Co1/Ru Single‐Atom Alloy Catalyst for Sustainable Polypropylene Hydrogenolysis to Long‐Chain Liquid Products Authors: Yuzhen Ge, Yimeng Jin, Alexandra Krestnikova, Sibei Zou, Antonio J. Martín, Gonzalo Guillén‐Gosálbez, Javier Pérez‐Ramírez Published: 2026 ABSTRACT ABSTRACTAchieving selective polyolefin waste hydrogenolysis into fuels and chemicals requires architectures that disrupt Ru─Ru ensembles without compromising the intrinsic activity of the state‐of‐the‐art Ru catalyst. While metal modifiers can beneficially tune selectivity, they have been shown for high promoter‐metal contents, with associated penalties in structural definition and potential practicality. Here, we engineer cobalt speciation on Ru through impregnation to disclose how promotion spans the entire compositional space realizing a titania‐supported Co1/Ru single‐atom alloy catalyst (Co10Ru90, 3 mol% total metal content) that resolves these trade‐offs. Ex situ and operando analyses reveal that these bimetallic ensembles containing Ru─Co─Ru sites display a distinctively high barrier for over‐hydrogenolysis at late reaction stages while preserving the intrinsic activity of Ru. As a result, virgin polypropylene and post‐consumer waste were almost fully converted with a consistently high C11+ yield of up to 60% (78% total liquids), compared to <5% over the monometallic Ru counterpart. Process‐level simulations predict economically viable and environmentally sustainable operation enabled by the high single‐pass yields of long‐chain liquid products. These results reveal a wide compositional landscape for Ru promotion and recommend its second‐metal‐content single‐atom alloy limit as the most promising strategy for further catalyst design efforts. Introduction As of 2024, global plastic production exceeded 450 million metric tons, with polyethylene (PE) and polypropylene (PP) accounting for 55%–60% [1, 2, 3]. Yet, only around 1% is chemically recycled on average and mainly via the well‐developed pyrolysis and gasification. These are energy‐intensive but highly effective routes for producing synthesis gas or short‐chain hydrocarbons with more challenging selectivity control toward longer products [4, 5, 6, 7, 8, 9, 10, 11]. Complementarily, hydrogenolysis of polyolefins into liquid alkane hydrocarbons under moderate conditions shows promise, but remains within laboratory walls mainly due to catalysts’ insufficient selectivity‐activity balance [12, 13, 14, 15]. State‐of‐the‐art supported Ru catalysts can efficiently activate polyolefin C─C bonds [16, 17, 18]. However, the strong affinity for reaction intermediates of contiguous Ru─Ru sites promotes over cleavage into low‐value gaseous and short‐chain liquid products [19, 20, 21] incompatible with economic and environmental feasibility [22]. Varying the geometric and electronic structure of Ru sites through suitable supports or metal modifiers can regulate intermediate adsorption and thus enhance selectivity toward longer‐chain products, though the extent is highly dependent on the nature of the polymer [23, 24, 25, 26, 27]. For example, alloying Ru with other metals (e.g., Mn, Fe, Co, and Ni) opens C─C cleavage pathways leading to less severe over‐hydrogenolysis [28, 29, 30]. In a previous work, our group demonstrated that titania‐supported Ru1Ni3 nanoparticles increased liquid product yields from 3% for monometallic Ru to ∼50% in consumer‐grade high‐density polyethylene (HDPE) hydrogenolysis, claiming favorable sustainability prospects. This performance was attributed to the in situ formation of Ru─Ni alloy domains that favor internal over terminal C─C cleavage through both hydrogenated and dehydrogenated intermediates [22]. Similar improvements in liquid product selectivity have recently been reported for CeO2‐supported Ru1M3 (M = Fe, Co, Ni) catalysts for low‐density polyethylene (LDPE) hydrogenolysis [31]. These gains in selectivity control, probably linked to disruption of contiguous Ru─Ru ensembles, are generally accompanied by reduced polymer conversion, while the precise nature of these alloys remains unclear. This strategy of disrupting Ru─Ru ensembles with a second metal in materials with high M:Ru ratios has been recently taken to its extreme by a catalyst consisting on isolated ruthenium atoms (ca. 1.3 wt.%) in a matrix containing cobalt nanoparticles (ca. 49 wt.