Energy 44 (2012) 944e954

Contents lists available at SciVerse ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Material constraints for concentrating solar thermal power

Erik Pihl a, Duncan Kushnir b, *, Björn Sandén b, Filip Johnsson a
a
Division of Energy Technology, Chalmers University of Technology, Göteborg, Sweden
b
Division of Environmental Systems Analysis, Chalmers University of Technology, Göteborg, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: Scaling up alternative energy systems to replace fossil fuels is a critical imperative. Concentrating Solar
Received 28 February 2012 Power (CSP) is a promising solar energy technology that is growing steadily in a so far small, but
Received in revised form commercial scale. Previous life cycle assessments (LCA) have resulted in confirmation of low environ-
25 April 2012
mental impact and high lifetime energy return. This work contributes an assessment of potential
Accepted 27 April 2012
Available online 30 May 2012
material restrictions for a large-scale application of CSP technology using data from an existing parabolic
trough plant and one prospective state-of-the-art central tower plant. The material needs for these two
CSP designs are calculated, along with the resulting demand for a high adoption (up to about 8000 TWh/
Keywords:
Thermal electricity
yr by 2050) scenario. In general, most of the materials needed for CSP are commonplace. Some CSP
Solar energy material needs could however become significant compared to global production. The need for nitrate
RES salts (NaNO3 and KNO3), silver and steel alloys (Nb, Ni and Mo) in particular would be significant if CSP
Material reserves grows to be a major global electricity supply. The possibilities for increased extraction of these materials
Resource scarcity or substituting them in CSP design, although at a marginal cost, mean that fears of material restriction
are likely unfounded.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction commercial scale deployment in Spain, USA, Australia, Egypt and

India [5e7].
The available solar flux on land is several thousand times higher CSP plants use reflective surfaces to concentrate sunlight,
than today’s anthropogenic primary energy conversion and is providing heat for a thermodynamic cycle, such as a steam turbine.
thereby the dominant potential source for renewable energy. The The physical principle is thus very different from photovoltaic
global solar market has been rapidly growing for the past decade, panels, which use the photons in sunlight to excite electrons and
but is still dwarfed when compared to conventional fossil fuel create currents in solid state matter. These differences mean that
power. So far, the main barrier to large-scale deployment of solar CSP will differ significantly from PV regarding properties such as
power has been higher costs of electricity, because of relatively environmental impact and material constraints.
small volumes and less historical investments in technology With projected strong growth in view, it is of interest to identify
development than presently dominant power generation technol- and quantify barriers to large-scale solar power deployment, other
ogies. Through development and continued strong growth, as solar than cost as mentioned above. One such barrier is restrictions in
technologies progress down the learning-curve, the cost per kWh either the reserves (extractable resources at a given cost) or annual
of solar electricity is projected to reach parity with peaking power supply of materials needed for solar power conversion devices.
in main markets by about 2020e2030 [1e4]. Such restrictions can imply increased raw material costs as the
So far, photovoltaic (PV) technologies have the largest share of technologies grow, or even set absolute limits to how much that can
the solar power market, but there is at present a relatively steady be built. The recent study on CSP by the EASAC [2] has pinpointed
share of concentrating solar thermal power (CSP, also sometimes a need to investigate the limits and potential bottlenecks and
referred to as Solar Thermal Power, STP). CSP has undergone manufacturing constraints for CSP production.
expansion from about 400 MW installed capacity in the early Material demand and constraints for low-carbon technologies
2000s, to about 1.3 GW in 2011, with another 2.3 GW under has been evaluated in several studies over the last fifteen years.
construction and 32 GW in planning. The technology is today in Some recent studies provide overviews of constraints for many
low-carbon technologies [8e11] while others analyse metal
resource constraints for specific technologies such as electric
vehicle batteries [12e14] and solar photovoltaics [12,15e21]. A
* Corresponding author. Tel.: þ46 317721197.
E-mail address: [email protected] (D. Kushnir). general conclusion is that no technology group (such as solar PV or

0360-5442/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2012.04.057
 E. Pihl et al. / Energy 44 (2012) 944e954 945

