A Sustainable Chemicals Manufacturing Paradigm Using CO2 and Renewable H2.

The chemical industry must decarbonize to align with UN Sustainable Development Goals. A shift toward circular economies makes CO2 an attractive feedstock for producing chemicals, provided renewable H2 is available through technologies such as supercritical water (scH2O) gasification. Furthermore, high carbon and energy efficiency is paramount to favorable techno-economics, which poses a challenge to chemo-catalysis. This study demonstrates continuous gas fermentation of CO2 and H2 by the cell factory, Cupriavidus necator, to (R,R)-2,3-butanediol and isopropanol as case studies. Although a high carbon efficiency of 0.75 [(C-mol product)/(C-mol CO2)] is exemplified, the poor energy efficiency of biological CO2 fixation requires ∼8 [(mol H2)/(mol CO2)], which is techno-economically infeasible for producing commodity chemicals. Heat integration between exothermic gas fermentation and endothermic scH2O gasification overcomes this energy inefficiency. This study unlocks the promise of sustainable manufacturing using renewable feedstocks by combining the carbon efficiency of bio-catalysis with energy efficiency enforced through process engineering.

Introduction

The chemical industry has been central to the modern world since the Industrial Revolution, converting raw materials such as fossil reserves into thousands of products through numerous continuous processes. Although the chemical industry has served economic growth well over many decades, much of the chemical industry has become misaligned with the United Nation's (UN) Sustainable Development Goals, notably Sustainable Industrialization and Climate Action (Axon and James, 2018). Current large-scale manufacturing processes suffer from a reliance on finite fossil reserves, high energy consumption, and poor overall catalytic selectivity. Circular economies are markedly absent and net greenhouse gas emissions exacerbate climate change (Keijer et al., 2019).

Given the chemical industry needs to decarbonize, the use of CO2 as a carbon feedstock for producing chemicals has significant synergy with UN Sustainable Development Goals, provided a renewable supply of reducing power is available from either H2 or H2O. Renewable H2 can be produced via a number of sustainable technologies, including (1) biomass pyrolysis-gasification (Dou et al., 2019), (2) dark fermentation of complex carbohydrates (Boboescu et al., 2016), (3) supercritical water gasification (Okolie et al., 2019), and (4) megawatt-scale water electrolysis (Schmidt et al., 2017). Such renewable H2 can be used to produce methanol from CO2 chemo-catalytically, where a typical reactor has a methanol outlet composition of ∼61% by mole of the total carbon products (Toyir et al., 2009). This renewable methanol can be converted to C2–C4 olefin chemical building blocks using SAPO-34 zeolite catalysts, noting a typical reactor has a propene outlet composition of 39% by weight of the total carbon products in commercial practice. In addition to low selectivity, the Methanol to Olefins (MTO) process suffers from rapid catalyst coking, necessitating continuous regeneration of the catalyst within the fluidized bed reactor (Tian et al., 2015). The MTO process' greater selectivity for propene creates opportunities to produce C3 and C4 alcohol solvents. Isopropanol can be produced via the hydration of the C3 propene fraction, whereas C4 alcohols can be produced through hydroformylation of propene using the OxoSM Process with ∼85% selectivity for the linear C4 product over the branched C4 by-product (Tudor and Shah, 2017). In addition to the techno-economic challenges posed by low overall selectivity and catalyst deactivation, the high temperature and pressure processing associated with the chemo-catalytic conversion of CO2 and H2 to C3 and C4 products is energy intensive (Toyir et al., 2009, Tian et al., 2015, Tudor and Shah, 2017).