%) and aluminum (Co:Ru = 65:1 mol/mol) [32]. This catalyst offered ca. 80% yield to liquid products from LDPE, assigned to hydrogen spillover from Ru centers to adjacent Co ones in a single‐atom alloy structure. These results reveal that promotional effects may extend across a wide compositional range over the correct atomic structures and sparks the question of whether the symmetric strategy, i.e., using atomically‐dispersed low‐activity metals as selectivity modifier for ruthenium may achieve high performance over a well‐defined structure allowing mechanistic understanding while removing the need of high metal contents. We impart speciation control through systematic compositional screening of the titania‐supported Ru─Co system for the hydrogenolysis of the under investigated polypropylene feedstock, identifying Ru─Co─Ru ensembles in a Co1/Ru single‐atom alloy (SAA) structure that retain the intrinsic activity of monometallic Ru while modulating late‐stage intermediate hydrogenolysis. This selective control strongly favors the formation of environmentally and economically optimal C11+ alkane [22] from the hydrogenolysis of post‐consumer goods. The SAA architecture enables a precise atomic description of sites [33, 34, 35, 36, 37], offering a strong platform to understand metal–metal interactions and derive mechanistic features able to tune selectivity in this complex reaction, while preserving the host metal's intrinsic activity. Results and Discussion Preparation and Structure‐Performance Relation of Supported Co x Ru y Catalysts TiO2‐supported Co x Ru y catalysts were synthesized by wet impregnation, where x and y indicate the molar ratios of Co and Ru, respectively [18, 38]. Catalysts with a total metal loading of 3 mol% and varying Co:Ru ratios were firstly prepared (Table S1) and systematically evaluated under standard conditions: 0.1 g of catalyst, 1.0 g of PP12 (M w = 12 kDa), 513 K, 20 bar H2, 12 h, 750 rpm. Monometallic Ru100 achieved complete PP12 conversion (defined as C1‐C35 yield) but predominantly produced gaseous products, including 58% methane and 13% C2‐C5 hydrocarbons, with only 4% C11+ long‐chain products (Figure 1a and Table S2). The incorporation of a small amount of Co in Co10Ru90 slightly reduced PP12 conversion to 87% but dramatically decreased gaseous product yield from 71% to 13% and simultaneously increased the C11+ yield to 55%, surpassing previously reported results (Table S3). Further increase of the Co content led to a pronounced decline in PP conversion until 12% for Co100, resulting in a volcano‐shaped dependence of C11+ yield on composition, with the maximum observed for Co10Ru90. Catalytic performance for PP hydrogenolysis over TiO2‐supported Co x Ru y catalysts. Effect of (a) Co:Ru ratios and (b) total metal loading for Co10Ru90, keeping the total amount of Ru and Co used in each reaction. Detailed compositions are available in Table S1. (c) Reusability of Co10Ru90 over multiple reaction runs. (d) Performance obtained with different PP‐based feedstocks. 1, Ru100 after 12 h reaction. 2, Co10Ru90 after 12 h reaction. 3, Co10Ru90 after 24 h reaction. Reaction conditions (unless otherwise specified): 1.0 g PP12, 0.1 g catalyst, 513 K, 20 bar H2, 750 rpm, 12 h. The catalytic behavior of Co10Ru90 samples with varying total metal content were then examined (Figure 1b and Table S2). Bare TiO2 showed negligible activity. Increasing metal loading first enhanced conversion, reaching 87% at 3.2 mol% then declining to 73% at 6.7 mol%. Long‐chain liquid products were in all cases the predominant fraction with the C11+ yield following a mild volcano trend, which could be attributed to geometric effects associated with nanoparticle size, influencing substrate adsorption and C─C scission efficiency [23, 24]. Notably, the Co10Ru90 catalyst with a total metal content of 1.6 wt.% (1.5 wt.% Ru and 0.1 wt.% Co) exhibited performance comparable to a previously reported Ru1/Co–LDH catalyst [32] (Table S4) exhibiting the inverse single‐atom alloy structure and developed for polyethylene processing, which contains a similar ruthenium content (1.3 wt.%) but a much higher cobalt one (49 wt.%). This result highlights the efficiency of the present catalyst design strategy leveraging low modifier contents. The structural effect of the support was investigated by replacing TiO2 with various oxides (SiO2, Al2O3, ZrO2, and CeO2) while maintaining a constant Co10Ru90 metal loading (Table S5). The C11+ yield followed the order TiO2 > Al2O3 ≈ ZrO2 > CeO2 > SiO2. This trend can be attributed to differences in metal‐support interactions [39], with TiO2 promoting the formation of uniformly distributed Ru nanoparticles of optimal size, as evidenced by the XRD results (Figure S1), thereby enhancing catalytic performance. The apparent differences among Ru reflections strongly suggest that both too mild (SiO2) and too strong (CeO2) interactions result in suboptimal Ru nano‐structuring. Reusability tests of the optimal Co10Ru90 catalyst (3.2 mol%) showed stable performance over four cycles (Figure 1c and Table S6), with PP12 conversion at 87%–94% and C11+ yields of 55%–63%, highlighting its potential for practical application. To evaluate the versatility of Co10Ru90 to valorize different feedstocks, control experiments with Ru100 as reference were tested for the conversion of PP12, PP250, and plastic waste (PP bottles, shampoo caps, and yogurt cups). Across