wind power) is hindered from reaching the TW-scale due to limited 2. GMC, material constrained growth: The maximum CSP growth
supply of materials, but scarcity of some specific metals such as per year (in TWh/yr2), constrained by the production of
tellurium, indium, ruthenium and silver may constitute a severe a specific material.
problem for specific designs at significantly lower levels of market
penetration. The analysis provided by Kleijn et al. (2011) indicates We use two measures of resources. By resources we denote
that the build-up of all energy infrastructures, regardless if it is material occurrence “in such form and amount that economic
nuclear power, carbon capture and storage (CCS) or renewable extraction of a commodity from the concentration is currently or
energy, will also have some impact on the societal flows of major potentially feasible.” [29]. This includes undiscovered resources.
materials such as stainless steel. Reserves are defined as “That part of the reserve base which could
There is currently a lack of studies of materials constraints on be economically extracted or produced at the time of determina-
CSP deployment. Some material needs and energy issues for CSP tion.” [29]. The term includes only demonstrated reserves. The
have been studied through life cycle assessment, see e.g. Burkhardt reserve base is the “in-place demonstrated (measured plus indi-
et al. [22], Lechón et al. [23], Viebahn et al. [24], May [25] and cated) resource from which reserves are estimated” [29]. The term
Weinrebe [26]. General conclusions are that CSP plants have energy production in most cases refers to mine production. For some
pay-back times of about one year, which can be compared to typical commodities (steel, glass, nitrate salts, cement) production refers to
lifetimes of about 30 years, and a relatively minor ecological foot- the output from a manufacturing process.
print, indicating resource effectiveness and low external costs. Yet,
they are overall significantly more material intensive at construc- 2.1. Technology diffusion scenario
tion (per kWh basis) than fossil fuel plants of equivalent capacity.
Water use has also been a contested issue, particularly as many In order to give context to the GMC values and analyse the
high-insolation areas suited for CSP are water stressed. The use of possible consequences of strong policy to promote solar technol-
water can be reduced by more than 90% by switching from wet to ogies, the results are applied to a scenario where CSP grows
dry cooling technologies and these design modifications have been according to the “Advanced Outlook” scenario of Greenpeace, IEA
included in some LCAs, e.g. Burkhardt et al. [22]. SolarPACES and ESTELA [30]. This scenario is cited as the highest
Estimates on constraints for steel, concrete and nitrate salts, growth function in the IEA CSP Technology Roadmap [31]. The
used for dish Stirling and parabolic trough plants, were included in function is exponential with a stepwise decreasing growth rate,
a study by García-Olivares et al. [27]. The study suggests that steel giving system capacities of about 120 TWh/yr in 2015, 360 TWh/yr
and concrete are not restricting, while the natural reserves of in 2020, 1500 TWh/yr in 2030 and close to 8000 TWh/yr in 2050. A
nitrate salts are relatively small and calls for synthetic production of constant yearly growth factor is assumed to describe the capacity
salts. Trieb et al. [28] have calculated the need for steel, glass, increase 2030e2050. The 2050 value is in close correlation to
aluminium, copper, lead and concrete for a growth scenario where Pacala and Socolow [32] who suggested that CSP could supply
CSP increases linearly in capacity over 30 years to cover 15% of the 8100 TWh/yr by 2050.
EU electricity in 2050. This scenario is found to require 1.6% of the
annual 2010 global production of glass, the corresponding figures 2.2. System boundaries and data sources
for the other materials are in the range 0.1e0.4%.
The aim of this work is to further assess possible material The material commodity needs for plants are determined by
constraints that will set limits for large-scale concentrating solar a bottom-up approach, identifying the amount and types of
thermal power (CSP) deployment. The main purpose of this study is materials required to build a given CSP capacity. This calculation is
to create inventories for the material commodity needs of a TW- based on case studies of two plants, one parabolic trough and one
scale capacity of CSP plants, as well as of the annual demands central tower design. Data were gathered from a literature review
required for the build-up of such a system. Further, these inven- and direct information from CSP plant operators and manufac-
tories are compared to the total available resources and current turers. Data on maintenance material flows (e.g. washing, replacing
production capacity of the materials in question. A special focus is mirrors) is taken from one of the companies in the solar tower case
on the materials found most restricted in production, compared to and from Viebahn et al. [24] in the parabolic trough case. The
the demand. These include nitrate salts (NaNO3 and KNO3), silver, system boundary includes only the materials used in the
steel alloys (Nb, Ni, Mo, Mg and Mn in particular) and to some construction and operation of the plants. Dismantling and indirect
degree glass and materials used for glass manufacturing. The material and energy use, e.g. for the production of capital facilities
demand for water during the operation phase is quantified but not (mirror factories etc.) and construction (cranes etc.), are not
compared to global availability, since water scarcity is a local issue included.
and demands a more detailed, site-specific analysis. We use data
representative for commercial designs of parabolic trough and 2.3. Component scaling
central tower (central receiver) plants, the two most widespread
CSP technologies. In cases when data are not available for the materials intensity
of a component in one of the two CSP plants, estimates are made
2. Method based on components with identical functions in other CSP plants
(reference plants). The size of the components is scaled to take into
The basis for evaluating material constraints is constructing an account differences in plant capacity. For instance, data from the
inventory of the materials used for producing a given production solar towers PS-10 and Gemasolar are scaled and used for the solar
capacity of plants and comparing the inventory and a scenario of tower in this work. Some of the scaled data is summarized in
adoption with the available stocks and flows of resources. Two Table 2.
ratios are of particular importance [15]: The mass of solar field components, mf, are scaled linearly based
on capacity:
1. SMC, material constrained stock: The total CSP capacity (in
C
TWh/yr) that can be built, given the amount of available mf ¼ m*f ; (1)
resources of a specific material. C*
946 E. Pihl et al. / Energy 44 (2012) 944e954

Table 1 The DNI adjustment is assumed only to affect the figures on

Plant Specifications, basis for non-DNI-adjusted data. From eSolar [40] and Cobra material use per TWh/yr, not the material use per GW. This is
Energi [36].
because higher yearly DNI typically means more hours of sunlight
Parabolic trough Central tower Unit but not that the maximum solar influx (i.e. nominal capacity for
Electric capacity 49.9 100 MW a given design) is significantly higher. The maximum influx is a sum
Operating hours 3640 6250 h/yr of the solar constant and atmospheric losses, showing differences
Storage 7.5 12.7 h
first when two sites differ greatly in latitude or altitude.
Site insolation, DNI 2200 2700 kWh/m2yr
Solar field size 512,000 1,472,000 m2
Land occupancy 1,950,000 5,000,000 m2 2.5. Replacement and recycling of materials
Annual production 0.182 0.625 TWh/yr
Over a longer timescale, material needs for CSP plants will
include replacing plants in addition to net capacity addition. The
where C is the thermal capacity of the studied plant and C* and m*f technical lifetime is here assumed to be 30 years. Some of the
are the thermal capacity and the mass of solar field components of materials for new plants can be recycled from old plants. For CSP
the reference plant, respectively. This is valid for trough fields. No plants reaching end of life, recycling of materials is assumed to be
scaling is done for tower field components, as this data has been 95% for aluminium and molten salts, 90% for steel (incl alloying
fully supplied in the right scale from a company source. material) and copper, 70% for glass and silver and 0% for the
Thermal cycle components do not scale linearly with remaining materials. The recycling flows are small during the build
throughput. The material commodity mass for a steam cycle phase considered here, but is likely to increase to form the majority
component, mc, is assumed to follow the same basic function as of the material requirements in the long term future where CSP
investment cost, which from Pihl et al. [33] is calculated as: capacity saturates.

