In the first segment of this research, the double hydrolysis of dialkyl phthalate esters (DP) into phthalic acid (PA) was investigated (Table 1). Under the studied conditions (3 mmol bis(2-ethylhexyl) phthalate, DEHP; 25 mL H₂O at 200 °C for 6 h), the diester exhibited a remarkable stability (<5% conversion), even when substantial amounts of homogeneous acid or base were added to the reaction mixture (Table 1, Entries 1–3) (Riemenschneider and Bolt, 2005). It is hypothesized that this arises from the very limited solubility of the organic plasticizer in the aqueous phase (Supplementary Table S2). Given the evident role of water in this reaction and the desire to avoid organic solvents, alternative heterogeneous catalysts were explored instead.

An evaluation of various Brønsted-acid zeolites was conducted (Table 1, Entries 4–13), leading to varying outcomes. For instance, aluminosilicates with too small pore sizes and/or a low Si/Al ratio (which suggests that their surface is rather hydrophilic) appear incapable of promoting the hydrolysis of the bulky and relatively apolar dioctyl phthalate ester (Table 1, Entries 4–8). In contrast, hydrophobic H-Y and H-Beta zeolites with high Si/Al ratios demonstrate enhanced hydrolysis rates, with PA yields up to 31% after 6 h (Table 1, Entries 9–13). Through the extension of the reaction time and/or an increase of the catalyst loading, complete phthalate conversion and quantitative yields of PA could be realized with both H-Y 40 and H-Beta 75 (Table 1, Entries 14–16). Remarkably, the intermediate monoalkyl phthalate (MP) was hardly detected during these experiments, implying a swift conversion of the second ester functionality once the bulky plasticizer molecule has made contact with the zeolite surface. In addition to promoting the hydrolysis of the phthalate ester bonds, these Brønsted-acid zeolites also catalyze the dehydration of the alcohols formed under current reaction conditions. (Liu et al., 2017; Chen et al., 2021). This conversion occurs rather fast once the alcohol (e.g., 2-ethyl-1-hexanol; 2-EH) is formed on the zeolite surface, resulting in a constant and remarkably high selectivity for octenes of ±90% during the hydrolysis of DEHP (Supplementary Figure S17). In other words, complete dioctyl phthalate conversion not only yields >99% PA, but also 10% 2-ethyl-1-hexanol and 90% isomeric octenes.

For the above-mentioned zeolite structures, stability issues have previously been reported when used in hot liquid water (Stuyck et al., 2020; Heard et al., 2020; Ravenelle et al., 2010; Ennaert et al., 2016). Therefore, recycling experiments were conducted with the most promising H-Y and H-Beta zeolites (Figure 2). Whereas H-Y 40 experienced a substantial loss in its hydrolysis activity per cycle (±30%), the reuse of the slightly less active H-Beta 75 resulted in stable outputs across four runs. In fact, H-Beta 75 outperformed the former H-Y 40 catalyst from the first recycle run onwards, proving more suitable for application. In line with these recycling results, powder X-ray diffraction (PXRD) demonstrated the loss of crystallinity of H-Y 40, while the characterization method confirmed the structural stability of H-Beta (Supplementary Figures S15, S16). The simultaneous loss of activity and crystallinity of the spent H-Y zeolite has been linked with the hydrolysis of framework bonds, producing an amorphous structure characterized by a highly reduced pore volume and a substantial decline in number of Brønsted acid sites (Stuyck et al., 2020; Heard et al., 2020). Based on the long-term stability of H-Beta 75, this commercial zeolite was selected for the remaining hydrolysis experiments.

