DOI: 10.1002/cssc.

200900190

Functional Materials for Sustainable Energy Technologies:

Four Case Studies
V. L. Kuznetsov and P. P. Edwards*[a]

44  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 44 – 58
The critical topic of energy and the environment has rarely mainly due to the materials’ intrinsic performance. We high-
had such a high profile, nor have the associated materials chal- light four areas of activity where functional materials are al-
lenges been more exciting. The subject of functional materials ready a significant element of world-wide research efforts.
for sustainable energy technologies is demanding and recog- These four areas are transparent conducting oxides, solar
nized as a top priority in providing many of the key underpin- energy materials for converting solar radiation into electricity
ning technological solutions for a sustainable energy future. and chemical fuels, materials for thermoelectric energy conver-
Energy generation, consumption, storage, and supply security sion, and hydrogen storage materials. We outline recent ad-
will continue to be major drivers for this subject. There exists, vances in the development of these classes of energy materi-
in particular, an urgent need for new functional materials for als, major factors limiting their intrinsic functional performance,
next-generation energy conversion and storage systems. Many and potential ways to overcome these limitations.
limitations on the performances and costs of these systems are

1. Introduction

It can be argued that the single most difficult scientific and tems and technologies. There continues to be a need for such
technological challenge facing humanity in this century is our incremental or evolutionary materials development. Set against
global energy future.[1, 2] Against such a backdrop of escalating this background one can also argue for an urgency in acceler-
concerns about our environment, our finite natural resources, ating fundamental research to identify the step-change, revolu-
and issues of energy storage and supply security, there is an in- tionary and disruptive research and technology that will pro-
creasing awareness of the importance of high-performance vide us with the new generation of energy materials to enable
functional materials for sustainable energy production, distri- ultrahigh efficiency, operation, and performance. Here, invest-
bution, and storage. It is recognized without doubt that re- ment needs to be made in high-risk, high-return projects. One
search and innovation in materials and materials’ technologies spectacular development, for example, could be the search for
will play a critical role in meeting these energy challenges for and discovery of a room-temperature superconducting materi-
the 21st century, played out within a global context. al with exceptionally high current-carrying ability. In both
In order to meet our (ever-)increasing energy demand, cases—incremental materials development and transformative
linked inextricably to our very human development, whilst materials discovery—such developments can lead to a major
minimizing both environmental impacts and supply security impact on reducing greenhouse gas emissions and, hence, be
risks all countries will require a balanced energy portfolio, in- a prominent part of a long-term solution towards sustainable
cluding a mix of fossil, nuclear, and renewable energy sources energy.
(Figure 1). Within all these technologies, the needs of and Materials are therefore set to play a central role in continu-
ing to improve the performance of what one might term “con-
ventional” energy systems, but also ensuring—through high
performance and low costs—the viability and uptake of new,
emerging sustainable energy technologies, such as photovolta-
ics, batteries, fuel cells, and hydrogen use. In these areas major
new discoveries are needed, and these will largely come from
basic research programmes that target an understanding—and
with that the anticipated enhancement—of the intrinsic per-
formance of energy materials. From such a vantage point, new
sustainable energy technologies will undoubtedly be forth-
coming.
The aim of this short Review, therefore, is to illustrate, with
Figure 1. The primary energy supply structure and a schematic of the cur- specific and rather detailed examples, how one can at least
rent and future positions of the energy mix. begin the process of gaining an improved understanding of
the fundamental factors limiting the performance of various
demand for future energy materials are critical. Whilst specific functional materials for specific sustainable energy technolo-
solutions may vary across the areas shown in Figure 1, there is gies. We have as our focus four important classes of functional
undoubtedly an overarching need for major improvements in energy materials: 1) transparent conducting oxides 2) solar
the functional performance of materials (be it thermal, electri-
cal, optical, or, indeed, “chemical” performance) supplemented [a] Dr. V. L. Kuznetsov, Prof. P. P. Edwards
of course by improvements in durability and reliability (and Inorganic Chemistry Laboratory, Department of Chemistry
University of Oxford
costs). Short-term “materials solutions” already under develop-
South Parks Road, Oxford, OX1 3QR (UK)
ment are generally based on current, known technologies, and Fax: (+ 44) 1865-272-656
involve incremental developments of existing materials’ sys- E-mail: [email protected]

ChemSusChem 2010, 3, 44 – 58  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 45
 V. L. Kuznetsov and P. P. Edwards

energy materials, 3) thermoelectric materials, and 4) hydrogen

storage materials. With this type of benchmarking approach,
one can hopefully see the emergence of a naturally multidisci-
plinary activity of energy materials, and of course, there are
other significant areas of research activity and development
not covered within this Review; for example, batteries, nuclear
materials, and others. Our view is that many of the issues re-
stricting advances across all areas center on the intrinsic per-
formance of functional energy materials, and our four case
studies will provide some guidance into how materials chemis-
try, materials physics, and materials science together can help
with such an understanding.

2. Transparent Conducting Oxides

Electrical conduction in an optically transparent solid is the Figure 2. Schematic energy-band model for SnO2-doped In2O3x for small x
rarest form of conductivity.[3] The seemingly contradictory (insulating) and large x (metallic) materials. Taken from Edwards et al.[9] as
properties of close-to-“metallic” conductivity in a material si- modified from the work of Fan and Goodenough.[7] Reproduced with per-
mission. Copyright The Royal Society of Chemistry.
multaneously exhibiting almost complete “nonmetallic” or in-
sulating optical transparency form the basis of numerous criti-
cal applications in contemporary and emerging energy tech- proximity of such impurity states to the host conduction band
nologies, such as photovoltaic devices, energy-efficient win- ensures facile thermal ionization into the band, developing, ul-
dows, flat-panel displays, light-emitting diodes, and others.[4–6]
timately, a degenerate, itinerant electron gas of current-curry-
In many respects the basic underpinning materials physics ing electrons which also gives rise to far-infrared (Drude-like)
can be reasonably well set out.[7–10] Thus, the high conductivity
absorption and high electrical conductivity, but at the same
of the prototypical transparent conducting oxide (TCO) SnO2- time the fundamental host bandgap is left intact, that is, the
doped In2O3 (ITO) derives from the presence of shallow donor
electrically conductive material remains optically transparent in
or impurity states located close to the conduction band of the the visible region. For doping levels above that set by the
host (In2O3), the donor states produced via chemical substitu- Mott criterion for metallization (i.e., n > nc ~ 1018–1019 cm3 for
tion of Sn4 + for In3 + or by the presence of oxygen vacancy im-
ITO),[11] one can consider ITO to have a full valence band and a
purities states[7] in In2O3x (Figure 2). At room temperature the host 5s conduction band partially filled by a degenerate free-
electron gas. Within the free-electron framework the optical
Dr. Vladimir Kuznetsov received his
and transparent properties are described well by a simple
Ph.D. in Chemistry from Moscow State
Drude model (the addition of a strongly frequency-dependent
University (Russia). He is a Senior Re-
electron scattering time t for Fermi surface electrons provides
search Fellow at the Inorganic Chemistry
a secondary level of sophistication; see below).[9] Conduction
Laboratory, University of Oxford (UK).
electron scattering in ITO thin films occurs from a number of
His main research interests lie in the
different scattering centers such as impurities, phonons, and
synthesis, processing, and characteriza-
defects. However, the most dominant process originates from
tion of novel functional materials for
the presence of the very impurity ions responsible for the
electronic and energy conversion appli-
doping, and this sets the scope for any interpretation of the
cations including thermoelectric materi-
limiting performance of such materials.[8]
als, transparent conductors, and elec-
A calculation of electrical resistivity due to this ionized im-
tronic materials.
purity scattering provides an important lower limit to the at-
tainable intrinsic resistivity (equivalently an upper limit to the
Prof. Peter Edwards is Head of Inorganic conductivity) of a TCO material;[8] this being a key performance
Chemistry at the University of Oxford indicator. The results of such calculations are shown as a func-
(UK), and leads a major solid-state tion of carrier concentration in Figure 3. The various data
chemistry group within the department. points are experimental values taken from the literature for
His very broad research interests encom- In2O3, ZnO, and SnO2 doped systems, and the solid line repre-
pass the electronic structure of solutions sents a calculation of electrical conductivity due to ionized im-
of alkali metals in non-aqueous solvents, purity scattering.[8] Importantly, this sets the upper perfor-
high-temperature superconductors, mance limit to the attainable conductivity at any particular car-
metal nanoparticles, materials for hydro- rier concentration.
gen storage, and transparent conduct- The similar behavior observed in all three different systems
ing oxides. strongly supports the idea that this impurity scattering mecha-

46 www.chemsuschem.org  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 44 – 58
Functional Materials for Sustainable Energy Technologies

tron mobility, me, recalling that me = et/m*). This means that for
a film to be nonreflective to light of free space l0 (i.e., wp <
2pe/l0); where l0 is the wavelength of light) the electron den-
sity n must satisfy:

4p2 e0 m
n< ð2Þ
m0 e2 l20

which for ITO yields the criterion n (cm3) < 1.6  1021/l0.
Therefore, for efficient transmission of the whole visible
spectrum (including red light of free-space wavelengths up to
780 nm) the electron density should not exceed 2.6 
1021 cm3.
Figure 3. Literature data for the three oxide systems In2O3, ZnO, and SnO2 To allow a suitable “safety margin” in our performance crite-
chosen as displaying high electrical conductivity at any particular carrier ria, we consider a plasma wavelength of just over 1 mm (i.e., in
concentration. The theory plot (solid line) is from results of the calculation
of conductivity (resistivity) due to ionized impurity scattering. (Modified the near-infrared) in which case the tolerable free electron
from Bellingham et al.[8]) density (nmax) is ca. 1.5  1021 cm3. Clearly, values of n > nmax are
severely detrimental to the performance transparency of the
films owing to the plasma edge creeping into the red part of
nism dominates the electrical resistivity (conductivity) in all
the visible spectrum. The power reflection coefficient R at an
three cases.
interface between a thick ITO layer and free space can be cal-
In an attempt to enhance the electrical performance by
culated from:
moving to higher conductivities, the films obviously need to
be doped to increase the carrier concentration n (Figure 3). pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2
 eðwÞ  1
However, increases in n will ultimately lead to a degeneration  
RðwÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3Þ
of the optical performance of the material. Specifically, al-  eðwÞ þ 1
though low resistivity (high conductivity) is highly desirable for
the performance of the materials and associated devices, the
free electron density in ITO cannot be increased beyond 2  An exact calculation of R using the full Drude formula for
1021 cm3 without pulling the plasma frequency into the red e(w) is shown in Figure 4 as a function of n for various incident
end of the visible spectrum—making it highly reflecting.[9, 10]
Furthermore, any increases in the film thickness cause re-
duced transparency due to a finite skin depth, d, emerging as
the electrical conductivity increases with doping.[10] However,
above the plasma frequency for n  1021 cm3, we find that
those increases in the electron mobility, me, cause increases in
d, and (approximately) d / me. This leads to the important con-
clusion that if me in TCOs can be increased/maintained above
the present state-of-the-art values of around 50 cm2 V1 s1 to
values of around 100 cm2 V1 s1, very substantial increases in
transparency are possible without sacrificing high electrical
conductivity. The clear message is that increasing the electron
mobility will have a significant impact on the key performance
characteristics of the power efficiency of light-emitting diodes Figure 4. The power reflection coefficient at the interface between a thick
TCO layer and free space as a function of free electron density n for various
and related devices.
incident wavelengths in the visible spectrum. (Modified from Edwards
The materials physics of the problem can be captured in the et al.[9])
following manner:[9, 10] for ITO in the limit of high electron den-
sity (i.e., n 0 1021 cm3) and high mobility (i.e., me = etm* 0
20 cm2 V1 s1) the plasma frequency, wp, can be written as: wavelengths in the visible spectrum.[9, 10] For frequencies below
the plasma frequency R is close to 1 (i.e., it is very reflective)
 1= with a small amount of electromagnetic absorption within the
ne2 2
wp ffi ð1Þ
e 1 e 0 m* material (approximately equal to (1R)). These results have im-
portant consequences if high-transparency films with very low
sheet resistance are required. This is best achieved by increas-
where e1 is the high-frequency dielectric constant and e0 is ing the electron mobility in preference to n, because increasing
pffiffiffi
the vacuum permittivity. Hence, in this limit wp / n and thus n leads to increasing conductivity and a proportional reduction
is determined almost exclusively by n and not t (i.e., the elec- in the skin depth as the plasma frequency is increased. This

ChemSusChem 2010, 3, 44 – 58  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 47
 V. L. Kuznetsov and P. P. Edwards

significant result illustrates the origin of me, so as to maximize The modern era of photovoltaic solar cells started in 1954
this quantity. Importantly, the intrinsic limit of me appears to be when scientists from Bell Laboratory developed and tested a
set around 90 cm2 V1 s1 as a result of ionized impurity scatter- single-crystalline silicon solar cell.[15] A solar cell is a semicon-
ing.[8] ductor device that converts light into electric energy. The basic
In summary, therefore, the task in any attempt to enhance steps in photovoltaic energy conversion are light absorption
the intrinsic performance of TCOs is centered around the fol- and charge-carrier (electrons and holes) generation, transport,
lowing three performance criteria for new materials: and injection/extraction into an external circuit. In convention-
al photovoltaic (PV) cells electron–hole pairs are formed in a
1) Maintain the condition of a wide electronic bandgap (>
semiconducting material by absorption of photons with ener-
3.5 eV) and low intergap absorption.
gies higher than the semiconductor bandgap. These light-gen-
2) Maintain the condition of moderate electron densities (n erated charge carriers are then separated by a potential differ-
 1021 cm3)—but certainly not higher than this value. ence across a p–n junction, leading to a flow of electric current
through an external circuit.
3) Maintain the highest possible carrier mobility; ideally close
The solar cell efficiency is defined as the maximum electrical
to or above 100 cm2 V1 s1.
power produced divided by the total energy of incident radia-
Thus, materials chemists must now be closely allied with mate- tion. There are several factors that limit the efficiency of stan-
rials physicists to understand how electron mobilities can be dard single-junction cells. One of the most fundamental limita-
understood, tuned, and thereby enhanced. tions on solar cell efficiency is related to the single bandgap
It appears that such performance limits are now regularly value of a semiconductor, which results in the (efficient) utiliza-
being approached in known materials, and this fact[8] sets the tion of only a limited portion of the solar spectrum. Light with
scene for major activities in the materials chemistry if new sys- an energy smaller than the semiconductor bandgap cannot
tems are to be discovered that optimize/enhance electronic generate electron–hole pairs and is therefore wasted as heat.
properties and materials performance. On the other hand, if the photon energy exceeds the bandgap
value any energy excess is also lost to heat because the photo-
excited electron–hole pairs quickly lose any energy they might
3. Solar Energy Conversion
have in excess of the semiconductor bandgap. These two loss
Among various renewable energy sources, solar energy is by processes limit the conversion efficiency of a PV cell to around
far the largest resource and the only one that has sufficient ca- 44 %. Shockley and Queisser showed[16] that if other important
pacity to fully meet the future global energy demand without loss processes are also taken into account, such as the grain
potentially destructive environmental consequences. A square boundary recombination of photoexcited electron–hole pairs,
meter of the Earth’s surface perpendicular to the sunlight from resistive losses, reflection from the surface, and some other
the zenith receives about 1 kW of solar energy. The sun pro- factors, the maximum theoretical efficiency a solar cell made
vides more energy to the Earth’s surface in one hour than the from a single material can achieve in full unconcentrated sun-
whole world’s present energy consumption in an entire year.[12] light is only 32 %.
Put another way, 20 TW of electrical power (nearly twice the To achieve the maximum conversion efficiency, a semicon-
current global production) can be generated by covering just ductor material should have a bandgap value of around 1.3 eV,
0.16 % of the Earth’s surface with 10 % efficient solar panels.[13] though the efficiency varies little over the range of 1.1–1.7 eV
There are three types of solar energy conversion systems, (Figure 5).[17] Many materials exhibiting bandgap values in this
depending on the primary energy product: solar electricity, range are very attractive for PV applications. The present effi-
solar fuels, and solar thermal systems. To become a significant ciencies of laboratory examples of silicon solar cells approach
contributor to the primary energy mix, solar energy faces two the theoretical limit of 32 % and reach 25 % for single-crystal Si
major technological challenges: first, cost-effective capture and and epitaxial GaAs photovoltaic cells (Figure 5). The single-
conversion of sunlight into useful energy, and second, storage junction crystalline and polycrystalline silicon solar cells repre-
and distribution of this energy to end users in an economical sent the first generation of solar cell technology, which still
and convenient form. While solar electricity could potentially dominates the commercial PV market today.
play a significant role among other electricity-generating tech- Cost reduction and improved efficiency are two major chal-
nologies, arguably the most attractive method for solar energy lenges facing the PV market. While solar cell production has
conversion and storage is in the form of chemical bonds by been growing by 30 %–40 % annually over the last decade,
direct production of cheap solar fuels. This would help to over- making it the world’s fastest-growing energy technology, the
come the intermittency of terrestrial solar radiation and also cost of photovoltaic electricity is still up to 10 times higher
enable the use of solar energy in the transportation sector. In than that of grid power. The materials costs dominate in solar
the present section the major scientific challenges in the devel- energy conversion applications because the installed PV power
opment of materials for photovoltaic electricity generation and capacity scales linearly with the area of solar cell modules. In
synthesis of fuels using solar energy are outlined. 2006 around one-third of the world’s electronic-grade silicon
The idea of capturing solar energy and turning it into elec- was used for the production of 0.2–0.3 mm thick single-crystal-
tricity is almost as old as the discovery of the photoelectric line solar cell wafers. A potential cost reduction as a result of
effect in 1839, by the French scientist Edmond Becquerel.[14] mass production of solar cells is currently limited by a high

48 www.chemsuschem.org  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 44 – 58
Functional Materials for Sustainable Energy Technologies