 0:89 3. Case study description and background assumptions

C
mc ¼ m*c ; (2)
C*
The two power plants chosen as cases in this study represent the
where m*c is the known mass data from a reference plant. For two dominant CSP technologies, parabolic trough and central
thermal power cycle components, C is in thermal capacity (MWth). tower. These are, however, not directly comparable, since they are
Masses for storage tanks, mt, are assumed to follow an exact area to at different stages of commercialisation. The trough plant can be
volume scaling: viewed as a design implemented on commercial scale, while the
specific tower design is e although similar to existing commercial
 2 tower designs e not to be viewed as fully commercialized. A
C 3
mt ¼ m*t ; (3) comparison of the main properties of the two plants is found in
C*
Table 1. The construction and materials needs for the different
Buildings are also assumed to scale this way. components of the CSP plants are given in Sections 3.3e3.6. For
components that are not specific to the plant design, such as
2.4. Normalisation of values buildings and some parts of the steam cycles, the material choices
are assumed to be identical for the two studied plants, if nothing
After producing the per-plant inventory, the material invento- else is specified.
ries are normalized to an energy production capacity of 1 TWh/yr.
This figure is adjusted to compensate for the varying solar 3.1. Parabolic trough plant configuration
resources. The two case study plants that are the source of the
material data are in different locations, varying in local “solar Parabolic trough fields are built up by Solar Collector Assemblies
resource”, as measured by yearly direct normal irradiation (DNI). In (SCAs), series of troughs of about 150 m length (Fig. 1). The plant
order for the plants to be comparable, the figures are adjusted to selected to represent parabolic trough technology in this report is
a given DNI of 2300 kWh/m2 yr. There is significant potential on all a 50 MWe plant in Spain, “as it would be constructed today”, with
inhabited continents to harness solar energy at this rate of irradi- data provided from manufacturer Cobra Energía [36]. The reflecting
ation, or higher [34]. aperture area of each collector is assumed to be 12  5.77 m2, with
Scaling functions for total material use as function of varying a mirror-aperture area factor of 1.10 [37,38]. The sunlight is
DNI have not been found, but there are available relations on cost concentrated on evacuated collector tubes in which heat is trans-
and DNI that could be applied with high precision. Kearney [35] ferred to synthetic oil. The oil, a mixture of biphenyl and diphenyl,
finds a cost decrease of 4.5% per 100 kWh/m2 yr. Using the input transfers the heat to the thermal cycle. Turbine operation can be
of material use, m*, at reference irradiation, I* (kWh/m2 yr), Eq. (4) smoothened and extended by heat storage in molten salt, a binary
gives the material use, m, at given irradiation, I (kWh/m2yr): mixture of NaNO3 and KNO3. Economic assessments often find that
large storage capacities are economically beneficial; for instance
m ¼ m* ð1  0:045ÞðII Þ=100
*
(4) Herrmann et al. [39] have found the lowest levelized electricity cost
at 12 h storage, but such capacities are rarely employed in present
installations. The parabolic trough plant selected for this study has
Table 2 enough storage for 7.5 h operation without sunshine. The plant is
Weights of steam cycle components for CSP plants. assumed wet cooled.
Component Mass (tons) Reference
Piping 176 55 MWth (PS-10) 3.2. Central tower plant configuration
Pumps 12.5 [45]
Steam drum 74.5 Central tower plants (Fig. 2), also called central receiver plants,
Steam turbine 160 50 MWe (Andasol 3) typically have a high tower with a large receiver on which light is
Generator 85 [46]
focused by a field of mirrors, called heliostats. This study assesses
 E. Pihl et al. / Energy 44 (2012) 944e954 947

Fig. 1. Part of a parabolic trough assembly (Plataforma solar de Almeria) similar to that chosen for this study.

the 100 MWe eSolar conceptual molten salt tower design [40]. It 3.3. Steam cycle equipment
uses small scale; flat heliostats of roughly 1 m2 size, with individual
tracking systems, each on an individual and mass produced steel Steam turbines and the piping, valves, tanks, pumps, heat
frame structure. eSolar has demonstrated their design with the exchangers, domes and other components constituting the steam
5 MWe direct steam Sierra SunTower demonstration plant. The cycles are the most complex part of the CSP plants. Viebahn et al.
molten salt design has a heliostat field very similar to the Sierra [24] show that the total material use of the entire steam cycle is
SunTower but larger towers and molten salt as heat transfer and typically a small part of CSP plant overall weight (<5%), but because
storage medium. The central tower case study plant is dry (air) of the high quality materials required, the composition is of
cooled. importance. Steam turbines are commonly built with a high

Fig. 2. View of a solar tower plant (Sierra SunTower), similar to that chosen for this study. With permission from eSolar.
948 E. Pihl et al. / Energy 44 (2012) 944e954

proportion of stainless 9%-12%-chromium steels, also containing proportions: silica sand 73%, soda ash 13%, lime 8%, others 6%. The
molybdenum (Mo), manganese (Mn), nickel (Ni), vanadium (V), reflective coating is assumed to be silver, in a 100 nm thick layer.
carbon and some other trace elements. Supporting steel structures for collectors/heliostats of both
In this study, the steam turbine is assumed to be composed by plants are assumed to be hot-dip galvanized steel. This material is
85.4% Fe, 13.0% Cr, 0.4% Ni, 0.13% Mn, 1.0% Mo and 0.06% V [41,42] mainly carbon steel (98% Fe, 1% C, 1% Mn) covered by a zinc layer of
(data also from an undisclosed steam turbine manufacturer). For 100 mm [48]. The steel used for the parabolic trough absorber tubes
the trough plant, steam cycle pipes and heat exchangers are is DIN 1.4541 stainless steel [49] with an approximated content of
assumed to be T22 low-chromium steel for high temperature 18% Cr, 10.5% Ni and 0.4% Ti [50].
(400  C) applications and carbon steel for low temperatures. For the
central tower plant, the heat exchangers are assumed to be built in
347H stainless steel for high temperatures, to withstand the molten 3.5. Heat transfer and storage
salts, and carbon steel for low temperatures [43]. Steam pipes for
the tower plant are assumed 12% Cr steel at high temperatures and Large amounts of liquid media are used in both plants for
carbon steel at low temperatures. Assumed data for key steam cycle transfer and sensible storage of heat. The tower plant uses molten
components are given in Table 2. Some of the data on pipe weights salt for both purposes, while the trough plant collects heat by
are supplied by the plant manufacturers [36,40], for wet and dry synthetic oil flowing through the absorber tubes and stores the
cooling equipment by GEA [44]. heat in a separate system with molten salt. Figures for the amounts
of thermal media used in the plant have been provided by plant
3.4. Collectors manufacturers [36,40]. Material needs for the storage tanks have
been estimated by using data for trough and tower plants in the
Mirrors are used for the collectors or heliostats of the solar report by [24] and scaled by Eq. (3) based on mass and density of
plants. It is assumed that these use low-iron glass as substrate [47]. the storage medium. When using nitrate salts there will be some
Raw materials used for the glass have approximately the level of decomposition to nitrite and other secondary products,

Table 3
Per GW and TWh/yr inventory for the two case plants of this study. The TWh/yr values are DNI-adjusted.

