You are on page 1of 9

Energy &

Environmental Science
COMMUNICATION

Sr- and Mn-doped LaAlO3—d for solar thermochemical H2


and CO production†
Cite this: Energy Environ. Sci., 2013, 6,
2424
Published on 04 June 2013. Downloaded by Purdue University on 28/10/2016 06:22:52.

a
Anthony H. McDaniel,* Elizabeth C. Miller,‡ab Darwin Arifin,§a Andrea Ambrosini,b
a c
Received 22nd April 2013 Eric N. Coker,b Ryan O'Hayre,c William C. Chueh{ and Jianhua Tong*
Accepted 4th June 2013

DOI: 10.1039/c3ee41372a

www.rsc.org/ees

The increasing global appetite for energy within the


transportation sector will inevitably result in the combustion of Broader context
more fossil fuel. A renewable-derived approach to carbon-neutral Efficiently storing solar energy in a chemical bond continues to be an
synthetic fuels is therefore needed to offset the negative impacts engineering challenge. Many technologies indirectly convert solar
of this trend, which include climate change. In this energy to fuel, e.g., water electrolysis coupled to solar-electric power to
communication we report the use of nonstoichiometric perovskite produce H2. However, direct conversion of solar energy to chemical
oxides in two-step, solar-thermo- chemical water or carbon potential energy is preferred. One promising route to making solar H2 is
a two-step thermochemical cycle operating at high temperature driven by
dioxide splitting cycles. We find that LaAlO3 doped with Mn and Sr
concen- trating solar power (CSP). Here, solar radiation in the form of
will efficiently split both gases. Moreover the H2 yields are 9× process heat drives an endothermic water-splitting reaction that
greater, and the CO yields 6× greater, than those produced by the produces a H2 mole-
current state-of-the-art material, ceria, when reduced at 1350 ◦C cule. The conversion of heat to chemical energy is accomplished using
non-volatile metal oxides as the working uid in an oxidation/reduction
and re-oxidized at 1000 ◦C. The tempera- ture at which O2 begins
cycle. CSP technologies for hydrogen production promise to be cost-
to evolve from the perovskite is fully 300 ◦C below that of ceria. competitive because of high conversion efficiency and simplicity of
The materials are also very robust, maintaining their redox activity implementing the process. However, success is predicated upon discov-
over at least 80 CO2 splitting cycles. This discovery has profound ering materials that operate more effectively than the current state of the
art. In this communication, we demonstrate that perovskite-type oxides
implications for the development of concentrated solar fuel
have the potential to revolutionize CSP fuel production because they
technologies.
exhibit high reaction extents at low reduction temperatures. And while
we offer a single example of this superior activity, the amenability of
the
Developing carbon-neutral, sustainable energy sources is
perovskite structure to modication heralds a vast composition space
necessary to mitigate the impacts of anthropogenic carbon
dioxide on global populations, 1 especially as the demand for
energy rises world-wide. 2 One solution is to exploit solar energy,
based on photosynthesis (articial or natural) and photo-
given its abundance and accessibility. The challenge, however,
electrolysis are under consideration. These aim to produce H 2
is to capture and store this relatively diffuse energy in an effi-
by water splitting (WS), CO by carbon dioxide splitting (CDS),
cient and cost-effective manner. Currently, several
or fungible fuels from biomass.3,4 An additional compelling
technologies
approach rapidly gaining attention is the conversion of solar
energy into fuel using solar-driven thermochemical processes.5,6
a
Sandia National Laboratories, Livermore, California 94551, USA. E-mail: amcdani@
sandia.gov; Tel: +1 925 294 1440 Thermochemical cycles rely on heat derived from concen-
b
Sandia National Laboratories, Albuquerque, New Mexico 87185, USA trating solar power (CSP) to drive endothermic gas-splitting
Metallurgical and Materials Engineering, Colorado School of Mines, Golden,
c
reactions, thereby converting solar energy directly into fuel. 7
Colorado 80401, USA. E-mail: jhtongm@gmail.com; Tel: +1 303 273 3053 Consequently, these cycles have the potential to realize greater
† Electronic supplementary information (ESI) available: Details of perovskite theoretical efficiency than methods based on photosynthesis,
synthesis, methods for composition and activity characterization, kinetic
photoelectrolysis, or conventional electrolysis coupled to
modeling, and additional gures (PDF). See DOI: 10.1039/c3ee41372a
‡ Present address: Department of Materials Science and Engineering, 2220
solar- electric.5,7,8 In addition, such processes are conceptually
Campus Drive, Northwestern University, Evanston, Illinois 60208, USA. much simpler and require no input of electrical energy,
§ Present address: Department of Chemical and Biological Engineering JSCBB, semiconductor processing, separation membranes, genetic
UCB 596, Boulder, Colorado 80303, USA. manipulation, or elaborate molecular self-assembly. 3 An
{ Present address: Department of Materials Science, Durand Building 496 Lomita example of a WS cycle using cerium oxide is outlined in the
Mall, Stanford University, Stanford, California 94305-4034, USA.