Thin-film based solar cells are considered as Generation II PV

technology and have the inherent advantage of easy module
assembly processes and drastically reduced material consump-
tion and manufacturing costs. The second-generation PV tech-
nology includes solar cells based on inorganic thin films such
as amorphous and polycrystalline Si, polycrystalline cadmium
telluride (CdTe), copper-indium-selenide (CuInSe2), and copper–
indium–gallium–selenium (Cu(In,Ga)Se2) solid solutions. These
thin films can be deposited by a variety of high-performance
deposition techniques using inexpensive large-area substrates
such as glass, stainless steel foil, or polymers.
Amorphous hydrogenated silicon thin films, first developed
in the early 1970s, have an advantage over bulk silicon be-
cause of a larger bandgap value, which can be tuned in the
range of 1.1 to 1.7 eV by varying the hydrogen content.[21] This
Figure 5. Solar cell efficiencies achieved for several PV materials and thermo- leads to a more effective absorption of solar radiation com-
dynamic limit of efficiency for single-junction solar cells. (Modified from Kaz- pared to bulk silicon, which exhibits a low absorption coeffi-
merski.[18]) cient due to its indirect bandgap of 1.1 eV. However, even for
amorphous silicon with a bandgap value of 1.7 eV, a film of up
and rapidly rising cost of sufficiently pure silicon, which has to 20 mm thick is required for effective light absorption. In con-
risen from US$ 9 kg1 in 2000 to US$ 60 kg1 in 2005.[19] On the trast, direct-gap semiconductors such as Cu(In,Ga)Se2, CdTe,
other hand, the efficiency of silicon single-crystal solar cells has GaAs, and InP have a very high optical absorption at energies
been optimized over many years and, hence, is unlikely to im- just above the bandgap; this resulted in maximum conversion
prove substantially in the future. efficiencies of 15–18 % for the best 2–3 mm thick films deposit-
Figure 6 displays the efficiency and installed cost of several ed on glass substrates.
generations of PV modules. The area labeled I refers to the The conversion efficiency of thin-film solar cells typically
first generation of solar cells based on crystalline Si wafers. The varies between 7 and 12 %, which is significantly below values
attained by corresponding single crystals and epitaxial thin
films. The main reason for lower efficiency is a fast recombina-
tion of the photogenerated minority carriers on grain bounda-
ries of polycrystalline thin films. Amorphous Si solar cells dem-
onstrate even lower conversion efficiencies in the range of 5–
7 % due to the inherent disorder and the large number of dan-
gling bonds acting as efficient carrier recombination centers. In
addition, amorphous silicon solar cells are also subject to a
performance degradation of up to 20–30 % after 1000 h of op-
eration under sunlight.[22] Nevertheless, despite these disadvan-
tages the high cost of In, Te, and Se and high toxicity of Cd
and As make amorphous silicon solar cells the preferred thin-
film technology, with a much larger market share than that of
chalcogenide and III–V semiconductor thin films.
Although thin-film solar cells are the most likely candidates
for low-cost PV power generation systems, none of them have
Figure 6. PV electricity cost (US$ W1) as a function of module efficiency and so far had a real breakthrough in performance/cost ratio,
installed cost (per 1 m2 of PV modules) for three generations of solar cell
technology. (Modified from Green.[20])
needed to become cost-competitive with other energy genera-
tion technologies. A significant cost reduction in comparison
to single-crystalline silicon solar cells is offset by a lower effi-
ciency of thin-film solar cells; as a result their cost/efficiency
cost of generated electricity is represented on the graph by ratio has improved only slightly compared to the first-genera-
dashed lines, assuming the maximum solar energy of 1 kW m2 tion PVs (Figure 6). The grand challenge is to achieve dramatic
at the Earth surface. For the first generation of solar cells the improvements in the economics of solar energy through the
electricity cost varies from US$ 2.0 W1 to US$ 3.5 W1 development of inherently high-efficiency and low-cost con-
(Figure 6). To become cost-competitive with other electricity- version concepts.
generating technologies the PV electricity cost should be re- The vast majority (> 99 %) of first- and second-generation
duced to around US$ 0.4 W1. This can only be achieved by terrestrial PV devices are based on single-junction configura-
new approaches with dramatically reduced PV module costs tions, which have a Shockley–Queisser conversion efficiency
and improved efficiencies. limit of 32 %.[16] A critical feature of rapidly developing Genera-

ChemSusChem 2010, 3, 44 – 58  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 49
 V. L. Kuznetsov and P. P. Edwards

tion III PV systems (Figure 6) is that their module efficiencies tween two conducting plates. Unlike conventional solar cells,
can potentially exceed the Shockley–Queisser limit by employ- in dye-sensitized solar cells the light absorption and charge
ing a multijunction design or exploiting novel physical phe- separation processes occur in separate molecular layers. When
nomena to separate the photogenerated charge carriers.[23] excited by light, the dye becomes oxidized and injects elec-
Several approaches exist to increase the efficiency of PV devi- trons into the TiO2 conduction band; the dye returns to its
ces to above 32 %, including multijunction solar cells, multiple original state by electron transfer from the electrolyte. This
energy level solar cells and multiple exciton generation solar charge separation through an insulating dye monolayer helps
cells. to avoid the charge-carrier recombination encountered with
Solar energy can be utilized much more efficiently if the en- silicon cells, simplifies cell design, and reduces the need for ul-
ergetic electrons and holes created by the high-energy pho- trahigh-purity semiconductors.
tons do not lose their excess energy to heat through excitation The energy generated by the Grtzel cells depends on the
of the lattice vibrations but instead contribute to electrical difference between the oxidation/reduction potentials of the
power output. This can be achieved in multijunction solar cells electrolyte and the dye/nanocrystal interface. The power con-
comprising several thin-film layers of semiconductor materials version efficiency of Grtzel cells is currently exceeding
with decreasing energy bandgaps. Ideally, such a structure 11 %.[29, 30] Coupled with an excellent output stability under op-
would have a continuously adjustable bandgap of 0–3.5 eV to erating conditions for more than 10 000 h,[28] the performance
cover the entire solar spectrum,[17] but in practice two or three of dye-sensitized cells rivals conventional devices and even ex-
thin layers of materials with different bandgap values are used. ceeds them in performance/cost ratio. Future improvements in
In these so-called tandem cells the first wide-bandgap material performance can be achieved through the development of
absorbs the high-energy photons but transmits photons with new dyes with improved optical absorption properties and
lower energies, which then get absorbed in the lower layers, photocurrent density and further optimization of the structure
just collectively utilizing the entire solar spectrum (Figure 6). and electrical properties of the nanoporous TiO2 layer for the
Due to the absorption of a broader portion of the solar spec- fabrication of large-surface-area PV modules.
trum and minimization of thermal relaxation losses, the ther- While dye-sensitized solar cells can be considered as organ-
modynamic efficiency limit of such a tandem cell can reach ic/inorganic hybrid cells, other types of purely organic solar
66 % for single-sun radiation with the thermodynamic limit cells based on bulk donor–acceptor heterojunctions have been
being as high as 85 %.[20, 24] At the moment tandem solar cells developed. The active layer of these cells consists of an inter-
based on III–V compound semiconductors hold the efficiency penetrating network of p- and n-type organic semiconductors,
record of 42.8 %.[25] However, the cost of these devices is very sandwiched between two electrodes with different work func-
high, limiting their use almost exclusively to space applications tions. Light absorption does not lead immediately to free
and terrestrial solar concentrators. charge carriers, but results in the creation of excitons (elec-
A different approach to efficiently absorb photons over a tron–hole pairs), which carry energy but have no net electrical
wide energy range is used in multiple-energy-level solar cells. charge. When these excitons diffuse to the bulk-distributed p–
In this type of solar cell additional levels are introduced in a n heterojunction region of the cell, they split and produce the
single-gap semiconductor by creating a quantum well struc- required free electrons and holes.
ture, leading to similar light absorption characteristics to those Organic solar cells have recently generated much excitement
exhibiting by multijunction solar cells. Another promising type owing to their potential for the construction of low-cost, light-
of third-generation solar cells are multiple-exciton-generation weight, large-area, flexible solar cells.[31] They can be processed
solar cells, which enable the conversion of a single high- from solutions at or near room temperature on flexible sub-
energy photon to multiple electron–hole pairs by employing strates using simple and inexpensive deposition methods such
semiconductor superlattices or quantum dot semiconductor as spin coating or printing, thereby yielding very cheap devi-
structures.[26, 27] This approach could overcome the central limi- ces. The current version of organic solar cells suffers from low
tation of existing solar cells exhibiting a one-to-one relation- efficiencies of less than 2 %. Calculations show that an efficien-
ship between absorbed photons and electron–hole pairs. cy of 6 % should be possible with the current state of technol-
While the basic physical phenomena for these novel types ogy, and the potential for reaching more than 10 % efficiency
of third-generation solar cells have been demonstrated, signifi- exists if the electronic properties of materials and cell architec-
cant additional challenges remain in developing materials and ture are optimized further.[31, 32] Even with the present low con-
structures to utilize these concepts. The implementation of version efficiency and poor durability of organic solar cells, the
these innovative solutions can lead to non-incremental step energy payback time is expected to be very short; this is the
changes in efficiency for solar electricity production. main driving force for further development. Organic solar cells
One of the key issues in PV solar systems is how to separate are in a relatively early stage of development and the main re-
charge carriers efficiently without using expensive, highly pure search task is to develop sufficient understanding of the phys-
semiconductors. A promising new technology in solar energy ics and chemistry of organic PV materials and systems to en-
conversion is the dye-sensitized cell, or Grtzel cell.[19, 28] These hance efficiency and improve the long-term stability under op-
cells consist of a highly porous nanostructured layer of titani- erating conditions.
um dioxide TiO2 (anatase) coated with a monolayer of a metal- Another attractive method of solar energy harvesting and
complex dye and a liquid electrolyte layer sandwiched be- utilization is the photocatalytic splitting of water to hydrogen

