Three biomass materials were employed in this study, including poplar wood, pine wood and corn stalk. Poplar wood was used as the major feedstock for pyrolysis experiments. Prior to experiments, these biomass materials were crushed, then particles in size of 0.10–0.20 mm was sieved and dried at 105°C for 24 h, stored in a desiccator for pretreatment and experiments. The component and elemental composition results of these materials on the dry basis are given in the (Table S1).

Poplar wood and other biomass materials were pretreated to impregnate K₃PO₄ (or K₂HPO₄, KH₂PO₄) by using the incipient wetness impregnation method. The typical procedure was as follows. K₃PO₄ solutions in different concentrations were obtained using certain amounts of K₃PO₄ and 50 mL deionized water. Afterwards, poplar wood in certain amounts were soaked in these K₃PO₄ solutions, and treated with ultrasonic stirring for 12 h. Finally, the samples were dried at 105°C for 4 h and kept in a desiccator for experiments. Five pretreated poplar wood samples were prepared with different K₃PO₄ contents of 1, 5, 8, 10, and 15 wt%, respectively. K₂HPO₄ and KH₂PO₄ impregnated samples were prepared in the same way.

Catalytic pyrolysis experiments were conducted using a lab-scale pyrolysis setup, which is shown in Figure 1. The details of the experimental setup can be found in our previous studies (Lu et al., 2016; Ye et al., 2017). For each experiment, the amount of raw biomass was 3.00 g, while the quantity of the pretreated biomass varied to ensure the pure biomass quantity of 3.00 g. For example, 3.24 g biomass was used for the biomass impregnated with 8 wt% K₃PO₄. The feedstock was stored in the container, and then fed into the quartz pyrolysis reactor that was vertically placed in a heating furnace. Three different carrier gases were employed for experiments, i.e., nitrogen, hydrogen (6.00%, argon as shielding gas), and mixed gas (4.00% H₂, 17.01% CO, 14.95% CO₂, 6.06% CH₄, helium as shielding gas), with the flow rate of 100 mL/min. The feeding of the feedstock lasted for 30 min with a strictly controlled rate, to ensure the temperature change in the reactor within ±3°C. The pyrolysis temperature was in the range of 400–700°C. A certain amount of quartz wool was placed in the quartz reactor to support the materials, preventing solid particles from falling into the condensation unit. The pyrolysis vapors were condensed in the condensation unit that was cooled by an ice and water mixture. The non-condensable gas was collected by a gasbag. In the end of the experiment, the pyrolysis quartz was cooled to room temperature within the carrier gas. The liquid and solid products were collected and weighed, and the quantity of the gas product was determined by difference.

The liquid products from different pyrolysis experiments were not always homogeneous. Hence, after each experiment, a certain amount of ethanol was added into the condenser to mix with the pyrolytic liquid before taking out of the condenser, to obtain a homogeneous liquid for further analysis. The water content was determined by the Karl-Fisher method, and thus, the water content of the original pyrolytic liquid could be calculated. In addition, the chemical composition of the liquid product was analyzed by GC/MS (Perkin Elmer, Clarus 560). GC separation was accomplished using an Elite-35MS capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness) with a split ratio of 1:80. Helium (99.999%) was used as the carrier gas (1 mL/min). The GC oven was set at 40°C (stay for 2 min), then heated to 280°C with a heating rate of 15°C/min, followed by maintaining at 280°C for 2 min. The GC injector temperature was kept at 300°C. The detected products were determined based on NIST, Wiley Library and previous studies (Lu et al., 2013; Zhang et al., 2015b). Moreover, since phenol was the target product of this study, its actual yield was quantitatively determined via an external calibration method. Different amounts of pure phenol were injected into GC/MS to record the peak area values, so as to obtain a calibration curve in terms of phenol quantity and its chromatographic peak area. Based on this calibration line and the pyrolytic liquid yields, the actual yields of phenol under different pyrolysis conditions could be calculated.

The pyrolytic solid residues were the mixture of chars and K₃PO₄, which could be used to prepare ACs, because K₃PO₄ is a well-known chemical activation agent for AC preparation. Therefore, ACs were prepared via typical CO₂ activation method. Thirty gram pyrolytic solid residue was heated from room temperature to 800^oC with the heating rate of 20°C/min in nitrogen atmosphere with the flow rate of 500 mL/min, and then CO₂ (500 mL/min) was used to replace N₂ for activation at 800°C for 1 h, followed with cooling down to room temperature in nitrogen. The activated sample was washed at 60°C to neutral and dried in an oven for 24 h, to obtain the final AC. In addition, raw poplar wood and pretreated poplar wood with 8 wt% K₃PO₄ were also employed for preparation of ACs via CO₂ activation with the same method.

The textural properties of the ACs were analyzed by the Autosorb-iQ-MP physisorption analyzer. The surface areas were determined using the Barrett–Emmett–Teller (BET) method. The pore volumes and pore size distribution were determined by the Barrett–Joyner–Halenda (BJH) method from the adsorption branch of the isotherms.