This journal is ª The Royal Society of Chemistry 2013 Energy Environ. Sci., 2013, 6, 2424–2428 | 1
following two reactions:

2 | Energy Environ. Sci., 2013, 6, 2424–2428 This journal is ª The Royal Society of Chemistry 2013
View Article Online

Communication Energy & Environmental Science

1
CeO2 þ 1 1
d !D CeO þ 2 (1)
O 2—
d d 2
1 —
CeO2 — þ H O ——!D 1 CeO (2)
d d
2

þH
2 2
d
where d represents the extent of oxygen nonstoichiometry in ow and heated at 16 ◦C s—1 to 1350 ◦C. A positive production
the solid. In this two-step cycle, the oxide is rst reduced (eqn rate indicates O2 released by the solid, and a negative rate
(1)) on- sun at high temperature (T1 > 1500 ◦C). The oxidation indicates O2 incorporation. It is clear by the magnitude of the
reaction area under the SLMA3 curve (shown in parenthesis) that the
Published on 04 June 2013. Downloaded by Purdue University on 28/10/2016 06:22:52.

(eqn (2)) resupplies O atoms to the solid, liberates H 2, and


completes the cycle. The oxidation reaction occurs off-sun,
typi- cally at T2 < 1100 ◦C. Substituting CO2 for H2O, and CO for
H2, in these equations yields the corresponding reactions for
CDS.
A small number of non-volatile metal oxides have been
proposed for this purpose and in general fall into three cate-
gories based on the type of solid state process that results in
reduction: nonstoichiometric oxygen loss,9,10 solid solution
decomposition,5,11,12 and displacement reaction.13,14 Undoped ceria
(CeO2) is a prototypical nonstoichiometric oxide with fast redox
kinetics. However, very high temperatures (T1 > 1500 ◦C)
are necessary to achieve the extent of oxygen deciency required
for efficient fuel production. Conversely, ferrite solid solutions
undergo deep reduction at 1400 ◦C, but the kinetics are
slow15,16 and ferrites must be combined with an oxygen-
conducting refractory ceramic to prevent sintering and
maintain high redox magnitudes.12,17 Although reactive
structures formed through nano-engineering methods can
dramatically increase ferrite reaction rates, nanostructural
instability above 700 ◦C limits their usefulness.18 Here, we
describe the discovery of novel
perovskite compounds that achieve much higher oxygen de-
ciencies than reduced ceria (CeO 2—d) at lower T1, exhibit WS and
CDS kinetics similar to ceria, and do not deactivate during the
course of 80 redox cycles.
Perovskite oxides (ABO 3) are a diverse class of materials that
are largely unexplored for solar fuel applications. We synthe-
sized single-phase powders of Sr xLa1—xMnyAl1—yO3—d (labelled
SLMA1-3, see Abbreviations for each of the three compositions)
by a modied Pechini method (see ESI† for details). Our
approach was to derive a redox-active structure from a perov-
skite that is stable under both oxidizing and reducing envi-
ronments, such as LaAlO 3, in which both cations maintain a
single oxidation state. We sought to introduce redox activity and
tolerance for oxygen nonstoichiometry by doping Mn 2+/Mn3+/
Mn4+ on the B-site and Sr2+ on the A-site. Using a stagnation
ow reactor (SFR)14,15,18 equipped with a 500 W near-infrared
laser for rapid sample heating (refer to ESI†), we determined the
oxidation rates of each material (SLMA1-3) and quantied their
extent of reduction and cyclability.
A critical rst assessment for material viability in solar fuel
applications is to determine the extent of oxygen deciency that
can be achieved under conditions of rapid heating and constant
oxygen activity. Presented in panel (a) of Fig. 1 is the oxygen
uptake and release in SLMA3 and CeO2. Both samples are
exposed to a constant 0.2 mbar O2 partial pressure in helium
View Article Online