50 www.chemsuschem.org  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 44 – 58
Functional Materials for Sustainable Energy Technologies

and oxygen, which enables solar energy storage in

the form of HH bonds.[12, 33–39] Direct production of
hydrogen from sunlight in a single photocatalytic
system represents an appealing alternative to photo-
voltaic electricity generation. It offers the opportuni-
ty to integrate energy capture, conversion, and stor-
age functions in a single system by converting solar
energy directly into fuels suitable for storage and
use in residential, industrial, and transportation sec-
tors, thus enabling to overcome the diffuse nature
and intermittency of solar energy. Alternatively, hy-
drogen can be produced from solar energy by water
electrolysis with electricity generated by solar cells,
but because of the high cost of PV electricity this
Figure 8. Relationship between band levels of semiconductors and redox potentials of
method is currently much more expensive than hy- water splitting relative to the normal hydrogen electrode (NHE).[28, 40]
drogen production from methane or water electroly-
sis using grid electricity.[38]
The main difference between photovoltaic solar
cells and photocatalytic cells is how the excited electron–hole electrons and the upper level of the valence band determines
pairs are used—either to produce an electric current or to the oxidation potential of photogenerated holes. To achieve
drive chemical redox reactions. The primary steps of photoca- overall water splitting, the bottom of the conduction band
talytic hydrogen production are absorption of solar photons in must be located at a more negative potential than the reduc-
a photocatalyst material and subsequent conversion of these tion potential of H + /H2 (0 V vs. the normal hydrogen electrode
photons into electron–hole pairs, which then drive chemical (NHE) at pH 0), while the top of the valence band must be
oxidation and reduction half-reactions at the semiconductor/ positioned more positively than the oxidation potential of O2/
electrolyte interface leading to the evolution of hydrogen and H2O (1.23 V vs. NHE). This energy difference is equivalent to a
oxygen (Figure 7). The solar-to-fuel energy conversion efficien- light wavelength of shorter than 1010 nm, indicating that visi-
ble light is energetically sufficient for the decomposi-
tion of water.
For a combination of cost, stability, and perfor-
mance, titanium dioxide is so far considered as one
of the best photocatalysts among other semiconduc-
tors.[41] However, its large bandgap value of approxi-
mately 3.0 eV leads to the best performance in the
violet to near-UV spectrum, yielding a low absorp-
tion coefficient and, consequently, low conversion ef-
ficiency in the visible region (Figure 5). Promising
methods to expand the light absorption of TiO2 to
Figure 7. Basic principle of overall water splitting on a semiconductor photocatalyst and the visible region are band structure modification by
schematic diagram of photoredox processes at a semiconductor particle. (Modified from doping with a suitable cation or anion, sensitization
Kudo and Miseki.[34]) of a high-surface-area TiO2 electrode with light-ab-
sorbing organometallic dyes, and the use of co-cata-
lysts loaded on the surface as nanoparticles.[34] The
cy of photocatalytic systems is limited by several factors, in- formation of heterojunction structures between TiO2 and semi-
cluding the inability to utilize visible light efficiently, a quick re- conductor co-catalysts with a small bandgap and matching
combination rate of the photoexcited electron–hole pairs, and band potentials can extend light absorption, improve charge
quick backward redox reactions. separation, increase the lifetime of charge carriers, and en-
Wide-bandgap semiconductors are usually much more effec- hance the interfacial charge-transfer efficiency.[34, 42] Using pho-
tive photocatalysts than metals, in which the rates of electron– tocatalyst and co-catalyst materials in nanocrystalline form
hole recombination are much faster. Important parameters af- offers the opportunity to minimize the distances (and thus the
fecting the performance of semiconductor materials for photo- times) over which charges have to survive and be transported
catalytic water splitting are the width of the semiconductor after excitation. Deposition of noble-metal nanometer-size
bandgap and the levels of the conduction and valence bands. clusters on the photocatalyst surface has been known to en-
Figure 8 illustrates the relationship between band levels of var- hance the photocatalytic water splitting due to increased
ious semiconductors and redox potentials of water splitting at transfer rates of photogenerated electrons to the absorbed
pH 0. The energy level of the bottom of the conduction band metal/semiconductor clusters,[43] thus decreasing the possibility
is a measure of the reduction potential of photogenerated of electron–hole recombination.[44]

ChemSusChem 2010, 3, 44 – 58  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 51
 V. L. Kuznetsov and P. P. Edwards

The tentative target for photocatalytic hydrogen production tricity, the main goal of becoming cost-competitive will only
is to develop a stable material that can achieve a quantum effi- be achieved by improving the efficiencies of solar cell materials
ciency of 30 % at 600 nm, which gives an overall solar energy toward their theoretical limits while maintaining low cost. This
conversion efficiency of about 5 %.[45] To utilize visible light can only be done through fundamental research to identify
energy more efficiently new materials with a bandgap as new materials, new physical phenomena, and new energy con-
narrow as 2 eV (corresponding to 600 nm) are required. There version concepts. For photocatalyst materials the key issue is
are several approaches in the search for a photocatalyst for to achieve the same conversion efficiency for visible-light-in-
water splitting under visible light irradiation: 1) finding new duced formation of solar fuels as already demonstrated for
multicomponent materials, 2) suitable band structure engineer- near-UV irradiation and, at the same time, significantly increase
ing by cation or anion substitutional doping of UV-active pho- stability against damage from light and attack of highly reac-
tocatalysts, and 3) development and tailoring of new nano- tive species formed during photocatalytic redox reactions.
structured photocatalysts utilizing quantum size effects. Sever-
al visible-light-driven photocatalyst materials loaded with co-
4. Thermoelectric Materials
catalysts such as CrxRh2xO3/(Ga1xZnx)(N1xOx), Ru/SrTiO3 :Rh-
BiVO4, and graphitic carbon nitrides have been developed re- Thermoelectric devices are unique solid-state heat engines in
cently with a photocatalytic response at about 500 nm, but which charge carriers serve as the working fluid.[53–55] They
their quantum yields are still only a few percent.[34, 46–48] offer a reliable, fully solid-state means of cooling and electrical
Photocatalytic hydrogen production by water splitting is an power generation and could potentially play a significant role
example of more general conversion processes that involve in sustainable energy systems. This technology is particularly
evolution of oxygen as one component and formation of suitable for the generation of electrical power from tempera-
chemical fuels as the other. An important example that is the ture gradients, for example for harvesting low-grade waste
focus of intense research is the photocatalytic reduction of heat into usable electrical energy in industrial plants, automo-
carbon dioxide to methane, methanol, or ethanol.[49–52] Success- tive exhaust systems, solar energy converters, and any other
ful development of efficient materials and systems capable of sources of waste heat. One of the important aspects of ther-
converting CO2 into solar fuels under sunlight conditions could moelectric technology is the easy scalability of both cooling
address the issues of both carbon sequestration and sustaina- and electrical-power-generating devices in the power range
ble and clean production of carbon-neutral fuels from solar from milliwatts to kilowatts. However, the relatively low energy
energy using abundant, nontoxic resources such as water and conversion efficiency of present thermoelectric materials limits
captured carbon dioxide. their performance and range of applications to specialized
The overall conversion efficiency of photocatalytic produc- niche areas in refrigeration and electrical power generation
tion of solar fuels is currently significantly lower than that of technologies. Although ingenious engineering approaches
photovoltaic solar systems. Unlike PV electricity generation, could extend the scope of applications, the real economic
where absorption of a photon results directly in an electron– impact of this technology depends on the availability of more
hole pair formation, all methods of producing solar fuels must efficient (higher performance) thermoelectric materials.
involve coupling of a one-electron light absorption process For a given temperature range of operation the thermoelec-
with multielectron redox reactions of hydrogen formation or tric efficiency of a material is determined by its dimensionless
CO2 reduction. In the case of solar hydrogen production the thermoelectric figure-of-merit ZT:
most significant problem is catalysis of the water oxidation re-
action, 2 H2O!O2+4 H + +4 e , which is particularly kinetically a2 d ð4Þ
ZT ¼ T
demanding because it is a four-electron, four-proton process. k
In the case of CO2 photoreduction using water as a reductant
the overall kinetics is two orders of magnitude slower than where a is the Seebeck coefficient, s is the electrical conduc-
that of hydrogen production, since both multielectron transfer tivity, T is the absolute temperature, and k is the total thermal
reactions of CO2 reduction and water oxidation have to occur conductivity (k = ke+kL, where ke and kL are the electronic and
simultaneously. As a result, current CO2 photoreduction cata- lattice contributions, respectively). For conventional bulk mate-
lysts are less effective than current solar water splitting cata- rials the quantities a, s, and k are interrelated in such a way
lysts.[52] that it is very difficult to control these variables independently
Although many materials have been developed for solar to increase the ZT value. Indeed, a high ZT value requires a
energy utilization to produce electricity and chemical fuels, combination of a large Seebeck coefficient, usually found in in-
none of them are currently economically competitive with sulators or intrinsic semiconductors, a high electrical conduc-
fossil fuels in terms of cost and performance. At present, solar tivity, which is characteristic of metals, and a low thermal con-
electricity is 5–10 times more expensive compared with fossil- ductivity, typical of amorphous materials. The need to optimize
derived electricity, while solar fuels are more expensive as a a variety of inherently conflicting and interrelated properties is
primary energy source than fossil fuels by factors of 20–40.[38] the major challenge in the field of thermoelectric materials. Al-
Development of cheap, reliable, and efficient methods for though significant improvements in the properties of state-of-
production of solar electricity and solar fuels represent the the-art thermoelectric materials have been achieved over the
major challenges in solar energy utilization. For solar cell elec- last thirty years, the maximum ZT value for the best bulk ther-