In a subsequent set of experiments, several historically relevant phthalate plasticizers (diisobutyl phthalate, DIBP; dibutyl phthalate, DBP; benzylbutyl phthalate, BeBP; DEHP and DINP) were subjected to the revalorisation process (Figure 3). The double hydrolysis of LMW phthalates other than DEHP (i.e., DBP, DIBP and BeBP) proceeded smoothly, resulting in complete conversions within only 3 h of reaction time. On the other hand, the hydrolysis of DINP required prolonged reaction times (up to 28 h) in order to achieve complete plasticizer conversion and high PA yields. It appears that the hydrolysis of the phthalate plasticizers becomes progressively more challenging as the length of the hydrocarbon side chains increases. These observations can be related to the solubility of the various phthalate plasticizers: as the alkyl chain length of the esters increases, their water solubility decreases (Supplementary Table S2), and this also affects their ease of adsorption on the zeolite surface.

Next, the influence of the formed products on the hydrolysis rate was studied by adding half equivalents of an aromatic carboxylic acid (i.e., benzoic acid, BA) and of a long-chain primary alcohol (i.e., 2-ethylhexanol, 2-EH) to the phthalate mixture (Figure 4; Entries REF, BA and 2-EH). The increased average PA yield when BA was present at the beginning of the reaction may suggest a modest accelerating effect of carboxylic acids on the hydrolysis rate; however, the uncertainty in the data suggests that this influence may not be statistically significant. In any case, if the effect does occur, it would be helpful when dealing with plasticizer mixtures extracted from real plastic waste, which contain both low and high molecular weight phthalates (see below). On the other hand, the cleaved alcohol fragments are quickly converted into alkenes but seem to have little to no effect on the hydrolysis rate.

Phthalate extracts obtained from waste plastics can contain impurities; therefore the effect and fate of the detected extract contaminants on the hydrolysis reaction was investigated. As described elsewhere, the end-of-life plasticizers are extracted from post-consumer vinyl flooring, via a dissolution-precipitation approach whereafter the used solvents are removed from the supernatant by evaporation. The crude extract consists of ±75 weight % of the previously discussed phthalate plasticizers (DBP, DIBP, BeBP, DEHP and DINP), ±20% aliphatic plasticizers and ±5% residual PVC. Additionally, unspecified sulfur-components were detected in low amounts on solids that had been in contact with the plasticizer extract (Windels et al., 2022; Wagner and Schlummer, 2020).

Bis(2-ethylhexyl) adipate (DEHA) was selected as representative for the aliphatic plasticizers, given its substantial share in the extract, also reflecting its historical use. Hence, a blend comprising 75 mol% DEHP and 25 mol% DEHA was formulated at room temperature and subsequently employed as the plasticizer fraction (Figure 4, Entry DEHA). While the conversion rate of the phthalate ester is minimally affected, the adipate ester is likewise transformed into adipic acid under the tested conditions.

Considering the excellent compatibility of phthalate esters with PVC (Bocqué et al., 2016; Howick, 2021), the deep removal of the polymer from such a plasticizer extract is known to be challenging. To evaluate the effect of residual polymer, some virgin PVC was dissolved in DEHP (5 wt% polymer, 3 h at 100°C) and after cooling back to room temperature used in the hydrolysis reaction (Figure 4, Entry PVC). Water is known to be an excellent anti-solvent for PVC (Grause et al., 2017; O’Rourke et al., 2023) and could cause the polymer to precipitate not only on the reactor walls, but also on the heterogeneous catalyst. However, the hydrolysis rate appears to be only slightly affected by this phenomenon, which suggests that the majority of the active sites of the heterogeneous catalyst remain available, indicative for only minimal fouling.

Thus, the dialkyl phthalate hydrolysis seems to be relatively insensitive to detected extract impurities. Therefore a crude plasticizer mixture, extracted from real PVC waste, was used in the following experiments (Figure 4, Entry Waste DP). With an equivalent yield of PA after 6 h (66% PA) compared to the reference DEHP hydrolysis (67% PA), it indeed appears that no waste extract impurity exerts a significant poisoning effect on the zeolite. As a consequence, quantitative yields of PA (>99%) could be obtained from real waste-extracted phthalate plasticizers by extending the reaction time to 22 h. The hydrolysis of legacy phthalate esters using a hydrophobic H-Beta zeolite in water emerges as an exceptionally robust detoxification method for these legacy plasticizers. Furthermore, the obtained aromatic product can be further revalorized into other drop-in chemicals such as benzoic acid and benzene (see below).