Microbial cell factories produce biocatalysts (enzymes) and use the cell's energy carriers to synthesize products via these non-native biochemical pathways. Such bio-catalysis presents opportunities to reinvent chemicals manufacturing using sustainable feedstocks and renewable energy, harnessing the high catalytic selectively of microbial cell factories at low temperature and pressure (Hedstrom, 2010). Although several acetogenic cell factories are able to fix CO2 using H2 anaerobically, these cell factory platforms suffer from energetic limitations (Molitor et al., 2017) and the production of fermentative by-products such as acetate (Hoffmeister et al., 2016). Conversely, Cupriavidus necator (formerly, Alcaligenes eutrophus and Ralstonia eutropha) is a chemolithoautotrophic bacterium capable of aerobic, autotrophic growth using CO2 as the sole carbon source, H2 as electron donor, and O2 as the electron acceptor (Brigham, 2019). When the genes producing C. necator's natural carbon sink, polyhydroxybutyrate (phb), are attenuated, the cell accumulates pyruvate as the central metabolite under nutrient limitation, which can be redirected to a number of different carbon products (Steinbüchel and Schlegel, 1989). Consequently, metabolic engineering of C. necator to produce chemicals from CO2 and H2 has demonstrated promise, most notably in the production of (1) 2-hydroxyisobutyrate for Plexiglas (Przybylski et al., 2012), (2) isobutanol (Brigham et al., 2013), (3) 3-methyl-1-butanol (Li et al., 2012), (4) methyl ketones (Müller et al., 2013), (5) isopropanol (Marc et al., 2017), (6) α-humulene (Krieg et al., 2018), and (7) acetoin (Windhorst and Gescher, 2019). These studies have contributed appreciably to advancing the metabolic engineering of C. necator as a platform for using CO2 as a carbon feedstock, noting that the focus was not on process engineering considerations. As such, these studies were demonstrated at low cell density in batch operation, which is not aligned techno-economically with continuous manufacturing. Additionally, these studies have not addressed the energy inefficiency of biological CO2 fixation, where the high H2 utilization makes the process techno-economically infeasible for producing commodity chemicals (Emerson and Stephanopoulos, 2019).

The objective of this study has been to demonstrate integrated, continuous production of chemicals from CO2 using C. necator as the microbial cell factory. Aligning with the continuous operating paradigm of the chemical industry, this paper is the first to demonstrate the stable and continuous bio-manufacture of chemicals from CO2 using C. necator as carbon-efficient cell factory. Furthermore, this study is the first to demonstrate the use of process engineering to overcome the techno-economic hurdle associated with the energy inefficiency of biological CO2 fixation. The following sections outline the metabolic engineering, continuous gas fermentation, and rigorous process simulation, which exemplify carbon and energy-efficient continuous bio-manufacture through two case studies producing (R,R)-2,3-butanediol (BDO) and isopropanol (IPA).

Results

Metabolic Engineering

Microbial cell factories are able to produce a large number of products via their biochemical networks from a variety of carbon feedstocks. BDO and IPA were selected as the two case studies for continuous autotrophic fermentation in this work, given both biochemical pathways have been extensively characterized (Ji et al., 2011, Marc et al., 2017). The biochemical pathways that convert the central metabolite pyruvate to either BDO or IPA are detailed in Figure 1A. The biochemical pathway genes depicted in Figure 1B were overexpressed as a single operon from either a plasmid or chromosomal integration, employing the synthetic biology methods described in the Transparent Methods section.

Figure 1

Biochemical Network and Pathway Operons for the Synthesis of (R,R)-2,3-butanediol and isopropanol

(A) Biochemical network outlining the synthesis of (R,R)-2,3-butanediol and isopropanol in the microbial cell factory, C. necator H16, converting CO2 and H2 to pyruvate via the Calvin Cycle and redirecting carbon flux from pyruvate to (R,R)-2,3-butanediol and isopropanol as case studies. Attenuated genes are contained in red text boxes. Genes overexpressed to allow (R,R)-2,3-butanediol synthesis are contained in gray text boxes, whereas genes overexpressed for isopropanol synthesis are contained in green text boxes.

(B) Pathway operons for BDO (BD2) and isopropanol (IPA4)-producing strains. Both operons rely on pBAD as inducible promoter with ribosome-binding sites as per Table S2 and genes as per Table S4, noting that phaA is the native open reading frame.

C. necator's natural carbon sink pathway to polyhydroxybutyrate (phb) was knocked out by deleting the operon phaC1AB1 for the BDO cell factory and the genes phaC1B1 for the IPA cell factory (Peplinski et al., 2010, Raberg et al., 2014, Müller et al., 2013), thereby redirecting carbon flux and reducing equivalents to the fermentation product. Also, given the reported degradation of BDO's precursor acetoin by C. necator via what would constitute a competing pathway (Fründ et al., 1989), the acoXABC gene cluster was deleted in the BDO cell factory.