all feedstocks, Ru100 produced predominantly gaseous products (>70%) with less than 10% C11+ hydrocarbons (Figure 1d and Table S7). In contrast, Co10Ru90 consistently suppressed gas formation below 13% and increased the C11+ product selectivity to above 50% under standard conditions (12 h). C11+ product yields varied across the series due to difference molecular weights and presence of additives interacting with the catalyst. Maximum values were 59% for PP12, 40% for PP250, 27% for PP bottles, 44% for shampoo caps, and 38% for yogurt cups. Catalyst characterization was carried out to preliminary elucidate the structure originating the enhanced performance resulting from Co incorporation into Ru. Synchrotron high‐resolution powder diffraction (HRPD), offering substantially higher resolution and sensitivity than conventional XRD, was employed to probe the crystallinity and for phase identification for the Co x Ru y catalysts (Figure 2a). Progressive shifts of the characteristic reflections from Ru100 to Co100 with increasing Co content preliminarily indicate Ru─Co interactions that were further explored by X‐ray absorption spectroscopy (XAS). Fourier‐transformed extended X‐ray absorption fine structure (EXAFS) analysis at the Ru K‐edge (Figure 2b and Figure S2), together with fitting results (Table S8), shows that the Ru─Ru coordination number (CNRu─Ru) decreases from 7.0 in Ru100 to 3.0 in Co70Ru30 with increasing Co content, while the Ru─Co coordination number (CNRu─Co) increases from 0.6 in Co10Ru90 to 4.8 in Co70Ru30. At higher Co contents (and thus lower Ru content), reduced spectral resolution precluded reliable fitting. These results early supported the formation of CoRu alloy nanoparticles. X‐ray absorption near‐edge structure (XANES) spectra at the Ru K‐edge (Figure S3) reveal a “white‐line” intensity (ca. 22117 eV) comparable to that of Ru metal foil and significantly lower than that of Ru oxide, indicating that Ru in the catalysts predominantly exists in the metallic state. Structural characterization of TiO2‐supported Co x Ru y catalysts and their correlation with catalytic performance. (a) HRPD spectra. The gray lines indicate the characteristic diffraction peaks of CoRu species. (b) Ru K‐edge EXAFS profiles. C11+ yield and conversion (defined as C1‐C35 yield) of PP12 as a function of (c) CNRu–Ru and (d) CNRu–Co:CNRu–Ru ratio. The color code for catalysts in the legend applies to all panels. A correlation between catalytic performance and coordination number for Co x Ru y catalysts gave an initial insight into structure‐performance relationships. PP12 conversion increases monotonically with higher CNRu─Ru, underscoring the role of contiguous Ru─Ru sites in promoting substrate conversion (Figure 2c). In contrast, the C11+ product yield exhibits a volcano‐shaped dependence on CNRu─Ru, reaching a maximum at CNRu─Ru = 6.3 and CNRu─Co = 0.6 for Co10Ru90, indicating that introducing a small fraction of Ru─Co bonds is critical for achieving high selectivity toward long‐chain products. Further increases in CNRu─Co or CNRu─Co:CNRu─Ru ratio did not improve C11+ selectivity and reduced PP12 conversion (Figure 2d and Figure S4). Based on these results, the Co10Ru90 catalyst was then examined in depth. Identification of Co1/Ru SAA in Co10Ru90 Further characterizations were performed to identify the speciation governing Co10Ru90 catalysts. High‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) reveals nanoparticles predominantly 2–5 nm in size, along with a minor population of smaller clusters (Figure 3a and Figure S5). Atomic‐resolution STEM images show a lattice spacing of 0.234 nm, corresponding to the (100) plane of hexagonal close‐packed Ru, indicating Ru‐rich nanoparticles (Figure 3b). Energy‐dispersive X‐ray spectroscopy (EDX) mapping of individual particles confirms a uniform Ru distribution, while Co appears as isolated angstrom‐scale signals (Figure 3c–e and Figure S6), further indicating the formation of a CoRu alloy. Structural characterization of TiO2‐supported Co10Ru90. (a–e) High‐resolution HAADF‐STEM images with corresponding EDX elemental mapping. EXAFS profiles and fitted curves (dashed lines) at the (f) Ru K‐edge and (g) Co K‐edge. XANES profiles at the (h) Ru K‐edge and (i) Co K‐edge. Corresponding metal foils and oxides were used as references for EXAFS and XANES analyses. (j) Steady state CO adsorption profiles for Ru100, Co10Ru90, and Co100 catalysts, with peak positions indicated. The coordination environment of Co was studied through XAS analyses at the Co K‐edge (Figure S7) for freshly reduced ex‐situ samples. The Co─Ru coordination number (CNCo─Ru) and Co─Co coordination number (CNCo─Co) were 4.1 and zero in the optimal Co10Ru90 catalyst, respectively (Figure 3g). These results indicate that Co in Co10Ru90 interacts exclusively with Ru atoms, proving single‐atom Co alloyed within