Sunteți pe pagina 1din 11

Resources Policy 37 (2012) 93–103

Contents lists available at SciVerse ScienceDirect

Resources Policy
journal homepage:

The time dimension and lithium resource constraints for electric vehicles
Duncan Kushnir n, Björn A. Sandén
Environmental Systems Analysis, Energy and Environment, Chalmers University of Technology, Göteborg 412 96, Sweden

a r t i c l e i n f o abstract

Article history: The availability of lithium resources for a transition to electric vehicles is a vital topic for transport
Received 1 July 2011 technology strategy. Recent debate seems to have concluded that there is ‘sufficient’ lithium available,
Received in revised form but for the purposes of a technological transition, time matters. It is not simply the quantity of resource
29 November 2011
that is relevant—the flow rate into society may be a much more difficult constraint and transient events
Accepted 29 November 2011
Available online 13 January 2012
have disrupted heavily concentrated material supply chains in the past. Furthermore, critical assump-
tions such as the presence of recycling systems may not be justified without policy support.
JEL classification: Complacency is therefore not an appropriate stance for a robust evaluation of material risks in the
Q30 case of lithium.
& 2011 Elsevier Ltd. All rights reserved.

Introduction opportunity cost of obtaining a resource and that the trend of

this cost is unpredictable. An excellent description of this debate
Issues of resource security and the end of cheap oil have sown for resources in general is put forward by Tilton (2003).
the seeds for a discontinuous change in transportation vehicles Much research and data show that scepticism is appropriate
and fuels. While there are many competing visions regarding towards claims of physical limits of resource availability; no case
what shape future transport systems can and should take, one of general mineral depletion has existed as of yet (Simpson et al.,
prominent narrative is the adoption of plug-in hybrids as the first 2005). Warnings of mineral scarcity have proven wrong time and
step toward a future of electric vehicles. The assumed technolo- time again because new discoveries and improving technology
gical core of these systems is the lithium ion battery, which have outstripped the depletion of high grade resources. The
possesses clear performance advantages versus other battery empirical proof for this is that the real cost of most mineral
technologies at present for plug in hybrid (PHEV) and full battery commodities has thus far been falling over time, while the
electric (BEV) vehicles (Khaligh and Zhihao, 2010). physical flow through society has increased (Govett and Govett,
One of the strongest arguments for using electricity as a fuel is 1978; Slade, 1982; Tilton, 2003). The theoretical underpinnings of
the promise of decoupling the production of transport energy this trend are very relevant to policy makers and have been
from its end use; electric vehicles allow the maximum flexibility recently reviewed (Svedberg and Tilton, 2006) and defended
in choosing appropriate sources and mixes of energy for the (Tilton and Lagos, 2007). Despite this, given the importance of
transport sector. Whilst this is correct in principle, policies aimed lithium to the vision of electric transport and of electric transport
at or hoping for a transformation of the transport sector towards to some visions of sustainability, it is vital to not remain
electricity presuppose that batteries can be supplied in overall complacent and dismiss resource availability issues out of hand.
quantity sufficient for the scale of the project and at a rate that Discussions regarding the uncertain future of lithium avail-
permits a sufficiently rapid change in the stock of vehicles. This ability have taken place for quite some time. With batteries now
has not gone unnoticed, yet there are divergent views on how looming as another vast new demand for lithium, this debate has
best to assess resource adequacy in the long run; one begins from recently resurfaced. The scale of material use implied by some
the truth that the planet and thus resources are finite and another scenarios for mass production of vehicle batteries is sufficiently
holds that a better (only) way to measure scarcity is the large that resource scarcity in the medium and long term cannot
be ruled out outright for a number of potential battery metals
(Andersson and Råde, 2001; Gaines and Nelson, 2009). With
Corresponding author. Tel.: þ46 31 772 1197. regards to lithium, generally accepted ideas of the magnitude of
E-mail address: (D. Kushnir). the resource seem to allow for a sizeable fleet of electric vehicles

0301-4207/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
94 D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103

Fig. 1. A general material flow model for lithium.

(Evans, 1978, 2008a, 2008b; Will, 1996; Garrett, 2004; Gaines and (2008b) and Yaksic and Tilton (2009) is approximately correct
Nelson, 2009; Yaksic and Tilton, 2009; Gruber et al., 2011). Many and that 25 million tons of lithium metal are viable (our word,
of the required resources are likely available without extreme discussed in Section 3) to recover from known resources plus
cost or new technology, yet a closer look will show that visions of another 5 million tons in marginal stocks, which we assume will
US or even Western European style mobility or vehicles with large be producible at prices that batteries will support. Beyond this,
batteries will test the boundaries of these assessments. new discoveries or ocean extraction will be needed for further
Furthermore, it is not sufficient to conclude that there is supply.
‘enough’ lithium for a given scenario because there are other
factors that complicate the issue with regard to lithium and
which could ultimately be more relevant than the size of the Societal stocks and flows—estimating the demand for virgin
resource or its projected cost of extraction. The rate of extraction lithium
needed to build up a large societal stock over a given time period
(Andersson and Råde, 2001; Tahil, 2008) has implications, and We begin with the assumption that electric vehicles based on
historical cases featuring resources with similarly concentrated lithium batteries will become heavily adopted for transport in the
supplies, such as cobalt, give reason to believe that institutional future. The first level of comparison in our framework is thus to
inefficiency can be a major mechanism driving transitory scarcity define what this means in terms of a societal stock. This will
(Alonso et al., 2007). While there is not necessarily reason for enable comparison with total and annual availability of virgin
alarm over lithium supplies, there are risks and policy decisions resources in following sections.
that can affect the outcome. A strategic outlook on material
availability demands explicit consideration of these risks and Stocks and flows related to current lithium applications
implications thereof, and thus it is our hope to continue the
discussion on this important topic. Availability of historically extracted lithium
Historical cumulative extraction (1940–2010) is estimated at
about 0.5 Mt obtained by updating the 0.32 Mt that Andersson
Analytic framework and paper layout and Råde obtained for the cumulative extraction in 1999 with the
annual mine production figures in the USGS lithium series for
We use a somewhat stylized model to demonstrate key factors 1999–2011 (Andersson and Råde, 2001; Jaskula, 2006, 2011a,
for the future of lithium. Fig. 1 illustrates the general stocks and 2011b). This is a small amount compared to projected lithium
flows of lithium as well as showing the layout of the argument demand and to resources still in the ground (indicating that
presented here. The flows begin with the conceptual movement of lithium is a fairly new metal in industrial society). The amount
a given bit of lithium from the resource base into the ‘pseudo- contained in lithium batteries is also currently very small com-
stocks’1 of resources and subsequently reserves from which they pared to any quantities projected for vehicles. Moreover, due to
are extracted. As reserves are extracted into society they can be the many dissipative forms of lithium use and the limited
used in ways that result in the lithium being dispersed (e.g. utilization of recycling we assume that this potential resource
grease), or they can be used in ways that form recoverable stocks can be neglected and that there is effectively zero recoverable
that can later offset part of the supply flow into society if lithium stock in society in 2011.
economics and technology allow. All such flows will be mediated
and routed over time by a set of transient processes (institutional Expected use and dispersion of lithium in other applications
forces). Current production of lithium is roughly 25 kt/yr as metal
We begin in Section 2, by outlining possible global mobility equivalent. As shown in Table 1, lithium has a variety of uses, but
scenarios and then work backwards from the implied vehicle many are experiencing only modest growth compared to the total
battery stocks to determine the necessary flows into society and or to that of batteries. It is very uncertain how other lithium
how they relate to the resources. The discussion on implications is applications will react to a large new demand that possibly
structured by the time dimension; Sections 3–5 discuss possible pushes prices up. To this end, no specific treatment of the
constraints at the stock, flow and transient level. Our base elasticity of other lithium applications to price signals exists to
assumption, discussed in Section 3, is that the work of Evans the authors’ knowledge. If the cumulative availability curve is
sound and prices stay near where they are, then other applica-
So called because it is a conceptual subset of the total resource stock, not tions could continue on their current trajectories, which are
physically distinct. generally growing with the exception of lithium used in some
D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103 95

