Selective phenol recovery via simultaneous hydrogenation/dealkylation of isopropyl- and isopropenyl-phenols employing an H2 generator combined with tandem micro-reactor GC/MS

The pyrolysis of bisphenol A (BPA), an essential process ingredient used in industry and many everyday life products, helps produce low-industrial-demand chemicals such as isopropenyl- and isopropyl-phenols (IPP and iPrP). In this study, tandem micro-reactor gas chromatography/mass spectrometry combined with an H2 generator (H2-TR-GC/MS) was employed for the first time to investigate the selective recovery of phenol via simultaneous hydrogenation/dealkylation of IPP and iPrP. After investigating the iPrP dealkylation performances of several zeolites, we obtained full iPrP conversion with over 99% phenol selectivity using the Y-zeolite at 350 °C. In contrast, when applied to IPP, the zeolite acid centres caused IPP polymerisation and subsequent IPP-polymer cracking, resulting in many byproducts and reduced phenol selectivity. This challenge was overcome by the addition of 0.3 wt% Ni on the Y-zeolite (0.3Ni/Y), which enabled the hydrogenation of IPP into iPrP and subsequent dealkylation into phenol (full IPP conversion with 92% phenol selectivity). Moreover, the catalyst deactivation and product distribution over repetitive catalytic use were successfully monitored using the H2-TR-GC/MS system. We believe that the findings presented herein could allow the recovery of phenol-rich products from polymeric waste with BPA macro skeleton.

Polycarbonate (PC) is currently the largest consumer of bisphenol A (BPA). Therefore, pyrolysis of PC waste has been widely studied 1-3 because it allows the recovery of oil and gas from polymeric waste 4,5 . This, by itself, represents a significant advantage to treating polymeric waste combined with other resins and organic additives, which cannot be treated by mechanical recycling and solvolysis [6][7][8][9][10][11][12][13] . However, there are several drawbacks associated with this approach. First, this method generates a mixture of various products like phenol, 4-isopropenylphenol (IPP), 4-isopropylphenol (iPrP) and other alkyl phenols [14][15][16] , some of which are important chemical feedstock that cannot be properly utilised due to their difficult purification. Secondly, the oil obtained through this approach contains oxygen as part of the phenolic compound, which reduces the calorific value when used as a fuel. Even though IPP and iPrP are the major products in this case, they have very low industrial demand. Moreover, these issues are also observed with other BPA-skeletal polymers such as epoxy resins, polysulfones, bismaleimides, triazines, polyarylates, and BPA-type flame retardants. Nevertheless, the global demand for BPA should increase annually 17 due to the rapid worldwide growth of automotive, construction, and electrical and electronic application markets. Therefore, it is crucial that new methods are developed for the recovery of useful chemicals from polymeric wastes with BPA skeleton via the pyrolytic approach.

Results and Discussion
Evaluation of zeolites for the conversion of iPrP into phenol. The effects of zeolite properties such as pore size and acid density on the dealkylation of iPrP were investigated using TR-GC/MS. The zeolite properties and abbreviations are summarized in Table 1. Structure optimisation of iPrP and IPP was carried out with Gaussian R 16W 58 using the density functional theory. The kinetic diameter of these molecules was calculated by the method reported by Wang and Frenklach 59 ( Fig. 3 and SI). Figure 4(a,b) present the GC/MS chromatograms obtained in the presence and absence of each zeolite, as well as the iPrP conversion and aromatics composition data for each catalytic reaction. The chromatographic data show that iPrP was thermally stable under the conditions used herein, and in the absence of zeolite, only iPrP was present ( Fig. 4(a)). Moreover, Z5-24 effectively converted iPrP into propylene and phenol (2 and 3, respectively, Fig. 4(a)) with 98% iPrP conversion rate and 96% phenol selectivity ( Fig. 4(b)) and without any side reactions. The detailed product distributions are summarised in Table S1 (SI). In contrast, the iPrP conversion rate in the presence of Z5-40 and Z5-1500 catalysts drastically decreased (50% and 17%, respectively). Plots of C iPrP vs. acid site density of the ZSM-5 zeolites show perfect linearity (Fig. S4, SI). These results revealed that the dealkylation of iPrP progressed on the acid sites, which worked as Brønsted acids ( Fig. 