Mark Z. Jacobson's 7 reasons why nuclear energy is not the answer to solve climate change

From ScienceForSustainability
Jump to navigation Jump to search

"The 7 reasons why nuclear energy is not the answer to solve climate change" is a piece by Mark Z Jacobson which is reproduced on Leonardo Di Caprio's website and others.

Much of the same arguments are given, with sources for his claims, in his piece "Evaluation of Nuclear Power as a Proposed Solution to Global Warming, Air Pollution, and Energy Security"[1]

The 7 reasons piece starts by asserting that "There is a small group of scientists that have proposed replacing 100% of the world’s fossil fuel power plants with nuclear reactors as a way to solve climate change."

Whilst Jacobson acknowledges that "Many others propose nuclear grow to satisfy up to 20 percent of all our energy (not just electricity) needs", the title of the piece and much of its argument is based at refuting an argument for 100% nuclear, a position not held by most of those who advocate for nuclear as part of mitigation solutions, and which is not supported by the scientific consensus.

There is a case for tackling one's opponents’ strongest argument, not their weakest one.

The stronger argument is not that nuclear energy is the solution to climate change, but that it is part of the solution. This position is supported by the IPCC, who find that "Nuclear energy could make an increasing contribution to low-carbon energy supply" and whose four illustrative pathways to staying below 1.5C all include various amounts of nuclear energy.

Thus Jacobson is positioning himself as rejecting the findings of the IPCC. There seems no obvious reason to treat Jacobson's rejection of the IPCC's findings differently from other rejections of IPCC findings, including those who deny climate change and its effects.

Jacobson's claims

Much of Jacobson's argument is based on the time taken to build nuclear power plants:

One nuclear power plant takes on average about 14-1/2 years to build, from the planning phase all the way to operation. According to the World Health Organization, about 7.1 million people die from air pollution each year, with more than 90% of these deaths from energy-related combustion. So switching out our energy system to nuclear would result in about 93 million people dying, as we wait for all the new nuclear plants to be built in the all-nuclear scenario.

It is true that millions of people die each year from energy-related combustion. This death rate can be reduced by replacing pollution-causing fossil fuels and biomass with clean energy sources including nuclear energy, hydro, wind, solar, geothermal, wave, tide, etc, or by fitting equipment to combustion power plants to capture pollution.

It is also true that people continue to die whilst these facilities are being built, until they are working.

However Jacobson goes on to claim:

Utility-scale wind and solar farms, on the other hand, take on average only 2 to 5 years, from the planning phase to operation. Rooftop solar PV projects are down to only a 6-month timeline. So transitioning to 100% renewables as soon as possible would result in tens of millions fewer deaths.

Clearly a rooftop PV project producing a few kilowatts peak power at a capacity factor of 10 or 20% is not equivalent to a nuclear power plant producing several GigaWatts with over 90% CF.

What about "utility scale wind and solar"?

The UK's largest wind farm to date, the Walney Extension[2], has a peak output of 660MW at around 40% capacity factor, thus producing about 264 MW. About 12 wind farms of this size would produce as much electricity, on average, as a nuclear power plant such as Hinkley Point C. Walney Extension took 34 months for actual construction (not including planning, which Jacobson includes in the time to build nuclear), a build rate equivalent to taking over 3 decades to build a Hinkley Point C-size power station.

It is true that we could build several large wind farms at the same time, but we can also build several nuclear power plants at the same time. There are constraints on how much we could increase production, such as the availability of jack-up rigs for placing offshore wind turbines, and large forging plants for constructing reactor pressure vessels, but to the extent that either technology is significantly limited by such constraints it makes sense to construct both, or all, available technologies simultaneously.

Wind and solar do have the advantage that they can start generating some energy sooner than nuclear can: whilst we have to wait until an entire reactor (and its turbines and generators etc) are completed to start generating energy from them, we can bring smaller wind and solar farms onto the grid as soon as they are ready and add more later as they are completed.

On the other hand there is the issue of capacity factor.

When most of our energy comes from air-polluting, greenhouse-gas-emitting sources such as fossil fuels and biomass, then adding any clean energy source - hydro, nuclear, wind, solar etc - can displace pollution, saving lives and mitigating global heating. Building whichever sources put units of clean energy onto the grid fastest saves most lives and does most for climate mitigation in the short term.

