A facile one-step hydrothermal reaction was conducted to synthesise LSP-Z100 using tetra(n-butyl)ammonium hydroxide as structure-directing agent (Methods, LSP-ZX refers to LSP zeolite with Si/Al ratio of X). It adopts MFI/MEL intergrowth structure with self-pillared layers, which feature a large external surface area and a series of mesopores (Fig. 2a–c, Supplementary Figs. 1 and 2 and Supplementary Table 2). A similar structure was found with LSP-Z75 but with reduced surface area and mesoporous volume (Supplementary Figs. 1 and 2 and Supplementary Table 2). A further decrease in the Si/Al ratio prevents the formation of LSP structures. Previous literature shows that strong acid sites and shape selectivity of zeolites are important features for conversion of PE29, but conventional zeolites such as HZSM-5 and HY alone showed little activity due to the narrow pores and limited acid sites on the external surface. Mesoporous materials, such as MCM-41 and SAB-15, have only weak acid sites, which are incapable of cleaving the C(sp3)–C(sp3) bonds under mild conditions. Two-dimensional structured materials are emerging catalysts showing unique advantages, such as high external surface area, but have not been used for PE upcycling. Conventional two-dimensional materials can hardly afford shape selectivity due to unrestricted surface. LSP zeolites have uniquely pillared structures and accessible acid sites for bulky molecules, distinctly different from conventional zeolites. Compared with the commercial HZSM-5, LSP-Z100 showed much higher N2 adsorption (Fig. 2c) and higher content of Q3 [Si(OSi)3(OH)] (−102 ppm) and Q2 [Si(OSi)2(OH)2] (−92 ppm) Si species (Fig. 2d and Supplementary Table 3), due to its distinct LSP network and the presence of mesopores (Fig. 2a). LSP-Z100 exhibits trace extra-framework Al sites (Fig. 2e) and a similar amount of acid sites to HZSM-5 (Supplementary Fig. 3 and Supplementary Tables 4 and 5), but shows higher amounts of strong Lewis acid sites (Fig. 2f, Supplementary Fig. 4 and Supplementary Table 5). Moreover, the acid sites of LSP-Z100 are significantly more accessible to bulky molecules as confirmed by 2,6-di-tert-butylpyridine (DTBPy) infra-red (IR) spectroscopy (Fig. 2g, Supplementary Fig. 5 and Supplementary Discussion 1). This is in contrast to the conventional porous materials (for example, HZSM-5, HY, MCM-41 and SBA-15), which have either limited accessibilities of acid sites or no strong acid sites.

a, High-resolution transmission electron microscopy images of LSP-Z100 and HZSM-5, showing layered self-pillared structure of LSP-Z100. Intermittent lattice fringes (white arrowheads) suggest that mesopores exist throughout LSP-Z100. Note that three sets of transmission electron microscopy images were taken at different magnifications. b, X-ray diffraction patterns of LSP-Z100 and HZSM-5. c, N2 adsorption/desorption isotherms of LSP-Z100 and HZSM-5. d, 1H–29Si cross-polarization/Magic Angle Spinning Nuclear Magnetic Resonance (MAS NMR) spectra of LSP-Z100 and HZSM-5, and de-convolution of 29Si-NMR spectra by Gaussian line shapes. e, Solid-state 27Al NMR spectra of LSP-Z100 and HZSM-5. f, IR spectra before and after adsorption of pyridine at variable temperatures on LSP-Z100 and HZSM-5. The dashed lines indicate vibrational peaks of pyridine molecules adsorbed on Brönsted acid sites (1545 cm–1) and Lewis acid sites (1455 cm–1). g, IR spectra before and after adsorption of DTBPy at 150 °C on LSP-Z100 and HZSM-5. The dashed lines indicate vibrational peaks of zeolite silanol group (3740 cm–1), Brönsted acid site (3160 cm–1) and adsorbed DTBPy molecule (3370, 1616 and 1530 cm−1). a.u., arbitrary units.