5(a)). Surprisingly, Y-zeolite and MOR achieved almost complete iPrP conversions, although they have lower acid site densities than Z5-40. However, these zeolites have bigger pore sizes than the ZSM-5 zeolites. These trends suggest that bigger pores than the kinetic diameter of iPrP (0.64 nm) are required for effective dealkylation. It has previously reported that the acid sites inside the pores have a significant role in the dealkylation, transalkylation and isomerisation reactions of diisopropyl-benzenes and -naphthalenes 36,60,61 . The high iPrP conversion and excellent phenol selectivity in the presence of Z5-24, which has slightly smaller pores than the kinetic diameter of iPrP, might be due to a dealkylation enhancement caused by the acid sites on the pore entrances 62 Table 1. Characteristics of the zeolites used in the present work. a Details were provided from Tosoh Corporation. b Determined using the area of the NH 3 temperature-programmed desorption (NH 3 -TPD) profile. NH 3 -TPD profile is summarised in Fig. S1 in the Supporting Information. Evaluation of zeolites for the conversion of IPP into phenol. The IPP conversion in the presence and absence of the zeolite catalysts was investigated using TR-GC/MS via the same procedure as for the iPrP conversion. The obtained GC/MS chromatograms, IPP conversion rates, and aromatics composition data are summarised in Fig. 4(c,d). The results confirmed that the bulk of IPP was rapidly evaporated in the first micro-reactor and carried into the second micro-reactor, while trace amounts of phenol, iPrP, other alkyl phenols and IPP dimers were also observed in the absence of a catalyst (Table S2). This suggests that a small-scale IPP pyrolysis reaction 20 occurred in the reactor. Moreover, the IPP conversion rate and phenol selectivity in the presence of Z5-24 were only 62% and 58%, respectively, much lower than the values obtained for the iPrP conversion. Although the IPP conversion and phenol selectivity decreased with decreasing acid site density of the ZSM-5 zeolites, the IPP conversion was not proportional to the acid site density (Fig. S4). FER, the catalyst with the highest acid site density and smallest pore size, showed higher IPP conversion rate (75%) and a lower phenol selectivity (44%) than the Z5 catalysts. Moreover, propylene was not detected in the presence of ZSM-5 and FER zeolites. These results suggested that the phenol in this reaction was not obtained from a direct IPP dealkylation, but from thermal decomposition of the IPP pyrolysates. The kinetic diameter of IPP (0.63 nm) is comparable with that of iPrP, suggesting that IPP mainly reacted in the acid sites on the surface of ZSM-5 and FER. In fact, Brzozowski and Skupiński 60,62 suggested that the external surfaces and pore entrances were also important reaction sites, though the molecules do not enter inside the pores. Moreover, various alkyl phenols were observed in the reactions examined herein in the presence of Z5-24 and Z5-40, while the very weak acid Z5-1500 did not enhance the IPP reaction. This suggest that the acid sites catalysed the IPP polymerisation and subsequently enhanced the cracking of the IPP polymers, resulting in diversification of the alkyl phenols ( Fig. 5(b)).
In the presence of MOR, relatively higher IPP conversion (76%) and phenol selectivity (58%) were observed. Moreover, propylene was formed simultaneously with the phenol production. This suggests that IPP could enter the pores, where it transformed into iPrP, which subsequently dealkylated to form phenol ( Fig. 5(a)). Both IPP conversion and phenol selectivity increased (over 99% and 74%, respectively) in the presence of the Y-zeolite, which has the largest pores (0.90 nm), thereby favouring the entrance of IPP molecules. The size limitations of the pores could arguably explain the different behaviours of the IPP molecules depending on the pore size. For ZSM-5 zeolites and FER, the IPP molecules mainly reacted at the acidic sites on the surface, which had no space limitations, thereby enhancing the IPP polymerisation. In contrast, the IPP molecules which entered into the pores of the Y-zeolite could not be enlarged by polymerisation due to the limited space inside the pores. According to the optimised structures and kinetic-diameter calculations (Fig. S3, SI), IPP dimers (kinetic diameter: 0.80 nm) were the largest oligomers that could enter the pores, with IPP trimers already being too large. Therefore, the Y-zeolite favoured the decomposition of IPP and IPP dimers into phenol rather than the polymerisation.