52 www.chemsuschem.org  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 44 – 58
Functional Materials for Sustainable Energy Technologies

moelectric materials does not exceed 1.4 over the temperature be inversely proportional to each other, which inevitably leads
range of 100–1200 K (Figure 9), even though thermodynamics to a compromise between high effective mass and high mobi-
does not place any upper limit on ZT value. lity.
An additional materials optimization conflict arises from the
necessity for minimizing the thermal conductivity. For achiev-
ing a high thermoelectric figure-of-merit, a material with high
electrical conductivity and low thermal conductivity is required
[Equation (4)]. However, the electronic contribution to the
thermal conductivity is proportional to the electrical conductiv-
ity according to the Wiedemann–Franz law:

k e ¼ LTs ð7Þ

where L is the Lorenz number L ¼ p2 kB2 =3e2. The optimization

of these interdependent and contradictory requirements repre-
sents the major challenge for materials chemists in the devel-
opment of new thermoelectric materials with maximum figure-
Figure 9. Temperature dependence of the dimensionless figure-of-merit ZT of-merit.
of some n-type thermoelectric materials. The figure was compiled by using Using Equations (4)–(7) the figure-of-merit Z can be ex-
various literature sources.[56–60]
pressed in terms of the fundamental transport parameters
through the so-called materials parameter b:[53, 54]
Generally, the transport coefficients that determine the ther-
moelectric efficiency of a material are more complex than, for Z / b ¼ mðm Þ3=2 =kL ð8Þ
example, optical or magnetic properties, which are very closely
related to the band structure. All established thermoelectric Therefore, to maximize the Z value, a material with large ef-
materials possess a complex energy band structure; their fective masses of charge carriers, high carrier mobility, and a
degree of degeneracy is high and invariably several types of low lattice thermal conductivity is required. The best combina-
chemical bonding are involved together with several co-exist- tion of these transport parameters is usually found in heavily
ing carrier scattering mechanisms. All these factors, together doped semiconductors with a doping level of around 1019–
with often strong anisotropy of the charge and heat transport 1020 carriers cm3.
properties and limitations of existing theoretical models, make Traditionally, the development of highly efficient thermo-
any theoretical predictions regarding the thermoelectric figure- electric materials was focused at maintaining the typical elec-
of-merit prohibitively difficult. Despite this, several useful con- tronic properties of a conducting crystal, while at the same
clusions on possible directions in the search for novel im- time minimizing the lattice thermal conductivity. The reduction
proved materials can be drawn from the classical Fermi–Dirac in the lattice thermal conductivity of a material was usually
statistics and a simplified model of electron scattering. achieved by the formation of solid solutions with isoelectronic
In the approximation of parabolic bands and energy-inde- elements, which creates a mass fluctuation in a crystal sublat-
pendent scattering of carriers, the Seebeck coefficient of tice and leads to an increase in phonon scattering. This
metals or degenerate semiconductors is proportional to tem- method has been successfully employed over the last 40 years
perature and effective mass m*, and is inversely proportional for the development of all state-of-the art thermoelectric mate-
to the charge carrier concentration n:[61] rials. Nevertheless, this approach has serious limitations, since
a high concentration of atomic point defects also inevitably
8p2 kB2   p 2=3 ð5Þ leads to a decrease in the electrical conductivity and carrier
a¼ mT
3h2 e 3n mobility.[54]
Recent progress in the development of thermoelectric mate-
where kB is the Boltzmann constant, e is the charge of an elec- rials with high ZT values resulted from the discovery of solids
tron, and h is Planck’s constant. Thus, a large Seebeck coeffi- that allow freedom to modify and tailor their crystal structure
cient requires a low carrier concentration and large effective in order to decouple the electrical and thermal transport prop-
masses of charge carriers, which is equivalent to large density- erties. Among novel promising thermoelectric materials are
of-states or flat bands at the Fermi level. On the other hand, two classes of inclusion compounds with skutterudite- and
high electrical conductivity requires large carrier concentration clathrate-type crystal structures. The unit cells of these com-
n and carrier mobility m: pounds contain large voids, which are occupied by loosely
bound atoms (Figure 10). The loose bonds between the filling
s ¼ nem ð6Þ atoms and the oversized voids cause anharmonic vibrations of
the former, dramatically lowering the lattice thermal conductiv-
However, as was shown in the section on transparent con- ity by effectively scattering phonons while weakly affecting the
ducting oxides, the carrier mobility and effective mass tend to electronic transport. This results in a combination of high elec-

ChemSusChem 2010, 3, 44 – 58  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 53
 V. L. Kuznetsov and P. P. Edwards

ample of large thermoelectric efficiency enhancement in low-

dimensional materials is the rough Si nanowire array in which
ZT = 0.6 was achieved at room temperature.[74] Although bulk
Si is a poor thermoelectric material at room temperature, Si
nanowires demonstrate 100-fold reduction in thermal conduc-
tivity while the Seebeck coefficient and electrical conductivity
values remain the same as those of doped bulk Si. The reason
for this dramatic reduction in thermal conductivity is the large
difference in mean free path lengths between electrons
(110 nm) and phonons (300 nm).[74] Consequently, for Si nano-
wires with critical dimensions between 110 and 300 nm the
Figure 10. Unit cell fragment of a type-II clathrate crystal structure with gen- thermal conductivity should be significantly reduced without
eral formula AII8 BIII16 BIV30 . affecting the electrical properties.
Novel approaches in designing a new generation of func-
tional thermoelectric materials offer a promise of direct elec-
trical conductivity typical of crystalline materials and, at the tricity production with efficiencies of 20 % using low-grade
same time, a very low lattice thermal conductivity typical of an waste heat with a temperature difference of a few hundred de-
amorphous solid (so-called “electron-crystal, phonon-glass” ma- grees, which would finally make such applications commercial-
terials).[62–65] The temperature dependence of ZT values of two ly viable.
n-type materials that belong to skutterudite (Yb0.2Co4Sb12) and
clathrate (Ba8Ga16Ge30) families of inclusion compounds are
5. Hydrogen Storage Materials
presented in Figure 9.
Remarkably low, glass-like thermal conductivity was ob- Reducing or eliminating our dependency on petroleum in
served in several new classes of p-type thermoelectric materials transportation systems is a major objective worldwide. The
with large and complex unit cells, such as Tl9BiTe6, Zn4Sb3, and combination of hydrogen and fuel cells in particular represents
Yb14MnSb11.[66–69] The structural complexity of these materials is a key enabling technology for a future sustainable energy
often combined with a high degree of disorder and local struc- economy which has the potential to revolutionize our energy
tural distortions at multiple length scales, leading to extremely system offering cleaner, more efficient alternatives to today’s
low lattice thermal conductivity. On the other hand, the re- energy technologies (fuel cells are not subject to limitations of
maining ordered framework of the crystal structure provides the Carnot cycle and have an energy efficiency twice that of in-
conditions for a good electrical conduction and large Seebeck ternal combustion engines).[76, 77]
coefficient, resulting in ZT  1 at elevated temperatures. Although hydrogen contains more energy on a weight-for-
Another promising class of novel materials with potentially weight basis than any other substance, it has a very low
large ZT values are low-dimensional quantum-confined nano- energy density per unit volume. Viable hydrogen storage is
structured materials, such as quantum dot superlattices and considered by many as one of the crucial and the technically
quantum wire arrays.[56, 57, 70–75] Nanostructured thermoelectric most challenging barriers to the widespread use and ultimate
materials have recently received considerable attention be- market success of hydrogen fuel cell vehicles.[78, 79] Typically, 4–
cause of their potential to overcome the efficiency of conven- 6 kg of hydrogen needs to be stored on board a vehicle to
tional bulk thermoelectric materials. With decreasing dimen- come close to the automotive requirements of a range of
sionality, the new variable of size becomes available, offering a some 300 miles (500 km) as a benchmark.
possibility to dramatically change the density of electronic There are four major options for on-board hydrogen storage
states and allowing new opportunities to vary electrical and systems, namely: 1) compressed hydrogen (typically at pres-
thermal properties independently. In quantum-confined struc- sures between 350–700 bar at room temperature); 2) cryogenic
tures the narrow electron energy bands produce high effective liquid hydrogen (operating at < 20 K and at a pressure of 1–
masses and, therefore, a large Seebeck coefficient. Further- 10 bar); 3) solid-state hydrogen storage materials, ranging from
more, the introduction of many interfaces which scatter pho- metal hydrides and complex hydrides to high-surface-area
nons more effectively than electrons enables a dramatic reduc- porous materials; and 4) hybrid solutions, incorporating at
tion in lattice thermal conductivity. This opens up new possibil- least two of the above technologies. These various options are
ities in designing novel materials with enhanced thermoelectric described in detail elsewhere.[76, 77, 79]
performance. Current hydrogen storage systems for vehicular transporta-
A ZT value of more than 2 was reported in PbSeTe-based tion are centered upon high-pressure gas containers and liquid
quantum dot superlattice structures[71] exceeding by several hydrogen tanks. However these storage options have a signifi-
times the corresponding ZT values of bulk materials of the cant energy penalty—up to 20 % of the energy content of hy-
same composition. It was demonstrated that in 0D structures it drogen is required to compress the gas and up to 35 % to liq-
is possible to both increase the numerator and decrease the uefy it. In addition, these storage options are inadequate to
denominator in Equation (8) relative to a bulk sample, thereby meet the necessary driving range requirements without signifi-
resulting in an increase in ZT from both processes. Another ex- cant intrusion into cargo or passenger space. The most viable