Source data

For a typical reaction, high-density PE (HDPE) was mixed with the catalyst in a mass ratio of 5:1 and the mixture was loaded into an autoclave. After purging with N2, the autoclave was heated to 240 °C for 4 h. The reaction was then ceased by cooling to room temperature, and the products collected for analysis. Among all the tested microporous (HZSM-5, HY and USY) and mesoporous (LSP, meso-HY, MCM-41 and SBA-15) catalysts, only LSP zeolites exhibited substantial catalytic activities (Table 1, entries 7–9, and Supplementary Table 6). Importantly, LSP-Z100 gave a high conversion of HDPE of 81.8% (Table 1, entry 8) and an unprecedented selectivity of >99% to C4–C12 compounds (gasoline range) with negligible C1–C3 compounds (<1%) (Supplementary Tables 7 and 8). This is attributed to the advantage that the formation of C4+ compounds, particularly branched C4+ compounds through A-type (tertiary–tertiary) or B-type (secondary–tertiary) β-scission, is energetically more favourable30,31 and can rapidly take place at mild reaction temperatures (Supplementary Discussion 2). With a prolonged reaction time of 24.5 h, the yield of gasoline increased to 87.2% (Table 1, entry 9). Compared with commercial gasoline, the liquid product from this reaction is of premium quality in terms of higher content of branched alkanes (44.1% versus 72.5%), comparable research octane number (86.6 versus 88.0, Supplementary Table 9) and notably lower contents of contaminating aromatics (20.3% versus 10.1%) and olefins (7.4% versus 1.4%) (Table 1, entries 8 and 10, and Extended Data Fig. 1). The reduced olefin content is probably related to the shortened diffusion path in LSP-Z100. This suggests that intermediate alkenes diffuse more rapidly within the catalyst, facilitating adequate interaction with active sites in a confined system. Consequently, this promotes the transformation of intermediate alkenes into alkanes (that is, the main component of gasoline). For instance, reducing the particle size of HZSM-5 from 4 μm to 200 nm resulted in a decrease in selectivity for alkenes (Supplementary Table 10, entries 1 and 2). A series of reported best-behaving catalysts (for example, fresh fluid catalytic cracking (FCC) catalysts27, spent FCC catalysts27, short b-axis ZSM-5 (ref. 28), [C4Py]Cl–AlCl3(ref. 23), Ru/C (ref. 16), Pt/γ–Al2O3 (ref. 24) and Pt/WO3/ZrO2 + HY(30) (ref. 21)) have also been tested in our system (Table 1, entries 11–17), but they show regrettable activity due to the lack of external hydrogen source or additives.

The time course of HDPE depolymerization over LSP-Z100 at 240 °C was studied by analysing the products using gas chromatography (GC), GC mass spectroscopy (MS) and by analysing the solid residues using elemental analysis, 13C MAS NMR, 1H NMR and glow discharge electrospray ionization (GD ESI) mass spectrometry. At 0.2 h (Fig. 3a), alkanes and alkenes were observed, indicating that LSP-Z100 is extremely active to crack PE. Between 0.2–4 h, the yield of alkenes experienced a slight increase and then decreased to <2%, whereas rapid formation of alkanes continued until the completion of reaction when the yield of alkanes reached 75%. From 4 h to 24.5 h, secondary cracking of the C6–C9 products took place (Extended Data Fig. 2), as well as cyclization and aromatization to produce only a small fraction of aromatics and cycloalkanes (Fig. 3a). The 13C NMR spectrum of the solid residue from the reaction showed a clear signal of methyl group (18 ppm), branched (21 ppm) and unsaturated species (120–150 ppm) (Fig. 3b, Supplementary Fig. 6, Supplementary Tables 11 and 12 and Supplementary Discussion 3). This indicates that part of PE undergoes partial cracking, isomerization and hydride transfer to form unsaturated oligomers (H/C <2) as internal hydrogen source. The SSH pathway is validated by elemental analysis of products and reaction residues. As shown in Fig. 3c, C and H content transfer from the solid residue to the product with the increase of reaction time. The H/C ratio of products raises from initially 2.03 to ~2.2 as the reaction starts, while the H/C ratio of residue decreases during the reaction, confirming the internal hydrogen transfer (Fig. 3d). The H/C ratio of total hydrocarbons (products and solid residues) maintained at ~2.03 throughout the reaction (Fig. 3c and Supplementary Table 13) and total mass balances for all hydrocarbons typically closed to within 1.8% (Supplementary Table 14). The SSH strategy inevitably generates a small amount (18.2%) of solid residues (H/C <2). The solid residues were treated with hydrofluoric acid (HF) to remove the zeolite, and the remaining hydrocarbons were separated through dichloromethane (DCM) extraction (Supplementary Table 15) for further characterization by 1H NMR, 13C NMR and GD ESI mass spectrometry. The fraction insoluble in DCM comprises unreacted PE (Supplementary Fig. 7). The DCM-soluble portion is predominantly composed of long-chain alkylaromatics and long-chain alkylcyclic compounds (unsaturated oligomers), with a molecular weight range between 100 and 1,000 (Supplementary Figs. 8–11 and Supplementary Tables 16 and 17). The observation of these unsaturated oligomers further corroborates the proposed SSH mode. Thermogravimetric analysis (TGA) shows that the yield of coke is only 0.58% (Supplementary Figs. 12 and 13 and Supplementary Discussion 4) and thus the residual comprises primarily unsaturated oligomers and uncracked PE, probably restricted by hindered diffusion.