Table 1
Assumptions for non-vehicle lithium applications (Yaksic and Tilton, 2009; Jaskula, 2011a, 2011b).

End use Recycled Growth scenario

Non-vehicle secondary batteries Yes (3 years) 15% growth until 2018, 10% next decade, 3% to 2050 then 1%
Non-vehicle primary batteries Yes (1 year) 8% growth until 2018, 5% next decade, 3% to 2050 then 1%
Pharmaceuticals No 4% growth until 2018, 2% next decade, then 1%
Glass/ceramics No 3% growth until 2018, 2% next decade, then 1%
Rubber/polymers No
Lubricants No 5% growth until 2018, 3% next two decades, then 1%
Aerospace alloys No 4% growth until 2018, 2% next decade, then 1%
Other No

Fig. 2. Vehicle population and electrification scenarios. The dotted line shows the percentage of new vehicles with a battery component reaching 95% by mid-century. The
solid lines show the evolution of vehicle population under our three transport adoption scenarios.

aluminium production techniques. Assuming that they will con- stabilization level of global per capita vehicle intensity from the
tinue growing would represent the ‘worst case’ scenario for 0.12 cars per capita that exist today (Fig. 2):
lithium demand.
We do not include non-vehicle battery demands for lithium in – High: Vehicle population stabilizes at Western European levels
the calculations presented in the rest of this paper. The effect of (0.5 cars/capita)
the assumptions in Table 1 would be to disperse just under 3 Mt – Medium: Vehicle population stabilizes at 0.33 cars/capita.
of lithium by the end of the century. In terms of stock and flow Taken to be the same as Yaksic and Tilton (2009) for
this amounts to less than 10% of our result for either variable: the comparability.
conclusion is that other applications of lithium certainly have a – Low: Vehicle population stabilizes at 0.2 cars/capita, repre-
role to play on the margin, but will not affect the viability of senting some coordinated efforts to limit the global car
lithium batteries one way or the other. population.

EV lithium demand scenarios

Electric vehicle adoption
From Section 2.1, the only societal stock of lithium that we The rate of either hybrid or full electric drivetrain adoption is
consider is that that will exist in car lithium batteries, either in assumed to rise along a logistic curve that reaches 95% by 2050
use or discarded and potentially recyclable. Knowing the precise (Fig. 2). This is an arbitrary assumption, but provides reasonable
form that battery technology will take is not possible, but an comparison with other studies.
assumption regarding lithium intensity is necessary (Sections
2.2.3 and 2.2.4). Combining this intensity with the scenarios for
Battery capacity per vehicle and battery lifetime
vehicle adoption discussed in Sections 2.2.1 and 2.2.2 gives rise to
We present scenarios based on two battery sizes here: a 9 kWh
the implication for societal stocks of lithium in PHEV or BEV type
battery for a ‘PHEV’ scenario and a 36 kWh battery for a ‘BEV’
scenario. The number is intended to represent an ‘average’ battery
size across all vehicles entering the fleet.
Vehicle population We use an average battery lifetime of 10 years in our model,
Our population scenario is taken from the UN baseline sce- and 100% collection for recycling. The most sensitive result to this
nario, stabilizing at 9.3 billion (UN, 2008). We do not consider assumption is the resource flow peak (Section 4). A two-year
vehicles other than cars, such as electric bicycles and scooters. We longer battery lifetime (or a delay in recycling stockpiles) can
put forward three potential scenarios for the ultimate exacerbate the peak required resource flow by more than 10%.
96 D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103

Table 2
Comparison of battery lithium intensity with other studies.

Battery High Low Sources

chemistry intensity intensity
(g/kWh) (g/kWh)

We use 160 160 Estimation based on Fig. 3

LiNiCo 300 155 Gaines (based on production
LiFePO4 batteries at Argonne)
LiNiCo-type 140 110 (Andersson and Råde, 2001)
140 (Yaksic and Tilton, 2009)
LiCo, LiNi, LiMn 114 114 (Gruber et al., 2011)

Fig. 3. Lithium content of several battery chemistries.