Energy & Environmental Science Communication

Fig. 1 (a) Rate of O2 released and absorbed by SLMA3 and CeO2 as a function
of time and temperature measured in a constant O 2 partial pressure of 0.2
mbar. Samples heated at 16 ◦C s—1 between 850 and 1350 ◦C, then held at
1350 ◦C for 120 s. The total amount of O2 released in mmoles per g
material is shown in parentheses. (b) Oxygen evolution at the onset of thermal
reduction in 100 mbar He showing SLMA3 reduces at temperatures 300 ◦ C
below CeO2.

perovskite is reduced to a larger extent than CeO 2 under these


conditions. Fully 8× more O2 is released and subsequently
re- absorbed by SLMA3 than ceria. This depth of reduction
directly equates to greater cycle capacity and hence
potentially more efficient fuel production provided the re-
oxidation reaction with water or carbon dioxide is sufficiently
favoured. Moreover, the onset of O2 evolution in SLMA3
under a He ambient (O2 pres- sure < 10—3 mbar) occurs at
temperatures 300 ◦C lower than CeO2, shown in panel (b) as
DT1, which implies that this compound likely possesses a
lower reduction enthalpy than CeO2. The data shown in panel
(b) of Fig. 1 was acquired by heating the sample at 10 ◦C min—1
using a furnace. At the slower heating rate and uniform
thermal environment, the powder sample is presumed to be
equilibrated and therefore the lagging
O2 onset temperature for CeO2 is likely not a kinetic limitation.
Recently, the reduction enthalpy has been identied as an
important metric for assessing the viability of candidate mate-
rials for use in CSP fuel production. In general, the
magnitude of the reduction enthalpy establishes boundaries
on tempera- ture and pressure for which WS and CDS
thermodynamic favourability is expected.19 Typical CSP
conditions for CeO2 place this compound near the upper limit
making it difficult to
reduce but easy to re-oxidize. We nd that the SLMA
compounds trend in a better direction because a solar fuel
production cycle using SLMA3 will operate at T1
temperatures well below those established for either ceria
(1550 ◦C) or ferrite (1400 ◦C) without losing the ability to
carry out reaction 2 at a reasonable T2 (800 ◦C). From a
practical point of view this has important implications for
reactor design and materials of construction.
The evolution of H2 (CO) when exposing the reduced SLMA
formulations to H2O (CO2) at conditions typically used in CSP
thermochemical cycles demonstrates that WS (CDS) is favour-
able. Shown in Fig. 2 are the H 2 and CO production rates for
SLMA1-3 and CeO2, measured during WS and CDS over previ-
ously reduced oxides. Nearly 9× more H2 and 6× more CO are
produced by the SLMA compounds as compared to CeO2
when
redox cycles making it difficult to assess long term trends).
Moreover, the number of perovskite compounds that may
exhibit WS and CDS activity is potentially many times larger
than what can be explored by modifying the ceria uorite
structure.
Although decreasing T1 while increasing cycle capacity is
signicant, fast oxidation kinetics are equally important to solar
fuel production because they directly inuence solar conversion
efficiency and ultimately govern reactor design. 7,12 Therefore, we
performed a detailed numerical analysis 15 of the H2 and CO
production rates for SLMA1-3 and CeO2 in order to compare the
Published on 04 June 2013. Downloaded by Purdue University on 28/10/2016 06:22:52.