a, The trends of conversion of PE and yield of products. Reaction conditions: catalyst, 0.09 g; HDPE powder, 0.45 g; reaction temperature, 240 °C; N2, 0.1 MPa. b, Solid-state 13C NMR spectra of the reaction residue, showing the presence of unsaturated species and methylene group in residue (32 ppm, Supplementary Fig. 6 and Supplementary Table 12). a.u., arbitrary units. c, Variation of C and H contents and H/C ratio of short chain products and solid residues as a function of reaction time. Bars with dots are attributed to unreacted PE. d, Scheme of PE depolymerization pathway and SSH pathway. Habs, hydride abstraction; β-scis, β-scission; Isom, isomerization; Htf, hydride transfer. e, SANS spectra for HDPE and reacted HDPE at 190 °C. All SANS spectra shown are after the subtraction of SANS spectrum of the empty cell. f, Mass fractal dimensions of LSP-Z100, mixture of LSP-Z100 and HDPE, and reaction of mixture at different reaction time obtained from SANS spectra. Simplified models of LSP-Z100 and HDPE are shown based on SANS results (LSP-Z100, orange; HDPE, grey; argon, cyan). g, A comparison of reactions 1–4 over LSP zeolites at 240 °C for 4 h. After each reaction, the catalyst was used directly after drying. Fresh PE was added to maintain the same mass of hydrocarbons in Rx,0 for each reaction. Rx,0, before reaction; Rx,t, after reaction, where x = 1,2,3,4. Atomic utilization = (sum of products/total added virgin PE) × 100%. h, The variation of C and H content and H/C ratio of products and solid residue in reactions 1–4 over LSP zeolites at 240 °C for 4 h. i, Comparison of gasoline yield over five cycles of reactions over LSP zeolite at 240 °C for 4 h. After each cycle, the catalyst is calcined at 550 °C under air.

Source data

In situ small-angle neutron scattering (SANS) was applied to investigate the diffusion of PE and intermediates to the catalysts during the reaction (Fig. 3e). The decrease on intensity of neutron scattering in the high Q region (>0.2 Å−1) is due to lower incoherent scattering caused by hydrogen atom, suggesting the cracking of HDPE and evaporation of generated light hydrocarbons (Fig. 3e). In addition, the change of mass fractal dimension in the Q region of 0.02–0.2 Å−1 reflects that the components of reaction mixture had altered during reaction (Fig. 3f and Supplementary Fig. 14). Before 4 h, the mass fractal dimension increased gradually, indicating the surface-assisted cracking of PE chains around active sites. In this period, the viscosity of the system decreases because the cracking of long-chain PE into shorter chain oligomers32. At 4 h, the mass fractal reached 2.68 (close to 2.69, the mass fractal dimension of LSP-Z100), indicating the approximate depletion of coil chain of PE upon reaction. This reveals the hindered diffusion of residual oligomers to the active sites, consistent with the catalysis and TGA results.

