Photovoltaic-driven microbial protein production can use land and sunlight more efficiently than conventional crops
Edited by William H. Schlesinger, Cary Institute of Ecosystem Studies, Millbrook, NY, and approved April 13, 2021 (received for review July 16, 2020)
Significance
The cultivation of microbial biomass, which is rich in proteins as well as other nutrients, can play a vital role in achieving food security while mitigating the negative environmental footprint of agriculture. Here, we analyze the efficiency associated with using solar energy for converting atmospheric CO2 derived from direct air capture into microbial biomass that can feed humans and animals. We show that the production of microbial foods outperforms agricultural cultivation of staple crops in terms of caloric and protein yields per land area at all relevant solar irradiance levels. These results suggest that microbial foods could substantially contribute to feeding a growing population and can assist in allocating future limited land resources.
Abstract
Population growth and changes in dietary patterns place an ever-growing pressure on the environment. Feeding the world within sustainable boundaries therefore requires revolutionizing the way we harness natural resources. Microbial biomass can be cultivated to yield protein-rich feed and food supplements, collectively termed single-cell protein (SCP). Yet, we still lack a quantitative comparison between traditional agriculture and photovoltaic-driven SCP systems in terms of land use and energetic efficiency. Here, we analyze the energetic efficiency of harnessing solar energy to produce SCP from air and water. Our model includes photovoltaic electricity generation, direct air capture of carbon dioxide, electrosynthesis of an electron donor and/or carbon source for microbial growth (hydrogen, formate, or methanol), microbial cultivation, and the processing of biomass and proteins. We show that, per unit of land, SCP production can reach an over 10-fold higher protein yield and at least twice the caloric yield compared with any staple crop. Altogether, this quantitative analysis offers an assessment of the future potential of photovoltaic-driven microbial foods to supplement conventional agricultural production and support resource-efficient protein supply on a global scale.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Food security is a critical issue that humanity faces in this century. The combined effect of population growth and increasing consumption of animal-based products are projected to cause a surge in demand for food which could severely challenge global production by 2050 (1, 2). Moreover, the regional impacts of climate change pose a threat to future food security in many countries (3). Although, historically, the food supply has expanded alongside increasing demand, major improvements to crops are now slowing as they approach biological constraints (4, 5). At the same time, agricultural land expansion has limited potential to increase supply, since food production currently occupies more than a third of the Earth’s terrestrial surface (6) and already exerts large environmental burdens (7–10). Therefore, addressing food security requires societal changes as well as innovations in the global food system that go beyond conventional agriculture. In the current study, we explore the potential for the cultivation of microbes to help address this global challenge.
Production of nutrient-rich foods derived from microbial biomass, better known as microbial protein or single-cell protein (SCP), offers a promising means to address food security without exacerbating pressure on the environment, as it utilizes water and nitrogen more efficiently than plants (11–14). Several companies are already producing SCP derived from algae, fungi, or bacteria at commercial scale destined for animal or human consumption (15). The feedstock used to cultivate these microbes is typically either agriculturally derived glucose or fossil-derived methane and methanol (11). Yet, a more sustainable alternative, which minimizes reliance on fossil carbons and agricultural land, is to use renewable energy (here, photovoltaics) to convert atmospheric carbon dioxide and water into molecules that can serve as electron donors for microbes (16, 17). Previous studies have considered the land requirements for SCP production using feedstocks derived from agriculture, fossil fuels (12), and, more recently, also renewable energy (18). Nevertheless, a comprehensive assessment of land and energy efficiency of fully photovoltaic-driven microbial food production is still lacking. Focusing on solar energy allows us to compare the potential of food production using microbes against contemporary agriculture on an even playing field, since both technologies rely on the same primary resources (i.e., land, sunlight, water, and fertilizers). More specifically, this study sought to answer how productive photovoltaic-driven SCP (PV-SCP) systems can be in terms of calorie and protein production per unit time and land area in comparison to other SCP systems and to conventional crops, focusing on the effect that solar irradiance has on PV-SCP yields. This quantitative comparison can assist in planning the future allocation of limited land resources toward feed and food production.
We used literature data to calculate the overall efficiency by which solar energy can be harnessed to generate SCP, considering different electron donors and metabolic pathways. We decided to focus on bacteria since they are flexible in their use of feedstock and reach higher protein content than other microorganisms (11). We assumed that all carbon requirements are met by direct air capture of carbon dioxide from the atmosphere (DAC) (19) in order to minimize the reliance on fossil fuels as well as to support a fair comparison with plants. We further took into account other energetic expenditures, such as production of macronutrients for microbial cultivation, bioreactor stirring and cooling, and downstream processing of biomass and proteins. We show that PV-SCP technologies can substantially outperform conventional staple crops in terms of both calorie and protein yield.
Results
Energetic Efficiency of SCP Production.
We considered a PV-SCP system that converts solar energy into energy stored in food by the following four generalized steps (Fig. 1):Process (1) corresponds to PV solar farms capturing solar energy and converting it to electricity. Process (2) represents the electrochemical conversion of electrical energy into chemical energy stored in an electron donor and/or carbon source. Process (3) refers to microbial growth which converts the chemical energy from the previous step into chemical energy stored in biomass. Process (4) describes a filtration step whereby nucleotides, fatty acids, and carbohydrates are discarded while only the protein is retained. The removal of nucleic acids is crucial when SCP serves as a human food since in too high of concentrations, their catabolism leads to an accumulation of uric acid, which cannot be easily degraded and can form gout (20). Unlike humans, all farm animals possess the enzyme uricase, which precludes this effect, therefore making nucleic acid removal unnecessary for feed production. Each of these processes is associated with different energetic efficiencies—ηpv, ηec, ηbio, and ηfilter (Fig. 1)—which we calculate according to available measurements, as explained in Methods. These four steps describe the direct transfer of energy from solar to biochemical storage in food. However, operating the SCP system also requires several electricity inputs not depicted in this linear chain, and we account for all of them by introducing another efficiency term η*, which is described below. For example, η* accounts for the energetic cost of operating DAC which supplies the CO2 required at steps (2) or (3).
Fig. 1.
With respect to the first step, solar energy capture, it is common to report the energetic efficiency of PV solar energy conversion to electricity (ηpv) as approaching 20%, which represents the solar cell efficiency under standard test conditions (21). However, this neglects numerous practical factors, most importantly PV ground coverage ratio, and losses due to power electronics, solar tracking, inverter, and temperature as well as surface soiling from dust, snow, and other debris (22). To obtain a more realistic view of solar farm efficiency, we used available data from >600 utility-scale sites (Dataset S1A). As explained in Methods, we found that ηpv ranges between 4.1% and 5.6% (30th to 70th percentiles), considerably lower than the solar cell efficiency. If PV technology is replaced by concentrated solar power (e.g., parabolic trough, power tower, or linear Fresnel reflector), the energetic efficiency is even lower (Dataset S1B); hence, we did not consider this latter technology further.
Following the conversion of solar energy to electricity, electrical energy is converted into chemical energy by producing simple molecules (electron donors), which support microbial growth. To provide a broad perspective on the properties of different electrochemical and biological processes, we considered three electron donors (hydrogen, methanol, and formate) and several microbial assimilation pathways. For the production of all of the three electron donors considered, water is first split and oxidized at an anode to provide electrons and oxygen (O2) to the processes. Carbon dioxide (CO2), which is the only primary source of carbon in the production process, is obtained via DAC of CO2, and, as shown in Fig. 1, there are two possible entry points for CO2 (SI Appendix).