the Ru lattice (Ru─Co─Ru site, Co1/Ru SAA). Its XANES profile at Co K‐edges (Figure 3i and Figure S3) further indicate partial electron transfer from Co to Ru in Co10Ru90, evidenced by a shift of the Co K‐edge white‐line toward higher unoccupied states relative to metallic Co and closer to Co oxide [34]. Owing to the much higher Ru abundance (Figure 3f), the corresponding increase in Ru electron density is not clearly resolved at the Ru K‐edge (Figure 3h). Although XAS confirms the Co1/Ru SAA structure of the Co10Ru90 catalyst, it does not provide information about the location of Co atoms. To further probe the existence of surface Co/Ru sites, steady‐state CO adsorption diffuse reflectance infrared Fourier transform spectroscopy (CO‐DRIFTS) was performed (Figure 3j and Figure S8). For Ru100, four CO adsorption bands of comparable intensity are observed: a low‐frequency band at 1986 cm−1 assigned to bridged CO on Ru─Ru sites, bands at 2017 and 2044 cm−1 corresponding to linear CO on contiguous Ru sites with high coordination [40, 41], and a higher‐frequency band at 2075 cm−1 attributed to linear CO adsorption at the Ru─TiO2 interface [42]. These features indicate the presence of diverse and continuous Ru surface sites. In contrast, Co10Ru90 exhibits a single CO adsorption band at 2061 cm−1, assigned to linear CO adsorption on Ru sites. The loss of multiple Ru─Ru‐related features indicates that Co incorporation disrupts Ru─Ru continuity, leading to a more uniform surface site distribution and the formation of Ru─Co─Ru ensembles. This observation is consistent with theoretical predictions of low surface segregation energies for CoRu alloys [43]. The upshift of the linear CO band in Co10Ru90 relative to Ru100 suggests weakened Ru→CO back‐donation due to isolation of Ru atoms by Co. Notably, no CO adsorption features associated with Co sites in Co10Ru90 are detected, consistent with the weak CO─Co interaction at room temperature and the absence of CO adsorption bands on Co100. Operando XRD and XAS measurements confirmed the structural stability of the reference Co10Ru90, and Ru100 catalysts (Figures S9–S13). The XRD results indicate the formation of CoRu alloy phases compared to pure Ru, with stable reflections maintained throughout the measurement period (Figure 4a). XANES spectra at the Ru K‐edge for Co10Ru90 exhibit adsorption profiles similar to those of Ru foil, indicating the persistence of a stable metallic Ru phase during the entire measurement (Figure 4b). EXAFS spectra at the Co K‐edge show only Co─Ru coordination without detectable Co─Co contributions for both freshly reduced and spent catalysts from the operando experiment, confirming the stability of the Co1/Ru SAA structure under reaction conditions (Figure 4c). Time‐resolved operando characterization of the Co1/Ru SAA structure in TiO2‐supported Co10Ru90. (a) HRPD spectra, with TiO2‐supported Ru100 included as a reference; characteristic reflections emerging from Ru and CoRu nanoparticles are indicated. (b) XANES spectra at the Ru K‐edge. (c) EXAFS spectra at the Co K‐edge. Corresponding metal foils were used as references for both XANES and EXAFS analyses. Operando HRPD and XAS measurements were conducted synchronously under the following reaction conditions: PP12:catalyst = 1:1 mass ratio, 513 K, 20 bar H2, 4 h. For comparison, the set of TiO2‐supported Co10Ru90 catalysts with varying total metal loadings were also analyzed. HRPD patterns show consistent characteristic reflection positions across all samples (Figure S14), while decreasing peak intensities with lower metal loadings indicate similar Co─Ru interactions but reduced nanoparticle sizes. Consistent with this trend, XAS analysis and fitting (Figures S15–S17) confirm the preservation of the Co1/Ru SAA structure across all Co10Ru90 catalysts, with the Ru─Ru coordination number (CNRu─Ru) decreasing from 7.1, 6.3, and 5.9 to 4.2 as the total metal loading is reduced from 6.7, 3.2, and 1.6 to 0.8 mol% (Table S9), respectively, indicating a progressive decrease in average particle size. Preferential C─C Bond Scission Over Co1/Ru SAA Further control experiments were conducted to elucidate the origin of the enhanced selectivity of Co10Ru90 catalysts toward long‐chain products. To probe intrinsic activity and selectivity, Ru100 and Co10Ru90 were evaluated at different PP12 conversions by varying reaction time (Figure 5a; Tables S10 and S11). At low PP12 conversion (<30%), selectivity toward gaseous products (C1−C5) remained minimal (<2%) for both catalysts, indicating that internal C─C bond cleavage within the polymer backbone initially dominates. As PP12 conversion increased to around 60%, gaseous product formation rose sharply for Ru100 but much more gradually for Co10Ru90. At this stage, C11+ selectivity remained around 50% for Ru100, while decreasing from 85% to 70% for Co10Ru90, reflecting