To promote the diffusion and atomic economy, fresh PE (81.8%) is mixed with the reaction residue (18.2%) after 4 h (R1) and undergoes a new cycle of reaction (R2) (Fig. 3g). The newly added PE acts as solvent to facilitate mass transport of residual oligomers as well as reaction substrate in the new reaction. Importantly, the result of the new cycle is comparable with the first reaction (Fig. 3g and Supplementary Table 18), suggesting that the previous reaction residues can be converted to products with fresh PE, although another solid residue (22.1%) was formed from the new cycle. This reveals the key role of promoted diffusion of residues, thus preserving the activity of LSP-Z100. After four cycles of reaction, the yield of gasoline is maintained at >70% and the total atomic utilization is as high as 91% (Fig. 3g). The H/C contents are balanced during the four reactions (Fig. 3h and Supplementary Table 19) and little coke was formed (~0.5%, Supplementary Table 20 and Supplementary Fig. 15) owing to the widely opened pores of LSP zeolites to prevent coke formation. The excellent stability of LSP-Z100 was further demonstrated by cycling testing (the residues were removed by calcination before next cycle), in which the yield of gasoline remained at 80% over five cycles (Fig. 3i). The characterization of used catalysts confirms little change to the structure and acidic sites, demonstrating the excellent structural stability of LSP-Z100 (Supplementary Figs. 16–18). Thus, LSP-Z100 enables the success of SSH mode for conversion of PE to gasoline with coke resistance and high atomic utilization.

The local environment of the acid sites Si–O–Al was interrogated through near-edge X-ray absorption fine structure (NEXAFS) analysis at the oxygen K edge. The absorption edges of LSP-Z100 were consistent with those of SiO2 and Al2O3 (Fig. 4a). The peak at 534 eV is due to adsorbed water in the materials33. The region between 538 eV and 543 eV (white line) in the O K-edge NEXAFS spectrum is attributed to transitions from O 1s to O 2p anti-bonding states hybridized with Si 3s, Si 3p, Al 3s or Al 3p character34,35,36. To understand the interaction between acid sites and guest molecules, the conversion of HDPE over LSP-Z100 was studied at different reaction stages. At stage 1 of reaction (190 °C for 2 h), LSP-Z100 showed decrease in the white line intensity (538–543 eV) relative to bare LSP-Z100 (Fig. 4b and Supplementary Fig. 19), suggesting a partial filling of the O 2p anti-bonding orbital. This confirms that the generated unsaturated species (for example, alkenes) during the reaction were bound to the Si–O–Al sites by donating electrons to the O 2p anti-bonding orbital. In addition, the increase in intensity in the region of 545–555 eV is due to scattering from the adsorbed unsaturated species on Si–O–Al sites. With the reaction ongoing (stage 2 of reaction, 230 °C for 1 h, Fig. 4b), the white line intensity increased, indicating the adsorbed alkenes are saturated because of hydride transfer reaction. Importantly, at the stage 3 of reaction (240 °C for 2 h), the intensity was closed to bare LSP-Z100, indicating the complete regeneration of Si–O–Al sites upon conversion of alkenes to alkanes that could desorb readily from the catalysts.

a, O K-edge NEXAFS for LSP-Z100, SiO2 and Al2O3. The absorption edges of LSP-Z100 were consistent with those of SiO2 and Al2O3. b, O K-edge NEXAFS spectra for LSP-Z100 before reaction and LSP-Z100 during the reaction. c, 31P MAS NMR spectra of LSP-Z100 and HZSM-5 loaded with TMPO. d, View of structure of Al sites with adsorption of TMPO. e, GC traces of i-C5H12/C6D14 hydride transfer reactions by LSP-Z100 and HZSM-5. I–V are mainly composed of i-C5H11D, C6D13H, i-C5D11H, i-C6D13H and C6D12H2, respectively (Extended Data Figs. 3 and 4). f, Mass spectra of 2-methylbutane after reaction on HZSM-5 and LSP-Z100. g, Mass spectra of hexane after reaction on HZSM-5 and LSP-Z100. h, Scheme of i-C5H12/C6D14 hydride transfer reactions. Habs, hydride abstraction; a.u., arbitrary units.