However sources such as wind and solar are intermittent, varying between their full, "nameplate", capacity and zero, depending on the weather, time of day and season, and in most parts of the world the difference between what they supply and what is required is made up by dirty - air polluting, GHG-emitting, people-killing - sources. For wind, with capacity factors between around 20% for poorly-sited onshore turbines and about 50% for the latest, biggest offshore units, this means that every unit of energy produced by wind is matched by at least one, and up to 5 units of dirty energy. For solar with CF of 10-20% each unit is matched by up to 9 units of dirty energy. This dirty energy is locked in to the system by the intermittent renewable source.

Renewables advocates propose various ways of getting around this, from wishful thinking and arm-waving to various schemes involving prodigious amounts of energy storage and/or heroic cross-continental interconnectors. None of these has yet been shown working at large scale, and small-scale projects have been disappointing.

Jacobson's "seven major problems with nuclear energy"

1. Long Time Lag Between Planning and Operation

The time lag between planning and operation of a nuclear reactor includes the times to identify a site, obtain a site permit, purchase or lease the land, obtain a construction permit, obtain financing and insurance for construction, install transmission, negotiate a power purchase agreement, obtain permits, build the plant, connect it to transmission, and obtain a final operating license.

The planning-to-operation (PTO) times of all nuclear plants ever built have been 10-19 years or more.

As discussed above time to build is not the only criterion to consider.

It is worth noting that Jacobson considers not just the time to build nuclear or wind or solar, but the time to plan and get approval for a plant. Whilst this is a constraint in the real world it is also largely a political one in which the anti-nuclear lobby creates significant delays in the planning approval process, which it then cites as a flaw of nuclear energy per se.

The distribution of nuclear power station build times from the IAEA PRIS database.

The actual mean time to build the NPPs in our current fleet is around 7.5 years, with a handful built in just 3 years[3].

From Climate Gamble

Jacobson cites NPPs such as Olkiluoto 3, Hinkley Point C, Vogtle, Haiyang and Taishan as examples, but these are all First Of A Kind designs never previously built; EPRs and AP1000s. Olkiluoto 3 was, along with Flamanville (which Jacobson oddly fails to mention) the first ever EPR builds started and have been plagued with delays and cost overruns. Even so the rate at which Olkiluoto 3 adds capacity to Finland's grid will - despite further delays leading to a revised start data of 2022 - probably still beat the best rate that wind and solar has achieved over a similar (15 year) period (Denmark 1999-2014 and Germany 2002-2017)

2. Cost

Jacobson cites Lazard's Levelised Cost Of Energy (LCOE) figures.

Levelized Cost Of Energy comparison -- Lazards.png

The figures Lazard produces are a guide for investors wishing to make money by investing in the electricity supply industry, not the costs of producing a decarbonised energy system.

Lazards' report notes that: "Solar PV and wind have become an increasingly attractive resource relative to conventional generation technologies with similar generation profiles; without storage, however, these resources lack the dispatch characteristics of such conventional generation technologies" and "We find that Alternative Energy technologies [wind and solar] are complementary to conventional generation technologies".

Lazards' figures show that wind and solar are more profitable than nuclear in a market which economists acknowledge is broken because it does not price the externalities of air pollution and CO
emissions — including the pollution and emissions caused by the fossil fuel plants which almost invariably back up wind and solar and make it profitable.

If we're interested in decarbonising at the lowest cost then we can't compare these LCOE figures directly. For example an offshore wind farm with a capacity factor of 40% on a grid running largely on gas with a carbon intensity of 350 g/kWh will (assuming wind is zero carbon) have a net carbon intensity of 210 g/kWh. For solar with a capacity factor of 10-20% the figure is worse still. The overall system is simply not low carbon and cannot be honestly compared with nuclear.

To properly compare wind and/or solar with nuclear requires a way of making the intermittent source as reliable as nuclear. This requires storage and/or interconnectors, plus overbuilding, and the price of whatever scheme is chosen has to be added to the price of the wind/solar/etc generators themselves in the comparison. An example of how one might build the equivalent of a Hinkley Point C or Sizewell C in the UK with renewables and storage is at renewable alternatives to nuclear in the UK.

3. Weapons Proliferation Risk

As Jacobson correctly points out the IPCC identify proliferation as a concern with nuclear energy.