Characterisation of the synthesised Ni/Y catalysts.
Among all the zeolite catalysts tested, Y-zeolite showed the best catalytic ability for iPrP dealkylation and phenol production from IPP from all tested zeolite catalysts; therefore, Ni-supported Y-zeolites with different Ni loadings were synthesised. Ni was loaded without any structural damages to the zeolite frame (Fig. S2). Fig. 6 shows the TEM images of the original Y-zeolite and the Ni/Y catalysts with different amounts of Ni (0.3Ni/Y, 1.0Ni/Y, and 3.0 Ni/Y). The TEM images of the Y-zeolite and 0.3Ni/Y were comparable because the Ni particles were not observed in the 0.3Ni/Y TEM image due to the small Ni loading. Simultaneously, the energy-dispersive X-ray (EDX) spectrum of 0.3Ni/Y confirmed the presence of Ni in the catalyst (Fig. 6(e)). In contrast, the Ni particles were clearly observed in the TEM images of 1.0Ni/Y and  3.0Ni/Y, with an average particle sizes of 8 nm and 13 nm, respectively, determined by equation (1) in the SI. The TPR profiles of the synthesised catalysts ( Fig. 6(f)) showed that the initial NiO reduction temperature decreased with decreasing Ni loading amount, indicating the presence of smaller Ni particles in 0.3Ni/Y than in 1.0Ni/Y and 3.0Ni/Y.    Fig. 4(e,f). Chromatograms obtained in the absence of a catalyst and in the presence of Y-zeolite were used for comparison. All the catalysts achieved >99% IPP conversion ( Fig. 4(f)). Surprisingly, both 0.3Ni/Y and 1.0Ni/Y showed full IPP conversion and 93% phenol selectivity, much higher than those obtained by the Y-zeolite alone. This was due to the hydrogenation of IPP into iPrP and subsequent dealkylation of iPrP by the Ni/Y catalysts (Fig. 5(c)). However, the phenol selectivity decreased substantially to 40% when 3.0Ni/Y was employed, while the amounts of iPrP and hydrodeoxygenated products such as benzene and alkyl benzenes increased. The rise in the iPrP amount suggests that the dealkylation ability decreased possibly due to the Ni covering reducing the number of acid sites. Moreover, the hydrodeoxygenation was enhanced substantially by the increase in H 2 concentration to 48 vol%, resulting in 66% selectivity for benzene and alkyl benzenes, and 22% selectivity for phenol in the liquid products. Thus, it can be concluded that higher H 2 concentration and higher Ni loading are unsuitable for recovering phenol in this reaction system. Notably, the shift in retention time under 48 vol% H 2 flow (Fig. 4(e)) was due to the lowered carrier gas viscosity. Thus, we confirmed that the reaction with 0.3Ni/Y under 10 vol% H 2 achieved 100% IPP conversion and 93% phenol selectivity.
Monitoring the deactivation of 0.3Ni/Y. The conversion of IPP into phenol using the 0.3Ni/Y catalyst under 10 vol% H 2 atmospheric flow was repeated 10 times to monitor the deterioration of the catalytic ability. The IPP conversion and liquid composition data obtained for each run are summarised in Fig. 7. The first run showed a complete IPP conversion with 94% phenol selectivity. Moreover, complete IPP conversion was observed until the end of the forth repetition and the amount of iPrP seemingly increased with every following repetition. However, the IPP conversion decreased significantly after the 7 th repetition, reaching 55% after 10 repetitions. The phenol selectivity in the liquid products also decreased to 7% in the 10 th repetition. Although optimization of the catalyst bed temperature, feed concentration, and carrier gas flow, as well as the reduction of catalyst/sample ratio, are needed for process improvement in the future, important information regarding catalyst behaviour has been provided herein using a new H 2 -TR-GC/MS system. Comparison between images of the catalyst bed before and after the 10 subsequent repetitions (Fig. 8(a)) indicates that coke was deposited on the catalyst. Since coke is preferably deposited on the acid centre, the acid sites in this case were preferentially covered by coke. Moreover, the fact that the dealkylation of iPrP was deactivated at an early repetition stage suggests that the zeolite pores were blocked by coke deposition during the repetitions. This, in turn, resulted in reduced availability of the acidic sites and lowered access of IPP to the Ni-support in the pores, which is one of the reasons for the decrease in IPP conversion with every repetition. A SEM image and EDX spectrum of the used 0.3Ni/Y (Fig. 8(b-d)) confirmed the coke deposition on the catalyst. However, the type of coke was indistinguishable. Moreover, the Ni particles were still not observed even after 10 repetitions (Fig. 8(d)). When Ni is sintered, it forms bigger particles after considerable amount of time 63,64 , implying that the conditions used herein did not produce significant Ni sintering.