Source data

The structures of active sites were further investigated by 31P NMR spectroscopy of trimethylphosphine oxide (TMPO) as a probe molecule37. LSP-Z100 possesses a large amount of FTAl sites, which act as Lewis acid sites to coordinate with lone pair electrons of basic probe molecule (64 ppm, TMPO ∙ ∙∙Al(OSi)3; 62 ppm, TMPO ∙ ∙∙Al(OSi)2(OH); 68 ppm, TMPO ∙ ∙∙Al(OSi)(OH)2) (Fig. 4c,d and Supplementary Table 21). In contrast, the spectrum of TMPO adsorbed on HZSM-5 shows higher intensity of TMPOH+ ions protonated by Brönsted acid sites (76 ppm) and less prominent signal of Lewis acid coordinated molecules (63–69 ppm). This result is consistent with the Py-IR experiment that shows higher L/B ratio of LSP-Z100 than HZSM-5 (Fig. 2f and Supplementary Table 5). In addition, the peaks related to FTAl sites of LSP-Z100 (62, 64 and 68 ppm) exhibit slight shifts compared with that of HZSM-5 (63, 65 and 69 ppm), since the Al sites on layered structure of LSP-Z100 are extensively open and have different chemical environment, consistent with the DTBPy IR spectroscopy analysis (Fig. 2g). Thus, the super-strong Lewis acid sites (strongly bound with pyridine even at 450 °C as shown in Fig. 2f) of LSP-Z100 originate from oFTAl. Moreover, the hydride transfer between 2-methylbutane (i-C5H12) and deuterated hexane (C6D14) has been studied at 240 °C to demonstrate the activity of oFTAl. On LSP-Z100, i-C5H12-xDx (x = 1–10), C6D14–yHy (y = 1–3) were observed (Fig. 4e–h, Extended Data Figs. 3 and 4, Supplementary Figs. 20 and 21, Supplementary Tables 22 and 23 and Supplementary Discussion 5), while HZSM-5 shows little activity, indicating that the hydride abstraction and hydride transfer occur primarily on oFTAl. By extending the reaction time for HZSM-5 to 8 h, a similar deuteration level of 2-methylbutane and selectivities of products were observed, compared with that of LZP-Z100 achieved in 2 h (Supplementary Figs. 22–25), thus confirming the higher efficiency of LSP-Z100 in hydride abstraction and transfer. Controlled experiments on LSP-Z100 with decreased accessible acid sites (poisoned by DTBPy) or oFTAl sites (partially removed by ammonium hexafluorosilicate) demonstrated a notable drop in PE conversion (Supplementary Fig. 26, Supplementary Tables 10 and 24 and Supplementary Discussion 6). Also, the contribution of silanol groups with weak Lewis acidity to the conversion of PE is limited (Supplementary Fig. 27 and Supplementary Table 10, entries 5–9). These results demonstrate that oFTAl and accessible Brönsted acid sites play the primary role in the PE conversion.

To further understand the reaction mechanism at an atomic level, in situ INS (Supplementary Fig. 28 and Supplementary Discussion 7), combined with density functional theory (DFT) calculations, was employed to investigate the conversion of HDPE on LSP-Z100. The INS spectrum of activated LSP-Z100 gave a clean background with no prominent features at 0–1,600 cm−1 (Supplementary Figs. 29 and 30). The mixture of HDPE with LSP-Z100 showed a similar spectrum to that of the bare HDPE, indicating little interaction between HDPE and LSP-Z100 upon mixing at room temperature (Supplementary Fig. 30). HDPE on LSP-Z100 underwent the first catalytic conversion at 190 °C for 2 h (stage 1 of reaction, Fig.5a and Supplementary Fig. 31). In comparison with the spectrum of HDPE, the spectrum of stage 1 of the reaction showed a decrease in intensity at 130, 201 and 726 cm−1, which are assigned to the in-plane and out-of-plane skeletal stretching of HDPE, and to the methylene (–CH2–) rocking, respectively. This suggests the scission of C–C bonds and depolymerization of PE chains. Meanwhile, a broad feature at low energy (below 200 cm−1) appeared, suggesting that the intermediate species were disordered over the catalyst surface showing restricted translational and rotational dynamics. In addition, new peaks at 212, 235, 255, 266, 336, 440, 913 and 971 cm−1 were observed. The appearance of peaks at 212–266 cm−1 (assigned to methyl torsion, Supplementary Table 25) further confirms the cleavage of PE chains. Notably, the overall spectrum profile of stage 1 of reaction is consistent with that of adsorbed unsaturated oligomers (Fig. 5b and Supplementary Discussion 8), which is in excellent agreement with the formation of oligomers upon cracking of PE as confirmed by 13C NMR in the time course study. Specifically, shoulder peaks at 212, 255 and 266 cm−1 were identified as methyl torsion of 2-methylpentane and 3-methylpentane (Fig. 5c, Supplementary Fig. 32 and Supplementary Table 25) and the peaks at 336 and 440 cm−1 as the skeletal stretching modes of short alkanes (Fig. 5b, Supplementary Fig. 32 and Supplementary Table 25). This result indicates that PE was activated via hydride abstraction on oFTAl and underwent β-scission/isomerization/hydride transfer to generate iso-alkanes. Although oFTAl are able to activate C(sp3)–H of PE/alkanes, the alkanes produced can readily desorb as supported by the INS spectra of adsorbed alkanes on the LSP-Z100 (Extended Data Fig. 5a,b). In contrast, alkenes bound strongly to active sites to promote their further conversion with SSH (Extended Data Fig. 5c).