Lithium battery material intensity

There is a minimum amount of lithium required for a given
battery capacity that is derived from the battery chemistry and
cell voltage. Although technology will continue improving, the
technically demonstrated and commercially viable lithium inten-
sities of each chemistry will remain higher than these funda-
mental limits. Purpose matters as well, BEV and PHEV batteries
will be either constructed differently or use differing materials
appropriate to the task and will thus be different with respect to
lithium content per kWh.
Fig. 3 shows the current use and range of lithium requirements
for a number of different lithium batteries (Kushnir and Sandén,
2011) An estimate of 200 g of lithium or roughly 1 kg of lithium
carbonate equivalent2 (LCE), per kWh of battery capacity is a Fig. 4. Cumulative lithium demand from vehicle scenarios with and without
reasonable approximation of current designs for BEV batteries. recycling. The figure shows the cumulative virgin lithium demand in million tons
(Mt), for PHEV (left scale) and BEV (right scale) adoption. The dashed lines indicate
With progress, 160 g of lithium per kWh may be a reasonable the 30 Mt resource used in this paper.
medium term estimate, and this is what we use here as an
average across all technologies. This could be compared
(Table 2) with 140 g per kWh used by Andersson and Råde are assumed, the left axis applies, while the right axis applies to
(2001) as well as Yaksic and Tilton (2009),3 or 250 or up to the BEV scenario. Assuming 80% recycling, the cumulative
400 g for current high power designs (Gaines and Nelson, 2009). If demand is almost twice as large as the lithium in the vehicle
lithium does become scarce, it may be possible to use as little as stock (100% recycling) by the end of the century. No recycling at
120 g per kWh by substituting sodium for a part of the lithium, all would multiply the cumulative demand over the century by a
although the resulting battery chemistry does not have the same factor of four compared to 100% recycling.
performance (Whittingham, 2004). It is thus both logical and understandable that virtually all
No additional processing losses in production are assumed; studies on this issue assume recycling to take place; in virtually
although there certainly will be some, they are likely to be very any scenario recycling will represent the largest cumulative
small in comparison with other flows (Andersson and Råde, source of lithium by the second half of the 21st century. Despite
2001), or indeed the uncertainty on the resource recovery factors. being universally assumed to occur, recycling is by no means a
The other key observation is that raw lithium contributes very given. The cost of recycled lithium is difficult to determine
little (between 1% and 2%) to the current cost of a battery. As it is because current battery recycling processes are aimed at recover-
the only critical ingredient, the willingness of battery applications ing many different materials, cobalt in particular. Recycling
to pay for it is presumably extremely high, and assessment of economics for lithium are currently not good4 and consequently
which resources will be available to society over the longer term recycling is only done in limited quantities as a by-product of
must be on generous terms. Furthermore, it seems reasonable recovering such more expensive materials from batteries or for
that batteries could outbid some existing applications of lithium if the purpose of minimizing toxic hazard from waste.
supply was tight. Nevertheless, not any price can be paid. Although large automotive batteries are expected to be tech-
nically easier to recycle than electronic equipment batteries and
Implied cumulative and annual demand for lithium in vehicle economies of scale in recycling will emerge, there are also reasons
batteries—the critical role of recycling to believe that the economics of recycling lithium batteries may
Fig. 4 shows the implied cumulative demand curves arising degrade instead, despite these points. Many new chemistries for
from the vehicle scenarios and assumed material intensity. Curves lithium batteries are being tested or in use for their unique
for three levels of recycling, 0%, 80% and 100%, are plotted. If PHEV chemical virtues. If future battery technologies minimize cobalt

2 4
1 unit lithium metal¼ 5.32 units lithium carbonate equivalent (LCE). Toxcos process cost was initially about $10/kg of battery but is estimated to
6.75 kg LCE for a 9 kWh battery (Yaksic and Tilton, 2009). be half that now (Gaines and Nelson, 2009).
D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103 97

Table 3
Costs and recoveries of materials from batteries.

Component Material—recovery Value Recovery Recovered value Recovered value—future?

(per kg battery) (g)

Metal casing Steel—100% $0.80/kg 20 $0.02 $0.02

Cathode/anode Cobalt—90% $17.50/kg 19 $0.33 $0.00
Lithium—90% $6.20/kg 18 $0.12 $0.12
Electrolyte LiPF6—70% $20.00/kg 11 $0.22 $0.22
Copper foils Copper—95% $4.00/kg (scrap) 105 $0.42 $0.42
Compare to $5/kg recycling cost at present $1.05 $0.72

Material compositions are taken from Kushnir and Sandén (2011). Prices for metals from (June, 2011). Electrolyte an estimate from lithium supplier
websites. Efficiency from text. The purpose is to illustrate how recycling economics could degrade.

use (or other expensive metals such as nickel), the apparent need price that has yet existed. Ocean refers to the very large quantity
for as well as the economic viability of recycling processes could potentially available from the ocean.
in fact decrease, despite improving technology. Furthermore, as The first 26 Mt of lithium are in our viable category. If we
the scale of production of the electrolyte increases, another source include marginal resources we reach 30 Mt. This abundance in
of recovery value will erode. As an example, a previous assess- comparison with demand as well as the reasonably certain
ment from Argonne in 2001 listed LiPF6 at $60/kg. Recycling production price of much of the curve and slow increase in cost
technology and systems are attempting to hit a moving target for the marginal unit of supply has been used to argue that ‘‘the
that is being driven by the need to make the batteries themselves shape of the lithium cumulative availability curve indicates that
cheaper (Table 3). depletion is not likely to pose a serious problem over the rest of
There is some data available for estimating material efficiency this century and well beyond’’ (Yaksic and Tilton, 2009). Similarly,
of recycling systems. Primary lithium batteries are currently only Gruber et al. (2011) argue that the resource is about 39 Mt (in-
handled by specialist companies (e.g. Toxco in North America) situ) and the maximum use until 2100 is roughly 20 Mt. These
that use cryogenic processes to minimize the explosive/fire conclusions are reassuring but also implicitly assume that recy-
hazard. Some commercial secondary battery recyclers with cling will be used and that the so-called marginal lithium and
lithium recovery capability exist, and there are also many even perhaps oceanic lithium will be available, conditions which
national research initiatives to implement advanced battery may not necessarily be realized (see Fig. 4 where the 30 Mt of
recycling. One example is the EU VALIBAT program,5 which viable and marginal resources are compared to cumulative
resulted in a process to recycle 95% of magnetic metallic com- demand).
pounds, 90% of lithium from the electrode material, 90% of metal In any case, the lithium to form the societal stock must initially
oxides and 70% of lithium salt. Cobalt recovery can be upwards of 93% come from resources currently in the ground, and which
(Xu et al., 2010), but these targets may be high for lithium recovery resources will be available is important for analysis. Currently
compared to commercial processes in use today (Bernardes et al., most mineral lithium is produced in a form that is used directly
2004; Dewulf et al., 2009). Over the medium term, however, they are for a specific end use (e.g. glass). This fact has been used to argue
certainly achievable and the 80% lithium recovery that is assumed in that not all sources are suitable for batteries (Tahil, 2008), but this
many studies seems to be a decent estimate of what can be achieved seems to be false. Although the final purification processes for
with purities sufficient for re-use in batteries (Xu et al., 2008). The many sources are uncertain, it should be assumed that all lithium
other key determinant of available recycled lithium is of course the outputs can be converted to lithium carbonate (and suitable
institutional recycling efficiency, e.g. the proportion of recyclable thereafter for purified battery materials) at a modest cost, even
artefacts actually collected. We assume 100%, but one reasonable for resources considered as marginal. This is true for recycled
estimate could be the 95% achieved for lead batteries in the US near sources (Xu et al., 2008), brines (Garrett, 2004; Boryta et al.,
the beginning of the millennium (Socolow and Thomas, 1997) or 96% 2006), and minerals (Holger et al., 2000; Garrett, 2004) and
today (BCI, 2009). asserted to be true (Ajie, 2010) although not demonstrated for
clays. As mentioned, the main future demand market will be able
to tolerate higher prices than exist today and therefore purifica-
tion and transformation economics may affect the preference, but
likely not the long run viability of mineral and brine resources
Lithium resource stocks and potential limits
beyond their individual merits.
The economic availability of the resources that we will need
Terrestrial lithium resources
for electric vehicle fleets is most eloquently captured with the
concept of the cumulative availability curve (as introduced by
Because of the heretofore arbitrary nature of the viable and
Tilton, 2003 and refined for the case of lithium in Yaksic and
marginal categories, it may be better to lump them simply as
Tilton, 2009). The cumulative availability curve refers to the
terrestrial, which can be further divided into brine (24.5 Mt) and
amount of resources available with current technology at a given
mineral resources (5.5 Mt). Economics of extraction seems to support
extraction price and is shown for lithium in Fig. 5. The graph is
the contention that the majority of the resources identified this far
divided into three sections denoted viable, marginal and ocean.
could be extracted with reasonably conceived technology, but the
‘Viable’ resources are defined as those where the midpoint of
estimates presented here for recovery may be optimistic. It is also
estimates for production cost are below the peak price of $6.4/kg
possible and even likely that new discoveries and technology will
lithium carbonate that occurred in 2008. Marginal resources are
continue increasing the resource. There is currently a wave of lithium
known resources with higher production price estimates than any
exploration activity that is defining costs for known resources and
identifying and examining the feasibility of bringing new ones not
EU program ID: G1RD-CT-2000-00232. counted here into production.
98 D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103