The performance evaluations of these BDO and IPA cell factories in heterotrophic shake flask culture are summarized in Table 1 and Figure 2, demonstrating the carbon split using fructose as carbon source. Given that chromosomal integration is associated with greater genetic stability in C. necator (Voss and Steinbüchel, 2006, Gruber et al., 2014), the performance of the expressed operons (Figure 1B) was assessed for both plasmid-based and chromosomally integrated cell factories. The cumulative specific fructose uptake rate was comparable across plasmid, integrated and control strains, where the biomass synthesis was controlled via nitrogen limitation. In shake flasks, the integrated BDO cell factory had similar yield and productivity performance to the plasmid-based cell factory, whereas the IPA plasmid-based cell factory outperformed the integrated cell factory. The greater genetic stability of integrated cell factories is best suited to continuous fermentation, and a single copy integration also provided for a better comparison of performance between BDO and IPA. Therefore, both the BDO and IPA integrated cell factories were taken forward into continuous fermentation using CO2 and H2 as per the Transparent Methods section.

Table 1

Microbial Cell Factory Performance in Heterotrophic Shake Flasks for (R,R)-2,3-Butanediol and Isopropanol Synthesis (Triplicate Biological Replicates)

Performance Parameter	Unit	2,3-Butanediol			Isopropanol
		Plasmid	Integrated	Host	Plasmid	Integrated	Host
Cumulative specific fructose uptake rate	[(mmol C)/((g DCW)·h)]	9.6 ± 1.3	9.7 ± 1.2	8.8 ± 1.1	10.2 ± 1.8	9.9 ± 1.6	9.1 ± 1.1
Cumulative molar carbon efficiency (yield)	[(C-mol product)/(C-mol fructose)]	Figure 2
Final product titer in liquid phase	[g/L]	1.9 ± 0.1	1.8 ± 0.1	nd	2.7 ± 0.2	1.8 ± 0.2	nd

nd designates not detected.

Figure 2

Carbon Efficiency for BDO and IPA Strains in Shake Flask Culture

Cumulative molar carbon efficiency (yield) of microbial cell factories in heterotrophic shake flasks (triplicate biological replicates), representing the carbon split using fructose as carbon source in [(C-mol product)/(C-mol fructose)]. The by-product for the BDO strains is meso-2,3-butanediol and for the IPA strains is acetone. The CO2 carbon split has been estimated from the carbon balance, noting that the analysis excludes losses of volatile products and by-products to the gas phase. The desired product comprises less than 25% of the carbon from fructose, given the required reducing power that needs to be derived from fructose. Table 1 summarizes additional performance parameters.

Continuous Gas Fermentation

Microorganisms often show reduced tolerance to the accumulation of solvent products such as BDO or IPA. Increasing IPA concentrations in fermentation are detrimental to C. necator's growth above 15 g/L (Marc et al., 2017), recognizing that stripping of IPA into the bioreactor's off-gas increases as the aqueous concentration increases. Also, in continuous fermentation, the IPA is further diluted from the bioreactor. The low volatility of BDO makes its accumulation in the bioreactor's aqueous phase a greater concern. Consequently, shake flask experiments revealed that, above a BDO concentration of 30 g/L, the growth rate of C. necator H16 is impaired (Figure S1). Therefore, without resorting to genetic modification, the BDO concentration in continuous bioreactors needs to be controlled through dilution alone.

The continuous, autotrophic fermentation results were generated using the bioreactor experimental setup shown in Figure 3, outlining the decoupled, multi-loop SISO (single-input, single-output) process control strategy for intensifying the process within the flammability safety constraints. The specific CO2 uptake rates and specific productivities on a biomass (DCW) basis are trended in Figure 4, thereby allowing for comparison between BDO and IPA synthesis against the ΔphaC1AB1 ΔacoXABC control strain. For the BDO cell factory, meso-2,3-butanediol is the principal by-product, indicative of promiscuous secondary alcohol dehydrogenase activity accepting acetoin as substrate. Acetone is the principal by-product for the IPA cell factory, indicating that the secondary alcohol dehydrogenase is the rate-limiting step in the assembled pathway. The control strain produces no products or by-products. The molar ratios of H2 and O2 uptake to CO2 uptake reflect the specific requirement for H2 as electron donor and O2 as electron acceptor (Figure 5). The steady-state CO2 uptake rate for both case studies demonstrated a specific uptake rate of 3–4 ([(mmol C)/(gDCW·h)], consuming ∼8 [(mol H2)/(mol CO2)] and 2–3 [(mol O2)/(mol CO2)]. An ∼8 [(mol H2)/(mol CO2)]) molar ratio reflects the reducing power required to fix CO2 and produce the product, generating significant heat owed to the exothermic reaction. Alongside this measure of energy efficiency, the carbon split to product, by-product, and biomass summarizes the carbon efficiency (Table 2 and Figure 6). As such, these data are the first to demonstrate genetically stable and continuous production in C. necator via chromosomal integration of non-native genes. The calculation methodology for gas uptake rates, steady-state dilution rate, and carbon fluxes for the continuous, autotrophic fermentations is described in Figure S2. Given the results in Figures 4 and 5 represent calculated data incorporating several sensors, analyses, and calibration standards, two thousand Monte Carlo simulations were undertaken to determine the 90% confidence limits denoted by error bars in Figures 4 and 5. The histogram outputs from the Monte Carlo simulations are contained in Figure S3.

