditions.In an ideal case scenario this voltage should be independent of the electrical current drawn.
In reality one cannot avoid losses and there are other limitations.The process is irreversible and therefore results in lower output voltages.A typical voltage and current characteristic of a cell is given in Fig.2.There are at least four different processes responsible in a fuel cell for the observed behaviour.
The ohmic resistance to both currents (electrons and protons)generates heat and results in the slow and linear drop of voltage when the current increases (middle part of the curve).
reaction to a book or an articleThis voltage drop can be presented as
D V r ¼i R e þR p ÀÁ
and
R e ¼R cont þR backpl ,
where i is the current,R e is the resistance to electron current,R p is the resistance to proton current through the PEM,R cont is the contact resistance between catalyst and
backplate,R backpl is the resistance of backplates.The nature of the resistances will be discussed later in the paper.The voltage and current relationship during the electro-chemical reaction itself on each catalyst layer are linked by a simple dependence named after J.Tafel,who is renowned for the formulation in 1905of this electrochemistry law for irreversible reactions.The voltage drop occurring on the electrode is known as the over-voltage (overpotential)and reflects the fact that some energy is needed to generate a reaction product.The dependence has the simple form D V OV ¼
2:3RT a nF
ln i =i 0M ;T ;S ðÞÀ
Á,where R is the gas constant,T is temperature,a is a symmetry coefficient (usually around 0.5),n is the number of exchanged electrons,i 0is a exchange current which is dependent on the materials involved,temperature T and material active area S .The dependence on active area only springs from the fact that,while we can relate current from the cell to the projected membrane area,the exchange current is directly proportional to the chemically active area,which can be much bigger (or smaller,if all surface is not active)than this projected area since surface areas are nearly always bigger than projected areas.The chemical meaning of i 0can be understood from the fact that,even without any applied external current,the electrochemical reaction takes place with a certain probability,but the reaction products at the same time collapse back to form the initial reactants.We can say that the reaction does not have a direction.So i 0is a balance current of the fully reversible reaction.When we apply external current the balance is shifted and this is reflected by generating reaction products.Two different electrodes in a fuel cell with two different reactions on them lead to two different current/voltage dependences.Even with the most active catalysts known,the reaction on the anode,as a rule,is much faster than the reaction on the cathode and,as a result of this,we lose much more voltage at the cathode.The way to deal with this is to increase the chemically
Fig 2.Schematic dependence of output voltage and power density on electrical current density from a h
ydrogen fed PEM fuel cell.
V.M.Vishnyakov /Vacuum 80(2006)1053–1065
1055
active area at the cathode.This can be done either by increasing the catalyst active area or,if this is not possible, by increasing the amount of catalyst.The amount of catalyst is usually called‘‘catalyst loading’’and,in SI units, would be measured in mg per square metre,but is normally quoted in mg/cm2.
Some hydrogen in molecular form always passes through a PEM and ends at the cathode area.The catalyst on the cathode very effectively splits this hydrogen to protons. The protons then react with oxygen and this means that there is a small electrical load on the cathode side even without external current.Since the activity of the cathode is not particularly high,this leads to a noticeable over-voltage even without external current.Our open circuit voltage will be less than the ideal EMF by this over-voltage amount. The Gibbs free energy of formation and,as a result the EMF of the reaction,changes with the pressure,or,more accurately,with partial pressure and temperature of the reactants.In fact the dependence is a bit more complicated and involves also the amounts of active area on which the reaction takes place.In a certain situation it is usually found to depend on reactant activities.The depen
dence is known as the Nernst equation.In many cases it is simplified to the form of just pressure and temperature dependence.Without going into detail,the result to remember is that higher pressures and temperatures produce higher EMF values[1].
Partial pressures of reactants and their variations across the membrane during the functioning of the fuel cell have a serious implication on functionality.While hydrogen supply and reactions on the anode side can be treated relatively simply in most cases,the situation on the cathode is much more complicated and special measures should be taken to avoid oxygen starvation at high currents.Since reactants should be supplied to the electrodes and starvation is associated with the reactant usage,the whole problem may usually be described in mass transport or concentration losses terms.Externally to the fuel cell those losses can be associated with a certain output voltage drop. While it is difficult to describe all the mechanisms in detail and partial pressure variation on a point-to-point basis,it is possible to describe the external voltage drop by a simple empirical equation:
D V los¼k expðniÞ,
where k and n are some empirical constants and have typical values of around3Â10À5V and8Â10À3cm2mAÀ1,respectively.The equation wasfirst de-rived by Kim et al.[2]and describes voltage losses at high currents.Different constants apply to every specific cell construction.