Fig. 5. Cumulative availability curve for lithium. Quantities are displayed on a recoverable basis. Based on Yaksic and Tilton (2009) and Evans (2008b).

The general method of production from brine resources is to resource expansion. More discussion of lithium resources can be
pump the brine out and then to subsequently concentrate it found in Evans (2008), Yaksic and Tilton (2009) and Gruber et al.
through evaporation. The major operations in the Andes and (2011).
Tibet, which together comprise the vast majority of the resource,
use solar evaporation for this initial step. Because of this, such
Oceanic lithium stocks as a backstop technology
brine resources are cheap and energetically efficient to extract,
and will be generally insensitive to input costs. The Salar de
The availability of a viable process to extract lithium from the
Atacama in Chile is outstanding in terms of size and cost as seen
ocean is a critical determinant of the lithium supply curve
in Fig. 5. Limits on how much of the brine lithium resource is
because of its sheer magnitude. The oceanic stock is dilute, but
available therefore stem from the technical ability to obtain the
greatly exceeds any conceivable societal need; the 170 ppb of
brine from the ground formations, as well as the degree of
lithium present as a solute means some 230,000 Mt of lithium
extraction deemed socially and environmentally acceptable. We
metal (Garrett, 2004; Evans, 2008b). A viable extraction process
have assumed that extraction will proceed to the technical limit,
for any portion of this would fit the classic definition of a
but there are potentially significant environmental implications
‘backstop’ resource technology (Nordhaus et al., 1973). As a final
such as effluent and reagent use; large amounts of hydrochloric
‘waste basket’ (Ayres, 1999) for lithium, the ocean would be an
acid has left barren toxic zones at Atacama (Regolado and
infinitely sustainable source for all practical purposes.
Friedman-Rudovsky, 2010); and this could be objectionable or
Extracting lithium from seawater requires only a trivial
expensive to deal with in areas such as Uyuni (Bolivia) or the
amount of energy in theory. The change in free energy is
Salton (USA) or Dead Sea (Jordan/Israel) that have protected
RT ln(C0/C1), or 30 kJ/mol Li (1.2 kWh/kg) to move lithium up a
wildlife or competing land use claims. Salars on the surface are
gradient from ocean concentration to 2% solution. This means that
obviously well known, but there are many underground salars
advanced technologies for concentrating it could theoretically be
with either unknown or assumed to be economically irrelevant
possible, but nothing convincing has yet been demonstrated.
chemical compositions that are not included in the resource
Some newer nanomaterial-based adsorption processes (Chung
estimates. Other brine sources include oil field and geothermal
et al., 2004; Wang et al., 2009) could fit this bill, but the scalability
brines that are harder to quantify.
of such processes is still very much open to question: engineered
Ore deposits come as minerals ranging up to  5% Li by weight
nanomaterials still imply high embodied energies, and the low
(usually as Li2O) along with a wide variety of impurities. Most
adsorption efficiencies (e.g. 2% in Wang) thus far achieved with
commercial ores are lower in concentration; however, 0.3% to
such methods would mean comparatively large supporting infra-
1.5% are more common grades (Garrett, 2004; Yaksic and Tilton,
structure. The low or uncertain reversibility of the process,
2009), indicating that the gross inputs and impacts of lithium ore-
meaning that the material would need to be replaced, rounds
mining could perhaps be vaguely analogous to those of copper
out this set of concerns.
mining in crude terms of energy input and material moving.
Moving from theoretical possibilities to more practical
Generally used assumptions for recovery are 50% for underground
arrangements, lithium can be extracted from the ocean via a
mining, and 75% for open pit methods (Hustrulid and Kuchta,
number of well-understood processes (Steinberg and Dang, 1975;
2006; Yaksic and Tilton, 2009). There are also deposits of lithium
Takeuchi, 1980; Schwochau, 1984; Marinsky and Marcus, 1995).
that has been adsorbed from brines into clay. The grades are
The estimates for the actual cost to do so vary wildly. Steinberg
generally lower, but there are some large deposits and the soft
and Dang (1975) calculate the full process cost as $15–22 USD/kg
clay makes mining relatively easy. As an example, a feasibility
LCE, if we use a factor of 3.89 to inflate 1976 dollars.6 Their
study at the King’s Lake, Nevada, USA deposit (Ajie, 2010)
calculated optimum electricity requirement was 0.1 kWh/g or
indicates cash costs around US $6/kg without by-product credits
0.1 MWh/kg of lithium metal which actually appears to be less
to produce lithium carbonate, indicating that large clay resources
energy than the processing of some ‘viable’ hard rock resources
could theoretically be produced at costs similar to those paid for
lithium carbonate today. Economic mineral deposits can occur
underground and thus there may be some scope indeed for 6
As done also by Yaksic and Tilton, 2009.
D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103 99

Fig. 6. Potential maximum mine output from mineral resources. Shaded area indicates one possible production profile peaking at the sum of the most likely outputs from
the mineral sources. The lines indicate the maximum and minimum profiles with the mineral sites producing at extreme rates.