54 www.chemsuschem.org  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 44 – 58
Functional Materials for Sustainable Energy Technologies

alternative to compressed and liquid hydrogen is the storage volumetric storage densities but do not meet other important
of hydrogen in solids or liquids; here we are concerned solely operating requirements for transportation applications, includ-
with the issue of solid-state hydrogen storage materials. ing low temperature, fast kinetics and reversibility of hydrogen
The main performance criteria for a hydrogen storage mate- uptake and release, and effective heat transfer.
rial are:[78, 80, 81] The weight requirements are clearly relaxed for stationary
applications, where interstitial hydrides (such as LaNi5H6 and
1) multicycle reversibility of hydrogen uptake/release (not less
solid solutions within the Ti-Zr-V-Fe-Ni system) are highly effec-
than 500 cycles)
tive hydrogen stores demonstrating good reversibility and low
2) fast kinetics of hydrogen uptake/release; temperature of hydrogen adsorption/desorption. Such a relax-
ation of weight percent targets also forms the basis for current
3) low operating pressure (not more than 10 bar) and low op-
hybrid solutions to onboard storage of hydrogen using a com-
erating temperature in the range of 50 to 120 8C;
bination of a high-pressure hydrogen tank partially filled with
4) high gravimetric and volumetric hydrogen densities (hydro- a reversible interstitial hydride. Such a hybrid solution can im-
gen storage capacity of at least 6.5 wt % and at least 65 g L1 prove the volumetric hydrogen storage density by 60 % com-
of hydrogen available from the material). paring to a compressed hydrogen tank.[82]
Of course, there are other key factors (e.g., cost, confine- Currently, four major approaches are being pursued to ach-
ment, toxicity) but we focus here on the points above because ieve multivarious performance parameters for the perfect hy-
they illustrate some of the major materials chemistry challeng- drogen storage material:
es. It should be noted that as of yet, there is no material
1) A combinatorial, high-throughput approach designed to
known to meet simultaneously all of these requirements, and
search for an entire range of materials derived from multinary
other important criteria for onboard hydrogen storage.
combinations of hydrides of light elements not previously in-
The objective is to tune the materials’ properties to obtain
vestigated (also, taking guidance from the story of high-Tc su-
reversible hydrogen storage materials that are comprised of
perconductors where entirely new phases are observed
light elements and are capable of operating at, or close to,
beyond binary and ternary systems).
room temperature and atmospheric pressure. An ideal hydro-
gen storage material releasing hydrogen via thermal decompo- 2) The thermodynamic destabilization of existing materials
sition should possess properties between those of cryogenic through alloy formation or more complex reaction schemes
high surface area adsorbents, which have hydrogen binding (see below); the aim here is to attempt to modify the thermal
enthalpies in the range of 4–15 kJ mol1 H2, and complex hy- characteristics of materials, while at the same time maintain a
drides, which have dehydrogenation enthalpies above high weight percent for hydrogen storage.[83–85]
40 kJ mol1. Tuning the decomposition enthalpy will enable to
3) The cryo-adsorption of hydrogen on high-surface-area ma-
modify key performance parameters, Tdec and Pdec, representing
terials (e.g., activated carbons, zeolites, or metal-organic frame-
the temperature and pressure at which thermal decomposition
works).[86, 87]
of a hydrogen storage material occurs.
Simple atomic-number-based calculations reveal the obvious 4) A hybrid solution combining low-DH interstitial hydride ap-
fact that only the light chemical elements can be strictly enter- proaches with a (moderately) high pressure compressed hydro-
tained if criterion (4) is to be met.[78] A summary of volumetric gen design (e.g., operating at 350 bar with a TiCrMn or related
and gravimetric energy densities of some hydrogen storage alloy).[82]
materials and carbon-based fuels is illustrated in Figure 11. We use the second path, that is, material destabilization, to
Light and complex hydrides have attractive gravimetric and illustrate how the decomposition temperature and pressure
can be modified; this is one of the most important practical
parameters connecting both the thermodynamic and kinetic
aspects of these materials. This leads us to a brief discussion
on how the decomposition temperature may be understood
and consequently tuned to optimize one performance parame-
ter for a hydrogen storage material.[78, 88]
Grochala and Edwards[78] have presented thermodynamic ar-
guments for understanding how the values of Tdec,—certainly
those for binary hydrides (MHn)—might correlate with the E0
value for the corresponding M + /M0 pair in aqueous solutions.
Such a correlation is presented in Figure 12, where we show a
plot of the experimental Tdec versus E0 data for a wide variety
of binary hydrides. As one can discern from Figure 12, there is
an excellent correlation between Tdec and E0 for a wide range
of (seemingly) chemically disparate metal hydrides.
Figure 11. Volumetric and gravimetric energy densities of various hydrogen This simple empirical correlation, having its roots in the ther-
storage options and energy storage materials. modynamics of MHn formation, forms almost a “sorting map“

ChemSusChem 2010, 3, 44 – 58  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 55
 V. L. Kuznetsov and P. P. Edwards

composition temperature by 23 kJ mol1 H2 and 240 K, corre-

spondingly, compared to pure LiBH4.[90, 92]
The number of possible destabilization reactions significantly
exceeds the number of existing chemical hydrides, which
offers a promising avenue for developing a viable hydrogen
storage system. A large number of destabilization reactions
have already been studied both theoretically and experimen-
tally; some of the results are presented in Figure 14. This sug-
gests that the most promising candidate for a destabilized
system meeting the mobile storage requirements is ammonia-
borane and its derivatives (Figure 14). Another key perfor-
mance challenge for many destabilized systems is to achieve
the reversibility of hydrogen storage.

Figure 12. A correlation of the temperature, Tdec, at which thermal decom-

position of binary hydrides MH2 to the constituent elements proceeds, and
the corresponding standard redox potential of the Mn + /M0 redox pair in
acidic aqueous solutions, E0. The ranges of the working temperatures for
prototypical fuel cells are also shown. The E0 values for H2/2H and H0/H
redox pairs are indicated. (Modified from Grochala and Edwards.[78])

of experimental systems. One can also clearly see the impor-

tance of Tdec in matching the operating temperatures of differ-
ent types of fuel cells, ranging from alkaline, through polymer
electrolyte membrane, to solid oxide and molten carbonate
cells. The monotonic behavior of the Tdec versus E0 relationship
for the binary hydrides has been rationalized on the basis of
thermodynamic and theoretical arguments.[78]
Given the importance of Tdec and its “matching” to the vari- Figure 14. Effect of destabilization on gravimetric hydrogen density and hy-
ous operating temperatures of fuel cells, the notion (2) above drogen decomposition temperature (Pdec = 1 bar).[83, 84, 93–95] (Modified from
arises from “destabilizing“ high-weight-percent hydrogen stor- Vajo and Olson.[91]) The 2010 and 2015 “DoE system target” values represent
the US Department of Energy targets for 2010 and 2015.[96]
age materials in an attempt to reduce decomposition tempera-
tures.[83–85] Many hydrides of the light chemical elements have
DH values larger than the desired range of ca. 20–40 kJ mol1 Hydride destabilization by the formation of new compounds
H2. Chemical destabilization enables one to precisely modify or alloys in the dehydrogenated state is an effective strategy
the dehydrogenation thermodynamics by substituting an ener- for lowering the overall dehydrogenation enthalpy and there-
getically unfavorable reaction with other reactions involving by lowering the dehydrogenation temperature and increasing
the formation of different compounds in the dehydrogenated the equilibrium hydrogen pressure of hydrogen-rich but
state. The formation of these compounds reduces the enthalpy strongly bound hydrides. Using this approach, the key perfor-
of the dehydrogenating reaction and lowers the dehydrogena- mance parameters Tdec and Pdec can potentially be tuned to a
tion temperature. An example[84, 89–91] of such an approach is il- larger extent than would be possible with individual com-
lustrated in Figure 13 for a LiBH4+MgH2-destabilized system pounds. LiBH4 in combination with MgH2 has been used above
exhibiting a drop in the enthalpy of decomposition and de- as one of the many examples of destabilized reactions. Recent
developments with ammonium borane, NH3BH3, with a hydro-
gen content of 19.6 mass % also highlight the great potential
of this technique to develop high-weight-percent hydrogen
storage compounds. The interaction between ammonia
borane and LiH leads to the formation of LiNH2BH3 with sub-
stantially different dehydrogenation behavior.[97, 98]
Maintaining high-weight-percent hydrogen storage, whilst
reducing Tdec and enhancing reversibility, will represent a major
milestone and a potential “show-stopper” for the transition to
Figure 13. Reduction in the decomposition temperature of destabilized a hydrogen economy. However, much more fundamental re-
LiBH4 + MgH2 system by 240 K comparing to pure LiBH4.[90] search is required to understand the physical and chemical