In summary,one can say that the voltage produced by the cell can be described as
V cellðiÞ¼EÀD V rÀD V OVAnodeÀD V OVCathodeÀD V los. More information about modelling and‘‘real cell para-meters’’can be found for example in Refs.[3,4].3.Proton exchange membranes
Essentially,the function of the PEM should only be to conduct protons and separate catalysts.For thefirst activity the membrane needs good proton conductivity, for the second function mechanical strength is required. Unfortunately,the list of requirements for the PEM is much longer.The membrane,for instance,should conduct protons,but not electrons.It should be as thin as possible, so that proton current is affected as little as possible and the voltage drop across the membrane is minimized.The PEM should survive in a highly acidic environment at elevated temperatures for thousands of hours.It also should have a reasonably low permeability for the fuel. There are few commercially available polymers devel-oped by different companies which can satisfy those requirements.For temperatures below1001C Nafion s developed by the Du Pont company is most probably widely known and used.Over the last40years Nafion s properties have become a benchmark for comparison with other materials.On the molecular level Nafion is a polytetrafluorothylene(widely known as PTFE or Te-flon s)with a long side chain ended by a sulphonic acid radical SO3ÀH+.Being highly polarised,the radical is bonded to the end of the side chain by an ionic force.The radicals are hig
hly hydrophobic while PTFE itself is highly hydrophilic.As a result,the material becomes like a sponge for water and can absorb it in high quantities.Owing to undulation of hydrophilicity within the material,the water tends to be retained in clusters.In some proton-conductiv-ity models it is believed that water clusters form a nearly continuous pathways for protons.From this simple model one can conclude that proton conductivity is proportional to water content.Highest conductivities of protons are only achieved for fully hydrated membranes.The best conductivities achieved for Nafion s usually are in the region of0.001S/m.Simple calculation shows that at 500mA/cm2and a membrane thickness of50m m the voltage drop across the membrane would be25mV. Dehydrated membrane conductivity can drop by more than two-orders of magnitude and this would make the voltage drop across membrane a limiting factor for the cell performance.On the other hand,if too much water is present in the backplate channels,condensation occurs and electrode areas becomeflooded.Accessibility of reactants to catalyst layers then diminishes and a situation of reactant starvation takes place with a significant drop in output power[5].In many cases the hydrophobicity of the membrane surface plays a decisive role in the water balance[6].
Hydration of the membrane is not helped by the process of the electro-osmotic drag,when up tofive molecules of water are transferred from an anode to a cathode with each passing hydrogen atom[7].Th
e anode side of the membrane can quickly become dehydrated and needs a source of water to function properly.Some water, produced on the cathode side,diffuses back and special
V.M.Vishnyakov/Vacuum80(2006)1053–1065 1056
measures can be taken to assist this diffusion[8].However, for high performance,external hydration of both sides is still necessary[9].
Hydration of a membrane is accompanied by membrane size changes.Fuel cells usually work under external mechanical compression in order to optimise the contact area,reduce contact resistance between materials at the electrodes and also to optimise gasflow and heat removal [10].When the membrane changes size this readjusts overall compression in the cell.Diffusion layers or back-plates can accommodate part of this movement,but the new set of working parameters does not guarantee optimum performance and constant readjustments lead to a drop-off of system performance over time.This situation is not helped by the fact that it is almost impossible to keep the same level of hydration at all membrane surface points during operation under dynamic load[11].Cell design for undulated high-power applications is a very challenging task.
Alternative membranes to Nafion have been developed. Perfluorinated ionomer(named Hyflon),with a
side chain shorter than in Nafion,has been investigated for a considerable time and shows very promising properties especially for operations above373K.The reader is referred to one of the latest reports on this material by Ghielmi et al.[12].Sulfonated aromatic polymers(for example sulfonated polyetherketone[SPEK]and sulfo-nated polyetherketone[SPEEK])have been widely inves-tigated,but they also need high levels of hydration and,at the membrane composition optimised for conductivity, may suffer from low mechanical stability[13,14].