but is still two orders of magnitude larger than the free energy a study using a sample of over 1000 mines and covering the full
change. To accomplish this however, the process assumed a high- range of modern mining methods over the range of mineral
insolation solar pond covering 400 square kilometres per 1 kt/ grades confirmed the validity of Taylors basic model (Long,
year facility capacity to provide the majority of the initial brine 2009). However, in this more comprehensive study, a and b were
concentration, quite a caveat indeed. At 2000 kWh/m2 yr as an estimated at 0.12 and 0.65 for open pit mines and 0.30 and 0.56
estimate for high insolation, such a facility assumes a ‘renewable’ for underground mines.
energy input of 2 TWh per square kilometre per year, or 800 MWh Applying this formula to the data set of mineral resource sites
of sun energy per kg lithium. The ponds required for a 100 kt/year included in the cumulative resource curve produces an estimate
capacity would thus be 40,000 km2, intercepting about 80 PWh in of 125 kt/yr capacity from the included mineral resources if all
a high insolation area, on a scale comparable with current global such resources were considered to be reserves (e.g., in a high
energy consumption. lithium price scenario) and with recoveries of 50% and 75% for
From a theoretical viewpoint, extracting lithium from the underground and open pit mining, respectively. The 95% con-
ocean is a thus a possibility, but there are massive problems fidence interval coefficients from the same data set produce a
and uncertainties surrounding practical incarnations of such a range of 60–330 thousand tons lithium metal per year production
technology. Scenarios ending this century with a significant rate capacity, or some three to 15 times current extraction rates
of virgin resource input implicitly assume it to be possible, but (which are from both brines and minerals).
the ocean should by no means be assumed to exist on the supply Fig. 6 shows the estimated mineral output over time fit to a
curve for planning purposes. logistic curve with magnitude and time constants resulting from
this estimation over the mineral resources which total 5.5 Mt of
recoverable lithium. The lines represent the production profile if
Extraction capacity limits each lithium resource site was represented by coefficients at the
95% high and low confidence boundaries from Long’s study, an
Lithium resource flows and potential limits exceedingly unlikely scenario.7

Roughly 25 kt of lithium metal equivalent is extracted

Estimating limiting factors for lithium flows from brines
annually at present from a number of locations. This rate will
The most important resource is the Salar de Atacama, and it is
need to rise a minimum of 8 times, to 200 kt for the low vehicle
difficult to think of physical constraints on the possible rate of
intensity scenario with 9 kWh batteries, and correspondingly
production. With three meters of evaporation per year, and an
more with the other scenarios (Fig. 7). The following section lays
average porosity of roughly 20% over the 200 m mining conces-
out some potential limits on flows from resource stocks, and the
sion (Evans, 1978; Garrett, 2004), 15 m depth, or 8% of the
implications for the potential speed of a transition.
resource could be evaporated per year by covering the entire
surface with evaporation ponds. As the evaporation slows with
Estimating production limits for ore and clay deposits concentration, and the current cycle takes two years, dividing this
There is no way to know the potential output or recovery from output by two gives a rough estimate of 500 kt/yr lithium metal,
as of yet unknown mineral resources, but the historical dataset of some 45 times the current rate; this could serve as a proxy for the
mines offers one possibility to estimate that from known sources. maximum conceivable rate of production. The ability to pump
Taylor (1977) noted that observed output rates of mines are more this much brine, and the potential to damage the porous salt
closely correlated to a logarithmic function of ore volume than structure or otherwise limit the recoverable resource is open to
any other variable and proposed a power law relating the question (Garrett, 2004; Tahil, 2008). Although the potential is
elasticity of mine capacity with ore reserves:

P ¼ aRb 7
Observe that the output from the individual mines could be added in many
ways resulting in many different shapes of the curve for total output, besides the
where P (tonnes/day) is the ore production, R (tonnes) is the bell shape of the logistic curve. This does not affect the basic argument, that if the
remaining ore reserves in the mine, and a and b are constants. most likely outputs of the individual mines, according to Long’s estimate, are
Taylor estimated a and b at 0.014 and 0.75, respectively. Recently, added one gets a maximum total output of 125 kt/yr.
100 D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103

Fig. 7. Implied resource extraction rates. The demand curves assume 80% recycling. The bottom end of each band corresponds to the 0.2 cars/capita scenario and the upper
end to the 0.5 cars/capita scenario. The estimated limited mineral extraction rate from Fig. 6 is shown, indicating the role of lithium from brines and ocean.

certainly high, it is also highly uncertain. Unhappily, the ultimate smaller lithium containing salars) not mentioned here. Each will
extraction rate from Atacama is of central importance to the likely have some local limit on extraction, but the overall picture
supply debate because as long as the production there can be (or remains the same: in general the sum of brine resources likely
is thought to be) easily increased, there will be a cap on the enables a very high resource production rate, but with the bulk of
incentives to start new production elsewhere. the potential at Atacama, possibly other undeveloped salars in
Zabuye salt lake, also spelled Chabyer, is now the largest South America and perhaps in Tibet.
lithium operation in China. Initially, extraction at the location
was considered difficult because of the high magnesium content Implications of limiting flow factors
(Evans, 2008; Zheng and Liu, 2009), but it is now proceeding with
a targeted production of 10 kt/yr lithium to be achieved some Fig. 7 demonstrates that there are two different problems related
time in the foreseeable future. The potential production rate is to annual lithium availability. The first relates to the build-up phase.
unknown, though presumably higher than this. Tibetan glaciers From the preceding section on mineral flow limitations, it is clear that
and rivers give ample chance for water supply, and thus the the rate production from brine sources will determine the timeframe
achievable rate of extraction could possibly be very high. and form of an electric vehicle transition. Although the potential flow
The Salar de Uyuni is the largest or second largest lithium resource from Atacama could reach as high as 500 kt/yr, Fig. 7 shows that the
on the planet, depending on the resource estimate used (Ericksen build-up of a BEV stock over the next couple of decades presents a
et al., 1978; Garrett, 2004; Evans, 2008). So far, the extraction has not huge challenge. On the other hand, the PHEV scenarios seem to be
been easy to make cost competitive. The high magnesium levels, and easily within reach.
the annual flooding, lower evaporation, and transport difficulties The second problem relates to maintaining the lithium stock. The
make logistics far harder than at Atacama and similar extraction scenarios in Fig. 7 assume 80% recycling. Making up for the 20% loss
rates are unlikely to be achievable. More importantly, there are in each cycle will then create a substantial demand in the second part
serious potential extraction limits at Uyuni stemming from limited of the century. If the recycling efficiency is not substantially increased
freshwater supplies8; the water use required for the extraction and if no major new resource flows are found, this will have to be
process is mutually exclusive with existing quinoa farms. Further- supplied from the ocean (see Section 3.2).
more, the water use at 10 kt/yr production is larger than the recharge
rate of the basin (Aguilar-Fernandez, 2009), and would give an
estimated aquifer draw down lifetime of only twenty years, while Institutional factors and transient events
displacing local agriculture. If this limit cannot be overcome then only
a small percentage of Uyuni’s resources could become available over a As the previous sections depict, there is wide latitude, but not
timeframe relevant for the electric vehicle expansion considered in unlimited space for lithium battery transition scenarios. Lithium
this paper. As a contrast, although Atacama’s region is already using does however have human factors that pose risk for supply
most of its total available water (Grosjean and Veit, 2005), there is no security, such as the geographic concentration of resources, the
similar issue because the actual processing takes place at Antofagasta, market structure of producers, time lags in supply system
which is on the coast and gets water from a different source. response and a lock-in effect leading to demand inelasticity. Some
Other geothermal and oilfield brines could add production, but of these aspects have been discussed for lithium (Andersson and
not with the potential of the Andean salars. As an example, the Råde, 2001; Ebensperger et al., 2005) as well as for some other
Salton Sea is a protected wildlife refuge, which may eliminate materials such as cobalt (Alonso et al., 2007), where they have
surface ponds as an option, but the seven existing geothermal been directly implicated in supply disruption.
power plants there could produce an estimated 3 kt/yr from their
brine throughput. There are many additional brine sources (e.g. Geographic concentration of resources

