CO₂-assisted dehydrogenation-hydroformylation cascade enables syngas self-sufficiency and carbon-efficient propane upgrading

Abstract

Propane upgrading is fundamentally constrained by the high energy demand of conventional dehydrogenation and the reliance of hydroformylation on fossil-derived syngas, limiting both efficiency and sustainability. We redesign this architecture by establishing a carbon dioxide (CO₂)–assisted oxidative dehydrogenation–hydroformylation (CO₂-ODH-HF) cascade that replaces propane dehydrogenation (PDH) with a CO₂-ODH reactor and circulates CO₂ to generate carbon monoxide (CO) and dihydrogen (H₂) internally. Aspen Plus simulations show that this shift in reaction route creates a syngas self-sufficient system in which propylene formation, CO₂ utilization, and hydroformylation become directly coupled. The integrated cascade enhances carbon-utilization efficiency, eliminates external CO procurement, and substantially reduces total production costs compared with PDH-HF. Life-cycle assessment further indicates ~42% lower greenhouse-gas emissions per kilogram of aldehyde produced, without triggering economic-environmental trade-offs. Sensitivity analyses reveal strong robustness against fluctuations in the feed price of liquefied petroleum gas (LPG) containing propane compounds, CO cost, and hydroformylation catalyst loss. By restructuring underlying reaction pathways and carbon flows, the CO₂-ODH-HF cascade establishes a scalable and carbon-efficient route for propane upgrading and aldehyde synthesis.

A CO₂-assisted oxidative dehydrogenation–hydroformylation cascade improves carbon efficiency in propane-to-aldehyde conversion.

INTRODUCTION

The upgrading of propane into higher-value oxygenates remains a persistent challenge for low-carbon chemical manufacturing (1–3). The primary obstacle lies in the structural disconnect between propane dehydrogenation (PDH) and downstream hydroformylation (HF), as well as inherent thermodynamic barriers (fig. S1). As the industrial route to propylene, PDH is thermodynamically constrained and requires high temperatures that increase fuel consumption (4, 5), generate carbon-rich by-product streams, and impose substantial separation demands. HF, although widely applied and atom efficient for aldehyde synthesis, relies almost entirely on externally purchased syngas (6–8). Because this syngas is predominantly produced from fossil-based reforming, its use embeds substantial upstream emissions into the overall process and intensifies the carbon footprint of four-carbon (C₄) products (9–11). With growing industrial demand for light-alkane functionalization, PDH-HF configurations face sustained pressure from high operating costs, carbon loss associated with PDH side reactions, and the lack of compatibility with emerging CO₂-reduction strategies (12, 13). Addressing these challenges requires not incremental improvements to individual unit operations but a reconsideration of how dehydrogenation and hydroformylation can be more effectively coupled at the process level.

Previous research has attempted to integrate PDH and HF through plant colocation, shared energy utilities, or modest intensification of separation steps. While these approaches improve operational efficiency, they do not modify the fundamental carbon-flow architecture that underlies existing limitations (14). More recent studies have explored cofeeding CO₂ into dehydrogenation (5, 15, 16), leveraging its mild oxidative character, the reverse water-gas shift (RWGS) reaction (17–19), and dry-reforming pathways to shift PDH equilibria and alter product distributions (20). These reaction-level benefits demonstrate the potential of CO₂ to participate directly in the dehydrogenation chemistry. However, current process designs do not translate these mechanistic advantages into system-level improvements (21). In conventional flowsheets, CO₂-derived intermediates, particularly CO and H₂, are typically separated or combusted rather than integrated into downstream hydroformylation (14). As a result, “integrated” PDH-HF schemes remain fully dependent on fossil-derived syngas and inherit the associated economic and environmental penalties. This persistent reliance on external syngas highlights a broader architectural gap: Existing designs treat CO₂ as a passive cofeed or additive rather than as a functional carbon mediator capable of linking dehydrogenation and hydroformylation through material recycling. Without reconfiguring this structural relationship, improvements in carbon utilization or life-cycle performance remain fundamentally constrained.

In this work, we introduce a CO₂-assisted oxidative dehydrogenation–hydroformylation (CO₂-ODH-HF) cascade that addresses these limitations by redesigning the connectivity between reaction pathways and material flows. The proposed strategy replaces conventional PDH with CO₂-assisted oxidative dehydrogenation (CO₂-ODH) and incorporates internal CO₂ circulation, enabling the system to generate CO and H₂ in situ and directly channel them into hydroformylation. This configuration eliminates dependence on externally sourced syngas and establishes a semiclosed carbon loop that links propane activation with aldehyde synthesis. In this architecture, CO₂ functions not merely as a thermodynamic aid but as a structural integrator that enables reaction-pathway coupling and more coherent carbon redistribution across the flowsheet. Through detailed Aspen Plus simulations and comprehensive techno-economic-environmental analysis, we demonstrate that this cascade design can overcome long-standing constraints of the PDH-HF value chain. The results show that integrating CO₂-driven dehydrogenation with downstream hydroformylation offers a viable route to enhance carbon utilization, reduce raw-material consumption, and mitigate greenhouse gas (GHG) emissions relative to conventional configurations. More broadly, this study establishes a process-level framework in which CO₂ becomes an active enabler of syngas self-sufficiency and an instrument for improving both economic and environmental performance in propane upgrading and aldehyde synthesis.