Next, we focus on the decarboxylation of PA to selectively obtain either benzoic acid (BA) or benzene (B). Although this specific catalytic reaction has not yet been documented, heterogeneous noble metals have repeatedly been reported for the decarboxylation of other organic acids (De Schouwer et al., 2017; Dawes et al., 2015; Verduyckt et al., 2016; De Schouwer et al., 2015) and were therefore evaluated for this reaction (Figure 5; Supplementary Figure S19). While the control experiment resulted in a minimal conversion of PA (<5% after 22 h; Supplementary Figure S18), all tested noble metal catalysts demonstrated varying degrees of activity for the decarboxylation under the employed reaction conditions (3 mmol PA, 2.5 mol% noble metal, 25 mL H₂O at 225°C for 22 h). More specifically, Pd displayed the lowest activity among the tested metals, showing only 9% conversion after the first hour, while Ru- and Rh-catalysts exhibited an intermediate and remarkably similar decarboxylation rate (45% conversion after 1 h). Pt demonstrated an even higher decarboxylation activity, achieving complete PA conversion within the first hour of reaction. Moreover, each catalyst proved effective for the successive decarboxylation of BA to B under the same reaction conditions. As a result, all time profiles follow the typical course of consecutive reactions (Supplementary Figure S19), wherein the yield of the intermediate product (BA) reaches a maximum before progressing to complete transformation into the final product (B). Consequently, extended reaction times lead to near-quantitative yields of B while the selectivity towards BA requires additional optimization. At 225°C, there appears to be no significant selectivity difference between the noble metals (Figure 5; right), leading to a catalyst choice - being Pt/C - primarily determined by the decarboxylation activity. This choice is further supported by the steady output (PA conversion and BA selectivity) observed over four consecutive decarboxylation experiments, indicative of a stable catalyst (Supplementary Figure S20).

Considering that the maximum concentration of the intermediate in consecutive reactions depends on the ratio of the reaction rate constants (Espenson, 1981), and may thus be influenced by the activation energies, the temperature dependency of both decarboxylation steps was examined. To that end, the initial conversion rates of PA and of BA were determined in separate reactions over a temperature range between 180 °C and 250°C, after which the obtained data were displayed in an Arrhenius plot (Figure 6). From this graph, it can be deduced that the decarboxylation rate of BA is more affected by the reaction temperature than that of PA, showing apparent activation energies of 63 and 32 kJ mol^-1, respectively. Indeed, the time profiles obtained in this temperature range show a steady increase in the maximum BA yield with decreasing temperature, a trend further highlighted in the corresponding selectivity profiles (Supplementary Figure S20). Based on these insights, tuning the reaction parameters (time and temperature) allows a straightforward optimization of the product selectivity. A maximum yield of 86% BA was achieved with 2.5 mol% Pt at 180 °C after 5 h, while the B yield exceed 99% under more rigorous conditions (2.5 mol% Pt at 250°C for 22 h).