The energy efficiency associated with the production of an electron donor (ηec) is determined by its combustion energy divided by the invested electrical energy. Hydrogen is the most commonly used electron donor, the production of which is relatively efficient via electrolysis, with an ηec of 70% (±5%) (23). Aside from hydrogen, only formate and carbon monoxide (CO) can be directly produced electrochemically at commercially relevant energetic efficiency, Faraday efficiency, and current density (24). We chose to focus on formate as it is miscible, therefore bypassing mass transfer issues that constrain the bio-consumption of low-solubility gaseous compounds such as CO. Moreover, formate can be assimilated into biomass much more efficiently than CO (25). Electrochemical reduction of CO2 to formate is characterized by an ηec of 40% (±10%) (24). Electron donors can also be produced from electricity indirectly, whereby electrolysis-derived hydrogen is reacted with CO2 to generate a reduced compound. As an example of this approach, we focused on the production of methanol, another miscible compound, as its two-step production from electricity is relatively efficient, with an ηec of 55% (±5%) (24).
As to the microbial assimilation pathways, we focused on bacteria as they are metabolically flexible in terms of electron donor and/or carbon source utilization and can reach higher protein content than algae and yeast (11). We considered only aerobic growth, since anaerobic growth (e.g., acetogenesis) diverts most of the carbon into nonprotein-excreted compounds (e.g., acetate) resulting in low biomass yield. Aerobic growth on H2/CO2 is mainly supported by the Calvin cycle. Due to the relatively high ATP cost of this pathway, the energetic efficiency associated with this growth is relatively low, ηbio of 32% (±5%) (25) (Methods). The Calvin cycle can also support growth on formate or methanol—with ηbio of 27% (±6%) and 21% (±2%), respectively (25) (Methods)—via complete oxidation of these electron donors to support de novo carbon fixation. Microbial growth on formate and methanol can instead proceed via direct assimilation of these carbon sources into biomass, which generally supports a higher energetic efficiency. The serine cycle assimilates formate with a high ηbio of 46% (±7%) and methanol with ηbio of 35% (±2%) (25) (Methods). The Ribulose Monophosphate (RuMP) cycle also supports a high efficiency of methanol assimilation, with ηbio of 45% (±3%) (25) (Methods). We note that the above figures are derived from laboratory experiments rather than industrial-scale facilities due to lack of data. However, we do not expect drastic changes to ηbio at scale since this factor is mostly determined by the stoichiometry of the metabolic pathways in each organism.
We considered two SCP production scenarios leading to an end-product of either feed or food. To produce feed, the entire microbial biomass is dried (by centrifugation and spray-drying) and directly used as feed for animals. To produce food, after dewatering by centrifugation, the proteins are extracted from biomass and purified to discard nucleic acids (e.g., by bead milling followed by microfiltration) and used as a food supplement. As previously mentioned, it is imperative to discard nucleic acids from SCP food products (11). We assumed that the usable cellular protein content lies between 55 to 75% on a dry mass basis, which is characteristic of microorganisms used for SCP production (11, 26). This translates to an energetic efficiency of converting biomass to protein, ηfilter, of 46% to 63% (Methods).
A simplistic expression for the overall energetic efficiency of converting solar energy to a feed or food product, ηscp, should be given by ηpv × ηec × ηbio × ηfilter (in which the last term is not needed in the case of animal feed). However, this neglects the energy required to support process operation, that is, the energy not directly transferred from the solar input into the final product and hence that is not accounted for in the above production chain. For the electrochemical process, the value of ηec, taken from the literature, already includes the energy required for process operation (Methods). Yet, as shown in Fig. 1, five processes have nonnegligible energy inputs that need to be accounted for: DAC of carbon dioxide (Dataset S1C), supply of macronutrients (Dataset S1D), bioreactor operation (Dataset S1D), and downstream processing of biomass (Dataset S1D) (Methods). We assumed that PV-derived electricity provides energy for all these processes (27). Hence, as depicted in Fig. 1, the initial electrical energy produced is split between the main production chain and the supporting processes. The fraction of electrical energy used for electrochemistry over the total electrical energy produced is equivalent to an effective electricity use efficiency, which we denote as η* (Methods). η* depends on other efficiencies, ηec and ηbio, and hence is not an independent factor. Note also that 1- η* is thus the fraction of energy diverted to the supporting processes. As shown in Table 1, η* ranges between 64% and 83% (i.e., between 17% and 36% of the electricity produced is consumed by the supporting processes) (Methods).
Table 1.
Refer to Fig. 1 and main text for the definitions of the different efficiencies. Each row represents a specific combination of electron donor, metabolic pathway, and type of final product (feed in orange or food in yellow). The overall energy efficiencies, ηscp, for feed and food production are presented in bold. Processing of biomass to produce feed entails zero loss (ηfilter = 100%), while food production requires the filtration of all nonprotein mass (therefore, ηfilter = 55 ± 9%). Values of ηpv, ηbio, and ηfilter are mean estimates (±uncertainty), in which the range represents the 30th and 70th percentiles. The sources for the values of ηec are described in detail in Methods. η* and ηscp were calculated using Eqs. 5 to 7, and their uncertainty is derived by error propagation (Methods).
The overall energetic efficiency of converting solar energy to the food product, ηscp, is therefore given by including η* in the previous equation (i.e., ηpv × ηec × ηbio × ηfilter × η*). In the case of feed as the final product, ηfilter is omitted. The energetic efficiency, ηscp, associated with different SCP production routes is reported in Table 1. Methanol production and assimilation via the RuMP cycle supports the highest ηscp (up to 0.8% for feed production and 0.4% for food production) due to the high energetic efficiency of microbial growth on methanol (ηbio). A similar ηscp value can also be achieved by growth on hydrogen, despite its reliance on the relatively inefficient Calvin cycle, due to the high efficiency by which this electron donor is produced electrochemically (ηec). Combining the Calvin cycle with other electron donors reduces ηscp by half (0.4% for feed and 0.2% for food). Finally, growth via the serine cycle, either on methanol or formate, results in a relatively high ηscp (up to 0.7% for feed production and 0.3% for food).
We note that our study included only the operational energy costs of SCP production (i.e., including N, P, and S provision), while energy expenditures for the construction of production facilities were not accounted for. In addition, the DAC process consumes reagents (e.g., sorbents) whose energetic costs of production are difficult to assess due to lack of data (28, 29), and these energy requirements were hence excluded.
Caloric Yield of Agriculture.
We compared PV-SCP with agriculture in terms of caloric and protein yield per unit land area. We selected the highest protein- and calorie-yielding crops, soybeans and sugar beet respectively, and the seven other staple crops whose total global production by mass is the highest: sugar cane, maize, rice, wheat, potato, cassava, and oil palm. We used a three-year average of the 2017 to 2019 Food and Agriculture Organization Corporate Statistical Database (FAOSTAT) dataset that contains crop mass yields (accounting only for the edible portion of crops) for over 180 nations (Dataset S1E). To obtain a representative value of agricultural productivity that integrates regional differences and avoids outliers, we used production-weighted yields, calculated by averaging each country’s yield weighted by its share of global production. Last, we used the FAO nutritional composition table, which reports the global average protein and kcal content of crops (also only considering the edible portion) to convert crop yields into protein and kcal yields (Dataset S1E). Importantly, food composition tables from various sources (30) suggest that the composition of any particular crop (e.g., due to the use of different cultivars) varies far less compared with regional differences in yield. Therefore, for consistency, we used the single FAO nutritional composition value to convert each crop’s mass yield into protein and caloric yields. We note that each plot of land is assumed to grow one type of crop per year.
Table 2 summarizes the calculated yields of the staple crops. We found soybean to have the highest protein yield, ≈115 g · m−2 · y−1. The crop with the highest caloric yield is sugar beet, ≈4,520 kcal · m−2 · y−1, followed by maize and oil palm, ≈2,640 and ≈2,650 kcal · m−2 · y−1, respectively. All other crops in Table 2 have lower energy and protein yields, and therefore, our conclusions apply to them as well.