RESULTS

Flowsheet architecture

The Aspen Plus flowsheet of the integrated CO₂-ODH-HF process is shown in Fig. 1, fig. S2, and tables S1 to S5. In this configuration, propane upgrading is restructured from a pair of loosely connected PDH and HF units into a single cascade in which dehydrogenation, hydroformylation, and CO₂ capture are linked through coordinated separation and recycle loops. After pretreatment in a C₄ separator, the three-carbon (C₃) liquefied petroleum gas (LPG) feed is combined with recycled propane and a high-purity CO₂ stream recovered from downstream pressure-swing adsorption (PSA) to form the feed to the CO₂-ODH section. The reactor sequence was modeled on the basis of the reference PDH design (figs. S3 and S4 and tables S6 and S7). Four RStoic modules were used to decouple parallel reaction pathways, enabling system-level process comparison. However, kinetics, equilibrium constraints, and recycle feedback were not considered. The stoichiometric contributions of PDH, CO₂-assisted oxidative pathways, propane dry reforming (PDR), and the RWGS reaction were accounted for separately. These parallel reactions generate a mixed gas stream containing C₃H₆, H₂, CO, H₂O, and unconverted C₃H₈ and CO₂.

Fig. 1. The scheme of the CO₂-ODH-HF cascade by the Aspen Plus simulation.

MPS, medium-pressure steam; LPS, low-pressure steam.

The ODH effluent is routed directly, without intermediate hydrogen purification, to the HF section (figs. S5 and S6 and tables S8 and S9). This contrasts with the conventional PDH-HF arrangement (figs. S7 and S8 and tables S10 and S11), where substantial hydrogen recycling and deep cryogenic separation precede HF. In the present cascade, the C₃H₆, CO, and H₂ produced in the ODH section serve collectively as the HF feed. The HF block itself retains the configuration of the industrial Rh-catalyzed process, producing n- and i-butyraldehyde. The reactor effluent passes through a light-ends column, a catalyst-recovery system, and multicolumn distillation train to generate the aldehyde products and regenerate the catalyst.

Separation and recycle loops establish the coupling between ODH and HF. Light gasses from HF are treated in a PSA unit that returns CO and H₂ to the HF reactor feed, while hydrocarbons are directed to the C₃ splitter. In this splitter, C₃H₆ is separated and recycled as HF feedstock, whereas high-purity C₃H₈ is forwarded to the ODH section. Carbon-containing off-gasses from the process are combusted and sent to a dedicated CO₂-PSA unit, which produces high-purity CO₂ for return to the reactor feed, thereby enabling internal CO₂ cycling. Because hydroformylation consumes a portion of the C₃H₆, CO, and H₂ generated in ODH, the cold-trap and C₃ separation duties are reduced relative to the conventional PDH-HF sequence, allowing their placement downstream of HF without overloading.

Together, the CO₂-ODH-HF flowsheet forms a tightly connected reaction-separation network in which the ODH section functions simultaneously as the propane activation unit and the internal source of CO and H₂ for hydroformylation, while PSA and distillation units regenerate the CO₂ and C₃ streams required to sustain operation. This architecture replaces the external CO supply and hydrogen-recycle–dominated PDH front end with a closed-loop system that coordinates carbon, hydrogen, and CO₂ flows across the entire cascade, establishing the structural basis for the process-level behavior analyzed in subsequent sections.

Reaction-network restructuring and carbon-flow behavior

The flowsheet architecture described above defines a set of material boundary conditions for the dehydrogenation and hydroformylation sections, and these conditions manifest directly in the distributions of intermediates and carbon flows obtained from the Aspen Plus simulations (Fig. 2 and tables S12 to S15). In the CO₂-assisted system, the ODH reactor no longer functions solely as a source of propylene and hydrogen, as in conventional PDH, but as a combined PDH-PDR-RWGS reaction environment. As a result, the outlet stream exhibits a composition fundamentally different from that of PDH.

Fig. 2. Process-level cascade diagram and carbon flow of the CO₂-ODH-HF cascade.

The feed to the ODH reactor consists of 49.1 mol % C₃H₈, 50.2 mol % CO₂, and only trace amounts of C₃H₆, H₂, and CO. After reaction at 550°C and 1 bar, the simulated effluent contains 15.8 mol % C₃H₆, 15.3 mol % H₂, and 16.0 mol % CO, together with unconverted C₃H₈ (18.6 mol %) and CO₂ (25.7 mol %). By contrast, the corresponding PDH reactor effluent contains 19.6 mol % C₃H₆, 49.6 mol % H₂, and essentially no CO. These differences reflect the prescribed contributions of RWGS and PDR in the CO₂-ODH block, which transform part of the incoming CO₂ and PDH-derived intermediates into CO, thereby shifting the outlet from a hydrogen-rich mixture to one containing near-equimolar C₃H₆, H₂, and CO.

Because the integrated cascade sends the dehydrated ODH effluent directly to HF, the HF feed composition is determined by this upstream reaction network. In the PDH-HF process, hydroformylation receives no carbonyl intermediates from PDH and therefore requires externally supplied CO to form the conventional 1:1:1 C₃H₆/CO/H₂ feed. In contrast, the CO₂-ODH-HF flowsheet draws its HF feed entirely from the ODH outlet (after cooling and removal of condensables), containing 17.1 mol % C₃H₆, 17.3 mol % CO, and 16.6 mol % H₂. The reactive components again appear in near-equimolar proportions, but they are now produced internally, owing to the combined action of the ODH reactions and the HF/PSA recycle structure. In the HF reactor, nearly all these components are consumed, reducing their mole fractions in the reactor effluent to 1.3 mol % C₃H₆, 0.5 mol % CO, and 1.5 mol % H₂, while the aldehyde fraction increases to 23.9 mol %. The remaining C₃H₈ and CO₂ are directed to the hydrocarbon and CO₂ recycle loops, respectively, sustaining the circulation structure built into the flowsheet.