intrinsic WS and CDS kinetics (refer to ESI†). It is important


to deconvolve the effects of dispersion, mixing, and detector
time lag that are inherent to the SFR experiment from the
observed H2 (CO) production rate transient in order to recover
the WS (CDS) rate constant.15 Reaction rate constants (k0) for
order- based processes derived from this analysis and the
percentage of H2 (CO) attributed to them are presented in
Table 1. The high
quality of the ts to our data with these kinetic models is
Fig. 2 (top) H2 and (bottom) CO production rates as a function of time evident from the solid lines in Fig. 2. Unlike CeO2, the SLMA
measured during oxidation in 40 vol% H2O or CO2 at 1000 ◦C (open symbols).
oxidation rates are governed by two mechanisms. The rst is an
SLMA1-3 and CeO2 thermally reduced at 1350 ◦C in He. The total amount of H2
or CO produced in mmoles per g material is shown in parentheses. Solid lines are order-based process (labelled F1 and F2 in Table 1) that
the results of kinetic modelling (see text). explicitly refers to the exponent of (1 — a)n in the kinetic model
where n ¼ 1 or 2, respectively. This process accounts for roughly
75% of the H2 (CO) produced. The second is a diffusion
reduced at 1350 ◦C and oxidized at 1000 ◦C. We note that the process that generates the long tails seen in the product decay
SLMA compounds and CeO2 were re-oxidized under a curves in Fig. 2 and is not listed in Table 1. Although it is
constant beyond the scope of this communication to discuss why WS
ow of water or carbon dioxide that swept the product gas (H2 or and CDS kinetic models differ between SLMA and CeO2,
CO) from the reactor. This created a condition that maximized experiments suggest ceria does not manifest a diffusion
the thermodynamic driving force for oxidation in the SFR limitation in our powdered samples. The WS and CDS rates
because the ratio of H2O/H2 or CO2/CO was always greater for SLMA compounds are comparable to CeO 2 at 1000 ◦C and
than 40 vol% oxidant concentration.
100. We also note that the SLMA compounds produce more A nal assessment of material viability for CSP fuel produc-
H2 and CO than CeO2 when ceria is reduced at 1500 ◦C, which tion is durability. Fig. 3 shows the results of a short-term
is a more commonly reported reduction temperature (refer to durability study on SLMA2 undergoing CO2 splitting at T1 ¼
ESI†). (La, Sr)MnO3—d is widely used in O2 separation 1350 ◦C, and T2 ¼ 1000 ◦C. The constant level of CO produced
membranes during the course of this experiment is encouraging and indi-
or as a cathode material in solid oxide fuel cells. And the effect cates there are no fast material degradation mechanisms at
of oxygen activity on nonstoichiometry is documented.20 play. X-ray diffraction patterns for the cycled material also
Therefore, we expected our SLMA compounds to exhibit indicate that SLMA2 retains the single-phase perovskite struc-
redox ture. This result is in contrast to a porous, monolithic ceria
activity. However, their ability to perform WS and CDS aer material that showed a rapid decrease in fuel production rate in
thermal reduction is not well understood, nor are the effects of the rst 100 cycles,25 or the degradation aer a single cycle seen
Al incorporation. WS activity aer chemical reduction in
methane was previously demonstrated using similar A-site
compositions of (La, Sr)MnO3—d and (La, Sr)FeO 3—d;21,22 also, a
Table 1 Oxidation reaction rate constants (k0) extracted from fits to experi-
recent report of CDS activity over La 0.65Sr0.35MnO3—d has been mental data in Fig. 2 using a coupled kinetic and dispersion model
published.23 However, to our knowledge this is the rst
demonstration of WS under thermochemical conditions, as well
as a comparative assessment of the reaction kinetics.
Putting this discovery in perspective, prior efforts to improve
the performance of ceria, which focused on destabilizing the rH f k0YH O(1 — a) rCO f k0YCO (1 — a)2
2 2 2

crystal structure through doping and substitution, succeeded in —1 a —1 a


Compound log k [s ] F1(%) log k [s ] F2(%)
increasing cycle capacity at a lower T1 of 1400 ◦C (notably for 0 0
Zr : CeO2 solid solutions mixed with Y, La, or Gd).24 However, SLMA1 —2.00 70 —1.90 76
the reduction and oxidation reaction rates decreased signi- production also dropped 33% by the second cycle questioning
cantly when compared to unmodied ceria. The enhanced H 2 the viability of this approach (data was presented for only 2
2426 | Energy Environ. Sci., 2013, 6, 2424– This journal is ª The Royal Society of Chemistry 2013
SLMA2 —2.11 75 —1.91 77
SLMA3 —1.81 84 —1.86 78
CeO2 —1.75 100 —1.67 100
a
F1 ¼ rst order reaction, F2 ¼ second order reaction.