Material Per GW (tons) Per TWh/yr (tons)

Trough Tower Trough Tower

Construction
Aluminium (Metal) 0 11,000 0 2200
Cement 250,000 72,000 65,000 13,800
Chromium 2200 3700 570 710
Copper 3200 1400 840 260
Aluminium (Elemental) 740 12,000 180 2300
Fibreglass 310 0 82 0
Foam glass 2500 1800 640 340
Glass 130,000 110,000 31,000 21,000
Iron 650,000 393,000 170,000 75,000
KNO3 220,000 150,000 59,000 28,000
Lime 11,000 9400 2800 1800
Limestone 170,000 49,000 44,000 9400
Magnesium 3000 2600 730 500
Manganese 2000 5700 540 1100
Molybdenum 200 56 52 11
NaNO3 340,000 220,000 88,000 42,000
Nickel 940 1800 250 350
Niobium 0 140 0 26
Oil 44,000 0 12,000 0
Polypropylene 500 0 130 0
Rock 1,300,000 5,000,000 340,000 950,000
Rock wool 4700 3400 1200 650
Sand 1900 1400 490 260
Silicon sand 92,000 81,000 22,000 15,600
Silver 13 16 3.1 3.0
Soda ash 18,000 16,000 4600 3000
Steel 240,000 400,000 63,000 77,000
Titanium 25 0 6.5 0,00
Vanadium 1.9 1.7 0.48 0.33
Zinc 650 1400 170 260
Operation and maintenance
(yearly, tons)
Aluminium 0.78 1.40 0.20 0.27
Glass 140 240 35 46
Lime 11 20 3.0 3.9
Magnesium 3.2 5.7 0.83 1.09
Oil 2000 0 510 0
Silicon sand 98 180 26 34
Silver 0.0140 0.034 0.0037 0.0065
Soda ash 19 34 4.9 6.5
Steel N/A 150 N/A 29
Water 12,000,000 160,000 3,200,000 31,000
 E. Pihl et al. / Energy 44 (2012) 944e954 949

mainly NOx. This loss rate is claimed to be low [51], and due to lack use in parabolic trough plants, compared to the 180 t/MW assumed
of reliable data, it has not been included in the assessment. by García-Olivares.
The receiver, hot salt pipes and hot salt tank of the tower plant The material breakdown graphs showed in Fig. 3 illustrates
are assumed to be made of 347H stainless steel, while other molten where in the plants some of the bulk materials are used. The main
salt pipes and the cold salt tank should be made of carbon steel use of most commodities is by far in the collector or heliostat fields.
[40,43]. The hot salt tank of the trough plant is assumed to be made The main exception is cement in the tower plant since the heliostat
of 316L stainless steel, while the oil pipes are low-Cr steel. Heat support structure does not use concrete in anchoring. The heliostat
exchangers use stainless steel 347H (tower) or low-Cr (trough) field design does, however, include ground preparation with large
steels for high temperatures and carbon steel for low temperatures amounts of gravel, meaning that site preparation is the dominating
(i.e. in the economizers). use of sand, rock and gravel. Molten salts are not included in the
graphs; they are in both cases used 100% in the storage system.
3.6. Foundations and buildings Aluminium is only used in significant amounts in the tower case.

Concrete, rock and gravel are the most common materials on

mass basis for the plants. Concrete is used for the solar field 5. Reserves and annual production of materials in
foundations, storage tanks, buildings and other miscellaneous comparison to CSP demand
structures. It is by approximation composed of 1/6 cement and 5/6
sand/rock, reinforced with rebar which is essentially 100% iron. This chapter discusses the relationship between material
Gravel is used in large quantities in the tower plant, to prepare the demands for CSP plants and the reserves and annual production of
ground under the heliostat fields. Data on materials use for foun- the corresponding materials. Table 4 shows the currently defined
dations, ground preparation and buildings are taken from plant reserves, resources and annual production of the minerals and
manufacturers and supplemented by data from Viebahn et al. [24]. commodities required for the CSP plants.
Table 5 gives the ratio of material reserve figures (Table 4) to the
4. Material inventories amounts available as obtained from the inventory (Table 3). Most
materials are widely available, and thus only materials found (from
Aggregated inventories of materials used in the two types of Table 4) to have relevant limits appear.
solar plants are shown in Table 3. As indicated previously, the table The values in Table 5 are significantly larger than the estimated
should not be seen as a direct comparison of the two technologies 20,000 TWh/yr of 2009 total electricity production [53] i.e. all
in general, since the two case plants are at somewhat different materials needed for the two types of CSP plant have a reserve that
stages of commercialisation. The reason that the tower plant is significantly larger than what would be required if all electricity
requires less molten salt per TWh/yr capacity, despite a greater at current level was to be produced by CSP units.
storage capacity, is because of the higher temperatures in the Table 5 paints an inadequate picture, though, since other
thermal cycle (higher DT). The difference in steel alloy use between applications are currently using these materials. If we take the
the two case plants is due to different steel compositions, mainly reserve life column from Table 4, the reserves for chromium, zinc,
because of varying steam cycle temperatures and heat transfer silver and copper will be exhausted at current mine rates before
media (molten salts are significantly more corrosive than thermal 2050. Chromium, zinc and copper appear to have significant room
oils). in their resource base for reserve expansion, but silver does not. As
A comparison with the findings of material use in the studies by noted, additional silver reserves will come from yet-to-be-
García-Olivares [27] and Trieb [28] shows no great discrepancy in identified polymetallic deposits (mostly lead-zinc and copper) but
results. This work finds 240 t/MW as a reasonable number for steel this availability is difficult to know in advance. The other materials

Fig. 3. Breakdown of bulk material mass distribution for the two CSP configurations, on mass basis.
950 E. Pihl et al. / Energy 44 (2012) 944e954