The temperature programmed oxidation (TPO) thermogram (Fig. 8(e)) of 0.3Ni/Y after 10 repetitions shows a two-step weight loss. The first step (until 200 °C) corresponds to the release of moisture from the catalyst 65  either the layered or the filamentous carbon deposited on the catalyst 67 . The fact that no filamentous carbon was detected in the SEM analysis suggests that the deposited coke is in fact layered carbon.
In this work, we employed H 2 -TR-GC/MS for the first time to investigate the possibility of selective phenol recovery via simultaneous hydrogenation/dealkylation of compounds with low industrial demand such as IPP and iPrP. Below, we have summarise the main results and conclusions from this work.
1. The Y-zeolite showed the best iPrP dealkylation ability, with >99% conversion and phenol selectivity at 350 °C. Since the pore size (0.90 nm) of the Y-zeolite was large enough to accept the iPrP molecules (0.64 nm), the dealkylation reaction was promoted on the acidic sites inside the pores. In addition, the Y-zeolite achieved IPP conversion of over 99%, while simultaneously providing lower phenol selectivity (74%) due to the polymerisation of IPP molecules and the subsequent cracking of the IPP polymers, which were simultaneously catalysed on the acid sites. 2. Complete IPP conversion and 92% phenol selectivity was obtained from the synthesised 0.3Ni/Y catalyst in the presence of 10 vol% H 2 atmospheric flow via a hydrogenation of the isopropenyl unit of IPP on the Ni-support and subsequent dealkylation on the acid sites in the Y-zeolite. The repetitive use of the catalyst led to coke deposition and subsequent deactivation of the hydrogenation and dealkylation abilities, which were successfully monitored using an H 2 -TR-GC/MS system. 3. In this study, IPP and iPrP (low industrial demand), which are the main pyrolysates of BPA skeletal polymers, could be selectively converted into phenol (high industrial demand) using relatively low-cost Ni/Y catalysts under mild conditions. The findings presented herein can allow the recovery of phenol-rich products from BPA skeletal polymeric waste and can also be expanded to recover phenol-rich products from alkyl phenols derived from lignocellulosic materials. 4. We believe that the novel methodology proposed herein, viz. the combination of tandem micro-reactor, H 2 -generator and GC/MS, can be employed to evaluate catalytic hydrogenation and dealkylation systems, and thus contribute substantially to the fields of green chemistry and reaction engineering.

Materials. Reagent-grade 4-isopropyl phenol (iPrP) and 4-isopropenyl phenol (IPP) were obtained from
Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan) and Kanto Chemical Co., Ltd. (Tokyo, Japan), respectively. Protonic zeolites HOA822 (Z5-24), HOA840 (Z5-40), HOA890 (Z5-1500), HUA360 (Y-zeolite), HOA640 (MOR), and HOA722 (FER) with particle sizes less than 10 μm were supplied from Tosoh Corporation (Tokyo, Japan). The characteristics of the zeolites used in this work are summarised in Table 1. The acid site densities of the zeolites were determined via ammonia temperature-programmed desorption (NH 3 -TPD) measurements, the detail analytical methods for which are summarised in SI. Even though zeolites Z5-24, Z5-40 and Z5-1500 have the same size pore (0.58 nm), they exhibited very different acid site densities (1.14, 0.51 and 0.08 mmol/g, respectively). In comparison, the Y-zeolite, which has the biggest pore size (0.90 nm) among the zeolites used herein, had relatively low acid site density (0.14 mmol/g). On the other hand, the zeolite with the smallest pores (0.48 nm) among the zeolites used herein, FER, exhibited the highest acid site density (1.45 mmol/g). Additionally, MOR had an acid density (0.49 mmol/g) comparable to Z5-40, while its pores (0.70 nm) were bigger than those of the ZSM-5 zeolites. Structure optimisation of iPrP and IPP was carried out with Gaussian R 16 W 58 , a computational chemistry software, using the density functional theory. The kinetic diameter of these molecules was calculated by the method reported by Wang and Frenklach 59 ( Fig. 3 and SI). In general, the kinetic diameter is the most important diameter to consider when examining the adaptability of molecules to the zeolite pores.  