Figure 3

Continuous, Autotrophic Bioreactor Experimental Setup

Continuous bioreactor experimental setup, outlining the decoupled, multi-loop SISO (single-input, single-output) process control strategy for intensifying the continuous autotrophic bioreactor within the flammability safety constraints. The steady-state dissolved oxygen (AIC101) is controlled via the phosphate addition rate, maximizing the process intensification under phosphate limitation. Avoiding a flammable atmosphere in the headspace of the reactor, AIC102 controls the O2 concentration in the headspace using the air flow.

Figure 4

Specific CO2 uptake rates and specific carbon productivities for continuous, autotrophic fermentations

Specific CO2 uptake rate (CUR) and specific carbon productivity on a biomass basis during the synthesis of (R,R)-2,3-butanediol and isopropanol in continuous autotrophic fermentation, compared with the ΔphaC1AB1 ΔacoXABC control. Specific productivities are comparable, noting the greater carbon overflow to by-product for the isopropanol cell factory. Error bars determined as per Figure S3.

Figure 5

Molar Uptake Ratios for Continuous, Autotrophic Fermentations

Molar uptake ratios for H2 (electron donor) and O2 (electron acceptor) to CO2 during the synthesis of (R,R)-2,3-butanediol and isopropanol in continuous autotrophic fermentation, compared with the ΔphaC1AB1 ΔacoXABC control. An ~8 [(mol H2)/(mol CO2)] molar ratio reflects the reducing power required to fix CO2 and produce the product, generating significant heat owed to the exothermic reaction. Error bars determined as per Figure S3.

Table 2

Microbial Cell Factory Performance in Continuous Autotrophic Fermentation for (R,R)-2,3-Butanediol and Isopropanol Synthesis

Performance Parameter	Unit	BDOa Microbial Cell Factory	IPAb Microbial Cell Factory
Specific CO2 uptake rate	[(mmol C)/(gDCW·h)]	3.04 ± 0.12	3.99 ± 0.11
Molar carbon efficiency (yield)	[(C-mol product)/(C-mol CO2)]	Figure 6
Carbon balance closure	[%]	102.6	93.7
H2/CO2Molar ratio	[(mol H2)/(mol CO2)]	7.91 ± 0.86	8.09 ± 0.74
O2/CO2Molar ratio	[(mol H2)/(mol CO2)]	2.92 ± 0.1	1.85 ± 0.06
Product in vapor phase	[-] mole fraction	0	0.75
Product in liquid phase	[g/L]	32.0 ± 0.1	7.7 ± 0.2

BDO is 2,3-butanediol.

IPA is isopropanol.

Figure 6

Molar Carbon Efficiencies for Continuous, Autotrophic Fermentations

Molar carbon efficiency (yield) of microbial cell factories in autotrophic fermentation, representing the carbon split using CO2 as carbon source in [(C-mol product)/(C-mol CO2)]. The by-product for the BDO strains is meso-2,3-butanediol and for the IPA strains is acetone. Table 2 summarizes additional performance parameters.