Table 2.
| Crop | Yield (weighted average) | World total | ||
|---|---|---|---|---|
| Protein [g · m−2 · y−1] | Energy [kcal · m−2 · y−1] | Area [Mha] | Fresh weight [Mtons · y−1] | |
| Soybeans | 115 | 1,010 | 123 | 346 |
| Sugar beet | 84 | 4,520 | 5 | 289 |
| Maize | 71 | 2,640 | 195 | 1,133 |
| Wheat | 51 | 1,390 | 216 | 757 |
| Potatoes | 40 | 1,680 | 17 | 368 |
| Rice, paddy | 31 | 1,450 | 157 | 713 |
| Sugar cane | 15 | 2,220 | 26 | 1,880 |
| Cassava | 13 | 1,560 | 27 | 291 |
| Oil palm | 5 | 2,650 | 28 | 405 |
The yield reported corresponds to a weighted average in which each nation’s yield is weighted by its contribution to the global output of the crop (Dataset S1E). Kilocalorie and protein yields are calculated using the FAO nutritional composition table (Dataset S1E).
We emphasize that these crops have varied roles in society, which may make SCP either more or less suitable to replace them. For example, soybean is primarily cultivated for the production of protein-rich animal feed and is well suited for substitution by SCP production (12). As sugar beet is primarily cultivated to produce sucrose and is rarely used as a stand-alone food source, it may be more difficult to substitute with SCP, though it too plays a role in feed production. Sugar beet exhibits relatively high protein yield due to its high biomass yield per unit land, yet its low protein content (≈1.3%) makes it impractical as a source of protein additive for animal feed. Maize is grown for food, feed, and bioenergy production. Oil palm is a major source of fatty acids in the food industry, but since in this analysis we only consider protein for food, it is not a prime candidate for replacement with SCP (31).
Caloric Yield of PV-SCP Production Can Surpass That of Crops.
To compare microbial foods with agriculture, we calculated the caloric yield Ycal of PV-SCP production in units of kilocalories per square meter per year. Ycal equals the previously calculated ηscp multiplied by the relevant irradiance levels (I), for which we assumed a global range of 700 to 2,700 kWh · m−2 · y−1. However, we identified a statistically significant (P value < 0.001) negative correlation between irradiance and the energetic efficiency of utility-scale solar farms (ηpv), which might be attributed to the known negative correlation between temperature and cell voltage (32). We derived a fitted unitless “correction function,” fC = 1.6 – I/(2,800 kWh · m−2 · y−1) (Methods) to account for the effect of irradiance on solar farm efficiency. Using this correction, the predicted ηpv efficiency at 1,000 kWh · m−2 · y−1 is 6.0%, while at 2,500 kWh · m−2 · y−1, it is reduced to 3.4%. Altogether, Ycal = I × fC × ηscp. We note again that in feed production, Ycal refers to the energy stored in all cellular components, including proteins, carbohydrates, and lipids, while for food production, Ycal refers only to the energy stored in the protein fraction of the microbial biomass.
Fig. 2 shows the expected Ycal for SCP production under the two scenarios, feed (above) and food (below), for different combinations of electron donor and assimilation pathways. Our results indicate that PV-SCP production can support higher energy yield than agriculture in both scenarios (i.e., when considering all cellular components or only the protein content). Specifically, at irradiance >1,600 kWh · m−2 · y−1, the PV-SCP technology can double, triple, or, in some cases, even quadruple agricultural caloric yield for the production of animal feed. It can also support a caloric yield for human food that is more than double that of cereals such as maize, rice, and wheat. A practical alternative SCP production route that bypasses the need for electrochemical reduction or specialized microorganisms (i.e., those that can utilize hydrogen/formate/methanol) is growing heterotrophic microbes on sugars derived from agriculture. For example, Quorn manufactures food products in this way by cultivating the fungus Fusarium venenatum on wheat-derived glucose (33). Here, we used sugar beet–fed SCP production as a reference to compare PV-SCP yields against agricultural-based SCP production. As shown in Fig. 2, the energy yield obtained in this case (yellow strip), too, is considerably lower than that of PV-SCP.
Fig. 2.
Protein Yield of PV-SCP Production Substantially Surpasses That of Crops.
Next, we focused on the production of proteins and compared the protein yield obtained from agricultural sources (Table 2) with that expected from PV-SCP production. We converted the caloric yield obtained from the previous analysis on food production to units of mass in grams protein (Methods). As demonstrated in Fig. 3, the protein yield of PV-SCP is much higher than that of soybean—the crop displaying the highest protein yield (≈115 g · m−2 · y−1)—regardless of the electron donor and metabolic pathway used. Growth on hydrogen (via the Calvin cycle), as well as growth on methanol or formate via the serine and RuMP cycles, respectively, could support an order of magnitude higher protein yield (>1,200 g · m−2 · y−1). Although data about the efficiency of converting solar energy into food for crops is not globally available, we calculated this efficiency for maize, sugar beet, and soybean for several countries (Dataset S1F and SI Appendix). We observed that the advantage of PV-SCP in terms of efficiency corroborates our findings in terms of caloric and protein yields. Even the cultivation of microbes on sucrose extracted from sugar beet (yellow strip) supports substantially higher protein yield than agriculture, though not as high as PV-SCP.
Fig. 3.
To give a more tangible picture of the energy demand for PV-SCP, we analyzed how a land area of one hectare (10,000 m2) that receives 2,000 kWh · m−2 · y−1 of solar energy should be divided so as to support the different components involved in PV-SCP. As shown in Fig. 4A, when using hydrogen as an electron donor, about 6,650 m2 should be devoted to PV electricity production for water splitting. The energy required for microbial cultivation (which includes the provision of nutrients, bioreactor operation, and biomass downstream processing) and DAC of CO2 would require ≈2,400 m2 and ≈900 m2 of PV arrays, respectively. We note that the land area required for the production facilities (e.g., electrolyzer, bioreactor, and DAC of CO2) is negligible compared with the area of the PV arrays needed to energize the processes and hence were not included in the calculation (12, 27). Overall, one hectare devoted to PV-SCP in this design could supply the protein requirement for more than 500 people (15 tons of protein per year; Fig. 4A; Methods).
Fig. 4.
To produce SCP from sucrose that is extracted from sugar beet, ≈9,400 m2 should be allocated for the cultivation of the sugar beet, while only ≈600 m2 would be covered with PV arrays to energize microbial cultivation, sugar beet cultivation, and sucrose extraction (Fig. 4B). Such a system is interesting as a simple intermediate solution which uses only mature technologies and does not rely on DAC or electrolysis. Furthermore, it can be viewed as a benchmark for existing microbial food facilities such as Quorn. For simplicity, here we assumed that the varied types of energy used in sugar beet cultivation are equivalent to an electrical energy input that is provided by PV. This production strategy could supply protein for 90 people (2.7 tons of protein per year). Finally, cultivation of soybean would produce only enough protein to feed 40 people per hectare (1.1 tons of protein per year; Fig. 4C), which is about a tenth of the people that can potentially be nourished by PV-SCP. The energy inputs into processes supporting soybean cultivation (e.g., for fuel and fertilizers) were not considered here, as they are not met by PV in conventional agriculture. Nonetheless, their inclusion would further lower the yield of soybean cultivation.
Discussion
We compared PV-SCP production and agriculture according to the expected annual nutritional yield per land area. Our analysis revealed that under the relevant irradiance levels, the total caloric yield obtained from PV-SCP production could be higher than that achieved by the agricultural cultivation of all staple crops. The potential advantage of PV-SCP production becomes even more pronounced when considering its protein yield, which is an order of magnitude higher than that of soybean (by far the highest protein-yielding staple crop; Table 2). We further analyzed an alternative SCP production system which relies on agricultural sugars to feed microbes—rather than electrochemistry, DAC, and specialized metabolic pathways. Although this system yields at least twice the amount of protein compared with soybean, it is still outperformed by PV-SCP by at least fivefold (Fig. 4). These results highlight the potential of next-generation SCP systems as highly land- and energy-efficient alternatives to plant-based protein.