This journal is ª The Royal Society of Chemistry 2013 Energy Environ. Sci., 2013, 6, 2424–2428 |
improvements in the redox thermodynamics, as well as a
dramatic increase in the redox capacity, by tuning the perov-
skite composition (e.g., combining Mn and Al on the B-site).
Further investigation into the structure–property relationship
in this particular system may yield even higher performing
compositions.
A tremendous challenge to advancing the use of thermo-
chemical cycles for water and carbon dioxide splitting is the
identication of viable materials. Commercial success is
pred-
icated upon materials composed of earth-abundant elements
Published on 04 June 2013. Downloaded by Purdue University on 28/10/2016 06:22:52.

that can operate at lower reduction temperatures than current


ferrite- or ceria-based systems. These materials must also have
sufficient activity to achieve high process efficiency. The non-
stoichiometric perovskite oxides described here are among the
Fig. 3 Total amount of CO produced per a cycle for SLMA2. Material was most promising thus far in this regard, exhibiting higher
reduced at 1350 ◦C in He (heated at 6 ◦C s—1), and oxidized in 40 vol% CO2
extents of reaction at signicantly lower reduction temperatures
at
1000 ◦C. Dwell times for thermal reduction and oxidation were 15 min each
than either ceria or metal-substituted ferrites such as NiFe 2O4 .
resulting in a 30 min redox cycle. Note that this cycle time is 50% shorter than the we have demonstrated that it is possible to achieve signicant
cycle time used for the data presented in Fig. 2.

in the modied Zr : CeO2 solid solutions explored by Le Gal and


Abanades.24

Conclusions
In general, the WS and CDS rates for the three SLMA
formula- tions we synthesized and characterized are
comparable to CeO2, a material known for fast redox kinetics.
In addition, these perovskites produce signicantly more
fuel than ceria at
reduction temperatures well below 1550 ◦C and maintain
constant activity for many cycles. This is key to market viability
because materials must remain active for many thousands of
redox cycles in order to avoid high costs associated with
frequent replacement (e.g., assuming 5 min per cycle, 8 h per
day, 300 days per year, 10 years material life equates to
~ 300 000 cycles). CSP materials must also have sufficient
activity to achieve high process efficiency (solar-to-fuel conver-
sion efficiency >20% on an annual average basis).8
Concern regarding process efficiency has been raised by
Scheffe et al. in their recent work with (La, Sr)MnO 3—d.23 They
claim that the LSM family of materials will likely lag ceria in
terms of solar-to-fuel efficiency under most CSP conditions
because they require lower oxidation temperatures (T2 ¼ 127 ◦C)
and large amounts of oxidant (H 2O or CO2) to achieve favour-
able redox behaviour. This conclusion was based primarily on
the need to heat LSM over a much larger temperature difference
between T2 and T1, thus requiring more solar energy input per
cycle than CeO2. However, assessing overall process efficiency is
a complex proposition that must take into account the specic
reactor design and method of operation.26 For example, it is
not impractical to operate a solar thermochemical reactor
using a large amount of excess steam, or even a large difference
between T2 and T1, as long as the sensible heat can be
recovered effec- tively. In addition, their assessment was based
on analyzing a narrow composition space of hole-doped
LaMnO3. In this study
We anticipate that the amenability of the perovskite structure
to doping and substitution will open a vast composition
space, within which even more effective materials can be
designed that will accelerate realization of economical fuel
production based on thermochemical cycles.

Abbreviations

SLMA1 Sr0.4La0.6Mn0.6Al0.4O3—d
SLMA2 Sr0.6La0.4Mn0.6Al0.4O3—d
SLMA3 Sr0.4La0.6Mn0.4Al0.6O3—d.

Acknowledgements
E.C.M. was supported by a DOE NNSA Stewardship Science
Graduate Fellowship, grant number DE-FC52-08NA28752.
W.C.C. was supported by an appointment to the Sandia
National Laboratories Truman Fellowship in National
Security Science and Engineering. Work conducted at
Colorado School of Mines was supported by the National
Science Foundation MRSEC program under grant no. DMR-
0820518. Work at Sandia was supported by the DOE Fuel Cell
Technologies Office as part of the Production technology
development area, and by Labo- ratory Directed Research
and Development at Sandia National Laboratories. Sandia is
a multi-program laboratory operated by Sandia Corporation,
a Lockheed Martin Company, for the United States
Department of Energy's National Nuclear Security
Administration under Contract DE-AC04-94AL85000.