PEM fuel cells usually function at temperatures below 360K.Many fuel cell parameters would be more optimised if the PEM can be made to function at higher temperatures. For instance,heat dissipation from the fuel cell becomes a smaller problem for high temperatures as the cooling system grows to be more weight and power efficient. Poisoning of contemporary catalysts by CO can be almost fully avoided at temperatures somewhere above430K as it is well known,that,if catalyst can tolerate10–20ppm levels of CO at350K,then it can tolerate a few1000ppm at 430K.Attempts to significantly raise the temperature above430K are limited by the stability of the catalyst and its support,which start to degrade rapidly at the higher temperatures.
Another barrier to a temperature rise is a‘‘hydration price’’.Unfortunately,the complexity and associated costs of appropriate hydration levels for standard membranes are very high at temperatures above373
K.There is also the problem of the Nafion glass transition[15].All this indicates that new PEM materials should be developed to work at lower levels of hydration,or even,without it at elevated temperatures[16].Encouraging results are achieved by impregnating Nafion with mineral materials, for instance silicon oxide[17].One way to achieve this is to use proton conductors in the PEM which do not require water.Some progress has been seen during the use of polymers doped with imidazole or N-methyl imidazole(see for example Refs.[18,19].Unfortunately,the proton conductivity of the membranes was still not as high as for fully hydrated Nafion s and there is a loss of dopant during operation.
Polybenzimidazone(PBI),widely used for durability in many applications,has a high glass transition temperature at above480K and can be doped with strong acids (phosphoric or sulphuric)to provide proton conductivity [20].The PBI derivative membranes allow humidification-free operation at temperatures reaching470K[21].High proton conductivity for PBI derivatives of almost 5Â10À4S/m has been reported[22],but practical imple-mentation for these materials has been slow,probably owing to the gradual loss of doping acids during operation. Complex mixtures of basic and acidic polymers have been shown to exhibit very promising properties in the mid-range of operational temperatures[23].
Solid acid membranes are in the development stage and have some promising properties though much work is needed before they can be used in practice[24].Membranes containing fullerenes and ca
rbon nanotubes are but another example[25–27].More details can be found in recent reviews on polymer membranes by Savadogo[28] and on polymer composite PEMs[29,30].
Despite these developments Nafion still retains its lead position.Water management in fuel cell is still very important and the current way forward is to use self-humidification.In this case much of the water management equipment would become redundant and the Nafion cells could be made much lighter if this can be made to work with high efficiency.It would also make the cell self-regulating and has been shown to significantly prolong a cell lifespan.Two approaches are possible for this purpose: to promote water diffusion back from cathode to anode and to produce water inside the membrane.Thefirst approach will be discussed later in the paper.The second approach is implemented by impregnating the membrane with a catalyst to promote oxygen diffusion into the membrane.Both hydrogen and oxygen cross-over are then bound to interact with each other and generate water inside the membrane.This wasfirst demonstrated by Watanabe et al.[31]and it is now under further development[32,33].
4.Catalysts
The development of catalyst layers for the PEM has a long history.In order to appreciate the amount of
work done and achievements made it is advisable to look at one of thefirst reviews on fuel cells[34].One can measure the progress from the simple fact that,during the last40years, the catalyst load(amount of catalyst on the membrane)has dropped from30–40m g/cm2to well below1m g/cm2whilst performance has increased a few times.
As mentioned earlier the role of the two catalysts is to split the hydrogen molecule on the anode side and combine proton with oxygen on the cathode side.Let usfirst
V.M.Vishnyakov/Vacuum80(2006)1053–10651057
concentrate on the anode side.As was mentioned earlier, under the given conditions there is a certain concentration of protons which is produced over each unit area of a catalyst.Some protons will recombine back to form a hydrogen molecule and some will diffuse into the available proton-conducting volume.In the situation,when there is a net drain of protons some equilibrium would be reached defined by proton production,recombination,diffusion and drainage.In the case of a fuel cell the PEM plays the role of the drain.Protons pass through the membrane and play their part in electricity and water production.To increase the overall reaction rate one needs to make sure that the maximum number of protons is produced by the catalyst,diffusion is as fast as possible,or,the diffusion
path to the drain is as short as possible.The number of protons produced is defined by the catalyst activity and the amount of active catalyst surface area(related to catalyst load).The desire to reduce the weight of the catalyst leads to the development of materials with the highest possible active surface area.Long searches for the best catalysts have produced very limited results so far.Platinum in pure form or alloyed with other elements is regarded as the only choice so far for high and stable performance[35]. Earlier pure metals and alloys(so named unsupported metal blacks)have been used as catalysts[34],but later it was discovered that catalyst particles could be attached to a finely dispersed carbon(the system is usually denoted as metal on carbon,Me/C).This not only increases the catalyst surface area,since small catalyst particles are separated,but also provides good electrical contact between the catalyst and the diffusion layer and,ultimately,the current collector. It is almost possible to spread the catalyst as a mono-atomic layer over the carbon particle in a quest for the highest catalyst area.However,it seems that there is a catalyst minimum thickness limit at around few nanometres below which there is no gain in catalyst activity[36,37].This is thought to be related to some surface reconstruction which is affected by the thickness.The reconstructed surface significantly affects catalyst activity.