Our study indicates that the choice of electron donor and metabolic pathway can affect the overall yields of PV-SCP production. We find that both hydrogen and methanol (in combination with the Calvin and RuMP cycle, respectively) can achieve the highest energetic efficiencies. On the other hand, hydrogen has very low solubility in water, limiting its transfer to the growing cells and thus constraining cell titer and productivity (25). Furthermore, its dissolution into the growth media requires intense mechanical stirring which generates additional heat and increases bioreactor cooling costs (34). Methanol and formate, which are liquid in atmospheric conditions and completely water miscible, bypass these constraints and can be easily stored and transported—unlike hydrogen and oxygen which require the implementation of expensive safety equipment and procedures. This facilitates the spatial and temporal decoupling of the electrochemical process from microbial cultivation, thus insulating it from the intermittent nature of renewable energy. The downside of using methanol and formate is their tendency to inhibit cell growth at high concentrations, which mandates regulated cultivation regime and could hinder high cell titers and productivities (25).
The possible order-of-magnitude improvement in protein yield stems from at least four facets of the PV-SCP production process: 1) the higher protein content of SCP biomass (>65%) compared with soybean grain (≈38%) (Dataset S1E), 2) greater allocation toward the edible portion of SCP biomass (55 to 75%) (11) compared with soybean (≈34%) (SI Appendix), 3) the higher energetic efficiency by which silicon-based PV cells combined with electrochemistry can convert light energy into energy stored in chemical bonds as compared with plant-based photosynthesis (21), and 4) the seasonality of plant growth, such that a large fraction of annual sunlight is not intercepted by crops.
The relevance of our results is strengthened by our use of empirical data to estimate the efficiency of the production steps (i.e., rather than relying on theoretical estimations of process parameters). For example, while it is commonly assumed that the conversion of solar energy to electricity can operate at close to the solar cell efficiency (≈20%), empirical data clearly show that the actual annual efficiencies at utility scale are much lower (≈5%). On the other hand, the nascent power-to-food SCP processes present pragmatic avenues for further efficiency improvements. For example, several emerging and maturing PV technologies (so-called third-generation PV) can surpass the Shockley–Queisser limit of 34% efficiency and have the potential to dramatically increase ηpv (35, 36). Multijunction PV has a theoretical upper limit of 68% for nonconcentrated sunlight (36). As there is a strong market incentive to improve PV efficiency outside of food production, such technological gains are expected to develop rapidly in the coming decades. In addition, the emergent “artificial photosynthesis” technologies which directly convert solar energy to chemical energy through photoelectrochemical reactions and particulate photocatalysts may one day improve the efficiency of harnessing solar energy for SCP production by combining ηpv and ηec into a single step (37, 38). Regarding ηec, we note that while water electrolysis is a relatively mature technology, electrosynthesis of formate and methanol are at an early stage of development. Hence, improvements in these technologies are expected to boost their ηec.
Similarly, on the biological side, recent progress in designing and engineering synthetic metabolic pathways for microbial assimilation of formate and methanol with higher energetic efficiency could assist in boosting ηbio (39). In addition, focusing on microbes with high protein content or, alternatively, engineering high protein content in industrial model organisms could increase ηfilter. Although crops may also be bred to improve yields, the gains in productivity throughout the Green Revolution were chiefly driven by higher partitioning of biomass to edible grain, and these are now near their upper theoretical limit (4). Hence, the next major improvements to plant yields are likely to be unlocked by profound redesigns of plant biology (e.g., by redesigning the photosynthetic apparatus to capture a larger spectrum of solar energy) or improving carbon fixation (21). Additionally, agriculture may become more sustainable if plants are engineered for more effective nitrogen assimilation, for instance, by enabling symbiosis between nitrogen fixing bacteria and cereal crops or by directly introducing nitrogenase into plants (with these latter two having the added environmental benefits of decreasing the nitrogen pollution and greenhouse gas [GHG] emissions associated with the use and synthesis of N-fertilizers) (4, 40–42). Despite the potential that genetically engineered crops and bacteria offer in terms of efficiency and mitigation of environmental impact, there is growing demand for genetically modified organism (GMO)-free products. Therefore, we focused only on naturally occurring microorganisms to demonstrate the relative benefits of SCP. It is worth noting, however, that many of the crops considered in our comparison are in fact typically cultivated as GMOs, especially soybean and US maize (43, 44).
Ultimately, wide-scale adoption of SCP technologies will be mainly contingent upon their economic competitiveness in the food and feed markets. Recent studies have shown that various SCP production pathways, including hydrogen- and sugar-based systems, could achieve costs that are in line with high-quality feed additives of animal origin such as fishmeal (45). We created preliminary cost estimates for the PV-SCP pathways envisioned in the present study, involving PV electricity, DAC of CO2, and green ammonia synthesis. Although our estimates do not represent a comprehensive techno-economic analysis, they enabled us to qualitatively compare the various PV-SCP pathways and draw general conclusions about the relative contribution of the main cost components. For instance, we find that the majority of costs are associated with the production of the electron donors. Our estimates for hydrogen-based SCP feed (≈$2.6 per kg-dw-biomass) are in agreement with findings from the above-mentioned studies (45) and show a similar range to that of methanol-based SCP feed produced through the RuMP metabolic pathway (≈$2.8 per kg-dw-biomass), whereas formate-based production is considerably more expensive (SI Appendix, Table S1 and Dataset S1G). When expressed on a per unit protein basis, these production costs are about $4 to $5 per kg-protein, which reveals costs higher than the market price of fishmeal (≈$2.5 per kg-protein) and soybean meal (≈$1 per kg-protein) (SI Appendix, Table S1). The food market, on the other hand, is characterized by higher market prices for protein, which could make SCP’s production costs viable (46) (SI Appendix, Table S1). For instance, the prices of commonly used whey protein ($7 per kg-protein) and vegetable proteins such as pea ($5 per kg-protein) are on par with our projected PV-SCP production costs. When compared with the prices of mycoprotein ($13 per kg-protein), the main building block of Quorn products, or with emerging alternatives such as insects ($40 per kg-protein), PV-SCP could more easily become price competitive.
We note that the range of energy efficiencies in PV-SCP discussed above is in part a function of technology choices and capital expenses, with larger investments unlocking higher efficiencies. Future trends in the availability of arable land may have a substantial effect on the economics of efficient but expensive solar harnessing technologies by making land-inefficient technology choices increasingly unprofitable.
Clearly, there are other challenges for marketing SCP as human food beyond just its price, such as meeting safety standards and demonstrating health benefits, overcoming regulatory hurdles, promoting consumer acceptance, and improving palatability. Although these market trends are difficult to predict, consumer acceptance does seem achievable since fermentation by bacteria and fungi has traditionally been used for processing many types of food (e.g., Saccharomyces cerevisiae for bread, Lactococcus for dairy products, and Aspergillus oryzae for soy sauce). Furthermore, contemporary companies such as Quorn have demonstrated successful marketing campaigns for fungi-based products (11). From a nutritional perspective, bacterial SCP has a high-quality amino acid profile, which is richer in essential amino acids than soymeal and close to the quality of fishmeal (47). Bacterial SCP is also rich in B vitamins (B1, B2, B3, and B8) (47) and can serve as a valuable complement for plant-based diets which tend to lack vitamin B3 (48). Furthermore, bacterial SCP provides an array of key minerals, including iron, zinc, calcium, phosphorus, potassium, sodium, magnesium, copper, and manganese (47), which encompass multiple micronutrients known to be deficient in the diets of several large populations worldwide (49). Follow-up studies should look further into which commodities are most readily substituted by SCP, incorporating a quantitative analysis of the micronutrients, as well as economic and societal considerations (e.g., cultural and taste preferences). Finally, microbial cultivation can be used for the production of various other commodities, beyond protein. For instance, some natural bacteria produce large amounts of palmitic and oleic acids (50), which are the primary components of palm oil. Hence, PV-driven microbial biomass could potentially also alleviate the environmental pressure of oil crops (51).