concurrent formation and secondary cleavage of liquid intermediates. Importantly, Ru100 preferentially converted shorter liquids (C6−C10) into gases, whereas Co10Ru90 favored the conversion of long‐chain products (C11+) into shorter liquids. At high conversions (60%−100%), Ru100 showed pronounced over cleavage, with methane selectivity increasing from 16% to 58% and C2─C5 from 8% to 13%, accompanied by a drop in C11+ selectivity from 50% to 4%. In contrast, Co10Ru90 maintained high C11+ selectivity (59%) even at near‐complete conversion, indicating that moderated C─C scission and suppressed excessive hydrogenolysis extent to gaseous products. Control experiments elucidate catalyst and polymer‐dependent C─C bond scission behavior in hydrogenolysis reaction. (a) Selectivity to methane and C11+ products as a function of PP12 conversion. PP12 conversion was controlled by varying reaction time. (b) Catalytic performance of Co10Ru90 using (1) PP12 and (2) HDPE100 as feedstocks. Reaction conditions (unless otherwise specified): 1.0 g PP12, 0.1 g catalyst, 513 K, 20 bar H2, 750 rpm, 12 h. TPSR profiles of Ru100, Co10Ru90, and Co100 using (c) PP12 and (d) 2,4‐dimethylpentane as substrates. MS signal at m/z 16 corresponding to methane was monitored. Of note, both Ru100 and Co10Ru90 catalysts exhibited methane formation even at low PP12 conversion, indicating that methane generation occurs during the initial activation stage. To examine whether demethylation is intrinsic to PP activation, hydrogenolysis reactions were performed under identical conditions using PP12 (branched structure) and HDPE100 (M w = 100 kDa, linear structure) over Co10Ru90 (Figure 5b and Table S12). PP12 reached 87% conversion with 11% methane and 13% total gaseous products yields, whereas HDPE100 was fully converted with markedly higher methane (77%) and total gas yields (94%) despite its longer average chain length. The substantially enhanced gas formation from HDPE100 indicates that internal C─C bond activation and scission along linear chains are rapid on Co10Ru90 catalyst. In contrast, for PP12, the presence of branched −CH3 groups might increase steric hindrance and weaken the interaction between catalyst and polymer. The removal of branched −CH3 groups likely constitutes the rate‐determining step for initial activation. Once demethylation occurs, subsequent main‐chain C─C cleavage may proceed readily. Given the critical role of demethylation, temperature‐programmed surface reaction (TPSR) experiments were conducted, with methane evolution monitored by mass spectrometry (MS, m/z 16). As shown in Figure 5c, methane formation for both Ru100 and Co10Ru90 commenced at around 361 K and increased with temperature, whereas Co100 exhibited negligible methane production, indicating that Ru100 and Co10Ru90 possess comparable and substantially higher demethylation activity during the initial reaction stage. To probe the reactivity of liquid intermediates formed at later stages, 2,4‐dimethylpentane was employed as a model compound due to its structural similarity to PP12. In TPSR experiments with 2,4‐dimethylpentane (Figure 5d), methane formation over Ru100 initiated at 328 K and peaked at around 433 K, whereas for Co10Ru90 it began at a higher temperature (369 K) and peaked near 480 K. These results demonstrate that contiguous Ru─Ru sites in Ru100 promote more facile cleavage of intermediate liquid products into gaseous species, while Ru─Co─Ru ensembles in Co10Ru90 suppresse secondary over‐cleavage, consistent with its enhanced selectivity toward long‐chain hydrocarbons. Operando DRIFTS was further employed to monitor the abundance of different C─H bonds during PP12 hydrogenolysis (Figure 6a–c). Bands at 3015, 2957, 2879, and 2839 cm−1 correspond to the asymmetric C−H stretch of methane, asymmetric −CH3, symmetric −CH−, and symmetric −CH2− vibrations of PP12, respectively [14]. Measurements were initiated at 323 K and increased stepwise by 20 K every 0.5 h up to 513 K, which was then held for 2 h. Time‐ and temperature‐resolved spectra (Figure 6d–f) show that methane intensity increased linearly with temperature up to 473 K for all catalysts, following the trend Ru100 > Co10Ru90 > Co100. Above 473 K, methane formation accelerated sharply on Ru100, increased moderately on Co10Ru90, and remained sluggish on Co100, in agreement with TPSR results. The appearance of methane at low temperatures indicates that branched −CH3 groups can be activated under mild conditions, although the reaction rate is strongly temperature dependent. Concurrently, the depletion of −CH2− and −CH− groups followed the same catalyst order, indicating that Ru100 promote more efficient C─C bond scission, while Co10Ru90 moderates excessive cleavage. Temperature‐ and time‐resolved bond scission tracking in PP12 under reaction conditions. Operando DRIFTS spectra collected at varying temperatures and