56 www.chemsuschem.org  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 44 – 58
Functional Materials for Sustainable Energy Technologies

processes governing hydrogen storage and release and to im- [19] M. Grtzel, in Energy…Beyond Oil (Eds.: F. Armstrong, K. Blundell),
prove the hydrogen absorption/desorption characteristics in Oxford University Press, Oxford, UK 2007, pp. 120 – 136.
[20] M. A. Green, Third Generation Photovoltaics: Advanced Solar Energy Con-
order to meet hydrogen storage requirements. version, Springer, Berlin, Germany 2004.
[21] X. Deng, E. Schiff, in Amorphous Silicon-based Solar Cells (Eds.: A. Luque,
S. Hegedus), Wiley, New York, USA 2004, pp. 505 – 565.
[22] A. D. Compaan, JOM 2007, 59, 31 – 36.
6. Summary [23] D. M. Bagnall, M. Boreland, Energy Policy 2008, 36, 4390 – 4396.
We have attempted to illustrate the importance of a basic un- [24] A. Marti, A. Luque, Next Generation Photovoltaics: High Efficiency through
Full Spectrum Utilization, Institute of Physics, Bristol, UK 2003.
derstanding of fundamental materials chemistry, materials
[25] B. Daviss, New Sci. 2007, 2633, 32 – 37.
physics, and materials science in four classes of functional [26] S. Kolodinski, J. H. Werner, T. Wittchen, H. J. Queisser, Appl. Phys. Lett.
energy materials: transparent conducting oxides, solar energy, 1993, 63, 2405 – 2407.
thermoelectric, and hydrogen storage materials—providing in- [27] R. D. Schaller, V. I. Klimov, Phys. Rev. Lett. 2004, 92, 186601.
[28] M. Grtzel, Nature 2001, 414, 338 – 344.
sights into the intrinsic performance limits of these systems
[29] M. Grtzel, Chem. Lett. 2005, 34, 8 – 13.
and how the delivery of intrinsic performance can be formulat- [30] F. Gao, Y. Wang, J. Zhang, D. Shi, M. Wang, R. Humphry-Baker, P. Wang,
ed. The field of energy materials aligns with, and is reliant S. M. Zakeeruddin, M. Grtzel, Chem. Commun. 2008, 2635 – 2637.
upon, all of these separate subdisciplines. One should add [31] B. Kippelen, J.-L. Brdas, Energy Environ. Sci. 2009, 2, 251 – 261.
that, despite the high sensitivity of materials’ properties to [32] B. Minnaert, M. Burgelman, Prog. Photovoltaics 2007, 15, 741 – 748.
[33] A. Fujishima, K. Honda, Nature 1972, 238, 37 – 38.
processing issues and conditions (e.g., grain boundaries, scat- [34] A. Kudo, Y. Miseki, Chem. Soc. Rev. 2009, 38, 253 – 278.
tering mechanisms), the electronic structure of a material is ar- [35] A. J. Nozik, Annu. Rev. Phys. Chem. 1978, 29, 189 – 222.
guably one of the most important factors for understanding [36] T. Bak, J. Nowotny, M. Rekas, C. C. Sorrell, Int. J. Hydrogen Energy 2002,
the unique interplay between chemical, electronic, magnetic, 27, 991 – 1022.
[37] R. Memming, Semiconductor Electrochemistry, Wiley-VCH, Weinheim,
optical, thermodynamic and kinetic properties of functional
Germany 2001.
energy materials. The search for new materials with superior [38] H. Tributsch, Int. J. Hydrogen Energy 2008, 33, 5911 – 5930.
performance characteristics must therefore always be under- [39] R. M. Navarro, M. C. Sanchez-Sanchez, M. C. Alvarez-Galvan, F. del Valle,
pinned by such a theoretical framework if improved combina- J. L. G. Fierro, Energy Environ. Sci. 2009, 2, 35 – 54.
[40] N. Serpone, E. Pelizzetti, Photocatalysis: Fundamentals and Applications,
tions of properties are to be discovered and utilized.
Wiley, New York, USA 1989.
[41] M. Ni, M. K. H. Leung, D. Y. C. Leung, K. Sumathy, Renewable Sustainable
Keywords: conducting materials · energy conversion · Energy Rev. 2007, 11, 401 – 425.
[42] C. Yang, W. Wang, Z. Shan, F. Huang, J. Solid State Chem. 2009, 182,
hydrogen · materials science · semiconductors 807 – 812.
[43] P. P. Edwards, H. B. Gray, M. T. J. Lodge, R. J. P. Williams, Angew. Chem.
2008, 120, 6860 – 6868; Angew. Chem. Int. Ed. 2008, 47, 6758 – 6765.
[1] a) D. A. King, Science 2004, 303, 176 – 177; b) N. Armaroli, V. Balzani,
[44] R. Baba, S. Nakabayashi, A. Fujishima, J. Phys. Chem. 1985, 89, 1902 –
Angew. Chem. 2007, 119, 52 – 67; Angew. Chem. Int. Ed. 2007, 46, 52 – 66.
1905.
[2] IPCC, Climate Change 2007: The Physical Science Basis. Contribution of
[45] K. Maeda, K. Domen, J. Phys. Chem. C 2007, 111, 7851 – 7861.
Working Group I to the Fourth Assessment Report of the Intergovernmen-
[46] K. Maeda, K. Teramura, H. Masuda, T. Takata, N. Saito, Y. Inoue, K.
tal Panel on Climate Change (Eds.: S. Solomon, D. Qin, M. Manning, Z.
Domen, J. Phys. Chem. B 2006, 110, 13107 – 13112.
Chen, M. Marquis, K. B. Averyt, M. Tignor, H. L. Miller), Cambridge Uni-
[47] H. Kato, M. Hori, R. Konta, Y. Shimodaira, A. Kudo, Chem. Lett. 2004, 33,
versity Press, Cambridge, UK 2007 Available from http://www.ipcc.ch/
1348 – 1349.
publications_and_data/publications_ipcc_fourth_assessment_report_
wg1_report_the_physical_science_basis.htm (accessed October 2009). [48] X. Wang, K. Maeda, A. Thomas, K. Takanabe, G. Xin, J. M. Carlsson, K.
[3] C. Kilic, A. A. Zunger, Phys. Rev. Lett. 2002, 88, 095501. Domen, M. Antonietti, Nat. Mater. 2009, 8, 76 – 80.
[4] C. G. Granqvist, Sol. Energ. Mat. Sol. Cells 2007, 91, 1529 – 1598. [49] M. M. Halmann, Chemical Fixation of Carbon Dioxide: Methods for Recy-
[5] C. G. Granqvist, A. Hultaker, Thin Solid Films 2002, 411, 1 – 5. cling CO2 into Useful Products, CRC Press, Boca Raton, USA 1993.
[6] Basic Research Needs for Solar Energy Utilization. Report of the Basic [50] J. S. Hwang, J. S. Chang, S. E. Park, K. Ikeue, M. Anpo, Top. Catal. 2005,
Energy Sciences Workshop on Solar Energy Utilization, April 18–21, 2005, 35, 311 – 319.
US Department of Energy, Office of Basic Energy Science, Washington, [51] G. A. Olah, A. Goeppert, G. K. S. Prakash, J. Org. Chem. 2009, 74, 487 –
USA 2005. Available from http://resolver.caltech.edu/CaltechAUTHOR- 498.
S:LEWsolarenergyrpt05 (accessed October 2009). [52] V. P. Indrakanti, J. D. Kubicki, H. H. Schobert, Energy Environ. Sci. 2009, 2,
[7] J. C. C. Fan, J. B. Goodenough, J. Appl. Phys. 1977, 48, 3524 – 3531. 745 – 758.
[8] J. R. Bellingham, W. A. Phillips, C. J. Adkins, J. Mater. Sci. Lett. 1992, 11, [53] A. F. Ioffe, Semiconductor Thermoelements and Thermoelectric Cooling, In-
263 – 265. fosearch, London, UK 1957.
[9] P. P. Edwards, A. Porch, M. O. Jones, D. V. Morgan, R. M. Perks, Dalton [54] H. J. Goldsmid, Electronic Refrigeration, Pion, London, UK 1986.
Trans. 2004, 2995 – 3002. [55] F. J. DiSalvo, Science 1999, 285, 703 – 706.
[10] A. Porch, D. V. Morgan, R. M. Perks, M. O. Jones, P. P. Edwards, J. Appl. [56] CRC Thermoelectrics Handbook: Macro to Nano, (Ed. D. M. Rowe), CRC
Phys. 2004, 95, 4734 – 4737. Press, Boca Raton, USA 2006.
[11] P. P. Edwards, M. J. Sienko, Phys. Rev. B 1978, 17, 2575 – 2581. [57] Semiconductors and Semimetals, Vol. 69–71 (Ed.: T. M. Tritt), Academic
[12] N. S. Lewis, D. G. Nocera, Proc. Natl. Acad. Sci. USA 2006, 103, 15729 – Press, New York, USA 2001.
15735. [58] G. S. Nolas, M. Kaeser, R. T. Littleton IV, T. M. Tritt, Appl. Phys. Lett. 2000,
[13] R. F. Service, Science 2005, 309, 548 – 555. 77, 1855 – 1857.
[14] E. Becquerel, C. R. Acad. Sci. 1839, 9, 145 – 149. [59] A. Saramat, G. Svensson, A. E. C. Palmqvist, C. Stiewe, E. Mueller, D. Plat-
[15] D. M. Chapin, C. S. Fuller, G. L. Pearson, J. Appl. Phys. 1954, 25, 676 – 677. zek, S. G. K. Williams, D. M. Rowe, J. D. Bryan, G. D. Stucky, J. Appl. Phys.
[16] W. Shockley, H. J. Queisser, J. Appl. Phys. 1961, 32, 510 – 519. 2006, 99, 023708.
[17] C. H. Henry, J. Appl. Phys. 1980, 51, 4494 – 4500. [60] J. P. Heremans, V. Jovovic, E. S. Toberer, A. Saramat, K. Kurosaki, A. Char-
[18] L. L. Kazmerski, J. Electron Spectrosc. Relat. Phenom. 2006, 150, 105 – 135. oenphakdee, S. Yamanaka, G. J. Snyder, Science 2008, 321, 554 – 557.