In all aspects except price, it seems that SCP is better suited to play a major role in substituting protein-rich animal feeds such as soymeal and fishmeal (12) rather than substituting human foods. Market penetration of SCP for feed is already validated by the success of SCP feed manufacturers such as Unibio and Calysta (11). Notably, several studies have shown that SCP provides health benefits to animals when included at up to ≈50% of their diet, particularly in the case of aquaculture (15, 52, 53). After extensive trials, the European Union approved bacterial SCP feeds (grown on natural gas) in 1995 as nutritional support for the diets of pigs, veal calves, and salmon (at levels of up to 8%, 8%, and 33% of their total feed, respectively) (47). Moreover, due to the rapid growth of the aquaculture sector and concomitant concerns of overfishing for fishmeal, there is an increasing commercial interest in microbial proteins destined for aquaculture (15, 54). Although the projected PV-SCP production prices are currently higher than soymeal or fishmeal, SCP could become more economically viable if the externalized costs linked to pollution, GHG, and ecosystem disruptions caused by extensive soy cultivation and fishing were reflected in prices by policy incentives (45). Furthermore, GMO-based technologies could open new markets by lowering the production costs and providing customized feed-additives (55), yet this may in turn prevent the sale of SCP in countries that heavily regulate GMOs.
Further research is needed to broaden our understanding of the overall impacts of substituting agricultural crops with PV-SCP. To our knowledge, no SCP studies have yet considered the life cycle of PV arrays within their analysis (56). Clearly, sustainable production and recycling of PV arrays are vital for PV-SCP to be environmentally friendly (57).
Yet the technology shows promise, as previous reports demonstrate that SCP uses water and nitrogen resources far more efficiently than crops (54, 58). Modern agriculture relies on large inputs of chemical nitrogen fertilizers to support crop productivity due to low N uptake by plants (≈50%) (59). In contrast, in-reactor SCP production can be tuned to use all the supplied reactive N, thereby performing conversion of inorganic N to edible protein without losses (59). Furthermore, plants and animals use ≈100 and ≈10,000 times more water than SCP (54, 60). This makes SCP an attractive option for regions facing both water and food security risks.
Taken together, production of SCP provides a compelling alternative for the sustainable supply of nutrients, which can rival and outperform contemporary agriculture in many aspects. Our analysis indicates that diverting land resources toward SCP production can help close the approaching “protein gap” (1) while curtailing further agricultural land expansion, thus safeguarding biodiversity and the carbon sink potential of forests and grasslands. Importantly, instead of directly competing for land resources, PV-SCP production can rely on land unsuitable for agricultural use (including urban areas and deserts), rendering a flexible and logistically efficient production process (58). In addition, PV-SCP production is considered to be climate independent and hence can mitigate food supply risks posed by climate change (56). Although SCP currently faces some challenges such as consumer acceptance in the food market or the competitive pricing of the feed market, its commercial viability is likely to improve as land resources become scarce and conventional food sources become increasingly expensive and unsustainable. Furthermore, the high resource efficiency that characterizes SCP production in terms of energy, land, water, and nutrient use also make it a prime candidate to support food production in future long-term missions in space and permanent settlements on extraterrestrial bodies (61, 62).
Methods
Solar-to-Electricity Energy Efficiency.
The first step of the solar-to-feed/food process is the conversion of solar energy to electricity. We calculated the energy efficiency of this process, ηpv, using available information on 628 utility-scale (>1 ha) PV solar farms (including 347 from the United States, 73 from Japan, 35 from France, and 28 from China; Dataset S1A):
[1]
where Eout is the annual electrical energy output of a solar farm, A is the total area of the solar farm, and Iin is the local annual irradiance energy incident per unit area. Electrical output and solar farm size were provided by Wiki-Solar as annual design output and, when available, by the US Energy Information Administration as average annual electrical output (Dataset S1A). Annual irradiance, for which we used Global Horizontal Irradiance, was determined using Solargis Prospect tool (63). The median ηpv was 4.9%, which implicitly incorporates several factors. We defined the lower and upper bounds of ηpv as the 30th and 70th percentile: 4.1% and 5.6%. We note that relatively low ηpv values, compared with ≈20% solar cell efficiency, are mainly attributed to ≈50% solar panel ground coverage ratio (to prevent interrow shading) (64, 65).
The median energetic efficiency associated with concentrated solar power is 4.1% (Dataset S1B), that is, lower than that of photovoltaic farms, and hence was not further considered.
Electricity-to-Electron Donor Energy Efficiency.
The energetic efficiency of converting electrical energy to chemical energy stored in an electron donor molecule (i.e., microbial feedstock) is described by ηec:
[2]
where YED is the electron donor yield per unit of input electrical energy and ΔH°ED is the electron donor heat of combustion on a lower heating value basis (reflecting stored energy). , , and correspond, respectively, to the energy efficiency of producing hydrogen, formate, and methanol. Based on a literature survey, (66), (16, 67–69), and (24, 70). We note that and include also peripheral energy requirements, such as heating, cooling, pumping, compression, airflow, etc. For formate electrosynthesis , no such data on peripheral energy requirements is available; hence, we used the relatively wide range given above for its energetic efficiency.
In each case, water supplies electrons to produce the electron donors.
For hydrogen and formate synthesis, 1 mol of water is required per mol of electron donor produced. Methanol synthesis requires two steps. In the first, 3 mols of H2O are electrolyzed to produce 3 mols of H2 and 1.5 mols of O2. Then, the H2 stream is channeled to a separate catalytic reactor and reacts with CO2 to generate 1 mol of methanol while also regenerating 1 mol of H2O. In the overall reaction, 2 mols of water are required to produce 1 mol of methanol.
Electron Donor-to-Biomass Energy Efficiency.
The efficiency with which microorganisms convert the chemical energy stored in electron donors (i.e., microbial feedstock) into biomass is described by ηbio:
[3]
where YB is the biomass yield per mol electron donor consumed, ΔH°B is the heat of combustion of bacterial biomass (i.e., stored energy in biomass) taken as 20 MJ · kg-dw−1 based on experimental result (71), and ΔH°ED is as previously defined in Eq. 2. YB values were taken from a recent study (25). For lower and upper bounds of ηbio, we took the 30th and 70th percentile of the values of each combination of electron donor and assimilation pathway.
Biomass-to-Food Product Efficiency.
The final step of food-grade SCP production is the conversion of wet biomass into protein by discarding all other cellular components, which we associated with the energy efficiency ηfilter. We assume that the filtering process maintains 100% of the protein. We calculated ηfilter as follows:
[4]
where ρ is the fraction of usable protein in biomass on a gram per gram cell dry weight basis, ΔH°P is the heat of combustion of protein in kJ per gram protein, and ΔH°B is as previously defined in Eq. 3. ρ is taken from literature and falls within a range of 55 to 75% (11, 72). ΔH°P is taken as 16.7 MJ · kg−1 (73). The energetic costs required to extract protein from biomass are included in the microbial cultivation energy, as described below. Note that ηfilter does not appear in the feed production scenario, since, in that case, all cellular components are retained in the final product.
Effective Electricity Use Efficiency, η*.
This represents the fraction of electricity that is used in the electrochemical process for the generation of the electron donor compound (i.e., microbial feedstock). The rest of the electricity produced is distributed among several supporting processes: DAC of CO2, provision of macronutrients for microbial cultivation, bioreactor operation, and biomass downstream processing.
We collected available information regarding the energetic demand of DAC of CO2 using multiple technologies. Dataset S1C shows all values, where 6 and 9 MJ · kg-CO2−1 are the 30th to 70th percentiles. We converted these values to represent energy demand per kg biomass. We assumed that all CO2 released from the bioreactor (e.g., from formate and methanol oxidation to provide cellular energy) is directly recycled without an additional energetic cost. In this case, the energy demand for CO2 capture is directly proportional to the carbon assimilated into the microbial biomass. As the weight fraction of carbon in CO2 is 27% and in biomass it is 48% [assuming biomass formula of CH1.77O0.49N0.24 (74)], we obtained an energy demand for CO2 capture between 11 and 16 MJ · kg-dw−1. As the combustion energy of biomass is 20 MJ · kg-dw−1, the normalized energy demand for CO2 capture, θDAC, ranges between 0.5 and 0.8. Low-temperature solid-sorbent DAC methods may also capture water in humid climates at a stoichiometry of 2 to 5 mol H2O per mol CO2 (19). This water could contribute to the water requirements of the electrochemical and cultivation processes.