There are a small number of critical resource locations and

From 300 to 600 t of purified water is used during processing per ton LCE. high rates of demand will only increase our reliance on them. If
Estimates from Garrett (2004) and Aguilar-Fernandez (2009). more than a moderately intense use of batteries is envisioned and
D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103 101

order to compete, but even if the current lithium producers were

to operate as a cartel, prices much above those prevailing today
would likely be met with new production sources.
This suggests another risk factor however: if the current brine
producers are able to keep prices limited, they can likely supply
the market for a very long time before it is considered profitable
to bring other sources online. If under such a scenario, a major
source, e.g. Atacama becomes unavailable due to either some
form of interference (e.g. natural disasters or sabotage) or through
production unexpectedly dropping (e.g. problems with pumping
rates, non-disclosure of exhaustion) then the production would
need to be replaced. This would be a major strain on the flow rate,
and rely on sources in less stable jurisdictions. Such a scenario is
not necessarily far-fetched; the loss of the main producer in the
cobalt industry produced years of large price swings and
shortages (Alonso et al., 2007).

Time lags for supply infrastructure change

Fig. 8. Geographic distribution of recoverable lithium stocks from Fig. 5.
The supply system takes time to change, and comparatively
rapid increases in demand could result in a comparatively limited
new resources prove more difficult to locate than anticipated then supply being rationed through high prices. A sustained price point
the demand structure will be reliant upon these locations with no high enough to preclude the gainful use of a resource for a given
other sources able to make up the rate of extraction required. purpose for which an alternative exists could shift demand and
Fig. 8 shows the distribution of the recoverable resources processes over time accordingly, and a perception of scarcity
included in this study (e.g. Fig. 5). The data from Yaksic, Tilton could be a real and critical determinant of how demand may
and Evans is based on a wider and more updated set of resources evolve.
than included in the latest USGS report (Jaskula, 2011a, 2011b), The lithium supply system currently has a limited fast (12
and are subsequently scaled down for their individual recovery months) response capability which lies mostly with the brine
factors, leading to a different distribution than the USGS in-situ producers. Constructing a new holding pond, and then pumping
values. more brine to begin the evaporation period is quite a fast under-
In Fig. 8, the large predominance of Chile has much to do with taking, and fastest at Atacama, where the high evaporation allows
the great size and amount of work that has been put into defining a roughly 12 month cycle through the first drying pond (Garrett,
the resource at the Salar de Atacama. Similarly, the USA has 2004). New processing plants take an average of two years for
undergone a great deal of work in defining deposits. Most notable construction if the process is known (Topp et al., 2008), indicating
however is that the other three of the top five countries (Bolivia, that almost any surge in brine output from the salars could be
China and Argentina) are associated with forms of institutional accommodated at the processing level by the same time it
risk as seen by the mining industry (McMahon and Cervantes, becomes available from the drying ponds.
2011). Examples include uncertain mineral rights, conflicting land Bringing a new mine online requires several years for permit-
use, security risks and other unquantifiable political exposure. ting and construction of facilities, and can easily take more than a
This could be problematic for development as they hold much of decade from resource delineation to production in the western
the potential for expansion of the highest quality brine resources. world. The technology to build a mine is largely fixed upon
Additionally, Zaire (DRC), Russia and a large component of the construction and is site specific. Upgrading the capacity from
‘Other’ category also routinely find themselves on the bottom of existing mines is possible in a shorter time frame, but subject to
such surveys. This serves to underline the importance of the logistic constraints; mine upgrades have been observed to fit into
resources in stable jurisdictions, and most particularly Atacama. the same function used to calculate output from the mines (Long,
These risks of distribution may be underappreciated when dis- 2009) which strongly implies decreasing returns to expansion.
cussing resources at the aggregate level. In the short term, e.g. less than two years, there is no easy way
to increase production past that which already has ‘shovels in the
Producer concentration ground’. Between two years and the length of time it takes to get
the next major mine into production, only increased brine
The lithium market is very concentrated at the production end, production, or higher throughput through producing mineral
with the top four producers representing approximately 90% or operations can take up slack. Unforeseen demand or problems
more of world production in 2009 as well as over the preceding with extraction are thus quite inelastic or difficult to fix at a
decade. Furthermore, as Ebensperger et al. (2005) note: ‘‘The high supply level.
level of concentration of producers, and the recent continuing
rises of lithium carbonate prices since 2001 seems to imply the
exertion of market power, significant barriers to entry and Conclusions
associated strategic behaviour to ensure long-term profits.’’ This
is not an irrational position where the by-far largest and cheap- Lithium resources and societal stocks
est-to-produce resources are controlled by an oligopoly with an
interest in maximizing profit. Over the longer term however, new The presently known lithium resources excluding the ocean
resources at the prefeasibility stage are already deemed profitable will only be exhausted this century if large scale use of predomi-
in the current price regime, and shut-in production in the US, nantly BEV sized batteries comes into play, or if batteries are not
Argentina and other countries may not require much investment recycled. This same resource will extend into next century for low
to restart. The evidence so far is that producers have cut prices in vehicle count scenarios with plug in hybrids and high recycling
102 D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103