Next, the integration of the decarboxylation reaction with the preceding hydrolysis of end-of-life phthalate plasticizers was attempted (Figure 7). In a first evaluation, 0.2 equivalents of 2-EH and 1.8 equivalents of 1-octene were added to the standard PA decarboxylation mixture in order to simulate the reaction blend obtained after the complete hydrolysis of DEHP (Figure 7, Entry Org. layer). Visual inspection of the liquid mixture at room temperature revealed the preferential localization of the Pt-catalyst in the organic top layer, both before and after the experiment (Supplementary Figure S23). Such behavior could explain the observed drastic decline of the decarboxylation rate when alcohol and alkene fragments are present in the liquid mixture. It may suggest strong competitive adsorption of the long-chain alcohol and alkene molecules on the carbon-supported noble metal catalyst, even if the liquid fractions were to become more miscible at elevated reaction temperatures. This catalyst inhibition phenomenon implies that a one-pot hydrolysis-decarboxylation process as such would be inefficient, demanding an intermediate purification step. Hence, from this point onwards the side-chain fragments (primary alcohols and alkenes) were removed from the hydrolysis mixture prior to the decarboxylation experiment. This separation was achieved by completely dissolving PA in the aqueous phase at room temperature through addition of NaOH (without base addition only ±30% PA dissolves), followed by separation of both immiscible liquids through decantation. Less than 5% of the PA was lost during this purification procedure. The following control experiment revealed that the added base has no significant negative impact on the decarboxylation reaction, rendering additional purification steps of the aqueous layer unnecessary (Figure 7, Entry NaOH).

Next, the effect of waste extract impurities, such as aliphatic plasticizers and residual PVC, on the Pt-catalyzed decarboxylation of phthalic acid was investigated (Figure 7). To simulate the real situation as accurately as possible, the decarboxylation test was preceded by the complete hydrolysis of the DEHP/impurity mixture (conditions as stated in Figure 4 with an extended reaction time of 22 h) and the abovementioned PA purification procedure. In a first evaluation, the fate of co-extracted adipate plasticizers (such as DEHA) was studied. Adipic acid, obtained from the hydrolysis of DEHA, remained unseparated from PA during the purification procedure and was consequently subjected to Pt-decarboxylation. The aliphatic carboxylic acid exhibited low stability under the studied reaction conditions, resulting in the formation of pentanoic acid and butane through single and double decarboxylation. On the other hand, the conversion rate of PA (and BA) increased compared to the reference experiment, despite an unchanged carboxylic acid to Pt molar ratio (Figure 7, entries Reference and Adipic Acid). In other words, it appears that co-extracted aliphatic plasticizers have no adverse effect on the hydrolysis-decarboxylation procedure to revalorize phthalate plasticizers, but are themselves stepwise broken down into more simple compounds.

In a final experiment prior to the evaluation of the coupled revalorisation process on real plastic waste-derived substrates, the effect of residual PVC in the plasticizer stream on the Pt-catalyzed decarboxylation was examined. This involved again the complete hydrolysis of a DEHP/PVC mixture (conditions as specified in Figure 4 with an extended reaction time of 22 h) and the subsequent purification of the aqueous phase. Exploiting the excellent antisolvent characteristics of water for PVC (Grause et al., 2017), the polymer was effectively separated from the aromatic substrate (the polymer could not be detected anymore using ¹H-NMR), and therefore not subjected to the decarboxylation experiment. However, the reduced decarboxylation rate observed in the obtained aqueous mixture implies that the Pt-catalyst is still hindered by (undetectable) traces of polymer chains that were transferred into the decarboxylation mixture Figure 7, Entry PVC).

Ultimately, waste plasticizers as such were subjected to the tandem hydrolysis-decarboxylation, as most of the co-extracted compounds demonstrated limited interference with the second step of the proposed revalorisation process. After the complete hydrolysis of the waste phthalate esters and the subsequent removal of the organic top layer, the carboxylic acids were decarboxylated almost as efficiently as when virgin PVC was added to the plasticizer mixture (Figure 7, Entry Waste PA and PVC). Thus, the retarding effects of residual chlorinated components on the PA conversion rate slightly outweigh the accelerating influences, such as the presence of adipic acid. Nevertheless, complete PA conversion is reached between 2.5 and 5 h of reaction time and a similar BA selectivity is obtained throughout the experiments under current conditions (Supplementary Figure S22). This observation underlines the inherent robustness of the proposed revalorisation process, although the lifespan of the zeolite and Pt-catalyst might be affected by the purity of the plasticizer stream.