These differences at the reactor and feed levels translate into a distinct carbon-flow topology for the CO₂-integrated process relative to PDH-HF (Fig. 2). In the PDH-HF configuration, all CO required for HF is imported, and a nonnegligible portion of carbon exits the system as low-value purge gas from hydrogen-rich separations. Carbon therefore enters the system as propane and external CO and leaves both as aldehyde products and as off-gas, forming an open-loop carbon boundary.

In contrast, in the CO₂-ODH-HF cascade, a portion of the CO₂ feed is converted to CO within the ODH section, and the modeled CO formation consistently meets the stoichiometric demand of HF (table S12). Carbon-flow analysis shows that a larger fraction of carbon is directed into aldehyde products, vent losses are reduced relative to PDH-HF, and a substantial CO₂ recycle stream appears as a defining feature of the flowsheet. Because CO that was previously imported now originates inside the reactor-separation network, carbon that would otherwise be discharged in off-gas remains available for productive conversion. The overall system therefore behaves as a partially closed carbon loop, in which propane and CO₂ enter, circulate among ODH, HF, and PSA units, and leave primarily as aldehydes.

Capital investment and cost structure

A comparative assessment of fixed capital investment (FCI) and annual operating costs for the PDH, HF, PDH-HF, and CO₂-ODH-HF processes is summarized in Fig. 3 and detailed in tables S16 to S23. Across all configurations, equipment purchases constitute the dominant share of total capital expenditure, representing ~42.1% of FCI (table S16). The capital cost estimates for the CO₂-ODH reactor and the CO₂-PSA unit are subject to higher uncertainty compared to conventional PDH and HF technologies, due to their lower level of industrial maturity. In this study, these units are evaluated using literature-based scaling relationships to enable a consistent comparison across process configurations. Hence, the reported capital costs should be interpreted as conceptual estimates, while the comparative trends remain primarily governed by differences in process design and raw material requirements. In addition, the integrated CO₂-ODH-HF configuration introduces increased process complexity due to the incorporation of CO₂ recycling, dual PSA systems, and tighter coupling between reaction and separation sections. This complexity brings higher technical uncertainty during scale-up and first-of-its-kind industrialization, including potential deviations in key equipment performance, increased difficulty in dynamic control, and extended commissioning periods. Consequently, higher contingency factors may be required to address potential design changes and rework, while investors typically demand higher risk premiums for previously unknown technology projects, leading to increased financing costs and a greater likelihood of deployment delays. Nevertheless, these factors primarily affect absolute capital investment and implementation timelines and do not alter the structural economic and environmental advantages established by this process in terms of reduced dependence on purchased CO and improved carbon utilization efficiency. As shown in Fig. 3A and table S17, distillation columns remain the major cost contributors in both integrated systems, reflecting the central role of C₃/C₄ splitting and butyraldehyde fractionation; they account for 26.3% of purchased equipment in PDH-HF and 21.4% in CO₂-ODH-HF.

Fig. 3. Capital investment.

(A) Purchased equipment costs and (B) annual production costs for different processes.

The total equipment cost of the PDH-HF system is effectively the sum of the stand-alone PDH and HF plants because both use the same reactor types, fired heaters, heat-exchange networks, compression stages, refrigeration systems, and storage capacities. In contrast, the CO₂-ODH-HF cascade incorporates an additional PSA unit dedicated to CO₂ capture and purification. The scale of this unit, sized to treat carbon-containing off-gasses and supply high-purity CO₂ to the ODH reactor, results in a PSA investment 6.7 times larger than that in PDH-HF, increasing total purchased-equipment cost by approximately $65.7 million.

Annual operating costs display the characteristic composition of large-scale chemical plants (Fig. 3B and table S18). Fixed charges, including maintenance, consumables, overhead, and labor, are comparable between PDH-HF and CO₂-ODH-HF because the overall plant architecture and number of major unit operations are similar. As in most hydrocarbon-processing systems, raw materials dominate annual operating expenditures, accounting for 76.5 and 66.2% of total production cost (TPC) in PDH-HF and CO₂-ODH-HF, respectively. The reduction in raw-material intensity in the CO₂-ODH-HF process arises from replacing purchased CO with syngas generated internally in the ODH section. Under identical C₃ feed rate and aldehyde production capacity, this substitution lowers raw-material costs by 34.5% relative to PDH-HF.

Relative to stand-alone PDH, the integrated CO₂-ODH-HF flowsheet increases raw-material cost by only 3.2%, due primarily to hydroformylation solvent use. Conversely, compared to stand-alone HF, which relies entirely on external syngas, the integrated cascade reduces raw-material cost by ~47%. These trends collectively demonstrate that using CO₂ as both an oxygen carrier and carbon source within the ODH-HF cascade generates a more favorable cost structure by minimizing dependence on purchased syngas components.