rates. The magnitude of the resource will clearly support the Transient events and disruptions
creation of a large number of electric vehicles, but visions of the
future that involve simply continuing present mobility trends Although there is not necessarily a physical logic for scarcity, it
with vehicles based on lithium ion batteries may involve opti- is not difficult to construct plausible supply scenarios that result
mistic assumptions about resource growth as well as recycling. in supply growth stalling. The arguments thus far have also
The known lithium resource is also expanding however and implicitly assumed foresight and a smoothly functioning market,
current prices seem more than sufficient to spur on exploration yet there are structural features of the lithium market that have
activity in even slightly more marginal deposits. The possibility been implicated as risk factors or as having contributed to other
for discovering resources is of course unknown; the salars are transient resource disruptions. It would therefore be unwise to
mostly found because they are surface features, but there are neglect them in an analysis of resource adequacy. Of particular
many lithium-containing lakes not included in USGS resource concern is the poor quality of relevant institutions in countries
figures or those presented here. Yet current and future explora- containing much of the prospective lithium resource.
tion will presumably result in the addition of more resources. The potential effects of certain actions or unknown risks could
Alternatively if environmental, political or social concerns suc- and perhaps should be evaluated via scenarios; for instance
ceed in preventing the exploitation of resources presented here scenarios can be constructed where producers could theoretically
then the outlook could also deteriorate somewhat; for instance control production output to achieve prices not quite high enough
the Salar de Uyuni, the second largest resource, has water and or too volatile for many marginal mineral resources to begin
wildlife issues with its exploitation as well as conflicting indi- extraction. If the main producers were disrupted, or they were
genous claims. Many other potential resources face similar simply unable to keep up with the logistics required to maintain
uncertainties as discussed in Section 5.2. output, then it could very easily be the case that there would not
For all BEV scenarios or PHEV scenarios not involving a drastic be enough alternative supply to make up for the production for
reduction in mobility trends, recycling will be a necessary option many years, if at all. Therefore the time lag response of the supply
if not for the build-up, then for maintenance of a societal lithium system, coupled with large information asymmetry poses definite
stock. Yet recycling economics are currently not good and may risks that could negatively impact supply, despite the lack of any
actually degrade somewhat instead of improving. The present physical reason for it to be constricted. In order to assure a
view that there is ‘enough’ lithium may also hinder recycling smooth transition, policy may be required to address these risks.
efforts. If, as the cumulative availability curve suggests, lithium
remains at a low cost for a long time then it may be necessary to Synthesis
design policy to encourage recycling rather than simply waiting
for it to evolve from commercial consideration. In the end, the It is not enough to look at lithium resource stocks and
sale price of recovered material plus any recycling fees must be conclude that there is enough. Although policy support may be
greater than processing and return/collection costs. If these required to enable recycling, if recycling does occur then resource
should fail to line up, strong policy support will be required to exhaustion does not appear to be credible threat. Yet the time
ensure recycling processes are available, and that their imple- dimension and its implications are more important in the case of
mentation is assured. lithium. Issues surrounding the required rate of lithium flows, and
particularly their dependency on a concentration of producers
and countries, will occur well before any limits to resource
Lithium extraction and societal flows quantity. Maintaining the present vision of personal mobility
through simply changing the technology of the car may thus be
The initial build-up of lithium stocks in society must be unrealistic unless the time scale for such a transition is extended.
accomplished from primary resources. There is nothing about A worldwide push for lithium batteries risks building up a large,
the conceivable rate of extraction that places a fundamental limit capital intensive stock of vehicles and associated production systems
on the timeline for a moderate transition, although this currently that are vulnerable to resources more concentrated than that of the
appears contingent upon an increasingly irreplaceable supply oil supply system existing today; more than two thirds of the
from lithium brines, most particularly the Salar de Atacama. resources considered here are concentrated in a small area shared
The peak virgin lithium demand will occur before 2050 in all by three countries and possibly to be exported via a single port. There
scenarios considered here. Recycling will not ease demand during is currently no battery technology able to compete with lithium for
the build-up stage of lithium stocks because annual lithium use large vehicle batteries, and no concrete indication that this will
will have to be increasing rapidly and the stock potentially change in the foreseeable future. If there are no readily scalable
available for recycling will be small compared to the demand. alternative lithium supplies or alternative vehicle energy technolo-
Extraction from known mineral deposits with typical logistics will gies, this would be a considerable risk to critical societal infrastruc-
not be able to cover the virgin resource demand for PHEV ture. This is a strong case for maintaining diversity at all levels of the
scenarios, and will assuredly leave a large shortfall in heavier system. Possible policy responses could be to maintain a portfolio of
adoption scenarios. The world will therefore be dependent on known lithium resources at the feasibility stage to minimize the time
brine extraction for the remainder in any scenario. of any prospective disruption as well as bringing other vehicle
This supply gap is almost certainly feasible to produce with brine technologies to competitive readiness.
resources over the peak for moderate PHEV scenarios. Larger bat- These obstacles and possible responses to them certainly
teries, quicker transitions or high vehicle growth will increase the deserve serious consideration, without which the promise of
burden. Full electric vehicle scenarios where the world approaches transport energy diversity risks running aground by shifting
Western European vehicle density by the end of the century will not energy source dependence to one of material dependence.
only strain the resource stock, but produce flow requirements with
no currently clear possibility of addressing them.
Towards the end of the century, a steady flow of new lithium Acknowledgements
will be necessary to make up for losses in the recycling loop and,
to a smaller extent, for losses from dispersive use in other The authors would like to thank Göteborg Energi, Mistra, IMIT
applications. This flow may have to come from ocean resources. and Chalmers Energy Initiative for their gracious support for this
D. Kushnir, B.A. Sandén / Resources Policy 37 (2012) 93–103 103