Process Simulation and Systems Biology

Heat Integration of Gas Fermentation with Supercritical Water Gasification

Gas fermentation is a highly exothermic process owed to the cascade of electrons through the biochemical network (Figure 1A) from the electron donor, H2, to the final electron acceptor, O2 (Tanaka et al., 1995). The CO2 is reduced to a number of carbon sinks, typically biomass, the fermentation product, and by-products. Supercritical water (scH2O) gasification is a hydrothermal technology converting renewable carbon feedstocks, such as wet lignin biomass, to CO2 and H2 at supercritical pressure and temperature, i.e., 240 bar (a) and 375°C (Rodriguez Correa and Kruse, 2018). The process is highly endothermic and the renewable H2 produced by the scH2O reactor originates from both the hydrocarbon feedstock and the scH2O. A heat pump can be employed to facilitate the energy (heat) flow from a low temperature (gas fermentation) to a high temperature (scH2O gasification) via a thermal cycle. The integrated process was rigorously simulated in Aspen HYSYS, summarized in Figure 7 and detailed in Figure 8 using the lignin model compound, guaiacol, as the waste carbon feedstock. From the heat pump cycle depicted in Figure 7, a suitable heat carrying fluid (isopentane) is evaporated at low pressure in an evaporator by the bioreactor's heat of reaction (4,004 kW/ton guaiacol), resulting in a substantial increase in the isopentane's enthalpy (energy) at constant temperature. The isopentane vapor is compressed to a higher pressure via a compressor, further increasing the isopentane's enthalpy owed to the heat of compression (175 kW/ton guaiacol). Further energy is transferred to the vapor via a series of heat exchangers. The scH2O Recovery Heat Exchanger recovers heat from the scH2O reactor's effluent (5,565 kW/ton guaiacol), and the Heat Pump Recovery Heat Exchanger recovers heat from the isopentane returning after heating the scH2O reactor feed to supercritical temperature. Thereafter, the temperature of the isopentane is greatly increased in a combustion chamber (3,952 kW/ton guaiacol), fired by a fraction of the renewable H2 generated in the scH2O reactor. The high temperature of the isopentane allows heat transfer to the subcritical aqueous feed (3,660 kW/ton guaiacol), raising the sub-critical feed to the scH2O reactor to supercritical conditions. After the Heat Pump Recovery Heat Exchanger, the vapor is condensed in a condenser (scH2O Reactor Pre-heater), which pre-heats the feed to the scH2O reactor (10,036 kW/ton guaiacol). Finally, the liquid isopentane is expanded to lower pressure over a valve for evaporation in the Heat Pump Evaporator by the bioreactor's heat of reaction. The thermal cycle linking the low-temperature energy source, i.e., the exothermic gas fermentation, and the high-temperature energy sink, i.e., the endothermic scH2O gasification, is subsequently repeated.

Figure 7

Schematic of the Heat Integration between Gas Fermentation and Supercritical Water Gasification

Schematic summarizing the heat integration between gas fermentation and supercritical water gasification via a heat pump using isopentane as enthalpy carrying fluid though a number of heat exchangers. The Heat Pump Evaporator recovers heat from the bioreactor at low temperature (4,004 kW/ton guaiacol), resulting in a reduction in the operating cost burden associated with cooling water use and electricity demand (see Figure 9). The cumulative recovery of heat energy within the heat pump cycle (purple cycle) minimizes the fraction of the H2 (pink arc) that needs to be combusted to heat the aqueous guaiacol fed (blue arc) to the highly endothermic gasification reactor via the Supercritical Heater (3,660 kW/ton guaiacol).

Figure 8

Process Flow Diagram for the Heat Integration between Gas Fermentation and Supercritical Water Gasification

Process flow diagram detailing the heat integration between gas fermentation and supercritical water gasification on a guaiacol feed basis of 1,000 kg/h. H2-rich gas (red) is produced from waste carbon in the aqueous media (blue) in a scH2O reactor. A heat pump using isopentane (purple) heat integrates the low-temperature gas fermentation with the high-temperature supercritical water gasifier through a number of heat exchangers. The H2-rich product (red) from the supercritical H2O gasification is feed to the gas fermentation and the combustion chamber as electron donor. Reducing the operating cost associated with compression for gas fermentation, the turbo-expander supplies air (green) with O2 as electron acceptor for gas fermentation and the combustion chamber, respectively. In addition to producing 148 kg of BDO per ton of guaiacol, the process generates 566 kW of renewable electricity per ton of guaiacol with a negligible requirement for cooling water.

Figure 9

Process Flow Diagram for Conventional, Heterotrophic Fermentation of Guaiacol to BDO

Process flow diagram for waste carbon as sole energy and carbon source, modeled on a guaiacol feed basis of 1,000 kg/h. A chilled ethylene glycol (EG, orange) loop provides for heat removal from the bioreactor via an ammonia refrigeration unit (purple). The compressor and cooling water duties reflect the operating cost burden associated with this conventional heterotrophic operating strategy, producing 158 kg of BDO per ton of guaiacol.