Redistribution of energy and utility demands

Figure 4A and table S19 summarize the utility consumption of the four process configurations, revealing distinct energy-distribution patterns shaped by their underlying reaction chemistry and separation requirements. In both the PDH and PDH-HF systems, direct fuel-fired heating represents a major portion of total utility load (28%), consistent with the strongly endothermic nature of the PDH reaction. In the CO₂-ODH-HF cascade, the reaction zone involves several endothermic pathways (PDH, PDR, and RWGS) and is modeled as a multistage reactor train with interstage reheating. This results in a 4.4-MW increase in direct fuel-fired heating, corresponding to 31% of its total load, where the additional duty is distributed across reactor stages to compensate for reaction endothermicity and restore the inlet temperature of each stage.

Fig. 4. Utility load distribution for different processes.

(A) breakdown by category and (B) aggregate total.

The PSA subsystems also exhibit substantial differences in duty and function. PDH and PDH-HF each use a single PSA unit for hydrogen purification, consuming ~10 MW of electricity. By contrast, the CO₂-ODH-HF flowsheet incorporates two PSA units: one for recovering H₂ and CO and one for CO₂ capture and purification. Because most H₂ and CO generated upstream are consumed within the HF reactor, the light-gas PSA handles only a small flow and requires just 0.2 MW of electrical power. The CO₂-PSA, sized for recycling CO₂ within the cascade, consumes 14.2 MW, bringing total PSA power demand for the integrated process to 14.4 MW. This elevated PSA consumption reflects the energetic cost of enabling internal CO₂ recovery and circulation.

Steam generation and recovery follow trends consistent with reactor temperature and off-gas composition. Both PDH and PDH-HF generate appreciable steam through off-gas combustion and heat integration across C₃H₈/C₃H₆ separations, aldehyde distillation, and HF reactor cooling. The higher operating temperature of the PDH section (620°C) induces more C₁-C₂ by-products, which enhances the heating value of off-gasses and increases recoverable steam. In contrast, the CO₂-ODH-HF process operates at 550°C, notably reducing light-gas formation and resulting in ~50% lower steam recovery relative to PDH-HF. Despite this difference, steam and cooling-water consumption still constitute more than 60% of total utility loads in all integrated systems due to high reflux ratios in the distillation sequence.

Differences in separation strategy strongly affect steam and refrigeration loads. In PDH-HF, hydrogen recycling requires a cold trap upstream of HF, introducing CO and H₂ into the C₃H₈-C₃H₆ splitter and increasing the column’s vapor traffic and steam demand. As a result, PDH-HF exhibits a higher steam load than the sum of stand-alone PDH and HF units. In the CO₂-ODH-HF configuration, hydrogen recycling is eliminated, allowing the cold trap to be placed after HF but before the C₃ splitter. This arrangement enables the splitter to process primarily C₃ hydrocarbons rather than multicomponent light-gas streams, reducing reflux ratios and lowering steam consumption by 35.9% relative to PDH-HF.

Refrigeration demand exhibits a similar trend. The PDH-HF process requires 57.3 MW of cooling, mainly for cryogenic separation of H₂ and two-carbon (C₂) hydrocarbons from the PDH effluent. Because CO₂-ODH-HF consumes most of the light components (C₃H₆, H₂, and CO) in the downstream HF reactor, the refrigeration duty is notably reduced to 26 MW.

Overall utility consumption (Fig. 4B) is lower for the CO₂-ODH-HF process than for PDH-HF, driven primarily by decreases in steam and cooling-water usage. These results show that integrating CO₂ as an active reactant modifies both the thermodynamic character of the reaction network and the composition of circulating streams, enabling more favorable separation duties and a more balanced energy-use profile across the cascade.

Economic performance and robustness

Under base-case economic assumptions [20-year lifetime, 12% discount rate, 7200 hours·annum⁻¹ (a⁻¹)], the HF, PDH-HF, and CO₂-ODH-HF processes generate comparable gross revenue because they deliver similar butyraldehyde production rates (Fig. 5 and tables S20 to S31). Their economic differentiation therefore arises from cost structure rather than product quantity. In the PDH-HF system, raw materials account for 76.5% of TPC, driven largely by C₃ LPG consumption and externally purchased CO. In the CO₂-ODH-HF cascade, however, syngas generated in the CO₂-assisted dehydrogenation section is fully consumed in hydroformylation, eliminating the need for external CO and lowering the raw-material share to 66.2%. This shift corresponds to a 34.5% reduction in raw-material expenditures and an overall 24% reduction in annual production cost relative to PDH-HF, despite modest increases associated with hydroformylation solvent demand and CO₂ handling.

Fig. 5. Economic performance metrics for different processes.

(A) Annual economic performance, (B) dynamic return on investment (DROI), and (C) dynamic analysis of cumulative cash flow.

The resulting profitability metrics reflect this rebalanced cost structure. The integrated cascade achieves an after-tax net profit of 261.5 M$·a⁻¹, which is approximately 1.6 times the value obtained for the PDH-HF process, and shortens the discounted payback period from 7.8 to 5.8 years. Over a 20-year horizon, the discounted return on investment increases from ~30% for PDH-HF to ~57% for the CO₂-ODH-HF process. These improvements originate directly from pairing high-value hydroformylation chemistry with a CO₂-enabled, syngas–self-sufficient dehydrogenation front end, which alters carbon-entry and carbon-exit pathways at the process scale.