article. We would also like to thank Bengt Kasemo, Roland Clift, Kushnir, D., Sandén, B., 2011. Multi-level energy analysis of emerging technolo-
and several others for insightful comments. gies: a case study in new materials for lithium ion batteries. Journal of Cleaner
Production 19, 1405–1416.
Long, K., 2009. A test and re-estimation of Taylor’s empirical capacity–reserve
References relationship. Natural Resources Research 18 (1), 57–63.
Marinsky, J., Marcus, Y., 1995. Ion Exchange and Solvent Extraction. Marcel Dekker
Inc., New York.
Aguilar-Fernandez, R., 2009. Estimating the Opportunity Cost of Lithium Extraction
McMahon, F., Cervantes, M., 2011. Fraser Institute Annual Survey of Mining
in the Salar de Uyuni, Bolivia. Nicholas School of the Environment, Duke, M.Sc.
Companies 2010/2011.
Ajie, J., 2010. King’s Valley NI 43-101 Technical Report: Preliminary Assessment
Nordhaus, W.D., Houthakker, H., et al., 1973. The allocation of energy resources.
and Economic Evaluation, Western Lithium Corporation.
Brookings Papers on Economic Activity 1973 (3), 529–576.
Alonso, E., Gregory, J., et al., 2007. Material availability and the supply chain: risks,
Regolado, A., Friedman-Rudovsky, N., 2010. The lithium rush. Technology Review
effects, and responses. Environmental Science & Technology vol. 41 (19),
6649–6656. Jan/Feb, 28–32.
Andersson, B.A., Råde, I., 2001. Metal resource constraints for electric-vehicle Schwochau, K., 1984. Extraction of Metals from Sea Water. Inorganic Chemistry,
batteries. Transportation Research Part D: Transport and Environment 6 (5), 91–133.
297–324. Simpson, R.D., Ayres, R.U., et al., 2005. Scarcity and Growth Revisited: Natural
Ayres, R.U., 1999. The second law, the fourth law, recycling and limits to growth. Resources and the Environment in the New Millennium. Resources for the
Ecological Economics 29 (3), 473–483. Future, Washington, DC.
BCI, 2009. National Recycling Rate Study. Battery Council International, Chicago. Slade, M.E., 1982. Trends in natural-resource commodity prices: an analysis of the
Bernardes, A.M., Espinosa, D.C.R., et al., 2004. Recycling of batteries: a review of time domain. Journal of Environmental Economics and Management 9 (2),
current processes and technologies. Journal of Power Sources 130 (1–2), 122–137.
291–298. Socolow, R.H., Thomas, V.M., 1997. Rejoinder to Lave, Hendrickson, and McMi-
Boryta, D.A., Kullberg, T.F., et al., 2006. Production of lithium compounds directly chael. Journal of Industrial Ecology 1 (2), 39–40.
from lithium containing brines. USPTO. US, Chemetall Foote Corporation. Steinberg, M. and V. D. Dang (1975). Preliminary Design and Analysis of a Process
7214335. for the Extraction of Lithium from Seawater, United States.
Chung, K.-S., Lee, J.-C., et al., 2004. Recovery of lithium from seawater using nano- Svedberg, P., Tilton, J.E., 2006. The real, real price of nonrenewable resources:
manganese oxide adsorpents prepared by gel process. Materials Science copper 1870–2000. World Development 34 (3), 501–519.
Forum. vol. 277. pp. 449–452. Tahil, W., 2008. The Trouble with Lithium 2: Under the Microscope, Meridian
Dewulf, J., Van der Vorst, G., et al., 2009. Recycling rechargeable lithium ion International Research.
batteries: critical analysis of natural resource savings. Resources, Conservation Takeuchi, T., 1980. Extraction of lithium from seawater with metallic aluminium.
and Recycling 54 (4), 229–234. Journal of Nuclear Science and Technology 17, 12.
Ebensperger, A., Maxwell, P., et al., 2005. The lithium industry: its recent evolution Taylor, H.K. (Ed.), 1977. Mine valuation and feasibility studies. Mineral industry
and future prospects. Resources Policy 30 (3), 218–231. costs. Northwest Min Assoc. Spokane, W.A.
Ericksen, G.E., Vine, J.D., et al., 1978. Chemical composition and distribution of Tilton, J.E., 2003. On Borrowed Time? Assessing the Threat of Mineral Depletion.
lithium-rich brines in salar de Uyuni and nearby salars in southwestern Washington, DC, RFF Press.
Bolivia. Energy 3 (3), 355–363. Tilton, J.E., Lagos, G., 2007. Assessing the long-run availability of copper. Resources
Evans, R.K., 1978. Lithium reserves and resources. Energy 3 (3), 379–385. Policy 32 (1–2), 19–23.
Evans, R.K., 2008a. An Abundance of Lithium. Topp, V., Soames, L., et al., 2008. Productivity in the mining industry: measure-
Evans, R.K., 2008b. An Abundance of Lithium. (part 2). ment and interpretation. Productivity Commission of the Australian
Gaines, L., Nelson, P., 2009. Lithium Ion Batteries: Possible Materials Issues. Government.
Argonne National Laboratory. UN, 2008, November 2009. World Population Prospects: The 2008 Revision.
Garrett, D., 2004. A Handbook of Lithium and Natural Calcium Chloride. Elsevier
Available from: /
Ltd., Oxford.
Wang, L., Meng, C.G., et al., 2009. Study on Li þ uptake by lithium ion-sieve via the
Govett, M.H., Govett, G.J.S., 1978. Geological supply and economic demand: the
pH technique. Colloids and Surfaces A: Physicochemical and Engineering
unresolved equation. Resources Policy 4 (2), 106–114.
Aspects 334 (1–3), 34–39.
Grosjean, M., Veit, H., 2005. Water Resources in the Arid Mountains of the
Whittingham, M.S., 2004. Lithium batteries and cathode materials. Chemical
Atacama Desert (Northern Chile): Past Climate Changes and Modern Conflicts.
Reviews 104 (10), 4271–4302.
Global Change and Mountain Regions, 93–104.
Will, F.G., 1996. Impact of lithium abundance and cost on electric vehicle battery
Gruber, P.W., Medina, P.A., et al., 2011. Global lithium availability. Journal of
Industrial Ecology 15 (5), 760–775. applications. Journal of Power Sources 63 (1), 23–26.
Holger, F., Joachim, P., et al., 2000. Method for Producing Highly Pure Lithium Salts. Xu, J., Thomas, H.R., et al., 2008. A review of processes and technologies for the
USPTO. Germany. BASF AG, 6,592,832. recycling of lithium-ion secondary batteries. Journal of Power Sources 177 (2),
Hustrulid, W., Kuchta, M., 2006. Open Pit Mine Planning and Design, 2nd ed. Taylor 512–527.
& Francis. Xu, Z., Wen-zhi, H., et al., 2010. Recycling of Electrode Materials from Spent
Jaskula, B., 2006. Lithium. 2005 Minerals Yearbook: Lithium. USGS. Lithium-Ion Batteries. In: Proceedings of the Fourth International Conference
Jaskula, B., 2011a. Lithium. 2009 Minerals Yearbook: Lithium. USGS. on Bioinformatics and Biomedical Engineering (iCBBE).
Jaskula, B., 2011b. Lithium. Mineral Commodity Summaries 2011. USGS. Yaksic, A., Tilton, J.E., 2009. Using the cumulative availability curve to assess the
Khaligh, A., Zhihao, L., 2010. Battery, ultracapacitor, fuel cell, and hybrid energy threat of mineral depletion: the case of lithium. Resources Policy 34 (4),
storage systems for electric, hybrid electric, fuel cell, and plug-in hybrid 185–194.
electric vehicles: state of the art. Vehicular Technology, IEEE Transactions on Zheng, M., Liu, X., 2009. Hydrochemistry of Salt Lakes of the Qinghai-Tibet Plateau,
59 (6), 2806–2814. China. Aquatic Geochemistry 15 (1), 293–320.