at 513 K over time for (a) Ru100, (b) Co10Ru90, and (c) Co100. (d–f) Corresponding evolution of peak intensities at indicated wavenumbers, attributed to C─H stretching vibrations in PP12. Reaction conditions: PP12:catalyst=1:1 mass ratio, 5 K min−1, 20 bar H2. To summarize the reaction mechanism, we propose a two‐stage pathway (Figure 7). In the initial stage, at low conversion, the large polymer chains adsorb or wrap on the catalyst surface and undergo demethylation of branched −CH3 groups to form methane, followed by main‐chain C─C bond scission (β‐scission) to generate long‐chain solid or liquid intermediates, with no pronounced difference between Ru and Co1/Ru SAA catalysts. As conversion increases, these intermediates are further processed into shorter liquid products. In the late stage, when long‐ and short‐chain liquid products dominate, catalyst‐dependent behavior emerges. On Ru, strong adsorption and C─C cleavage at contiguous Ru─Ru sites drive extensive over‐cleavage to gaseous products, limiting long‐chain liquid selectivity at high conversion. In contrast, isolated Ru─Co─Ru ensembles in the Co1/Ru SAA catalyst moderate scission, suppressing excessive cleavage and enabling sustained high yields of long‐chain liquid products under optimized conditions. Proposed main mechanistic differences between Ru and Co1/Ru SAA for PP hydrogenolysis. The reaction proceeds in two conversion‐dependent stages: at low conversion, PP undergoes non‐selective demethylation and main‐chain β‐scission to form long‐chain intermediates on both Ru and Co1/Ru SAA catalysts; at high conversion, contiguous Ru─Ru sites promote over‐cleavage to gaseous products, whereas isolated Ru─Co─Ru ensembles in the Co1/Ru SAA suppress excessive C─C scission and preserve long‐chain liquid products. Techno‐Economic and Environmental Analyses The pronounced selectivity differences between Ru100 and Co10Ru90 catalysts toward long‐chain liquid hydrocarbons motivated an environmental and economic assessment of a conceptual 20 t h−1 hydrogenolysis plant for PP waste conversion. Based on the process layout in Figure S18, the analysis includes catalyst regeneration and product separation units designed to reflect the experimentally observed product distributions. Scenarios using renewable hydrogen from wind‐ or solar‐powered electrolysis were compared with fossil‐derived hydrogen via steam methane reforming, while a business‐as‐usual, BAU case represents conventional fossil‐based hydrocarbon production and polypropylene life cycles. For the techno‐economic assessment (Figure 8a), revenues were estimated assuming all products were sold at market prices, while annual production costs included raw materials (H2, catalyst, and waste PP) and operational expenses associated with separation and recycling (Tables S13–S18). Profit was defined as the difference between total revenues and costs. The dependence of revenue, cost, and profit on PP12 conversion was evaluated using the optimal Co10Ru90 catalyst (Figure 8b and Table S19). At low conversion, costs exceeded revenues; however, as conversion increased, both rose until break‐even was reached at approximately 45%–50% conversion due to the high selectivity to long‐chain liquid products exhibited by Co10Ru90. Further increases in conversion led to steadily higher profits, reaching a maximum of 0.4 USD kgproduct −1. These results set a minimum conversion of ca. 50% under high selectivity to C11+ products to reach economic viability. Techno‐economic and environmental analysis of large‐scale hydrogenolysis of PP items. (a) Example calculation of profit margin, defined as the difference between total product revenues at market prices and production costs. (b) Calculated revenues, costs, and profits at different conversions of PP12 using Co10Ru90 catalyst. (c) Example calculation of CO2 emissions reduction, defined as the difference between the global warming potential of the hydrogenolysis process (GWPCR) and that of the equivalent business‐as‐usual process (GWPBAU). (d) Calculated GWPCR, GWPBAU, and CO2 emissions reduction at different conversions of PP12 using Co10Ru90 catalyst. Calculated profits (e) and CO2 reduction (f) for different PP substrates using Ru100 and Co10Ru90 catalysts. Product yields were obtained from Figure 1d at a reaction time of 12 h for Ru100 and 24 h for Co10Ru90. Hydrogen produced through water splitting powered by wind energy was used in all analyses. On the other hand, the environmental assessment indicates that producing liquid fuels via PP hydrogenolysis enables sequential utilization of fossil carbon: carbon atoms in polymers first serve as plastics and subsequently as fuel components. In contrast to the BAU scenario, which relies on separate fossil inputs for plastics and fuels, this circular pathway reduces overall fossil feedstock demand and lowers end‐of‐life emissions. By extending