Notes and references


1 M. L. Parry, O. F. Canziani and J. P. Palutikof, et al.,
Technical Summary. Climate Change 2007: Impacts,
Adaptation and Vulnerability. Contribution of Working
Group II to the Fourth Assessment Report of the
Intergovernmental Panel On Climate Change, Cambridge
University Press, UK, 2007, pp. 23–78.
2 International Energy Outlook 2011, U.S. Energy Information
Administration, DOE/EIA-0484(2011), September, 2011.
3 J. J. Concepcion, R. L. House, J. M. Papanikolas and
14 D. Arin, V. J. Aston, X. Liang, A. H. McDaniel and
T. J. Meyer, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 15560–
A. W. Weimer, Energy Environ. Sci., 2012, 5, 9438–9444.
15564.
15 J. R. Scheffe, A. H. McDaniel, M. D. Allendorf and
4 R. E. Blankenship, D. M. Tiede, J. Barber, G. W. Brudvig,
A. W. Weimer, Energy Environ. Sci., 2013, 6, 963.
G. Fleming, M. Ghirardi, M. R. Gunner, W. Junge,
16 E. N. Coker, J. A. Ohlhausen, A. Ambrosini and J. E. Miller,
D. M. Kramer, A. Melis, T. A. Moore, C. C. Moser,
J. Mater. Chem., 2012, 22, 6726.
D. G. Nocera, A. J. Nozik, D. R. Ort, W. W. Parson,
17 T. Kodama, N. Gokon and R. Yamamoto, Sol. Energy, 2008,
R. C. Prince and R. T. Sayre, Science, 2011, 332, 805–809.
82, 73–79.
5 T. Kodama and N. Gokon, Chem. Rev., 2007, 107, 4048–
18 J. R. Scheffe, M. D. Allendorf, E. N. Coker, B. W. Jacobs,
4077. 6 M. Romero and A. Steinfeld, Energy Environ. Sci.,
A. H. McDaniel and A. W. Weimer, Chem. Mater., 2011, 23,
2012, 5,
2030–2038.
Published on 04 June 2013. Downloaded by Purdue University on 28/10/2016 06:22:52.

9234–9245.
19 B. Meredig and C. Wolverton, Phys. Rev. B: Condens. Matter
7 R. B. Diver, J. E. Miller, M. D. Allendorf, N. P. Siegel and
Mater. Phys., 2009, 80, 245119.
R. E. Hogan, J. Sol. Energy Eng., 2008, 130, 041001.
20 S. B. Adler, Chem. Rev., 2004, 104, 4791–4844.
8 N. P. Siegel, J. E. Miller, I. Ermanoski, R. B. Diver and
21 A. Evdou, V. Zaspalis and L. Nalbandian, Int. J. Hydrogen
E. B. Stechel, Ind. Eng. Chem. Res., 2013, 52, 3276–3286.
Energy, 2008, 33, 5554–5562.
9 W. C. Chueh and S. M. Haile, Philos. Trans. R. Soc., A, 2010,
22 L. Nalbandian, A. Evdou and V. Zaspalis, Int. J. Hydrogen
368, 3269–3294.
Energy, 2009, 34, 7162–7172.
10 S. Abanades, A. Legal, A. Cordier, G. Peraudeau, G. Flamant
23 J. R. Scheffe, D. Weibel and A. Steinfeld, Energy Fuels, 2013,
and A. Julbe, J. Mater. Sci., 2010, 45, 4163–4173.
DOI: 10.1021/ef301923h.
11 M. D. Allendorf, R. B. Diver, N. P. Siegel and J. E. Miller,
24 A. Le Gal and S. Abanades, J. Phys. Chem. C, 2012, 116,
Energy Fuels, 2008, 22, 4115–4124.
13516–13523.
12 J. E. Miller, M. D. Allendorf, R. B. Diver, L. R. Evans,
25 W. C. Chueh, C. Falter, M. Abbott, D. Scipio, P. Furler,
N. P. Siegel and J. N. Stuecker, J. Mater. Sci., 2008, 43, 4714–
S. M. Haile and A. Steinfeld, Science, 2010, 330, 1797–1801.
4728.
26 I. Ermanoski, N. P. Siegel and E. B. Stechel, J. Sol. Energy
13 J. R. Scheffe, J. Li and A. W. Weimer, Int. J. Hydrogen Energy,
Eng., 2013, 135, 031002.
2010, 35, 3333–3340.

You might also like