The process integration into the heat pump cycle of (1) the combustion chamber, (2) air compression via a turbo-expander, and (3) renewable energy generation via a turbine is detailed in Figure 8. The integrated process was rigorously simulated in Aspen HYSYS, feeding the waste carbon from the High-Pressure Pump (339 kW/ton guaiacol) to the scH2O Reactor, where CO2 and H2 are generated from the carbon feedstock modeled as the lignin model compound, guaiacol. The high-pressure CO2 & H2 stream is expanded over the turbo-expander, using the generated 575 kW/ton guaiacol to compress air as the oxygen source for gas fermentation and the combustion chamber. A fraction of the depressurized CO2 and H2 and a fraction of the compressed air are fed to the loop bioreactor, where the bleed and permeate fermentation products are produced. The unreacted CO2 & H2 in the off-gas from the loop bioreactor is combined with the remaining fraction of the depressurized CO2 and H2 to fire the combustion chamber. The combustion chamber's off-gas is fed to a turbine producing renewable electricity (566 kW/ton guaiacol). The scH2O reactor's effluent is depressurized to release the CO2 that remained soluble at high pressure, after which the hot aqueous solution can be used for biomass hot water extraction.

Systems Biology for BDO Synthesis Using CO2 & H2 and Guaiacol as Sole Energy and Carbon Sources

Rather than gasifying lignin, a next best alternative technology within the context of more conventional heterotrophic fermentation would be a process that converts lignin to guaiacol, thereafter converting the guaiacol to BDO via the 3-oxoadipate pathway. Shen et al. (2020) demonstrated that lignin can be selectively converted to guaiacol as an alternate technology to gasification. C. necator is capable of degrading lignin monomers such as catechol (Wang et al., 2014). However, a microbial cell factory using the lignin model compound guaiacol as the sole energy and carbon source has not been reported. Mallinson et al. (2018) uncovered the O-demethylase reaction that allows for guaiacol catabolism in bacteria, characterizing kinetics that suggests guaiacol catabolism is constrained by the conversion of guaiacol to catechol. The required biochemical network for the synthesis of BDO from guaiacol in C. necator H16 is shown in Figure S4. Guaiacol is catabolized via the 3-oxoadipate pathway to succinyl-CoA and acetyl-CoA, after which the carbon flux is directed to the cell's TCA cycle. Pyruvate is produced from malate via the malic enzyme as the metabolite precursor to BDO synthesis. Using a genome scale model for C. necator (Unpublished Data), which advances on the existing model proposed by Park et al. (2011), comparative Flux Balance Analysis (FBA) simulations were run using (1) guaiacol as the sole carbon and energy source and (2) CO2 and H2. For CO2 and H2, the FBA simulation accurately predicted the O2 uptake rate, H2 uptake rate, and the BDO productivity as detailed in Table 2, providing confidence in the predictive power of the genome scale model. The FBA simulation for guaiacol predicted a molar carbon yield of 0.12 [(C mol product)/(C mol guaiacol)] (Table S6). Accordingly, the upstream processing for the guaiacol case was scaled to a guaiacol feed basis of 1,000 kg/h and simulated in Aspen HYSYS as shown for the conventional heterotrophic process flow sheet in Figure 9. Similarly, the upstream processing for the CO2 & H2 case was scaled to a guaiacol feed basis of 1,000 kg/h and simulated in Aspen HYSYS as shown for the autotrophic process flow sheet in Figure 8. The bioreactor scale-up for the guaiacol and CO2 & H2 Aspen HYSYS simulations is summarized in Table S7.

Discussion

High carbon and energy efficiency is essential to achieving favorable techno-economics when converting CO2 to chemicals, which poses a significant challenge to conventional chemo-catalysis. Principally, the low overall selectivity of the Methanol to Olefin (MTO) process would hamper the sustainable production of C3 and C4 alcohols from CO2. For the integrated cell factories producing BDO and IPA (Table 1 and Figure 2), the batch experiments using fructose as carbon source showed low molar carbon yields of 0.16–0.18 [(C mol product)/(C mol fructose)]. From Figure 2, the carbon sunk into biomass and by-products was low and a large fraction of carbon was released as CO2, indicative of the reducing power required for the synthesis of BDO and IPA. Although fructose makes no net contribution to CO2 emissions, a biogenic carbon source such as lignin would be more cost-effective. However, for the lignin model compound, guaiacol, systems biology simulation for heterotrophic catabolism predicts a similar molar carbon yield of 0.12 [(C mol product)/(C mol guaiacol)]. Scaling this conventional heterotrophic process as in Figure 9 produces 158 kg of BDO per ton of guaiacol, requiring 744 kW of electricity for air compression and 1,638 kW of electricity for the ammonia chiller per ton of guaiacol. The total cooling tower duty amounts to 7,702 kW, adding to the operating cost burden. Despite using renewable feedstocks, such a process could not be described as sustainable.