The energy requirement for the supply of macronutrients—ammonium, phosphate, and sulfate—was calculated, per kg of dry weight biomass, based on life cycle assessment literature and cell stoichiometry (Dataset S1D). The approximate stoichiometry for producing 1 kg of dw biomass was taken as 1.76 kg CO2 (calculated in Dataset S1C), 0.112 kg NH3 (12), 0.02 to 0.03 kg H3PO4 (75), and 0.01 kg H2SO4 (76). The largest energy input is for provision of green NH3 (29.6 to 39.7 MJ per kg NH3) (77). The total energy cost for the provision of nutrients accounts for 4.8 to 6.6 MJ · kg-dw biomass−1 (Dataset S1D). Normalizing these values by the energy of combustion of 1 kg biomass gives θnut, which lies in the range of 0.24 to 0.33.
Energy demand for the operation of the bioreactor, mainly stirring and cooling, was calculated to lie in the range 7.7 to 15.6 MJ · kg-dw−1 (Dataset S1D and Methods). Such a wide range of values derived from the fermentation industry allowed to consider a broad variety of possible reactor configurations and the related stirring and cooling requirements. The latter might differ greatly based on the substrates used and their solubility, and while this was out of the scope of this study, the considered range encompassed possible variations in this sense. Normalizing the 7.7 to 15.6 MJ · kg-dw−1 by the energy of combustion of 1 kg biomass gives θbioreactor, which lies in the range of 0.39 to 0.78.
The energy demand for biomass downstream processing (i.e., converting the wet biomass into the final product) was calculated differently for feed and food production. In the case of feed production, which includes centrifugation and spray drying, the energy demand was calculated to be 8.4 to 9.1 MJ · kg-dw−1 (Dataset S1D). The energy demand for food production, which includes the additional steps of bead milling and microfiltration, was calculated to be 10.5 to 21 MJ · kg-dw−1 (Dataset S1D). Normalizing these values by the energy of combustion of 1 kg biomass gives θdsp, which lies in the range of 0.42 to 0.46 for feed production and 0.52 to 1.1 for food production.
We note that the overall energy demand for provision of macronutrients, bioreactor operation, and biomass downstream processing is 21 to 31 MJ · kg-dw−1 for feed production and 23 to 43 MJ · kg-dw−1 for food production, which generally agrees with a previous study reporting energy demand of 28 to 32 MJ · kg-dw−1 (78).
The effective electricity use efficiency η* is defined as the fraction of electrical energy used for electrochemistry out of the total amount used in all processes. In order to produce 1 kg biomass, the energy required for electrochemistry is ΔH°B/(ηec x ηbio) and for all other processes is ΔH°B x (θDAC + θnut + θbioreactor + θdsp). Therefore:
[5]
Overall Solar-to-Feed/Food Energy Efficiency.
The overall energy efficiency of the solar-to-feed/food process, ηscp, is calculated by the product of the above-defined efficiencies. For the production of feed, the formula is:
[6]
For the production of food, the extra step of microfiltration discards calorie-containing biomass. Hence, this additional factor is multiplied into the production chain, reducing the overall efficiency:
[7]
Estimation of Error.
To estimate the confidence interval for the ηscp in each scenario, we used standard propagation of uncertainty assuming all variables are independent. The uncertainty of each independent variable was taken as the difference between the 30th and 70th percentiles in our collected datasets (as shown in Table 1). The final uncertainty is the square root of the sum of squares of each variable’s uncertainty times the partial derivative of ηscp with respect to it.
Solar-to-Electricity Efficiency Correction Function.
We identified a statistically significant negative correlation between annual irradiance and solar farm energy efficiencies (P value < 0.0001, n = 628), probably reflecting the known fact that solar cells become less efficient at higher temperatures (32). We fitted a regression equation of irradiance to solar farm efficiency PVR, while systematically discarding outliers. This was performed iteratively, wherein each iteration data point for which the measured efficiency was different from the predicted one by more than 0.03 were discarded, and the regression was recalculated. We found the regression function to be:
[8]
where I is the annual irradiance in the range of 700 to 2,700 kWh · m−2· y−1. We divided PVR by median solar farm efficiency to quantify the deviation from median as a function of irradiance. The resulting equation is the solar correction function, fC:
[9]
At low irradiance (I < 1,680 kWh · m−2 · y−1), fC is greater than 1, and at high irradiance, fC is less than 1.
Yield of Food Energy and Protein.
The yield of nutritional calories per land area and time, Ycal, is given by:
[10]
where I is irradiance ranging from 700 to 2,700 kWh m−2 y−1, ηscp is as in Eq. 6, and fC is as in Eq. 7.
In the case of food production, the energy contained in the final product, Ycal, reflects only the energy in the protein fraction of the biomass. Hence, the SCP system yield in terms of protein mass for food production, Yprot, is determined by dividing the food energy yield by the energy content per unit protein:
[11]
where Yprot is the yield of protein in g · m−2 · y−1 and is as defined in Eq. 4.
Yield of food/feed energy for SCP grown on sucrose extracted from sugar beet.
The production of SCP from sucrose extracted from sugar beet requires two plots of land. In one plot, sugar beet is cultivated. In the other, PV arrays are placed to generate the electricity needed for the extraction of sucrose and cultivation of microbes.
Upon derivation (described in the SI Appendix), we obtain an energy yield of SCP produced via sugar beet as follows:
[12]
Yscal is the energetic yield of sucrose derived from sugar beet, which was calculated as 6.5 kg · m−2 · y−1 (Dataset S1E) multiplied by 16%, the characteristic extractable sucrose content per fresh weight sugar beet (79), and converted to units of energy (considering that 16.7 MJ · kg-sucrose−1) to give 4.8 kWh · m−2 · y−1. θscult, the energy cost of sugar beet cultivation (11 to 28 GJ · ha−1) normalized by sucrose yield and the energy of combustion of sucrose (16.7 MJ · kg−1), ranges between 0.13 and 0.19 (Dataset S1H). θsx is the energy cost of sucrose extraction (0.2 MJ per kg-sugar beet (80) which converts to 1.25 MJ · kg-sucrose−1) normalized to the combustion energy of sucrose, which results in 0.07. θnut, θbioreactor, θdsp, and fC are as defined above. We note that the contribution of and thus the energy required to extract sucrose is negligible compared with the energy required for microbial cultivation. Note that in the case of feed production, the term is omitted from the numerator, which increases the overall yield.
To calculate the protein yield of SCP produced via sugar beet in units of mass per land area and time, we divide YSB-SCP by the combustion energy of protein (16.7 MJ · kg−1).
Protein consumption.
Data Availability
All study data are included in the article and/or supporting information.
Supporting Information
Appendix (PDF)
- Download
- 995.38 KB
Dataset_S01 (XLSX)
- Download
- 376.70 KB
Dataset_S02 (XLSX)
- Download
- 3.72 MB
References
1
M. Henchion, M. Hayes, A. M. Mullen, M. Fenelon, B. Tiwari, Future protein supply and demand: Strategies and factors influencing a sustainable equilibrium. Foods 6, 53 (2017).
2
M. Berners-Lee, C. Kennelly, R. Watson, C. Hewitt, Current global food production is sufficient to meet human nutritional needs in 2050 provided there is radical societal adaptation. Elem Sci Anth 6, 52 (2018).
3
S. S. Myers et al., Climate change and global food systems: Potential impacts on food security and undernutrition. Annu. Rev. Public Health 38, 259–277 (2017).
4
D. R. Ort et al., Redesigning photosynthesis to sustainably meet global food and bioenergy demand. Proc. Natl. Acad. Sci. U.S.A. 112, 8529–8536 (2015).
5
S. P. Long, A. Marshall-Colon, X. G. Zhu, Meeting the global food demand of the future by engineering crop photosynthesis and yield potential. Cell 161, 56–66 (2015).