It also has been shown that only catalyst particles connected to the PEM surface contribute to the pow
er production.This has led to a development where a catalyst layer is usually impregnated with an ionomer(for example perfluorosulfonic acid)in order to create three phase contact(gas/catalyst/membrane)for each particle in the catalyst layer.This arrangement provides easy enough gas access to the catalyst and a readily available diffusion path for a proton.The importance of this short diffusion path manifests itself in the fact that the catalyst layer thickness, from an overall performance point of view,has an optimum value at below100m m.A thicker layer,produced by loading more catalyst,does not improve performance significantly.Methods for catalyst loading have been recently reviewed by Lister[38].
Operation of a fuel cell in a real-life situation means that the purity of the gases cannot be maintained at very high levels at a reasonable cost.Some impurities would come from the air intake and some would come with the hydrogen.It seems that most concerns are centred on carbon monoxide and carbon dioxide.Both gases are produced during hydrogen extraction from fossil fuels.It is possible to reduce their concentration in the hydrogen stream to a very low level,but this usually means additional costs.The acceptable concentration of carbon monoxide in reformatted hydrogen is usually in the region of10–100ppm,whereas the concentration of carbon dioxide can often be as high as40%.This does not mean, that these10–100ppm levels do not affect the performance at all,but it is an acceptable
compromise.The overall strategy in this case is either to produce catalysts with high impurity tolerance levels or to develop techniques for catalyst activity restoration.
The poisonous influence of carbon monoxide on platinum catalyst layers is well [39–42]). It results from the attachment of a CO molecule onto the catalyst surface thereby reducing the surface available for hydrogen reactions.Linear or bridge-bonded CO species are formed on the catalyst surface[39,42,43].Externally, this leads to a significant reduction of cell voltage under load.At current densities around0.5A cmÀ2the voltage can be reduced by as much as0.5V for a pure Pt catalyst. Losses arising from carbon dioxide presence are smaller but still can be significant depending on the load and feed gas composition[44].To reduce poisoning and the associated losses,Pt is usually combined with ruthenium in different proportions approaching a50/50at%value. Much work has been done on alloying Pt with other metals and oxides to increase poisoning resistance.However,only molybdenum and,probably,gold[40]seems to work well during long term trials and the performance of PtMo/C (PtMo on carbon)catalysts can be comparable or better than PtRu/C.Some reported work and critical reviews on this topic have been published recently by Ralph[45–47] and Urian et al.[48].
Another way to approach the poisoning issue is to clean the catalyst surface during fuel cell operation.It is possible to add some oxidant(such as air,oxygen or hydrogen peroxide)into the hydroge
n stream[45–47]in order to oxidise the carbon monoxide to dioxide which is bonded much more weakly to the surface.The increase of temperature on the catalyst surface and accelerated membrane failure are drawbacks of this method if moderation is not observed.On the other hand,it is possible to pulse periodically the load current[49],which reduces the potential on the cell to a value low enough to promote electro-oxidation of CO on the surface.In fact,if the cell is kept at constant current in certain conditions it would start oscillations by itself[50].The problem arises in accommodation of both of those techniques into a fuel cell stack.It is most likely that some blend of using a catalyst with high resistance to poisoning and a dynamic cleaning technique could be used in future cells.
Catalyst activity on the cathode is a few times lower than on the anode.This usually means that higher catalyst loads
V.M.Vishnyakov/Vacuum80(2006)1053–1065 1058
版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。
发表评论