Parametric sensitivity analysis further clarifies the robustness of these trends. Because PDH-HF depends simultaneously on LPG and purchased CO, both of which are fossil derived and price volatile, its discounted return on investment declines sharply as either price increases, crossing its viability threshold at moderate LPG or CO price levels (fig. S9, A and B), highlighting its economically dominant role. In contrast, the CO₂-ODH-HF cascade remains sensitive to LPG price but is largely immune to CO-price fluctuations because CO is supplied internally via the dehydrogenation network. Further analysis shows that only when the CO price decreases to below ~0.1 k$·tonne⁻¹ does the PDH-HF process achieve a dynamic return on investment (DROI) comparable to that of the CO₂-ODH-HF process, highlighting the strong dependence of the conventional route on CO supply. By comparison, the increase in electricity price leads to only a moderate reduction in DROI for both processes (fig. S9C; CO₂-ODH-HF: from 58 to 51%) and does not alter their relative ranking, indicating that the impact of feedstock savings outweighs that of increased utility costs.

Product price sensitivity analysis (fig. S10A) yields the same distinction in economic dominance: Both processes benefit from increases in n-butyraldehyde price, but PDH-HF becomes marginally profitable below ~1.2 k$·tonne⁻¹, whereas the CO₂-ODH-HF process remains viable down to ~1.0 k$·tonne⁻¹. Sensitivities to HF solvent loss, Rh loss, flue-gas CO₂ concentration, and operational availability (figs. S10B, S11, and S12) exhibit small magnitudes and do not change relative rankings. These variables introduce only second-order effects because, once scaled, the economics are governed primarily by carbon feedstock pathways; that is, whether syngas must be purchased or can be generated in situ.

Sensitivity analyses of key process parameters (figs. S13 to S16) indicate that, although variations in operational and catalytic factors affect the DROI, they do not alter the economic advantage of the CO₂-ODH-HF process over PDH-HF. Operational time shows a moderate positive effect on DROI due to improved capital utilization and dilution of fixed costs, while the relative gap between the two processes remains nearly unchanged. Variations in conversion and product selectivity in both the CO₂-ODH and HF sections lead to gradual increases in DROI, as higher conversion improves feedstock utilization and higher selectivity reduces by-product formation and separation costs. DROI is slightly more sensitive to selectivity because it directly affects product yield and downstream processing.

In addition, longer catalyst replacement cycles improve DROI by reducing operating and maintenance costs, whereas higher ODH catalyst (Pt) prices slightly decrease DROI due to increased capital and catalyst expenses; both effects are small because catalyst costs represent a limited fraction of total expenditures. Changes in propane and CO₂ partial pressures also lead to marginal variations, as they only moderately influence reaction equilibrium and feed composition to the HF section (6, 22–25).

Overall, the CO₂-ODH-HF system exhibits both stronger economic performance and notably greater resilience to market volatility. By internalizing syngas formation and reducing exposure to fossil-derived CO markets, the cascade fundamentally alters the sensitivity landscape of propane upgrading and enables a more robust, high-return platform for sustainable aldehyde production.

Life-cycle environmental performance

The system boundary for the economic input-output life cycle assessment (EIO-LCA) is shown in fig. S17, and the evaluation procedure is summarized in fig. S18. Following the EIO-LCA framework (26), we quantified both direct and indirect atmospheric, water, and solid-waste emissions arising from annual operation (tables S32 to S35) and capital investment (tables S36 to S40). Sectoral mapping of capital and operating expenses and the environmental characterization factors (tables S41 and S42) were used in the analysis.

Under identical raw-material input and annual operation, the CO₂-ODH-HF cascade emits 456,900 tonnes CO₂·a⁻¹ during operation, which is 42.1% lower than PDH-HF (788,600 tonnes CO₂·a⁻¹) (table S32). This reduction originates from two process-level features: (i) elimination of H₂ preseparation and recycling, which decreases steam and refrigeration demand; and (ii) internal capture and purification of CO₂ from both flue gas and process off-gas, enabling its reuse as a reactive feedstock. Operational emissions of other atmospheric pollutants (CO, NO_x, PM_2.5, and SO₂) are slightly higher (table S33), reflecting additional PSA electricity consumption and increased direct-fired heating associated with the broader endothermic reaction network.

Waterborne emissions (NH₄⁺, NO₃⁻, and PO₄³⁻; table S34) and solid waste (table S35) are lower for the CO₂-ODH-HF system because in situ syngas generation avoids the upstream environmental burdens associated with externally supplied CO. Reduced steam and refrigeration duty further decrease freshwater consumption during operation.

In contrast, environmental impacts associated with capital investment (tables S36 to S40) are higher for the CO₂-ODH-HF process, primarily due to the embodied emissions of the additional PSA unit required for CO₂ capture and purification. Thus, the integrated process presents a characteristic trade-off: a larger initial environmental footprint but substantially improved operational sustainability.

We normalized the various indicators using product output and process net profit, respectively, including global warming potential (GWP₂₀), acidification potential (AP₂₀), eutrophication potential (EP₂₀), particulate-matter formation potential (PMFP₂₀), and maximum incremental reactivity (MIR₂₀) (Table 1 and tables S43 to S47), we compared the life-cycle burdens of PDH-HF and CO₂-ODH-HF. The PDH-HF system exhibits higher operational GHG emissions [789,900 tonnes CO₂-equivalent (eq)·a⁻¹], exceeding those of stand-alone PDH (575,000 tonnes CO₂-eq·a⁻¹), stand-alone HF (698,400 tonnes CO₂-eq·a⁻¹), and even their aggregated emission (774,700 tonnes CO₂-eq·a⁻¹). Relative to this baseline, the CO₂-ODH-HF cascade reduces operating-phase GWP by 42.1%. Although its capital-phase GWP is 24.5% higher, this burden is offset early in the operational lifetime, demonstrating a decisive long-term advantage.