6
J. A. Foley et al., Solutions for a cultivated planet. Nature 478, 337–342 (2011).
7
A. Arneth et al., “Summary for policymakers” in IPCC Special Report on Climate Change and Land (IPCC, 2019).
8
S. Díaz et al., Summary for Policymakers of the Global Assessment Report on Biodiversity and Ecosystem Services. Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES, 2019).
9
J. Poore, T. Nemecek, Reducing food’s environmental impacts through producers and consumers. Science 360, 987–992 (2018).
10
M. Springmann et al., Options for keeping the food system within environmental limits. Nature 562, 519–525 (2018).
11
A. Ritala, S. T. Hakkinen, M. Toivari, M. G. Wiebe, Single cell protein—State-of-the-art, industrial landscape and patents 2001-2016. Front. Microbiol. 8, 2009 (2017).
12
I. Pikaar et al., Decoupling livestock from land use through industrial feed production pathways. Environ. Sci. Technol. 52, 7351–7359 (2018).
13
T. Cumberlege, T. Blenkinsopp, J. Clark, The Carbon Trust (Carbon Trust, 2016).
14
S. Matassa, W. Verstraete, I. Pikaar, N. Boon, Autotrophic nitrogen assimilation and carbon capture for microbial protein production by a novel enrichment of hydrogen-oxidizing bacteria. Water Res. 101, 137–146 (2016).
15
S. W. Jones, A. Karpol, S. Friedman, B. T. Maru, B. P. Tracy, Recent advances in single cell protein use as a feed ingredient in aquaculture. Curr. Opin. Biotechnol. 61, 189–197 (2020).
16
M. Jouny, W. Luc, F. Jiao, General techno-economic analysis of CO2 electrolysis systems. Ind. Eng. Chem. Res. 57, 2165–2177 (2018).
17
A. J. Martín, G. O. Larrazábal, J. Pérez-Ramírez, Towards sustainable fuels and chemicals through the electrochemical reduction of CO2: Lessons from water electrolysis. Green Chem. 17, 5114–5130 (2015).
18
J. Sillman et al., Bacterial protein for food and feed generated via renewable energy and direct air capture of CO2: Can it reduce land and water use? Glob. Food Secur. 22, 25–32 (2019).
19
M. Fasihi, O. Efimova, C. Breyer, Techno-economic assessment of CO2 direct air capture plants. J. Clean. Prod. 224, 957–980 (2019).
20
X. W. Wu, C. C. Lee, D. M. Muzny, C. T. Caskey, Urate oxidase: Primary structure and evolutionary implications. Proc. Natl. Acad. Sci. U.S.A. 86, 9412–9416 (1989).
21
R. E. Blankenship et al., Comparing photosynthetic and photovoltaic efficiencies and recognizing the potential for improvement. Science 332, 805–809 (2011).
22
M. M. Fouad, L. A. Shihata, E. I. Morgan, An integrated review of factors influencing the perfomance of photovoltaic panels. Renew. Sustain. Energy Rev. 80, 1499–1511 (2017).
23
L. Bertuccioli et al., Study on Development of Water Electrolysis in the EU (E4tech Sàrl with Element Energy Ltd, 2014).
24
S. Szima, C.-C. Cormos, Improving methanol synthesis from carbon-free H2 and captured CO2: A techno-economic and environmental evaluation. J CO2 Utilizat. 24, 555–563 (2018).
25
N. J. Claassens, C. A. R. Cotton, D. Kopljar, A. Bar-Even, Making quantitative sense of electromicrobial production. Nat. Catal. 2, 437–447 (2019).
26
S. Upadhyaya, S. Tiwari, N. Arora, D. P. Singh, “Microbial protein: A valuable component for future food security” in Microbes and Environmental Management, J. S. Singh, D. P. Singh, Eds. (Studium Press, 2016), p. 21.
27
C. Breyer, M. Fasihi, A. Aghahosseini, Carbon dioxide direct air capture for effective climate change mitigation based on renewable electricity: A new type of energy system sector coupling. Mitig. Adapt. Strategies Glob. Change 25, 43–65 (2019).
28
A. Sinha, L. A. Darunte, C. W. Jones, M. J. Realff, Y. Kawajiri, Systems design and economic analysis of direct air capture of CO2 through temperature vacuum swing adsorption using MIL-101(Cr)-PEI-800 and mmen-Mg2(dobpdc) MOF adsorbents. Ind. Eng. Chem. Res. 56, 750–764 (2017).
29
X. Zhang, C. Bauer, C. L. Mutel, K. Volkart, Life cycle assessment of power-to-gas: Approaches, system variations and their environmental implications. Appl. Energy 190, 326–338 (2017).
30
M. Smith, Data from "Food composition tables for GENuS". Harvard Dataverse, V4. https://doi.org/10.7910/DVN/GNFVTT. UNF:6:Ubt+oDYQT6psn7qvTECuMA== [fileUNF]. Accessed 1 June 2021.
31
OECD/Food and Nations, OECD-FAO agricultural outlook 2020-2029 (2020).
32
S. Dubey, J. N. Sarvaiya, B. Seshadri, Temperature dependent photovoltaic (PV) efficiency and its effect on PV production in the world – A review. Energy Procedia 33, 311–321 (2013).
33
J. Whittaker, R. Johnson, T. Finnigan, S. Avery, P. Dyer, The Biotechnology of Quorn Mycoprotein: Past, Present and Future Challenges (Springer, Cham, Switzerland, 2020), pp. 59–79.
34
D. Rusmanis, R. O’Shea, D. M. Wall, J. D. Murphy, Biological hydrogen methanation systems–An overview of design and efficiency. Bioengineered 10, 604–634 (2019).
35
A. Polman, H. A. Atwater, Photonic design principles for ultrahigh-efficiency photovoltaics. Nat. Mater. 11, 174–177 (2012).
36
M. A. Green et al., Solar cell efficiency tables (version 50). Prog. Photovolt. Res. Appl. 25, 668–676 (2017).
37
Q. Wang, K. Domen, Particulate photocatalysts for light-driven water splitting: Mechanisms, challenges, and design strategies. Chem. Rev. 120, 919–985 (2020).
38
T. Wang, J. Gong, Sacrificing nothing to reduce CO2. Nat. Energy 5, 642–643 (2020).
39
O. Yishai, M. Bouzon, V. Döring, A. Bar-Even, In vivo assimilation of one-carbon via a synthetic reductive Glycine pathway in Escherichia coli. ACS Synth. Biol. 7, 2023–2028 (2018).
40
J. S. Amthor et al., Engineering strategies to boost crop productivity by cutting respiratory carbon loss. Plant Cell 31, 297 (2019).
41
V. C. S. Pankievicz, T. B. Irving, L. G. S. Maia, J.-M. Ané, Are we there yet? The long walk towards the development of efficient symbiotic associations between nitrogen-fixing bacteria and non-leguminous crops. BMC Biol. 17, 99 (2019).
42
J. Yang, X. Xie, M. Yang, R. Dixon, Y.-P. Wang, Modular electron-transport chains from eukaryotic organelles function to support nitrogenase activity. Proc Natl. Acad Sci. USA 114, E2460–E2465 (2017).
43
USDA, Recent Trends, in GE Adoption (2020). https://www.ers.usda.gov/data-products/adoption-of-genetically-engineered-crops-in-the-us/recent-trends-in-ge-adoption/. Accessed 6 November 2020.
44
J. Fernandez-Cornejo, W. D. McBride, Genetically engineered crops: U.S. adoption and impacts. Agricultural Outlook 294, 24–27 (2002).
45
S. Matassa et al., Upcycling of biowaste carbon and nutrients in line with consumer confidence: The “full gas” route to single cell protein. Green Chem. 22, 4912–4929 (2020).
46
Z. Bashi, R. McCullough, L. Ong, M. Ramirez, Alternative Proteins: The Race for Market Share Is on (McKinsey & Company, 2019).
47
M. Øverland, A.-H. Tauson, K. Shearer, A. Skrede, Evaluation of methane-utilising bacteria products as feed ingredients for monogastric animals. Arch. Anim. Nutr. 64, 171–189 (2010).