the lifetime of waste PP through hydrogenolysis, carbon release to the atmosphere is delayed, resulting in a decreased cradle‐to‐grave global warming potential (GWPCR, Figure 8c), where the entire life cycle is considered. The effect of PP12 conversion on reduction of CO2 emissions was further evaluated using Co10Ru90 (Figure 8d, Table S20). CO2 emissions were reduced across all conversions and increased only slightly at higher conversion, highlighting that economic and environmental performance for PP hydrogenolysis are only loosely coupled. Similar techno‐economic and environmental analyses were applied to various PP feedstocks over Ru100 and Co10Ru90 (Figures S19 and S20) based on results presented in Figure 1d. Processes using Ru100 consistently yielded negative profits (Figure 8e and Tables S21), whereas Co10Ru90‐based systems were economically favorable due to higher product yields, improved product distributions, and lower hydrogen consumption. Concurrently, Co10Ru90 exhibited substantially lower GWP across all hydrogen supply scenarios (fossil, wind, and solar) compared with Ru100 (Figure 8f and Table S22), largely reflecting its reduced hydrogen and electricity requirements. Sensitivity analyses further reveal that Co10Ru90 maintains profitability and CO2 reductions under variable raw‐material and energy prices, market conditions, and life cycle uncertainties for PP caps (Tables S23 and S24), whereas performance is moderately feedstock‐dependent for bottles and cups (Figures S21 and S22). These results underscore that the Co1/Ru SAA design not only enables selective C11+ production but represents a solid starting point toward practical PP recycling. Conclusions In summary, this work shows that the promotional effects in Ru catalysts extend beyond modifier‐rich materials across a wide compositional range encompassing the opposite low‐content extreme up to the atomic dispersion. More broadly, it demonstrates how engineering of bimetallic catalysts can fundamentally reshape polypropylene hydrogenolysis, steering it toward economically and environmentally beneficial long‐chain liquid hydrocarbons. Multimodal structural and spectroscopic analyses identify isolated surface Co atoms embedded within Ru (Co1/Ru SAA) as the active architecture in TiO2‐supported Co10Ru90 catalysts. Operando and kinetic studies reveal that these Ru─Co─Ru ensembles moderate C─C bond scission and suppress late‐stage over‐cleavage compared to contiguous Ru─Ru sites, sustaining C11+ long‐chain product formation and extending performance to real‐world plastic waste feedstocks. Process modeling shows how atomic‐scale catalyst engineering translates directly into improved economic viability and reduced environmental impact. This low‐modifier single‐atom alloy limit overcomes key limitations of earlier architectures and emerges as the most advantageous design point for advancing both fundamental understanding and practical, selective, and sustainable polyolefin upcycling. Experimental Section Materials Co x Ru y catalysts supported on TiO2 were prepared via a wet impregnation method, where x, y correspond to the molar percentage of Co and Ru, respectively. Typically, TiO2 (1.0 g, anatase, Sigma–Aldrich) was impregnated with calculated amounts of cobalt chloride precursor solution (40 mm CoCl2, Sigma–Aldrich) and ruthenium nitrosyl nitrate solution (Ru(NO)(NO3) a (OH) b , 0.015 g cm−3, a + b = 3, Sigma–Aldrich). The resulting mixture was vigorously stirred at 333 K (200 rpm) until complete removal of the solvent. The obtained solid was vacuum‐dried at 343 K overnight and reduced at 673 K for 2 h under a flow of 50 vol% H2 in He (30 cm3 min−1). Catalysts with different compositions were synthesized by varying the precursor quantities. Co90Ru10 catalysts supported on SiO2, Al2O3, ZrO2, and CeO2 were synthesized using the same protocol. The Ru1Co‐LDH catalyst was prepared via a reported one‐step hydrothermal method [32]. Characterization XRF measurements were conducted using a Rigaku ZSX Primus IV spectrometer. TPSR were performed using a Micromeritics Autochem II 2920 instrument equipped with a TCD and a mass spectrometer [44]. The catalyst (90 mg) was intimately mixed with PP12k (10 mg, Sigma–Aldrich) and loaded into a U‐shaped quartz microreactor, followed by drying in He at 313 K for 1 h and heating in 5 vol% H2 in Ar (20 cm3 min−1) to 873 K at 5 K min−1 [45]. TPSR of the model molecule 2,4‐dimethylpentane were conducted following the same experimental setup and procedure. Prior to testing, the catalyst (100 mg) was impregnated with 2,4‐dimethylpentane and left in fume hood until it dried. HAADF‐STEM and EDX were acquired using either an aberration‐corrected JEM‐ARM300F microscope (GrandARM, JEOL) or a Talos F200X microscope (ThermoFisher Scientific) [45]. Synchrotron XAS and HRPD measurements were performed at the Swiss‐Norwegian beamlines (SNBL, BM31) of the European Synchrotron Radiation Facility (ESRF) [46]. Absolute energy calibration was conducted using metallic Ru and Co references. Both Ru and Co edges were continuously scanned with a 0.5 eV step size. HRPD patterns were collected using a Pilatus3 X CdTe 2M detector and averaged over 60 s [45, 47]. CO‐DRIFT were applied using a Bruker Invenio S FTIR [22]. The sample was first dried in He and then exposed to CO (10 vol% CO in He) for 1 h at room temperature to reach saturation coverage. The sample was then purged by He for 1 h to remove gas phase and physiosorbed CO. The spectra were continuously recorded throughout the entire procedure at 0.5‐min interval. Operando Analyses Operando DRIFT spectroscopy during PP12 hydrogenolysis was applied using a Bruker Invenio S FTIR [48] following this procedure: (i) the catalyst was mixed with PP12 in a 1:1 mass ratio and loaded in the cell, (ii) the sample was flushed with pure H2 and pressurized to 20 bar, (iii) background spectra were collected after stabilization of pressure and signal, and measurements were performed starting at 323 K, increasing by 20 K every 0.5 h until reaching 513 K, which was then maintained for 2 h; (iv) the system was naturally cooled down to room temperature. Spectra were recorded at 0.5‐min intervals with a resolution of 4 cm−1. Operando XAS and HRPD experiments were conducted under the same settings for ex‐situ samples at the same BM31 beamline. Catalyst and plastic were loaded between quartz wool plugs in a quartz capillary reactor (1.5 mm diameter, 0.01 mm thickness) [22]. The experiment procedure was as follows: (i) the freshly reduced catalyst was mixed with PP12k in a 1:1 mass ratio and loaded in the cell; (ii) flushed with H2 and pressurized to 20 bar; (iii) data collection was initiated at room temperature, followed by heating to 513 K and stabilized for 4 h; (iv) the temperature was then decreased to RT. Consecutive XRD and XAS spectra at both the Ru K‐edge and Co K‐edge were acquired for the whole procedure. Catalyst Evaluation Catalytic performance was evaluated in a parallel pressurized batch reactor setup, as reported previously [19]. Typically, 1.0 g of PP and 0.1 g of catalyst were mixed in a glass inset and used for the reaction. Before the reaction, the reactor was purged sequentially with nitrogen and hydrogen before being pressurized with hydrogen to the desired pressure. The reaction mixture was then heated to the target temperature under mechanical stirring and held for the specified time. Upon completion, the reactor was cooled to ambient temperature by circulating chilled water. Product Analysis The products were analyzed following a previously reported method [22]. The glass inset was removed from the reactor and weighed to quantify the total gas produced. The composition of gaseous product was analyzed by GC (Agilent 6890 GC) equipped with an Agilent J&W PoraPlot Q column and FID. The liquid products and solid residue were dissolved by dichloromethane, sonicated and filtered using a syringe. The weight of tube, filter and syringe were measured before and after using in completely dry conditions, with the difference in mass corresponding to the amount of combined solid residue and catalyst. The total amount of liquid products was calculated by subtracting the gas product and solid residue from the amount of substrate used. The filtered liquid samples were analyzed in an Agilent 6890 GC equipped with an HP DB‐5 HT column to determining the distribution of products [19]. In the catalyst reusability test, the liquid and solid products retained in the insert were dissolved in cyclohexane instead of dichloromethane, with all other steps of the analytical procedure maintained unchanged. Process Modeling The PP hydrogenolysis was modeled in Aspen Plus v15 using the Peng‐Robinson thermodynamic package. The reaction was defined based on experimental results for conversion and selectivity over Ru100 and Co10Ru90 catalysts. The PP items were represented as a polyolefin model of consumer‐grade polymers, with a molecular weight of 250 kDa and an average waste PP feed rate of 20 t h−1. Additional details are provided in the Supporting Information. Author Contributions Y.G. and Y.J. synthesized the catalysts, performed the catalytic evaluations and characterizations, and data analysis. Y.G. designed and performed operando experiments. A.K. conducted environmental and techno‐economic assessments under the supervision of G.G.‐G. S.Z. carried out electron microscopy investigations. A.J.M. coordinated the work. J.P.‐R. conceptualized and supervised the entire project and managed resources and funding. Y.G., A.J.M., and J.P.‐R. drafted and revised the manuscript. All authors approved the final version of the manuscript. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Supporting File: adma73643‐sup‐0001‐SuppMat.pdf.