Table 4 Table 6
Reserves, resources and production of component materials (2010 values). The Material constrained growth for solar plants showing most constrained materials
reserve life shows how many years the present reserves can supply the current (GMC < 50,000 TWh/yr2 for either plant, sorted by minimum value).
demand before being exhausted.
Materials Material constrained growth, GMC, TWh/yr2
Reserve Resource Production Reserve life
Parabolic trough Central tower
(Mtons) (Mtons) (Mtons/yr) (years)
NaNO3 520 1100
Metals
KNO3 560 1200
Iron (Fe) 77,000 Abundanta 2300 33
Nb N/A 2400
Aluminium (Al) 32,000 Abundant 58 552
Ni 6300 4500
Titanium (Ti) 690 Abundant 6.3 110
Mo 4500 22,000
Copper (Cu) 630 3000 16.2 39
Si sand 4800 6900
Manganese (Mn) 630 Largeb 13.0 48
Glass 5500 8100
Chromium (Cr) 350 12,000 22.0 16
Ag 6700 7600
Zinc (Zn) 250 1900 12.0 21
Mg 7700 11,000
Nickel (Ni) 76 130 1.55 49
Soda ash 11,000 15,000
Vanadium (V) 14 63 0.056 250
Mn 24,000 12,000
Molybdenum (Mo) 9.8 14 0.23 43
Fe 13,500 31,000
Niobium (Nb) 3 Unknown 0.063 48
Cu 19,000 63,000
Silver (Ag) 0.54 Unknownc 0.022 25
Al 320,000 25,000
Steel Produced from Iron 1500
Steel 36,000 30,000
Minerals
Cr 39,000 31,000
Limestone Abundant Abundant
Cement 43,000 200,000
Lime Abundant Abundant 280
Zn 70,000 46,000
Silica Sand Abundant Abundant 120e
Soda ash 24,000 Abundant 46f 522
Potash 9500 Abundant 26 365
Magnesium (Mg)d 2400 Abundant 5.6 429 temperature corrosion resistant steel (Nb, Mo, Cu, Ni), and glass
Cement Produced from 2800 along with the silicon sand and refractory magnesium used to
limestone, etc.
produce it.
Glass Produced from silica 171g
sand, Mg salt A comparison of materials needs for a strong CSP growth
Potassium nitrate Produced from potash 33 2010e2050 with current annual commodity production is shown
Sodium nitrate Produced from soda ash 46 in Figs. 4e6. This can also be interpreted as the increase in
a
Some materials are so abundant as to have no practical limit on their use.
b
Manganese resources are deemed “large” but are quite irregular with South
Africa holding 75% and Ukraine holding 10% or more.
c
Silver resources will presumably be found with new polymetallic (Cu, Pb)
deposits.
d
Magnesium salts, not magnesium metal are used for glass production.
e
Industrial silica sand (quartz sand) and gravel production.
f
This figure is including synthetic production.
g
Detailed value not found. Based on silica sand production, assuming all silica
sand used for glass (silica comprising 70% of glass).
Sources: USGS Material Data Sheets [52] (Various authors).

in Table 5 still have enough ‘excess’ reserves at present mine rates

to substitute all current electricity generation with CSP.
The current material constrained growth, GMC, is shown in
Table 6. The table is again filtered to show only materials with
possibly relevant limits. The table shows generally high GMC, i.e. the
production rates are sufficient for a very rapid build-up of capacity
for most materials. The most potentially rate constrained materials
are the molten salts, silver, the alloy elements used for high

Table 5
Ratio of reserve of each material to the required amount for 1 TWh/yr capacity (DNI-
adjusted), for materials with SMC  1 million TWh. Sorted by minimum value.

Material Material constrained stock, SMC by reserves

(TWh/yr total capacity)

Parabolic trough Central tower

Niobium N/A 110,000
KNO3 160,000 340,000
Silver 170,000 180,000
Molybdenum 190,000 920,000
Nickel 310,000 220,000
NaNO3 270,000 570,000
Chromium 620,000 490,000
Manganese 1,200,000 570,000
Copper 750,000 2,400,000
Fig. 4. Ratio of material needs to material production, yearly, for molten salts in
Zinc 1,500,000 980,000
parabolic trough and central tower plants. According to growth scenario (cf 2.1).
 E. Pihl et al. / Energy 44 (2012) 944e954 951

6. Production constraints, market impact and substitution of

critical materials

The actual possibility for constraints to arise for each material

will depend on whether CSP represents a net new demand for the
material, how easily material supply can be increased and substi-
tution options for the materials in CSP designs or in their other
applications if necessary. This chapter discusses these factors for
the set of materials identified as having the largest requirement
relative to their supply: Glass and its constituents, the steel alloy
elements, silver, and the thermal salts.

6.1. Glass, magnesium and silicon sand

As shown above, a large diffusion of the CSP technology will

Fig. 5. Ratio of material needs to material production, yearly, for various materials in eventually require a large input of glass, and glass related
parabolic trough plants. According to growth scenario (cf 2.1). compounds (Si sand, soda, refractory magnesium). The material
constituents of these components are widely available, and adding
more manufacturing capacity for such common components is
yearly world production of materials that is required to make
unlikely to present a problem. Should there somehow be difficul-
the CSP growth scenario possible without decrease in demand
ties, there are possible ways to reduce or substitute the glass, such
for other purposes. Only materials for which demand would
as using thin glass, aluminium-based or polymer-based reflector
reach at least 5% of worldwide production are included. The
designs [47]. With large material supplies, little barrier to
demand in this scenario compared to current production is
increasing manufacturing, and the possibility of substitution, these
significantly higher for molten salts than for other materials.
components will not restrain CSP in any way.
Specifically for the trough plant, demand for both salts (NaNO3
and KNO3) is high in comparison to world production. Within
6.2. Steel alloy elements
two decades, CSP plants could become a big consumer of nitrate
salts with 15e35% of the global market. Around mid-century,
The elements that are alloyed with steel for enhanced corrosion
CSP plants could possibly use more than 10% of current world
resistance and strength at high temperature (Cr, Ni, Zn, Mg, Mo, Mb,
production of common materials such as nickel, glass and silica
Nb, V) would be required in large quantities for a CSP build-up. The
sand. The 10% level is reached already 2035e2040 for niobium
use of high temperature steel in CSP would however presumably
in tower plants.
substitute the use of the same alloys in other thermal electricity
In summary, the thermal salts would quickly require attention
production sites. The net effect of CSP on the demand for these
to manufacturing and extraction capacity while CSP demand for all
materials themselves is therefore unclear. Even if CSP growth was
other materials would remain a small part of the annual production
not a substitute for other thermal technologies and thus did not
for several decades, with none reaching 10% before 2030 if constant
reduce steel demand for other energy purposes, it would not
production is assumed.
require more than 5% of the current output of any of the materials
for several decades, with the exception of niobium.
The highest material needs relative to production are for
niobium and molybdenum. Molybdenum (as well as chromium) is
classed as having notably poor substitutability in thermal steels
[54]. Molybdenum availability could be an issue because the
reserve life of Mo is 43 years, while the resource base is not
significantly larger than the reserves. The importance of niobium
(for molten salt pipes in the tower design) is however debatable. It
can be omitted if other steels are used, possibly in combination
with coatings to protect the steel surface from corrosion (for non
heat transfer applications).
Although a large number of alloys may be substituted for one
another, it would be difficult to fully substitute high temperature
steel alloys in general by either using other materials or by less
material intense designs. Given the uncertain net impact on
demand for these materials, the generally low requirements rela-
tive to production and the long timescale involved, steel alloy
elements are unlikely to be a constraining issue for CSP. Molyb-
denum is the most likely candidate should issues arise.