could evaporate from the beaker. The recovered solid was calcined at 200 °C for 16 h in air to obtain NiO/Y. The synthesised Ni/Y compounds with 0.3, 1.0 and 3.0 wt% Ni loadings were named 0.3Ni/Y, 1.0Ni/Y, and 3.0Ni/Y, respectively. Before all the following experiments were conducted, the NiO/Y catalysts were converted to Ni/Y in a reactor (more details available in the next section). The synthesised catalysts were then characterised via temperature-programmed reduction (TPR), X-ray diffraction (XRD), scanning electron microscopy-energy dispersive X-ray spectroscopy (SEM-EDX) and field emission-scanning transmission electron microscopy (FE-STEM). The detailed analytical conditions are summarised in the SI. (Koriyama, Japan); GC: 7890 A, Agilent Technologies (Tokyo, Japan); Column: Ultra ALLOY ® metal capillary column UA + -1, Frontier Laboratories Ltd. (Koriyama, Japan)) reported in our previous papers 56,57 , was combined with an H 2 generator (HG270, H 2 purity: over 99.99%, GL Sciences (Tokyo, Japan)) and an H 2 mass flow controller (T1000, Fujikin, Osaka, Japan). The resulting system was abbreviated as the H 2 -TR-GC/MS system (Fig. 2). The detailed conditions are summarised in the SI. iPrP and IPP conversion using different zeolites. A sample holder, filled with 0.5 mg iPrP or IPP, was placed in the upper part (outside the heating zone) of the first micro-reactor. The flow rate was set at 104 mL/min (100:1 of split ratio; column flow rate of 1 mL/min; septum purge rate of 3 mL/min). The combination of 0.5 mg sample and the split ratio 100:1 was suitable for obtaining reliable peak intensity and good peak shape. A quartz tube reactor in the second micro-reactor was charged with the catalyst (15 mg). The catalyst amount used was the minimum amount needed for the characterization of spent catalysts after the experiments. For the non-catalytic experiments, only quartz wool was filled in the second micro-reactor. The temperature of the first micro-reactor was set at 300 °C because this is adequate to rapidly evaporate both iPrP and IPP 20 . The temperature of the second micro-reactor was set to 350 °C because it has previously been reported that n-propylphenol is effectively dealkylated using ZSM-5 at this temperature 37 . When the conditions were stabilised, the sample holder was dropped into the heating zone of the first micro-reactor, where iPrP and IPP were rapidly vaporised and carried into the second micro-reactor. The products generated from the TR were directly introduced into GC/MS with a separation column (UA + -1) and analysed without a product recovery process.

Synthesis and characterisation of the
IPP conversion using Ni/Y catalysts with different Ni loadings. The same amounts of IPP and synthesised catalysts were filled into the TR. The second micro-reactor was heated to 400 °C under 48 vol% H 2 /He flow (H 2 : 50 mL/min + He: 54 mL/min) for 90 min in order to reduce NiO/Y to Ni/Y. Then, the carrier gas was switched to 10 vol% H 2 /He flow (H 2 : 10 mL/min + He: 94 mL/min) or left uncharged, and the temperatures of the first and second micro-reactors were set at 300 and 350 °C, respectively. Then, IPP was supplied to the first micro-reactor. The remaining procedure is the same as the described in section (i).
Repeated IPP conversion using the 0.3Ni/Y catalyst. The IPP conversion using 0.3Ni/Y was repeated 10 times without a catalyst replacement. The procedures for the catalyst pretreatment, sample injection, and product analysis by GC/MS were the same as in the previous sections. Catalyst reduction between each run was avoided.

Calculations
The conversion rate of iPrP and IPP (C iPrP/IPP [%]), and the phenol selectivity (S [%]) were defined as follows: where A iPrP , A IPP , A phenol , A arom , and A all are the peak areas of the chromatograms of iPrP, IPP, phenol, all products containing aromatic rings, and all products standardized by the input sample weight, respectively.