Based on the above results, possible mechanism for phenol production can be deduced. Due to the synergistic effects of K₃PO₄ and hydrogen, phenol was obtained in the catalytic fast pyrolysis process. Previous studies have analyzed the catalytic effect and mechanism of K₃PO₄ in inert atmosphere (Lu et al., 2013; Zhang et al., 2015b). Fast pyrolysis of lignin mainly follows the free radical mechanism, during which a large amount of radical species are produced from lignin decomposition (Demirbaş, 2000; Kim et al., 2014). Generally, these radical species can combine with H radicals to form stable phenolic compounds, also can undergo radical-coupling to form large molecule oligomers. Under non-catalytic conditions, very limited H radicals can be released from the formation of coke and oligomers. Therefore, due to the lack of H radicals, it is difficult for lignin-derived radicals to form phenolics. Whereas, the addition of K₃PO₄ significantly altered the product distribution. Hence it is reasonable to deduce that K₃PO₄ would catalyze the biomass pyrolysis process to generate hydrogen donors to stabilize the lignin-derived radicals (Lu et al., 2013; Zhang et al., 2015b). Previous works reported that holocellulose could act as hydrogen donors (Omori et al., 1998), and alkali metals had promoting effect on heterolytic cleavage reactions to help create more hydrogen radicals for stabilization of lignin-derived radicals (Julien et al., 1991; Yu et al., 2014). Therefore, K₃PO₄ could enhance the formation of hydrogen radicals during the pyrolysis process. As a result, yield of mixed phenolics was promoted, whereas the selectivity for phenol was still poor. Hence, it can be inferred that the presence of hydrogen would help to eliminate the substitutes of phenolics to obtain phenol as the predominant single product. In this study, discussion will be mainly concentrated on explaining the possible mechanism of substitute removal from phenolic compounds in the catalytic process in hydrogen atmosphere.

Substitutes of phenolics are usually linked via C-O and C-C bonds to the aromatic ring of phenolics. Cleavage and removal of these substitutes are usually difficult and requires overcoming high energy barriers in inert atmosphere. Take 2-methoxy-4-methyl-phenol as a model phenolic compound, density functional theoretical (DFT) calculations based on Gaussian 09 software were performed to compare the substitute removal in the presence or absence of H radicals. Calculation details can be found in the Supplementary Material, and results of the pyrolytic pathways with or without H radicals are given in Figure 3. According to Figure 3, for removal of methoxyl group, the energy barrier of demethoxylation reaction without H radical is as high as 419.6 kJ/mol (Path 1), which means that this reaction is difficult to take place during the real pyrolysis process. Whereas, the energy barrier of demethoxylation reaction is significantly decreased to a very low value of 31.6 kJ/mol in the presence of H radical (Path 2), suggesting the reaction is very easy to occur. Similarly, removal of methyl group from 2-methoxy-4-methyl-phenol needs to overcome much lower energy barrier with H radical (Path 4) than without H radical (Path 3). The calculation results confirmed the viewpoint that the presence of H radicals would accelerate the removal of substitutes on the aromatic rings of phenolic compounds (Asmadi et al., 2011; Liu et al., 2014b). Therefore, combining above calculation results and previous studies, possible formation mechanism of phenol can be briefly explained. During the catalytic fast pyrolysis process, although K₃PO₄ was able to catalyze the decomposition reactions to generate hydrogen donors for stabilization of lignin-derived radicals (Lu et al., 2013; Zhang et al., 2015b), the radical amount was far from sufficient to eliminate the substitutes of phenolics to form phenol in the inert atmosphere. The utilization of hydrogen atmosphere in the catalytic process provided excessive hydrogen radicals that could be used to facilitate the elimination of substitutes on phenolics to form phenol. It should be stated that, as aromatic ring is stable, hydrogen radicals preferentially react with the substitutes of phenolics rather than aromatic rings, to produce phenol.

Considering that pyrolytic solid residues contained K₃PO₄, conventional utilization as normal chars was infeasible before removing K₃PO₄. Whereas, K₃PO₄ is a good fertilizer for soil in agricultural application. But given the limitation of experimental conditions, test of pyrolytic solid residues for soil application cannot be achieved in this study. In addition, K₃PO₄ is also a well-known chemical activator for preparation of AC, which has already been applied in commercial scales, Therefore, activation experiments were further conducted by using the pyrolytic solid residue obtained from catalytic fast pyrolysis of poplar wood with 8 wt% K₃PO₄ at 550°C in hydrogen atmosphere, to confirm whether the solid residue was suitable for AC preparation. In addition, raw poplar wood and poplar wood impregnated with 8 wt% K₃PO₄ were also subjected to CO₂ activation for AC preparation for comparison.

The textural properties of the ACs prepared from three different feedstocks are listed in Table 5. The specific surface area of the pyrolytic solid residue was only 36 m²/g. After CO₂ activation, the specific surface area was as high as 1,605 m²/g, much higher than that of the AC from raw poplar wood (518 m²/g) and also a little higher than that of the AC from K₃PO₄-impregnated poplar wood (1,481 m²/g). Such results clearly indicated that pyrolytic solid residue was suitable for AC preparation, and agreed well with previous studies that activation of biomass with combined K₃PO₄ and CO₂ would produce ACs with high surface area. The mechanism of the combined activation process has been clearly explained in previous studies (Laine and Calafat, 1991; Laine and Yunes, 1992). When heated in CO₂ atmosphere, the interaction between potassium phosphate and the carbonaceous material would result in the formation of a potassium carbonate active center, while the remained free phosphate would compose a polymer net. The polyphosphates were supposed to be prone to form a “cover” which had good capability of keeping CO₂ erosion from the surface of carbonaceous material to avoid excessive burn-off. Later, these phosphates were possible to locally interact with the activated carbon reactive sites, also might be along the edges of the twisted aromatic sheet linearly to form the slit-like micropores. In addition, this interaction allowed the aromatic sheet to be used for progressive CO₂ attacks to form mesopores.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.