6.3. Silver

The use of silver in mirrors for CSP requires a closer look as it is

difficult to substitute, would constitute a large new demand for
silver and the metal is potentially constrained in rate and available
Fig. 6. Ratio of material needs to material production, yearly, for various materials in reserves. As Fig. 7 shows, there has been a significant supply deficit
central tower plants. According to growth scenario (cf 2.1). in terms of the difference between silver mining and fabrication
952 E. Pihl et al. / Energy 44 (2012) 944e954

adjusted somewhat and calcium salts are being tested as a possible

component/substitute, but for a given salt composition, there is an
almost linear relationship between the desired amount of thermal
storage and the necessary mass of salt. If thermal storage is desired
for CSP, there is therefore no foreseeable possibility to substitute
this demand, with the technologies analysed in this work. The
molten salt use per unit energy can be reduced by employing
mixtures that store heat by phase change [60] or substituted by
concrete storages. The large availability of the primary elements
involved indicates that there should be a good scope for a supply
response for molten salts, but it is difficult to say at what speed and
cost.

7. Discussion and implications

Fig. 7. Silver supply and demand [55,56]. The results of this work should be viewed as order of magnitude
estimates of material availability for large diffusion of CSP tech-
nologies. Thus, rather than exact predictions and forecasts, it is
demand for more than a decade. This deficit has been filled by
useful to check for ‘show-stoppers’ and for judging the worst case
drawing down government stockpiles and by recycling scrap and
backdrop situation, since it covers neither expected efficiency
jewellery.
increases in power plants nor increases in ultimate reserves and
The dwindling use of silver in photography has been offset by
resources from improved technology, increased prices or new
the increase in electronic, photovoltaic, medical and nanomaterial
discoveries.
demand, applications which have a high ability to pay for silver and
In terms of the plants, the main sources of error in this work are
which do not typically result in a recyclable stock [55,56] The
lacking or insufficient data for some components and in data as
industrial demand for silver is thus very competitive at present, and
basis for the estimates for scaling of thermal components. It is also
represents 55% of fabrication demand and 75% of mine supply. The
an open question as to what degree the chosen power plant designs
remaining demand is for new jewellery, coins, silverware and
are representative for the total potential stock. One specific diffi-
bullion.
culty is assessment of the use of materials such as copper,
Diminishing recycled silver supplies may be difficult to
aluminium and rare metals in turbines, generators and electric
compensate through mining; roughly two thirds of silver produc-
motors. These are typically complex components and precise data
tion occurs as a by-product of mining other base metals, predom-
on metal composition is not readily available. It is unlikely that
inantly copper and lead [56]. Furthermore, silver has been mined
motors and generators contain more copper than the wiring in the
for thousands of years, and there is not a large potential for new
collector fields (for which there is good data) viewing the weights
primary silver mines. This situation implies that the mine supply
of these components compared to the sheer weight of the wiring.
response to higher prices would be muted for silver. There is thus
This notion has been confirmed by an expert from industry.
a potential for a large increase in price that all prospective silver
The long run availability of resources is not something that can
users should be considering in strategic plans.
be known with certainty. Tilton [61] gives a comprehensive
Reducing silver use for mirrors is a difficult challenge since it is
overview of the factors and reasons underlying this issue. The crux
already applied in extremely thin layers of about 100 nm. Alter-
of the matter is that the supply-demand balance of materials is
native materials for reflective coating have been investigated, but
a complex process, with many feedbacks on both sides. ‘Reserve’
none offer the same broadband reflection qualities [57]. The silver
and ‘resource’ figures will change over time based on the price
layer thickness could possibly be slightly reduced but there are
evolution of the material, the development of extraction tech-
durability and manufacturing issues strongly prohibiting layers
nology and the success of exploration efforts in uncovering
thinner than about 50 nm [58]. A possible substitute is to instead
heretofore unknown deposits. Conversely, high prices or
use aluminium as reflective layer, on an aluminium substrate with
impending shortages of a commodity will induce efforts at mini-
a covering layer of oxides or polymer to protect from corrosion.
mization or substitution, and technological change can alter
Changing from silver to aluminium reflectors typically decreases
demand exogenously (e.g. through the emergence of a competitor
the maximum reflectivity from w95% to w90% [59]. This decrease
technology).
could be compensated by scaling up the reflector area which would
increase the use of other less constrained materials and degrade the
8. Conclusions
plant economics, but would not rule out feasibility. As silver is
a small component of cost, the silver price would have to increase
This study has compared the material demands for construction
by multiples to make the increased reflector area needed for
and large-scale application of existing or near-term Concentrating
aluminium mirrors a cost-effective substitution.
Solar Power (CSP) technology with the present available production
and reserves of those materials. In conclusion, there is no material
6.4. Thermal salts that sets any relevant limit on how much of the two studied CSP
types could be built in the foreseeable future, based on available
The use of thermal salts for storage would represent a significant reserves. The reserves of every required material are many times
new demand for the nitrate compounds and is difficult to substi- higher than what would be required to substitute all electricity
tute. The material requirement would exceed the total current generation with CSP. Current usage patterns would indicate that
production capacities for KNO3 as well as NaNO3 at roughly 500 and the current reserves of zinc, chromium, silver and copper will be
1000 TWh/yr capacity addition for the trough and tower plants consumed with or without CSP by mid-century. The possibility of
respectively. In an aggressive adoption scenario, this could occur replacing these reserves for silver is unclear. The material flows in
before mid-century. The relative proportion of K and Na salts can be operation, with the exception of water, are negligible compared to
 E. Pihl et al. / Energy 44 (2012) 944e954 953