Barriers to and risks associated with an increasing use of nuclear energy include operational risks and the associated safety concerns, uranium mining risks, financial and regulatory risks, unresolved waste management issues, nuclear weapon proliferation concerns, and adverse public opinion (robust evidence, high agreement).

However the IPCC do find that "Nuclear energy could make an increasing contribution to low-carbon energy supply" and consistently refer to "low-GHG energy supply technologies such as renewable energy (RE), nuclear power, and CCS", and nuclear energy is a significant part of representative pathways in their Special Report on staying below 1.5C.

Jacobson claims that a country which uses civil nuclear energy can import Uranium and "it can secretly enrich the uranium to create weapons grade uranium", as if the controls imposed by the IAEA and the international community by which it is backed simply do not exist. As Iran has shown, attempting to secretly enrich uranium to 90% weapons grade under the pretence of doing so to 5% reactor grade are readily detected. Most countries without an existing nuclear industry which acquire nuclear energy do so as part of a package in which the country building the reactor also supplies fuel for it, in the form of already-fabricated fuel rod bundles. For a country to seek to enrich uranium to make its own fuel, as Iran claims to be doing, is itself a red flag for possible military uses.

Jacobson goes on to claim that a country could "harvest plutonium from uranium fuel rods for use in nuclear weapons". If any country were trying to do this with a pressurised (PWR) or boiling water reactor (BWR) - the most common types - they would be shutting it down for refuelling every few weeks, and the activity which this requires would be obvious to inspectors. The reactors which pose a proliferation risk are those which allow online refuelling; the Soviet RBMK (of Chernobyl notoriety) which was never built outside the former Soviet Union, and Britain's now-defunct Magnox (a design which North Korea built for its weapons programme).

Lastly Jacobson claims that "Small Modular Reactors (SMRs) may increase this risk further." Apart from the risk being practically non-existent to start with, some SMR designs such as Terrestrial Energy's have a sealed reactor core which does not give access to nuclear fuel material, either as fresh uranium or as spent fuel[4].

4. Meltdown Risk

Jacobson claims:

To date, 1.5% of all nuclear power plants ever built have melted down to some degree.

Meltdowns have been either catastrophic (Chernobyl, Russia in 1986; three reactors at Fukushima Dai-ichi, Japan in 2011) or damaging (Three-Mile Island, Pennsylvania in 1979; Saint-Laurent France in 1980).

Of the two "catastrophic" meltdowns Jacobson cites, one killed fewer than 50 people[5], and the other caused no deaths at the time; a non-death toll which is not expected to rise[6].

The nuclear industry has proposed new reactor designs that they suggest are safer. However, these designs are generally untested, and there is no guarantee that the reactors will be designed, built and operated correctly or that a natural disaster or act of terrorism, such as an airplane flown into a reactor, will not cause the reactor to fail, resulting in a major disaster.

The only accident which caused any deaths at all involved a reactor design which was unique to the former Soviet Union, quite unlike any built elsewhere or since. It lacked the basic safety feature of a containment vessel so would not have been permitted to be built anywhere else.

This "catastrophic" accident killed fewer people than an outbreak of food poisoning caused by Organic beansprouts[7][8].

Amongst reactors of the types generally used elsewhere several survived the fourth most violent earthquake to hit the planet in more than a century that records have been kept. Of the four reactors, world-wide, which have suffered melt-downs, including 3 classified as the most severe - level 7 - accidents, none killed anybody.

Even if the lessons learned from Three Mile Island, Chernobyl, and Fukushima had produced absolutely no improvements in the safety of reactor designs since the 1960s when the TMI and Fukushima reactors were designed, nuclear energy would still be remarkably safe. Claiming as Jacobson seems to be doing that improvements to reactor design are untested suggests an ignorance of engineering principles on which far more than just nuclear safety depends. Jacobson's own "100% Wind Water and Sun" plans depend heavily on hydroelectricity using dams, whose design, construction, and operation pose huge safety challenges, as evidenced by the worst ever accident involving an energy system which killed up to almost a quarter of a million people.

As for "an airplane flown into a reactor" it is a specific requirement of reactor design to withstand such a crash[9].

5. Mining Lung Cancer Risk


Uranium mining causes lung cancer in large numbers of miners because uranium mines contain natural radon gas, some of whose decay products are carcinogenic.