Table 1. Environmental impact analysis results.

Item	Capital investment section [kilotonne/(M$/a)]		Production section (kilotonne/M$)		Capital investment section [kilotonne/(kt production/a)]		Production section (kilotonne/kt production)
	PDH-HF	CO2-ODH-HF	PDH-HF	CO2-ODH-HF	PDH-HF	CO2-ODH-HF	PDH-HF	CO2-ODH-HF
GWP20 (CO2-e)	9.8	7.7	4.8	1.8	3.6	4.4	1.7	1.0
AP20 (SO2-e)	7.8 × 10−3	6.1 × 10−3	8.9 × 10−3	6.6 × 10−3	2.9 × 10−3	3.5 × 10−3	3.3 × 10−3	3.8 × 10−3
EP20 (NO3-e)	6.7 × 10−3	5.3 × 10−3	7.5 × 10−3	5.5 × 10−3	2.4 × 10−3	3.0 × 10−3	2.7 × 10−3	3.1 × 10−3
PMFP20 (PM2.5-e)	1.2 × 10−3	9.7 × 10−4	7.9 × 10−4	6.0 × 10−4	4.5 × 10−4	5.6 × 10−4	2.9 × 10−4	3.4 × 10−4
MIR20 (O3-e)	1.2 × 10−1	9.5 × 10−2	1.3 × 10−1	9.7 × 10−2	4.4 × 10−2	5.5 × 10−2	4.8 × 10−2	5.6 × 10−2
Total water	2.0 × 10−4	1.6 × 10−4	134.2	79.1	7.3 × 10−5	9.4 × 10−5	48.9	45.5
Waste residue	9.0 × 10−4	6.6 × 10−4	0.1	6.7 × 10−2	3.3 × 10−4	3.8 × 10−4	3.9 × 10−2	3.9 × 10−2

Other impact categories (AP, EP, and MIR) show moderate increases during both construction and operation, attributable to the energy intensity of CO₂ handling and the endothermic nature of the reaction network. However, these contributions are small relative to the orders-of-magnitude larger reductions in GHG emissions. Notably, the integrated process reduces total freshwater consumption by 6.8%, reflecting more efficient steam use and lower refrigeration loads, an important advantage for deployment in water-scarce regions. Solid-waste emissions are comparable between the two systems. We note that the choice of allocation method (mass, energy, or economic) substantially affects the normalized environmental impact per kilogram of product. To verify the environmental advantages of the CO₂-ODH-HF process, we performed normalization based on product output and process net profit, respectively. As shown in Table 1, neither normalization method altered the relative comparison between the PDH-HF and CO₂-ODH-HF processes, indicating that the environmental superiority of the CO₂-ODH-HF process does not depend on subjective choices of a specific evaluation methodology. In other words, whether assessed from the physical perspective of material production or from the economic perspective of capital value creation, this process demonstrates clear advantages over the conventional PDH-HF process.

A carbon-tax sensitivity analysis (fig. S19) further highlights the robustness of the CO₂-ODH-HF cascade. As carbon tax increases, TPC rises for both processes, but the CO₂-ODH-HF process remains substantially less sensitive due to its lower direct CO₂ emissions and reduced upstream emissions associated with syngas production. This reflects an intrinsic coupling between economic and environmental performance, where lower emission intensity directly translates into improved economic resilience under carbon pricing.

Overall, while the CO₂-ODH-HF process incurs a higher environmental burden during construction, its operational benefits, substantial GHG reduction, lower water consumption, and partial elimination of upstream syngas-related externalities dominate the life-cycle balance. When weighted by standard impact factors, the increases in AP, EP, PMFP, and MIR are negligible relative to the large GWP reduction, establishing the cascade as an environmentally superior alternative to conventional PDH-HF.

DISCUSSION

In summary, this work demonstrates that coupling CO₂-assisted oxidative dehydrogenation with hydroformylation reshapes the fundamental reaction-separation architecture of propane upgrading. The resulting CO₂-ODH-HF cascade forms an internally syngas–self-sufficient system in which CO₂ participates not merely as a dilute oxidant but as a transferable carbon vector linking the two reaction steps. The Aspen-based flowsheet analysis shows that this restructuring drives coordinated improvements in carbon utilization, energy distribution, and utility demand, while techno-economic and life-cycle assessments confirm substantial gains in production cost, GHG mitigation, and operational resilience relative to the conventional PDH-HF route.

These system-level benefits arise directly from altering the reaction network at the molecular scale, yet fully unlocking the potential of the cascade requires further advances in reaction engineering. Achieving precise kinetic control over PDH, PDR, and RWGS is essential for regulating the C₃H₆/CO/H₂ ratios delivered to hydroformylation and minimizing energy-intensive recycle streams. In parallel, the development of stable, selective heterogeneous catalysts for the HF step and reactor designs capable of integrating endothermic and exothermic subsystems within a unified thermal envelope will be necessary to reduce utility loads and broaden industrial applicability.

By linking reaction chemistry with process architecture, the CO₂-ODH-HF cascade provides a blueprint for converting CO₂ from a passive impurity into an active enabler of integrated reaction pathways. Although detailed reaction kinetics or equilibrium constraints are not explicitly captured in this study and the integrated process may introduce additional engineering complexity in terms of process control under recycle conditions, dynamic operation, and thermal management of coupled endothermic reactions, with continued progress in catalyst design, kinetic control, and integrated reactor engineering, this approach offers a scalable, carbon-efficient, and economically robust platform for low-carbon chemical manufacturing.