the construction phase. The impact of water use is a complex [13] Gruber PW, Medina PA, Keoleian GA, Kesler SE, Everson MP, Wallington TJ.
Global lithium availability: a constraint for electric vehicles? Journal of
question and highly specific to the particular region, it has been
Industrial Ecology 2011;15:760e75.
outside the scope of this study. [14] Kushnir D, Sandén BA. The time dimension and lithium resource constraints
The production rates of most materials are also sufficient to for electric vehicles. Resources Policy 2012;37:93e103.
enable a massive growth rate in installed CSP capacity. Despite the [15] Andersson BA. Materials availability for large-scale thin-film photovoltaics,
progress in photovoltaics: research and applications, vol. 8; 2000. pp. 61e76.
fact that there is little issue of material scarcity in an absolute sense, [16] Andersson BA, Azar C, Holmberg J, Karlsson S. Material constraints for thin-
some components and materials required for large-scale CSP film solar cells. Energy 1998;23:407e11.
adoption will stretch current production and manufacturing [17] Candelise C, Spiers JF, Gross RJK. Materials availability for thin film (TF) PV
technologies development: a real concern? Renewable and Sustainable
capacities. Nitrate salts and silver, both crucial in the present design Energy Reviews 2011;15:4972e81.
of CSP plants, could face supply shortage and increased prices in the [18] Feltrin A, Freundlich A. Material considerations for terawatt level deployment
coming few decades. With the growth scenario applied in this of photovoltaics. Renewable Energy 2008;33:180e5.
[19] Green MA. Price and supply constraints on Te and in photovoltaics. In:
work, CSP plants could be consuming 15e35% of the global nitrate Conference record of the IEEE photovoltaic specialists conference; 2010. p.
production within two decades. The silver use in CSP plants could 550e5.
be problematic both in the short term because of higher costs due [20] Green MA. Ag requirements for silicon wafer-based solar cells, progress in
photovoltaics: research and applications, vol. 19; 2011. pp. 911e916.
to a potential supply shortage that could arise irrespective of CSP [21] Wadia C, Alivisatos AP, Kammen DM. Materials availability expands the
growth, and in the long term because of reserve limitations. Issues opportunity for large-scale photovoltaics deployment. Environmental Science
with glass and some steel alloys would be comparatively minor and and Technology 2009;43:2072e7.
[22] Burkhardt JJ, Heath GA, Turchi CS. Life cycle assessment of a parabolic trough
are not likely to arise for several decades, if at all. Finally, these
concentrating solar power plant and the impacts of key design alternatives.
factors will only apply during a rapid increase in installed capacity. Environmental Science & Technology 2011;45:2457e64.
Operational material flows are negligible and the material flows [23] Lechón Y, de la Rua C, Saez R. Life cycle environmental impacts of electricity
production by solar thermal power plants in Spain. Journal of Solar Energy
available from decommissioning older units will be largely recy-
Engineering 2008;130:021012e7.
clable. Over the medium and long term, as CSP capacity saturates, [24] Viebahn P, Kronshage S, Trieb F, Lechon Y. NEEDS RS 1a e D12.2 final report
the net amount of new material required will decrease towards that on technical data, costs, and life cycle inventories of solar thermal power
required to make up for losses in recycling. plants. DLR/CIEMAT/Sixth Framework Programme; 2008.
[25] May N. Eco-balance of a solar electricity transmission from north Africa to
For silver, there is some potential for substitution, although with Europe. Diploma thesis. Technical University of Braunschweig and German
negative effects on plant costs. Silver use can already be substituted Aerospace Center (DLR), 2005.
by aluminized reflectors but with a corresponding about 5% [26] Weinrebe G. Technical, ecological and economic analysis of solar thermal
power towers. University of Stuttgart; 1999.
decrease in collector field efficiency. There are potential substitutes [27] García-Olivares A, Ballabrera-Poy J, García-Ladona E, Turiel A. A global
for the steel alloys used, by selecting other steel types or through renewable mix with proven technologies and common materials. Energy
using coatings instead to protect steel from corrosion. The nitrate Policy 2012;41:561e74.
[28] Trieb F, Schillings C, Pregger T, O’Sullivan M. Solar electricity imports from the
salts have little potential for substitution but have a large potential middle east and north Africa to Europe. Energy Policy 2012;42:341e53.
for a supply response. [29] USGS. Principles of a resource/reserve classification for minerals. Geological
With technology improvements, increase in extraction and Survey Circular 1980;831.
[30] Greenpeace, SolarPACES, and ESTELA. Concentrating solar power global
production capacity of strategic materials and research into
outlook 09; 2009.
substitution, materials availability will not hinder CSP from [31] IEA. Technology roadmap concentrating solar power. Paris: International
replacing all of today’s fossil fuelled electricity generation. The most Energy Agency; 2010.
important challenges concerning materials for CSP in the coming [32] Pacala S, Socolow R. Stabilization wedges: solving the climate problem for the
next 50 years with current technologies. Science 13 august 2004;305:968e72.
decades, will be to scale up nitrate salt production and develop [33] Pihl E, Heyne S, Thunman H, Johnsson F. Highly efficient electricity generation
good substitutes for silver in reflective surfaces. These issues may from biomass by integration and hybridization with combined cycle gas
affect the cost of CSP on the margin, but are not particularly severe. turbine (CCGT) plants for natural gas. Energy 2010;35:4042e52.
[34] Trieb F, Schillings C, O’Sullivan M, Pregger T, Hoyer-Klick C. Global potential of
concentrating solar power, presented at the SolarPACES 2009. Berlin; 2009.
[35] Kearney A. Solar thermal electricity 2025. ESTELA; June 2010.
References [36] Doyle Gutiérrez L, Cobra Energía. Personal communication, e-mail. 2011.
[37] Akyol S, Ahrens S, Jahr F, Rehberger C, Lüpfert E. Cost impact model for using
[1] Bhandari R, Stadler I. Grid parity analysis of solar photovoltaic systems in polymer film based lightweight mirror construction in CSP plant. presented at
Germany using experience curves. Solar Energy 2009;83:1634e44. the SolarPACES 2010, Perpignan, France; 2010.
[2] EASAC. Concentrating solar power: its potential contribution to a sustainable [38] Kearney DW. Parabolic trough collector overview, presented at the parabolic
energy future. European Academies Science Advisory Council; 2011. trough workshop 2007. Golden, CO, USA: NREL; 2007.
[3] Hernández-Moro J, Martínez-Duart JM. CSP electricity cost evolution and grid [39] Herrmann U, Kelly B, Price H. Two-tank molten salt storage for parabolic
parities based on the IEA roadmaps. Energy Policy; 2011. trough solar power plants. Energy 2004;29:883e93.
[4] Lund PD. Boosting new renewable technologies towards grid parity e [40] eSolar. Personal communication (undisclosed person), mail. 2011.
economic and policy aspects. Renewable Energy 2011;36:2776e84. [41] Bloch HP. A practical guide to steam turbine technology. McGraw-Hill; 1995.
[5] CSPToday. CSP world plant locations, CSPToday.com; 2011. [42] Klueh RL, Harries DR. MONO3 high-chromium ferritic and martensitic steels
[6] CSPToday. India map 2012, http://www.csptoday.com/india/pdf/CSP_ for nuclear applications. Ch 2: development of high (7e12%) chromium
IndiaMap_V2.pdf; 2012. martensitic steels. ASTM Standards and Engineering Digital Library; 2001.
[7] Iberdrola. Iberdrola engineering starts up the Kuraymat power plant in Egypt, [43] Moore R, Vernon M, Ho CK, Siegel NP, Kolb GJ. Design considerations for
with a total capacity of 150 megawatts e press release. Available from: http:// concentrating solar power tower systems employing molten salt. Albu-
www.iberdrola.es/webibd/corporativa/iberdrola? querque, New Mexico and Livermore, California: Sandia National Laborato-
IDPAG¼ENMODPRENNAC11&URLPAG¼/gc/prod/en/comunicacion/ ries; 2010.
notasprensa/110705_NP_01_Kuraymat.html; 2011, 05 Jul. [44] GEA heat exchangers. Personal communication, e-mail. 2011.
[8] Jacobson MZ, Delucchi MA. Providing all global energy with wind, water, and [45] García-Sobrinos G, Salvador-Villà I, Serradilla-Echarri J. Tower of power.
solar power, part I: technologies, energy resources, quantities and areas of Available from: http://www.asce.org/Content.aspx?id¼28974&css; 2010.
infrastructure, and materials. Energy Policy 2011;39:1154e69. [46] Flagsol-gmbh. Important progress in construction of andasol 3 solar power
[9] Kleijn R, Van Der Voet E. Resource constraints in a hydrogen economy based plant: turbine and generator installed. Available from: http://www.flagsol-
on renewable energy sources: an exploration. Renewable and Sustainable gmbh.com/flagsol/company/news/2010_09_20.html; 2010, Oct 14.
Energy Reviews 2010;14:2784e95. [47] Kennedy CE. Advances in concentrating solar power collectors: mirrors and
[10] Kleijn R, Voet E v d, Kramer GJ, Oers L v, Giesen C v d. Metal requirements of solar selective coatings. In. Presented at the 21st international vacuum web
low-carbon power generation. Energy 2011;36:5640e8. coating conference. Scottsdale, Arizona; 2007.
[11] Lund PD. Upfront resource requirements for large-scale exploitation schemes [48] Nucor-Fastener. Hot dip galvanizing e technical data sheet, http://www.
of new renewable technologies. Renewable Energy 2007;32:442e58. nucor-fastener.com/Files/PDFs/TechDataSheets/TDS_007_Hot_Dip_
[12] Andersson BA, Råde I. Metal resource constraints for electric-vehicle batteries. Galvanizing.pdf; 2011.
Transportation Research Part D 2001;6:297e324. [49] Schott. Schott PTR 70 receiver e setting the benchmark. Schott Solar; 2011.
954 E. Pihl et al. / Energy 44 (2012) 944e954