The largest single deposit of Uranium in the world is the Olympic Dam in Australia[10] which is primarily a copper mine, so whatever hazards are associated with its uranium mining are also associated with its mining of copper, which is widely used in the electrical industry and is a major component of wind turbines. The mine also produces silver which is used in solar panel production [11]

Clean, renewable energy does not have this risk because (a) it does not require the continuous mining of any material, only one-time mining to produce the energy generators; and (b) the mining does not carry the same lung cancer risk that uranium mining does.

(b) As noted above, mining some of the materials for renewable sources may carry the same lung cancer risk because they are mined alongside uranium.

(a) Renewables do not only require one-time mining: wind turbines and solar panels have finite lives (20 and 30 years respectively, according to the Lazards report Jacobson cites earlier) and whilst the copper in wind turbines is probably recycled, solar PV panels are currently not recycled and so they require fresh materials.

Mining and processing materials not only for nuclear fuel but for practically all other materials incurs more or less potential or actual harm to humans, non-human species, and the environment.

Baotou toxic waste lake
  • Production of rare earth metals such as the neodymium, praseodymium, dysprosium and terbium used in wind turbines and electric vehicle motors results in severe environmental pollution such as in Baotou in Inner Mongolia[12]
  • Some of the Cobalt used in electric vehicle batteries is produced by child labour in conditions in which fatal accidents are common and desperately poor miners frequently clash with the security personnel of large mining firms[13].
  • Reports indicate that members of China's Uighur minority are forced to work in factories in Xinjiang province which supply solar panels to the rest of the world[14][15].

By Jacobson's logic the response to these problems should be to shut down not only nuclear energy but wind and solar, electric vehicles, miroelectronics, and much more.

Alternatively, society can continue, and maybe even try to accelerate, the more complex task of improving safety and environmental standards in the mining and mineral extraction industries, increasing living standards and prosperity for desperately poor people in countries such as the DRC, exerting diplomatic and economic pressure to promote human rights, and so on.

6. Carbon-Equivalent Emissions and Air Pollution

Jacobson claims:

There is no such thing as a zero- or close-to-zero emission nuclear power plant. Even existing plants emit due to the continuous mining and refining of uranium needed for the plant. Emissions from new nuclear are 78 to 178 g-CO2/kWh, not close to 0. Of this, 64 to 102 g-CO2/kWh over 100 years are emissions from the background grid while consumers wait 10 to 19 years for nuclear to come online or be refurbished, relative to 2 to 5 years for wind or solar. In addition, all nuclear plants emit 4.4 g-CO2e/kWh from the water vapor and heat they release. This contrasts with solar panels and wind turbines, which reduce heat or water vapor fluxes to the air by about 2.2 g-CO2e/kWh for a net difference from this factor alone of 6.6 g-CO2e/kWh.

There are four different things Jacobson is talking about here:

  1. the emissions of a running nuclear plant due to emissions embodied in mining and processing uranium for its fuel (to which one could add emissions from vehicles nuclear plant workers and contractors use to travel to and from their workplace, etc);
  2. emissions from other generators supplying energy which the nuclear plant will eventually displace, during the time the plant is being planned and built, or is down for refuelling, maintenance, inspection, etc (which Jacobson groups together as "being refurbished");
  3. the greenhouse effect of water vapour emitted by nuclear plants;
  4. the global heating effect of the heat produced by the nuclear fission reaction itself, which ends up in the earth's atmosphere or oceans.
Nuclear power plant with cooling towers
Nuclear power plant cooled by sea water

Jacobson's 3rd point, that "all nuclear plants emit 4.4 g-CO2e/kWh from the water vapor and heat they release" is incorrect for two reasons:

  1. it would only apply to nuclear power plants which use wet cooling towers (e.g. left), not to those cooled by river or sea water (e.g. right).
  2. water vapour emitted by power stations and other anthropogenic sources does not add significantly to the amount of water vapour in the atmosphere, which is determined principally by the air temperature and thus the moisture-holding capacity of the air.

Incurred and averted emissions

Jacobson's assessment of the overall GHG intensity of nuclear differs most radically from that of the IPCC and others in the second factor he considers: that during the time from a decision being made to build a nuclear power plant to the point at which it is connected to the grid and supplying power, other generators are supplying power and creating GHG emissions. Jacobson argues that these emissions should be attributed to the nuclear plant.

This seems a valid approach, and a valuable one, allowing us to account for the reality that no power plant is created overnight, and giving us the means to compare the real carbon costs of different generators in practice.