In contrast, the BDO and IPA cell factories in continuous gas fermentation have substantially improved carbon efficiency, given the reducing power from H2 is fed separately from the oxidized CO2 feed. The BDO and IPA cell factories achieved high molar carbon yields of 0.75 [(C mol product)/(C mol CO2)] and 0.61 [(C mol product)/(C mol CO2)], respectively (Table 2 and Figure 6), recognizing that the IPA cell factory would benefit from further metabolic engineering optimization. Approximately 0.15–0.2 [(C mol DCW)/(C mol CO2)] is invested into the continuous production of bio-catalyst, negating the need to have a separate unit operation to regenerate catalyst as for the MTO process. Despite the stable and high overall carbon selectivity (Figures 4 and 6) aligned with the chemical industry's production paradigm, the poor energy efficiency of biological CO2 fixation requires considerable renewable H2 at ∼8 [(mol H2)/(mol CO2)] (Table 2 and Figure 5), which is techno-economically infeasible for producing commodity chemicals.

Several researchers have recognized that the energy inefficiency associated with biological carbon fixation is a hurdle to creating techno-economic processes based on gas fermentation (Bar-Even et al., 2012, Emerson and Stephanopoulos, 2019). The Calvin-Benson-Bassham (CBB) cycle is the dominant carbon fixation pathway given its prevalence within photosynthetic organisms. The CBB cycle's dominance is owed to its advantaged kinetics over other carbon fixation pathways such as the reductive acetyl-CoA pathway prevalent in anaerobic acetogens. This kinetic advantage to reduce CO2 into biomass and other metabolites comes at a substantial energy cost. For example, to produce 1 mole of acetate, 7.5 moles of H2 is required by the CBB cycle as opposed to 4 moles of H2 for the reductive acetyl-CoA pathway (Emerson and Stephanopoulos, 2019). Consequently, a number of augmented and artificial CO2 fixation pathways have been proposed as a means of improving the energy efficiency of biological C1 fixation cycles. Yu et al. (2018) proposed a malyl-CoA-glycerate (MCG) pathway to augment the CBB cycle, which reduces energy requirements by 22.5% to produce the central metabolite, acetyl-CoA. Implementing the MCG pathway would thus have potential benefit to this study's IPA, but not the BDO, cell factory. More generally, Gleizer et al. (2019) were the first to introduce an entire CO2 fixation pathway into a heterotrophic cell factory by relying on CBB cycle enzymes and formate assimilation. This significant achievement in metabolic engineering imparted chemolithotrophic metabolism to E. coli. Given formate needs to be oxidized to CO2 at a molar ratio of ∼8 [(mol formate)/(mol CO2 fixed)] to provide reducing power for the CBB cycle, the carbon efficiency of this formate-assisted pathway is low and the cell's energy efficiency no better than the conventional CBB cycle. Furthermore, the production of formate from CO2 via electrolysis is challenged by appreciable electricity demand and by electrode poisoning of the noble metal catalyst (Lee et al., 2019), where megawatt-scale implementation would be as capital intensive as for H2O electrolysis (Schmidt et al., 2017). Despite their apparent thermodynamic promise, no artificial CO2 fixation pathways have been successfully implemented in a cell factory. The problem of poor energy efficiency needs to be solved another way.

Low-cost renewable H2 production is essential to achieving favorable techno-economics, although this is only part of the solution. Biomass pyrolysis-gasification needs to be implemented at considerable scale to justify the required solids handling capital investment, which may be mismatched with the more distributed nature of smaller bio-manufacturing facilities (Dou et al., 2019). Dark H2 fermentation of complex carbohydrates suffers from low process intensification (Boboescu et al., 2016), whereas megawatt-scale H2O electrolysis is too capital intensive making large-scale H2 production prohibitive (Schmidt et al., 2017). Supercritical H2O gasification needs to overcome corrosion and fouling challenges but can be implemented cost-effectively at a smaller scale, demonstrates high reaction rates, and is not prohibitive from a capital investment perspective (Okolie et al., 2019). Cost-effective catalysis is key to economically viable H2 production at temperatures ∼400°C, where scH2O presents opportunities to produce nano-catalyst in situ (Huang et al., 2019) and for the recycle of valuable metals from spent catalyst (Grumett, 2003). Pertinent to this study, in addition to serving as a source of renewable H2, the highly endothermic scH2O gasification reaction provides a heat sink for the highly exothermic gas fermentation.