[50] M-WOITE. DIN 1.4541 steel specification, http://www.m-woite.de/en/ [57] Jaworske DA. Reflectivity of silver and silver-coated substrates from 25degC to
materials/14541.shtml; 2011. 800degC [for solar collectors]. In: Energy conversion engineering conference,
[51] Mar RW, Swearengen JC. Material issues in solar thermal energy systems. 1997. IECEC-97, proceedings of the 32nd intersociety, vol. 1; 1997. p. 407e11.
Solar Energy Materials 1981;5. [58] Peters PN, Sisk RC, Brown Y, Gregory JC, Nag PK, Christl L. Measurements of
[52] USGS (various authors). Minerals commodity summaries. U.S. Geological the optical properties of thin films of silver and silver oxide. Marshall Space
Survey National Minerals Information Center; 2011. Flight Center, NASA; Feb 1995.
[53] IEA World. Energy outlook 2011. Paris: International Energy Agency; [59] Fend T, Hoffschmidt B, Jorgensen G, Küster H, Krüger D, Pitz-Paal R, et al.
2011. Comparative assessment of solar concentrator materials. Solar Energy 2003;
[54] Morley N, Eatherly D. Ensuring material availability for the UK economy. 74:149e55.
London: E. Agency/BERR; 2008. [60] Oró E, Gil A, de Gracia A, Boer D, Cabeza LF. Comparative life cycle assessment
[55] Klapwijk P, Walker P, Ryan P, Newman P, Meader N, Tankard W, et al. World of thermal energy storage systems for solar power plants. Renewable Energy
silver survey 2010. Washington, DC: The Silver Institute; 2010. 2012;44:166e73.
[56] Hilliard HE. USGS mineral commodity summaries e silver. U.S. Geological [61] Tilton JE. On borrowed time? Assessing the threat of mineral depletion.
Survey National Minerals Information Center; 2011. Washington, D.C: RFF Press; 2003.