ChemSusChem 2010, 3, 44 – 58  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 57
 V. L. Kuznetsov and P. P. Edwards

[61] M. Cutler, J. F. Leavy, R. L. Fitzpatrick, Phys. Rev. 1964, 133, A1143 – Bagley), 2007, pp. 27–35. Available from http://www.matuk.co.uk/docs/
A1152. 4_Alternative%20Energy%20FINAL.pdf (accessed October 2009).
[62] G. A. Slack, in CRC Handbook of Thermoelectrics, (Ed. D. M. Rowe), CRC [82] Y. Kojima, Y. Kwaai, S. Towata, T. Matsunaga, T. Shinozawa, M. Kimbara, J.
Press, Boca Raton, USA 1995, pp. 407 – 440. Alloys Compd. 2006, 419, 256 – 261.
[63] G. S. Nolas, J. L. Cohn, G. A. Slack, Phys. Rev. B 1998, 58, 164 – 170. [83] S. V. Alapati, J. K. Johnson, D. S. Scholl, J. Phys. Chem. B 2006, 110,
[64] A. Bentien, M. Christensen, J. D. Bryan, A. Sanchez, S. Paschen, F. Steg- 8769 – 8776.
lich, G. D. Stucky, B. B. Iversen, Phys. Rev. B 2004, 69, 045107. [84] S. V. Alapati, J. K. Johnson, D. S. Scholl, Phys. Chem. Chem. Phys. 2007, 9,
[65] V. L. Kuznetsov, L. A. Kuznetsova, A. E. Kaliazin, D. M. Rowe, J. Appl. Phys. 1438 – 1452.
2000, 87, 7871 – 7875. [85] S. V. Alapati, J. K. Johnson, D. S. Scholl, J. Phys. Chem. C 2007, 111,
[66] B. Wçlfing, C. Kloc, J. Teubner, E. Bucher, Phys. Rev. Lett. 2001, 86, 4350 – 1584 – 1591.
4353. [86] L. C. Rowsell, O. M. Yaghi, Angew. Chem. 2005, 117, 4748 – 4758; Angew.
[67] H. J. Kim, E. S. Božin, S. M. Haile, G. J. Snyder, S. J. L. Billinge, Phys. Rev. B Chem. Int. Ed. 2005, 44, 4670 – 4679.
2007, 75, 134103. [87] K. M. Thomas, Catal. Today 2007, 120, 389 – 398.
[68] S. R. Brown, S. M. Kauzlarich, F. Gascoin, G. J. Snyder, Chem. Mater. 2006, [88] W. Grochala, P. P. Edwards, J. Alloys Compd. 2005, 404, 31 – 34.
18, 1873 – 1877. [89] J. J. Vajo, S. L. Sheith, F. Mertens, J. Phys. Chem. B 2005, 109, 3719- 3722.
[69] G. J. Snyder, M. Christensen, E. Nishibori, T. Caillat, B. B. Iversen, Nat. [90] J. J. Vajo, T. T. Salguero, A. F. Gross, S. L. Sheith, G. L. Olson, J. Alloys
Mater. 2004, 3, 458 – 463. Compd. 2007, 446–447, 409 – 414.
[70] L. D. Hicks, M. S. Dresselhaus, Phys. Rev. B 1993, 47, 12727 – 12731. [91] J. J. Vajo, G. L. Olson, Scr. Mater. 2007, 56, 829 – 834.
[71] T. C. Harman, P. J. Taylor, M. P. Walsh, B. E. LaForge, Science 2002, 297, [92] M. Aoki, K. Miwa, T. Noritake, G. Kitahara, Y. Nakamori, S. Orimo, S.
2229 – 2232. Towata, Appl. Phys. A 2005, 80, 1409 – 1412.
[72] M. S. Dresselhaus, Y.-M. Lin, S. B. Cronin, O. Rabin, M. R. Black, G. Dressel- [93] Y. Nakamori, S.-I. Orimo, J. Alloys Compd. 2004, 370, 271 – 275.
haus, in Semiconductors and Semimetals, Vol. 71 (Ed. T. M. Tritt), Academ- [94] J. Lu, Z. Z. Fang, H. Y. Sohn, Inorg. Chem. 2006, 45, 8749 – 8754.
ic Press, New York, USA 2001, pp. 1 – 121. [95] Y. W. Choa, J.-H. Shima, B.-J. Lee, CALPHAD Comput. Coupling Phase Dia-
[73] R. Venkatasubramanian, E. Siivola, T. Colpitts, B. O’Quinn, Nature 2001, grams Thermochem. 2006, 30, 65 – 69.
413, 597 – 602. [96] Basic Research Needs for the Hydrogen Economy, Report of the Basic
[74] A. I. Hochbaum, R. Chen, R. D. Delgado, W. Liang, E. C. Garnett, M. Najar- Energy Sciences Workshop on Hydrogen Production, Storage, and Use,
ian, A. Majumdar, P. Yang, Nature 2008, 451, 163 – 167. May 13 – 15, 2003, US Department of Energy, Office of Basic Energy Sci-
[75] M. S. Dresselhaus, G. Chen, M. Y. Tang, R. G. Yang, H. Lee, D. Z. Wang, ence, Washington, USA 2003.
Z. F. Ren, J.-P. Fleurial, P. Gogna, Adv. Mater. 2007, 19, 1043 – 1053. [97] X. Kang, Z. Fang, L. Kong, H. Cheng, X. Yao, G. Lu, P. Wang, Adv. Mater.
[76] S. G. Chalk, J. F. Miller, J. Power Sources 2006, 159, 73 – 80. 2008, 20, 2756 – 2759.
[77] P. Hoffmann, Tomorrow’s Energy: Hydrogen, Fuel Cells and the Prospect [98] Z. Xiong, C. K. Yong, G. Wu, P. Chen, W. J. Shaw, A. J. Karkamkar, T.
for a Cleaner Planet, MIT Press, Cambridge, USA 2002. Autrey, M. O. Jones, S. Johnson, P. P. Edwards, W. I. F. David, Nat. Mater.
[78] W. Grochala, P. P. Edwards, Chem. Rev. 2004, 104, 1283 – 1315. 2008, 7, 138 – 141.
[79] R. von Helmolt, U. Eberle, J. Power Sources 2007, 165, 833 – 843.
[80] P. P. Edwards, V. L. Kuznetsov, W. I. F. David, Philos. Trans. R. Soc. A 2007,
365, 1043 – 1056.
[81] P. P. Edwards, V. L. Kuznetsov, M. Gersch, S. Ellis, Z. X. Guo, in Materials Received: August 7, 2009
UK Energy Review. Alternative Energy Technologies (Eds.: J. Oakey, C. Published online on November 26, 2009

58 www.chemsuschem.org  2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 44 – 58