For example if we have two types of generator with identical characteristics except that one takes longer to build than the other, then clearly the one which is quicker to build will incur less emissions while we are waiting for it to come online and avert more emissions overall. Jacobson's approach also allows us to calculate a trade-off between, say, one generator which can be built more quickly but has higher emissions and another which takes longer but has lower emissions.

In order to make meaningful comparisons between different energy sources we do, of course, need to apply the same analysis to each.

Wind, solar PV, wave, and tidal energy are variable sources; their energy production varies (cyclically in the case of tidal, intermittently for the others) between a maximum - the nameplate capacity of the source - and zero. The ratio of the mean power output, averaged over a long enough period to be representative, and the maximum is known as the source's capacity factor (also known as load factor). It is usually expressed as a percentage.

The capacity factor for the latest, largest wind turbines, installed offshore, can exceed 50%, smaller turbines installed onshore in good locations may have a CF of around 30%, and in less favourable locations it can be below 20%. The capacity factor of solar PV varies from around 30% to 10% or so.

For a given amount of installed nameplate capacity of wind or solar on an electricity grid there has to be some other, dispatchable, power - one whose output can be turned up and down at will - to supply demand when the variable source is producing nothing, i.e. at night for solar, or in windless conditions for wind (and windless nights for a mixture of both sources).

At any given instant when the variable source is not producing its maximum output, the remainder of demand has to be met by the dispatchable source.

In most real world electricity supply grids the dispatchable sources which make up shortfalls in the output of variable renewable sources are fossil fuel generators: gas and even coal.

In order to consider the real, effective, life cycle emissions of a variable renewable source we need to combine its own life cycle emissions with that of the dispatchable source backing it up, in proportion to their contributions. For example for a wind farm with a capacity factor of 30% backed up by a Combined Cycle Gas Turbine, we would add 30% of the emissions of the wind farm for the electricity it produces to 70% of the emissions of the CCGT for the amount of electrical energy it produces. Jacobson calculates total 100 year CO
emissions for wind of 6.8-14.8 g/kWh, and 7.85-26.9 for utility scale solar PV, and assumes that the background carbon intensity of the grid before adding the nuclear, wind, or solar source is 557.3 g/kWh[1].

Emissions of clean energy source + backup.png

The table (below) and chart (right) show how the total 100-year CO
e emissions for various mixtures of VRE and fossil backups compare with Jacobson's calculations for nuclear[16]:

Mix Clean source Backup source
Nuclear (min) 78 -
Nuclear (max) 178 -
Wind 50% CF 5 279
Wind 30% CF 3 390
Solar 10% CF 2 502
Backup only 0 557

The Nuclear (min) and Nuclear (max) figures correspond to the range of total 100-year CO
e emissions Jacobson calculates for nuclear energy.

Storage, interconnectors, and demand management

Jacobson and others assert that the need for fossil fuel or other emissions-producing backup for intermittent renewables can be reduced or eliminated by the use of various types of storage, long-distance interconnectors, and/or demand management. These technologies can be easily incorporated into Jacobson's assessment method by assigning to the lifetime emissions of variable renewables the carbon cost of the dispatchable source(s) used to back them up, from the time the renewable (e.g. wind or solar farm) is planned until the time that the storage, interconnection or demand management system abates the need for backup.

Time available to abate VRE backup emissions.png

If it takes too long to implement a system to firm up VRE and abate emissions incurred from backup sources then the emissions from VRE + backup will be greater than emissions from nuclear. We can calculate how long this will take: see table (below) and chart (right). These calculations are approximate: they completely ignore the actual emissions from the VRE source and only factor in the emissions from the backup/background grid. Times are in years. The Nuclear (min) and Nuclear (max) figures are time to implement abatement to keep emissions below level of nuclear power at the minimum and maximum of the range Jacobson estimates for nuclear energy.

VRE CF Nuclear (min) Nuclear (max)
50% 28 64
30% 20 46
20% 17 40
10% 16 35

From this we find:

  • Best case: 50% capacity factor (e.g. latest offshore wind turbines) & maximum of estimated range for nuclear gives 64 years to implement full abatement
  • Worst case: 10% capacity factor (e.g. solar in cloudy, high latitudes) & minimum of estimated range for nuclear gives only 16 years to implement full abatement

In practice these time constraints will be shorter due to the non-zero emissions of the VRE sources themselves.