a, A comparison of INS spectra for solid HDPE and reacted HDPE over LSP-Z100. b, A comparison of INS spectra of unsaturated oligomers, adsorbed 2-methylpentane (2-MP), adsorbed 3-methylpentane (3-MP) and reacted HDPE (stage 1 of the reaction) over LSP-Z100. The INS spectrum of adsorbed unsaturated oligomer was generated by 1-butene adsorbed on LSP-Z100. The dosed 1-butene oligomerizes over LSP-Z100 and unsaturated bonds of oligomer bind to Brönsted acid site to form carbenium ions. c, Enlarged spectra of b at 0–400 cm−1. d, A comparison of INS spectra of adsorbed 2-MP, adsorbed 3-MP and reacted HDPE (stage 3 of the reaction) over LSP-Z100. All INS spectra shown here are after the subtraction of INS spectrum of the empty cell and zeolite. e, Selected vibrational modes of PE, oligomers, 3-MP and 2-MP (C, grey; H, white) observed by experiments. f, The proposed main reaction pathway for the conversion of PE to iso-alkane over LSP-Z100, including initiation, cracking/isomerization and hydride transfer. Habs, hydride abstraction; Prot, protonation; β-scis, β-scission; Isom, isomerization; Htf, hydrogen transfer.

Source data

Then, the stage 2 of reaction was studied (230 °C for 1 h, Fig. 5a). The very strong peak at 235 cm−1, corresponding to the torsional mode of oligomers, decreased notably in intensity. Meanwhile, the peaks at 212 and 255 cm−1, corresponding to the methyl torsion of iso-alkanes, grew in intensity. This result suggests the rapid conversion of activated HDPE or oligomers to iso-alkanes on the zeolite via β-scission/isomerization/hydride transfer reactions. However, the peaks of HDPE (130, 201 and 726 cm−1) were still observed. Finally, the stage 3 of reaction was also investigated (240 °C for 2 h, Fig. 5a). The peaks at 235 and 726 cm−1 disappeared and the peaks at 212, 255 and 266 cm−1 increased dramatically in intensity. The spectrum of stage 3 of reaction showed notable similarity with iso-hexanes (Fig. 5d,e), confirming the efficient transformation of HDPE into iso-alkanes.

A full catalytic circle can be established (Fig. 5f). Upon adsorption on the LSP zeolite, PE is activated via hydride abstraction38,39 on oFTAl or via protonation on the Brönsted acid sites40 to give carbenium ions or carbonium ions as the initiation step. The carbonium ions formed on the Brönsted acid sites are unstable, rapidly collapsing to afford carbenium ions along with H2 or alkanes. The isotope-labelling reactions (Fig. 4e–h) suggest that the activation via hydride abstraction on oFTAl dominates at 240 °C. Then the activated PE (that is, the carbenium ion originated from hydride abstraction or carbonium ion collapse) binds strongly to the Brönsted acid sites and undergoes rapid β-scission and skeletal isomerization via a three-membered ring structure41,42,43, followed by intramolecular hydrogen transfer to generate stable tertiary carbenium ions. Meanwhile, oFTAl abstract hydrogen from PE or oligomers. Subsequentially, tertiary carbenium ions react with the SSH through hydride transfer to yield iso-alkanes for facile desorption and the acid sites regenerated.