MATERIALS AND METHODS

Technical modeling assumptions

Process simulations for PDH, HF, PDH-HF, and CO₂-ODH-HF were performed in Aspen Plus V12. The CO₂-assisted dehydrogenation step was modeled by extending the Oleflex PDH framework, which uses a moving-bed Pt catalyst. All models were constructed using RStoic modules with prescribed conversions derived from literature-reported catalytic performance ranges. This simplified approach represents steady-state catalytic behavior but does not explicitly resolve detailed reaction kinetics or equilibrium limitations; therefore, the adopted conversions should be interpreted as representative operating windows rather than intrinsic kinetic predictions. Nevertheless, this framework enables consistent comparison of alternative process configurations at the system level. The prescribed conversion and selectivity were set on the basis of the steady-state catalytic performance ranges reported in the literature (Supplementary Text), allowing the process simulation to evaluate the system-level feasibility of the proposed CO₂-ODH-HF cascade. These findings, together with laboratory data for homogeneous HF catalysis (Supplementary Text) (27), served as the basis for selecting kinetic assumptions and reaction extents.

The four flowsheets involve a broad set of gaseous and liquid species (C₃H₈, C₃H₆, H₂, CO₂, CO, H₂O, n-C₄H₈O, and i-C₄H₈O) operating primarily under low-to-moderate pressures. PDH and CO₂-ODH occur exclusively in the gas phase, whereas HF proceeds in the liquid phase. Given the nonideal nature of hydrocarbon/oxygenate mixtures, the nonrandom two-liquid (NRTL) model was selected as the thermodynamic framework. NRTL provides accurate activity-coefficient predictions across a wide composition range and is extensively validated for nonideal reactive systems (24, 28).

Unit operations common to all flowsheets include reactors, heat exchangers, compressors, pumps, flash separators, PSA units, and multistage distillation columns. Aspen modules were selected to reflect industrial practice and available experimental constraints: RStoic for reaction modeling, Heater and HeatX for thermal exchange, Pump and Compr for liquid and gas compression, Flash2 for vapor-liquid disengagement, RadFrac for rigorous distillation calculations, and Sep blocks for PSA representation. Mixer and FSplit modules were used for stream combination and division (29, 30).

Economic analysis assumptions

Building on the technical simulations, economic evaluation was conducted for PDH-HF and CO₂-ODH-HF, with stand-alone PDH and HF used as reference benchmarks. The analysis accounts for both total capital investment (TCI) and TPC. TCI includes FCI, preparation and commissioning, and associated site-development charges (31, 32). FCI incorporates purchased equipment, installation, and construction costs.

Equipment purchase costs were estimated using the standard scaling relation, as shown in Eq. 1

$$C_2=C_1{\left(\frac{S_2}{S_1}\right)}^R$$ (1)

where $C_1$ and $S_1$ correspond to reference equipment cost and scale and $C_2$ and $S_2$ represent the scaled design. A scaling exponent R = 0.67 was adopted following established chemical-engineering cost correlations (33). Reference data for PDH-HF were taken from Agarwal et al. (32). and Xie and Subramaniam (34).

TPC includes variable production costs, labor, overhead, and other operating expenses, with all cost data expressed in 2025 US dollars. The cost parameters were adopted from literature and publicly available sources corresponding to similar time frames, and all process configurations were evaluated using a consistent cost basis. No additional cost escalation or inflation adjustments were applied to ensure internally consistent comparison across different process scenarios. Variable costs were calculated assuming 7200 hours·a⁻¹ of operation, with utility consumption extracted directly from Aspen simulation results. All remaining operating costs were estimated via standard multipliers applied to FCI or variable production costs (VPC), consistent with industrial practice

The DROI was computed using Eq. 2

$$DROI=\frac{{\displaystyle {\sum }_{i=n}^m}{N}_i/{(1+r)}^i}{{\displaystyle {\sum }_{i=0}^m}{C}_i/{(1+r)}^i}$$ (2)

where m = 20 years represents the evaluation period, n = 2 the first full operating year, $N_i$ is the net profit in year iii, $C_i$ is the capital expenditure in year $i$, and r = 12% represents the benchmark discount rate.

Environmental impact analysis assumptions

Environmental performance was assessed using an EIO-LCA framework (35), applying a cradle-to-grave system boundary. This boundary encompasses direct operational emissions (e.g., utility consumption and process off-gasses) and embodied impacts from material extraction, catalyst manufacture, equipment fabrication, and end-of-life disposal. The analysis uses the Ecoinvent database and quantifies annual operational emissions assuming 7200 hours·a⁻¹.