This data shows that whilst the Planning-To-Operation (PTO) times of nuclear power plants limits their emissions mitigation potential, regardless of whether or not Variable Renewable Energy plants can scale more quickly, consideration also needs to be given to the PTO times of the storage, interconnection, and/or demand-management systems needed to compensate for their fluctuations and abate the emissions caused by the background generators backing them up.

In practice this implies that if faced with a choice of either nuclear or VRE, high capacity factor renewables such as offshore wind, coupled with realistic plans for firming measures to abate backup emissions, may well produce better emissions mitigation than a First Of A Kind (FOAK) nuclear plant, especially one constructed by an inexperienced workforce in an environment without existing supply chains. However for lower capacity factor renewables, especially solar in higher-latitudes it becomes increasingly unlikely that the necessary grid-scale firming systems could be implemented quickly enough to produce lower emissions than any but the slowest to build nuclear power plants.

There is also the uncertainty factor: decades of experience show that large projects involving technological innovations in environments carried out by organisations lacking experience in the technology and without established supply chains to back up production, result in significant delays and cost overruns. The building of new EPR and AP1000 reactors which Jacobson cites are examples of this, but the same problems have beset many other large projects in other industries. There is no reason to think the same effects would not be seen in attempting to construct grid scale storage technologies, grid-capacity transcontinental or even intercontinental transmissions lines, and pervasive deep demand management and load shifting. Nuclear power plants may take longer to build than might be wished, but they are subject to at least reasonably well-known constraints. Grid-scale intermittency firming technologies are subject to largely unknown complicating, delaying, and cost-multiplying factors.

7. Waste Risk

According to Jacobson:

Last but not least, consumed fuel rods from nuclear plants are radioactive waste. Most fuel rods are stored at the same site as the reactor that consumed them. This has given rise to hundreds of radioactive waste sites in many countries that must be maintained and funded for at least 200,000 years, far beyond the lifetimes of any nuclear power plant. The more nuclear waste that accumulates, the greater the risk of radioactive leaks, which can damage water supply, crops, animals, and humans.

This is a misunderstanding. Dry cask storage of used fuel on site is an intermediate stage. Final disposal of spent nuclear fuel may be in deep geological repositories, deep boreholes, or recycling in fast neutron spectrum reactors to create more fuel. This is discussed in the article on this site on spent nuclear fuel.

Footnotes and references

  1. 1.0 1.1 Evaluation of Nuclear Power as a Proposed Solution to Global Warming, Air Pollution, and Energy Security
  2. Walney Extension Project Summary Walney Extension website (via Internet Archive Wayback Machine)
  3. Figures correct as of mid-2016 How long does it take to build a nuclear power plant? Euan Mearns; Energy Matters; 27 July 2016
  4. Integral Molten Salt Reactor: Replaceable Core Unit Wikipedia
  5. According to IPCC WG3 AR5 Chapter 7 the Chernobyl accident killed 31 people immediately, 28 of whom died of Acute Radiation Syndrome according to UNSCEAR's assessments of the radiation effects of The Chernobyl accident, and 15 died of Thyroid cancer (Balonov et al., 2011 cited by IPCC op cit); a total of 46 people.
  6. A decade after the Fukushima accident: Radiation-linked increases in cancer rates not expected to be seen UNSCEAR; 9 March 2021
  7. The 2011 Germany E. coli O104:H4 outbreak killed 53 people.
  8. To date there has been no campaign to close down the Organic industry because of this catastrophe.
  9. Aircraft impact assessment US Nuclear Regulatory Commission
  10. Olympic Dam mine Wikipedia
  11. Amount of silver needed in solar cells to be more than halved by 2028, Silver Institute says PV magazine; July 2018
  12. The dystopian lake filled by the world’s tech lust Tim Maughan; BBC Future; 2nd April 2015
  13. Why Cobalt Mining in the DRC Needs Urgent Attention Council for Foreign Relations; 21 Oct 2020
  14. Chinese Solar Companies Tied to Use of Forced Labor Ana Swanson and Chris Buckley; New York Times; 8 Jan 2021
  15. Fears over China’s Muslim forced labor loom over EU solar power AITOR HERNÁNDEZ-MORALES, KARL MATHIESEN, STUART LAU AND GIORGIO LEALI; Politico.EU; 10 Feb 2021
  16. Calculations are on this Google sheet