The consumption of H2 to fuel the CBB cycle's kinetics generates a significant amount of heat at low temperature in bioreactors, which is conventionally removed via a chiller unit at the expense of electrical energy and high cooling water duty as shown in Figure 9. A heat pump can be employed to facilitate the energy (heat) flow from a low temperature (gas fermentation) to a high temperature (scH2O gasification) via a thermal cycle as outlined in Figure 7 and detailed in Figure 8. Comparing the capital intensity of a conventional flowsheet (Figure 9) and the heat integrated flowsheet (Figure 8), Figure 8 has (1) a turbo-expander rather than a megawatt-scale air compressor, (2) a heat pump rather than a chiller thermal cycle, and (3) further energy recovery via a turbine. Reducing the operating cost burden associated with compression for gas fermentation, a turbo-expander supplies air for gas fermentation and the combustion chamber with no intrinsic electrical power consumption. In addition to producing 148 kg of BDO per ton of gasified guaiacol, comparable with guaiacol as sole carbon source (Figure 9), the process generates 566 kW of renewable electricity rather than consuming significant electrical power. From Figure 7, the overall heat duty of the scH2O gasifier amounts to 13.7 MW/ton guaiacol. Without heat integration with gas fermentation, the combustion chamber would need to supply 58% of this heat duty, severely limiting the supply of H2 per ton of guaiacol to gas fermentation, whereas with heat integration only 29% of this heat duty needs to be obtained by combusting a fraction of the gasifier's H2 product. Recovering the heat generated to fuel the CBB cycle removes the thermodynamic inefficiency of CO2 fixation as a significant burden to the process techno-economics. Although lignin is an abundant and low-cost feedstock, its recalcitrant, complex structure makes its direct exploitation in fermentation challenging. The process engineering solution in Figure 7 is not only energy efficient but also provides an innovative solution via scH2O gasification to using renewable feedstocks such as lignin to produce chemicals. This study unlocks the promise of sustainable manufacturing using renewable feedstocks by combining the carbon efficiency of bio-catalysis with energy efficiency enforced through process engineering.

Limitations of the Study

In the design of a sustainable bio-manufacturing facility, capital cost is directly proportional to the productivity in the bioreactors, which impacts significantly on achieving favorable techno-economics. Although this work is the first to demonstrate the stable and continuous bio-manufacture of chemicals from CO2 using C. necator as a carbon-efficient cell factory, productivity in gas fermentation will be limited by O2 transfer constraints in light of H2 flammability. Therefore, process intensification toward higher O2 mass transfer remains an important continued area of research. Although this study demonstrated stable and continuous gas fermentation experimentally, the integration of gas fermentation with scH2O gasification was verified through process simulation. Notably, process simulators, such as Aspen HYSYS, provide for rigorous simulation that enables effective process design. Although this study is the first to demonstrate the use of process engineering to overcome the techno-economic hurdle associated with the energy inefficiency of biological CO2 fixation, this work will benefit from the future demonstration of this integrated, continuous process at large laboratory scale.

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Rajesh Reddy Bommareddy (rajesh.bommareddy@nottingham.ac.uk).

Materials Availability

Materials generated in this study are available from the Lead Contact under Material Transfer Agreement (MTA).

Data and Code Availability

The ScrumPy package of metabolic modeling tools (build OMICS_20,375) was used for all systems biology simulations (Poolman, 2006). Aspen HYSYS V11 (build 37.0.0.395) from Aspen Technologies, Inc. was used for all process simulations in this study. Experimental sensor and analytical data from shake flask and continuous fermentation experiments (please see Transparent Methods section) were processed in Microsoft Excel 2016. Graphical representations of the processed data were produced using MATLAB R2019b (9.7.0) version 19.0 from Mathworks, Incorporated. Data and code generated in this study are available from the Lead Contact under MTA.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.