48
A. Shepon, G. Eshel, E. Noor, R. Milo, The opportunity cost of animal based diets exceeds all food losses. Proc. Natl. Acad. Sci. U.S.A. 115, 3804–3809 (2018).
49
Development Initiatives, 2018 Global Nutrition Report: Shining a light to spur action on nutrition (Development Initiatives, Bristol, UK, 2018).
50
X. Meng et al., Biodiesel production from oleaginous microorganisms. Renew. Energy 34, 1–5 (2009).
51
S. Parsons, S. Raikova, C. J. Chuck, The viability and desirability of replacing palm oil. Nat. Sustain. 3, 412–418 (2020).
52
A. Zamani, M. Khajavi, M. H. Nazarpak, E. Gisbert, Evaluation of a bacterial single-cell protein in compound diets for rainbow trout (Oncorhynchus mykiss) fry as an alternative protein source. Animals (Basel) 10, E1676 (2020).
53
T. S. Aas, B. Grisdale-Helland, B. F. Terjesen, S. J. Helland, Improved growth and nutrient utilisation in Atlantic salmon (Salmo salar) fed diets containing a bacterial protein meal. Aquaculture 259, 365–376 (2006).
54
S. H. El Abbadi, C. S. Criddle, Engineering the dark food chain. Environ. Sci. Technol. 53, 2273–2287 (2019).
55
T. Linder, Making the case for edible microorganisms as an integral part of a more sustainable and resilient food production system. Food Secur. 11, 265–278 (2019).
56
J. Sillman et al., A life cycle environmental sustainability analysis of microbial protein production via power-to-food approaches. Int. J. Life Cycle Assess. 25, 2190–2203 (2020).
57
Z. Zhou, M. Carbajales-Dale, Assessing the photovoltaic technology landscape: Efficiency and energy return on investment (EROI). Energy Environ. Sci. 11, 603–608 (2018).
58
I. Pikaar et al., Microbes and the next nitrogen revolution. Environ. Sci. Technol. 51, 7297–7303 (2017).
59
S. Matassa, D. J. Batstone, T. Hülsen, J. Schnoor, W. Verstraete, Can direct conversion of used nitrogen to new feed and protein help feed the world? Environ. Sci. Technol. 49, 5247–5254 (2015).
60
D. Pimentel, M. Pimentel, Sustainability of meat-based and plant-based diets and the environment. Am. J. Clin. Nutr. 78(suppl. 3), 660S–663S (2003).
61
J. F. Foster, J. H. Litchfield, A continuous culture apparatus for the microbial utilization of hydrogen produced by electrolysis of water in closed-cycle space systems. Biotechnol. Bioeng. 6, 441–456 (1964).
62
K. A. Alvarado, J. B. García Martínez, S. Matassa, J. Egbejimba, D. Denkenberger, Food in space from hydrogen-oxidizing bacteria. Acta Astronaut. 180, 260–265 (2021).
63
Solargis Prospect, GHI map. https://apps.solargis.com/prospect. Accessed 10 October 2018.
64
C. Deline, A. Dobos, S. Janzou, J. Meydbray, M. Donovan, A simplified model of uniform shading in large photovoltaic arrays. Sol. Energy 96, 274–282 (2013).
65
C. Deline, B. Marion, J. Granata, S. Gonzalez, “A performance and economic analysis of distributed power electronics in photovoltaic systems” (Tech. Rep. NREL/TP-5200-50003,National Renewable Energy Laboratory, 2011).
66
S. Satyapal, “017 Annual progress report: DOE hydrogen and fuel cells program” (NREL/BK-6A20-69077; DOE/GO-102017-5042, National Renewable Energy Laboratory, 2018).
67
K. L. Dubrawski et al., Microbial battery powered enzymatic electrosynthesis for carbon capture and generation of hydrogen and formate from dilute organics. ACS Energy Lett. 4, 2929–2936 (2019).
68
A. R. T. Morrison et al., Modeling the electrochemical conversion of carbon dioxide to formic acid or formate at elevated pressures. J. Electrochem. Soc. 166, E77–E86 (2019).
69
J. Li et al., Electroreduction of CO2 to formate on a copper-based electrocatalyst at high pressures with high energy conversion efficiency. J. Am. Chem. Soc. 142, 7276–7282 (2020).
70
D. Mignard, C. Pritchard, Processes for the synthesis of liquid fuels from CO2 and marine energy. Chem. Eng. Res. Des. 84, 828–836 (2006).
71
J.-L. Cordier, B. M. Butsch, B. Birou, U. von Stockar, The relationship between elemental composition and heat of combustion of microbial biomass. Appl. Microbiol. Biotechnol. 25, 305–312 (1987).
72
I. Pikaar et al., Carbon emission avoidance and capture by producing in-reactor microbial biomass based food, feed and slow release fertilizer: Potentials and limitations. Sci. Total Environ. 644, 1525–1530 (2018).
73
FAO, Food Energy - Methods of Analysis and Conversion Factors (Food and Agriculture Organization of the United Nations, Rome, 2003).
74
J. P. Torella et al., Efficient solar-to-fuels production from a hybrid microbial-water-splitting catalyst system. Proc. Natl. Acad. Sci. U.S.A. 112, 2337–2342 (2015).
75
H. C. de Boer, M. M. van Krimpen, H. Blonk, M. Tyszler, Replacement of soybean meal in compound feed by European protein sources- Effects on carbon footprint. Wageningen, Wageningen UR (Livestock Research Report 819, University & Research Centre, Livestock Research, 2014).
76
J. H. Litchfield, Advances in Applied Microbiology, D. Perlman, Ed. (Academic Press, 1977), vol. 22, pp. 267–305.
77
L. Wang et al., Greening ammonia toward the solar ammonia refinery. Joule 2, 1055–1074 (2018).
78
C. W. Lewis, Energy requirements for single cell protein production. J. Appl. Chem. Biotechnol. 26, 568–575 (1976).
79
L. Panella, Sugar beet as an energy crop. Sugar Tech 12, 288–293 (2010).
80
N. López, E. Puértolas, S. Condon, J. Raso, I. Álvarez, Enhancement of the solid-liquid extraction of sucrose from sugar beet (Beta vulgaris) by pulsed electric fields. Lebensm. Wiss. Technol. 42, 1674–1680 (2009).
81
C. E. Berryman, H. R. Lieberman, V. L. Fulgoni III, S. M. Pasiakos, Protein intake trends and conformity with the dietary reference intakes in the United States: Analysis of the National Health and Nutrition Examination Survey, 2001-2014. Am. J. Clin. Nutr. 108, 405–413 (2018).
Information & Authors
Information
Published in
Cover image: Pictured is the battleship USS Iowa. Complex systems, including battleships, biological organisms, and financial markets, follow universal themes of innovation. Approaches for upgrading complex systems tend to gradually improve over time but can be rendered obsolete by radical innovation, such as when airpower rendered dreadnoughts like the USS Iowa obsolete. Articles from the Evolutionary Models of Financial Markets Special Feature offer insights from the natural sciences that could help illuminate how innovation occurs in financial markets. See the Introduction to the Special Feature by Simon A. Levin and Andrew W. Lo, e2104800118. Image credit: Wikimedia Commons/PH1 Jeff Hilton.
Classifications
Copyright
Copyright © 2021 the Author(s). Published by PNAS. This open access article is distributed under Creative Commons Attribution License 4.0 (CC BY).
Data Availability
All study data are included in the article and/or supporting information.
Submission history
Published online: June 21, 2021
Published in issue: June 29, 2021
Change history
January 10, 2022: Dataset S1 has been updated.
Keywords
Notes
This article is a PNAS Direct Submission.
Authors
Competing Interests
The authors declare no competing interest.
Metrics & Citations
Metrics
Citation statements
Altmetrics
Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
Cited by
Loading...
View Options
View options
PDF format
Download this article as a PDF file
DOWNLOAD PDFGet Access
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginRecommend to a librarian
Recommend PNAS to a LibrarianPurchase options
Purchase this article to access the full text.