Life-cycle inventories include feedstock production and transport, catalyst preparation and regeneration, CO₂ purification, and manufacturing of major subsystems such as reactors, PSA units, and distillation trains. Impact metrics include GWP₂₀, AP₂₀, EP₂₀, PMFP₂₀, and MIR₂₀, calculated using Eqs. 3 to 7

$$GW{P}_{\mathit{20}}=(m_{C{O}_{\mathit{2}}}\times 1)\mathit{+}(m_{C{H}_{\mathit{4}}}\times x_{\mathit{20}\mathit{,}C{O}_{\mathit{2}}\mathit{-}e}^{C{H}_{\mathit{4}}})$$ (3)

$$A{P}_{\mathit{20}}=(m_{S{O}_{\mathit{2}}}\times 1)\mathit{+}(m_{{\mathit{\text{NO}}}_x}\times x_{{\mathit{\text{20,SO}}}_{\mathit{2}}\mathit{-}e}^{N{O}_x})$$ (4)

$$\begin{array}{c}E{P}_{\mathit{20}}=(m_{{{\mathit{\text{NO}}}_{\mathit{3}}}^{\mathit{-}}}\times 1)+(m_{{\mathit{\text{NO}}}_x}\times x_{\mathit{20}\mathit{,}{\mathit{\text{NO}}}_{\mathit{3}}^{\mathit{-}}\mathit{-}e}^{{\mathit{\text{NO}}}_x})\hfill \\ \mathit{+}(m_{{{\mathit{\text{NH}}}_{\mathit{4}}}^{\mathit{+}}}\times x_{\mathit{20}\mathit{,}{\mathit{\text{NO}}}_{\mathit{3}}^{\mathit{-}}\mathit{-}e}^{{{\mathit{\text{NH}}}_{\mathit{4}}}^{\mathit{+}}})\mathit{+}(m_{{{\mathit{\text{PO}}}_{\mathit{4}}}^{\mathit{3}\mathit{-}}}\times x_{\mathit{20}\mathit{,}{\mathit{\text{NO}}}_{\mathit{3}}^{\mathit{-}}\mathit{-}e}^{{{\mathit{\text{PO}}}_{\mathit{4}}}^{\mathit{3}\mathit{-}}})\hfill \end{array}$$ (5)

$$\begin{array}{c}PMF{P}_{\mathit{20}}\mathit{=}(m_{{\mathit{\text{PM}}}_{\mathit{2}\mathit{.5}}}\times 1)\mathit{+}(m_{\mathit{\text{CO}}}\times x_{\mathit{20}\mathit{,}{\mathit{\text{PM}}}_{\mathit{2}\mathit{.5}}\mathit{-}e}^{\mathit{\text{CO}}})\hfill \\ \mathit{+}(m_{{\mathit{\text{NO}}}_x}\times x_{\mathit{20}\mathit{,}{\mathit{\text{PM}}}_{\mathit{2}\mathit{.5}}\mathit{-}e}^{{\mathit{\text{NO}}}_x})\mathit{+}(m_{{\mathit{\text{SO}}}_{\mathit{2}}}\times x_{\mathit{20}\mathit{,}{\mathit{\text{PM}}}_{\mathit{2}\mathit{.5}}\mathit{-}e}^{{\mathit{\text{SO}}}_{\mathit{2}}})\hfill \end{array}$$ (6)

$$\begin{array}{c}MI{R}_{\mathit{20}}\mathit{=}(m_{O_{\mathit{3}}}\times 1)\mathit{+}(m_{{\mathit{\text{CH}}}_{\mathit{4}}}\times x_{\mathit{20}\mathit{,}{O}_{\mathit{3}}\mathit{-}e}^{{\mathit{\text{CH}}}_{\mathit{4}}})\mathit{+}(m_{\mathit{\text{CO}}}\times x_{\mathit{20}\mathit{,}{O}_{\mathit{3}}\mathit{-}e}^{\mathit{\text{CO}}})\hfill \\ \mathit{+}(m_{{\mathit{\text{NO}}}_x}\times x_{\mathit{20}\mathit{,}{O}_{\mathit{3}}\mathit{-}e}^{{\mathit{\text{NO}}}_x})\hfill \end{array}$$ (7)

where “₂₀” denotes the chosen time horizon and “$m_i$” (i = CO₂, CH₄, CO, NO_x, PM_2.5, SO₂, NH₄⁺, NO₃⁻, and PO₄³⁻) represents the mass of the emitted substances. “$x$” represents characterization factors in life cycle impact assessment. The key techno-economic assumptions are consolidated in Table 2, which summarizes the plant basis, economic parameters, and cost inputs that were applied consistently across all process configurations.

Table 2. Summary of key techno-economic analysis assumptions.

Category	Parameter	Value	Notes
Plant basis	Operating time	7200 hours·a−1	Consistent for all processes
	Plant lifetime	20 years	Used for DROI calculation
	Discount rate	12%	Standard industrial benchmark
	Capacity basis	n-C4H8O: 454 kilotonne/a	Ensures fair comparison
Economic framework	Cost estimation method	Scaling law (R = 0.67)
Capital cost	Reference processes	PDH and HF industrial data
Operating cost	Capital expenditure components	Purchased equipment costs (PEC), installation, construction, indirect costs	Tables S16 and S17
	Operating expenditure components	Raw materials, utilities, labor, overhead	Table S18
Utilities	Electricity price	Aspen default	Table S20
	Steam and heating costs	Aspen default	Includes recovery credits
Raw materials	C3 LPG price	$610 per tonne	Market data 2025 Q1
	CO price	$650 per tonne	Market data 2025 Q1
	H2 price	$4000 per tonne	Market data 2025 Q1
	n-Butyraldehyde price	$1350 per tonne	Market data 2025 Q1
	i-Butyraldehyde price	$1450 per tonne	Market data 2025 Q1
Financials	Tax rate	13%	Applied to operating profit
Methodology	Simulation tool	Aspen Plus V12	Process modeling
	Thermodynamic model	NRTL	For vapor-liquid equilibrium (VLE) accuracy

Supplementary Materials

This PDF file includes:

Figs. S1 to S20

Supplementary Text

Tables S1 to S47

References