Catalytic Conversion of Biomass for Aromatics Over HZSM-5 Modified by Dawson-Type Phosphotungstic Acid

Biomass

Catalyst

The modified catalyst was prepared by testing dipping method. The HZSM-5 powder (particle size of 5.0 μm) was calcined at 550 °C for 2 h to remove possible template agent and some moisture, and then immersed in a certain amount of phosphotungstic acid solution. Stirred at 80 °C for 4 h, and transferred to a drying oven and dried at 110 °C for 4h, and then roasted at 550 °C for 2h. Finally, the modified catalyst was obtained and three kinds of catalysts with different loads were prepared, including 5%PW/HZSM-5, 10%PW/HZSM-5, and 20%PW/HZSM-5.

Experiments

The TGA of rapeseed shell without and with catalysts was conducted in a thermo-gravimetric analyzer (Mettler Toledo TGA 1, Swiss). High-purity N₂ (99.999%) was used as carrier gas at a flow rate of 50 mL/min. The samples of biomass, catalyst, and biomass mixed with catalyst were placed in Al₂O₃ crucibles and pyrolyzed by duplicate at the heating rates of 10 °C/min, 20 °C/min, and 40 °C/min. In order to compare the catalytic performance of different catalysts, the same heating rate of 20 °C/min was adopted. Blank runs at the respective heating rates were performed, and the results were used to subtract from the sample runs to remove any instrument artifact. During the experiments, TG and DTG curves were recorded synchronously.

Kinetic Analysis

where m₀ is the initial mass of biomass [mg], m is the actual mass [mg], and m_z is the final mass [mg].

Then, the initial and final stages of biomass pyrolysis were removed since the mechanism of both stages is different from the pyrolysis. And the derivative conversion data, which should be used and analyzed in the kinetic calculation, was obtained from the numerical differentiation of the conversion data (dα/dT). It should be noted that the derivative conversion data usually contains data noises, and consequently, it tends to numerically unstable. Hence, to reduce the impact of data noises on the further analyses, a smoothing operation is needed. Before the calculation of E and A, the n should be given according to the Kissinger method. For a case of nth order reaction, n could be calculated by Kissinger index of shape equation (\( n=1.26\sqrt{s} \)) [23]. The shape index s is defined as the absolute value of the ratio of the slope of tangents to the curve at inflection points on the derivative conversion data.

when T is temperature [K], R is constant [8.314 × 10⁻³ kJ·(K mol)⁻¹], and the heating rate β is a constant. 2RT/E is far less than 1 for biomass pyrolysis, and \( -\mathit{\ln}\left[\frac{AR}{\beta E}\left(1-\frac{2 RT}{E}\right)\right] \) can be regarded as a constant. Hence, interpolating \( -\ln \left[\frac{-\ln \left(1-x\right)}{T^2}\right] \) to \( \frac{1}{T} \) when n = 1, and interpolating \( -\ln \left[\frac{1-{\left(1-x\right)}^{1-n}}{T^2\left(1-n\right)}\right] \) to \( \frac{1}{T} \) when n ≠ 1. The least square method was employed to make linear fitting of the data. After the correct n was determined through the Kissinger method, a straight with high linear correlation can be obtained, whose gradient is E/R and intercept is \( -\mathit{\ln}\left[\frac{AR}{\beta E}\left(1-\frac{2 RT}{E}\right)\right] \). Finally, the values of E and A can be calculated.

Experiments

The catalytic pyrolysis of rapeseed shell with catalysts was performed in a vacuum pyrolysis system, as shown in Fig. 1. The pyrolysis reactor was a fixed-bed type heated by an electric heating sleeve. A blind tube was installed in the center to help control the reaction temperature. The gas products formed by catalytic pyrolysis were pumped by vacuum pump and filtered into a cold bath. The medium of the cold bath was ethylene glycol, and the cooling temperature was controlled at around − 10 °C. The condensable gas liquefied to form liquid products. The non-condensable gas was gone under the suction of the vacuum pump. Simultaneously, the non-condensable gas was collected by the gas sampling device. The moisture content of liquid product was determined by Karl-Fischer titration (ASTM E203-08), and the ultimate analysis was carried out in an elemental analyzer (FLASH 1112A, Italy).

Open image in new window

Then, the obtained liquid products were weighed, extracted, and separated, and the organic phase was obtained. The dichloromethane (CH₂Cl₂) reagent was used to for extraction and separation. The organic products were detected and analyzed by FT/IR, GC/MS, and ¹H/¹³C NMR, while the gas products were analyzed by GC. During the experiments, ground reactants were prepared by evenly mixing the biomass and catalysts with a biomass to catalyst ratio of 5 which was the intermediate value based on above literatures [26, 27] and tested in the pre-experiments. Besides, the reaction temperature of 500 °C, heating rate of 20 °C/min and system pressure of 5 kPa were employed, which was proven to be better in previous studies [28, 29]. The organic phases obtained by using 0%PW/HZSM-5, 5%PW/HZSM-5, 10%PW/HZSM-5, and 20%PW/HZSM-5 were respectively labeled as OP-I, OP-II, OP-III, and OP-IV. After the experiments, the spent catalysts were separated to subject to the TGA analysis. In addition, the spent catalyst with the optimal performance was regenerated through 700 °C high-temperature roasting (12 h) and the regenerated catalyst was subjected to the BET analysis to evaluate the activity recovery.

Gaseous Product Analysis

The gas products were analyzed by GC (Agilent 7890A, USA) with carrier gas of high purity argon (Ar), gas flow rate of 45 mL/min, injection volume of 20 mL, column temperature of 55 °C, TCD detector of 60 °C, and bridge current of 60 mA. The main gas products of CO, CO₂, CH₄, and H₂ were determined respectively.

Organic Phase Analysis

The chemical functional groups of organic phases were characterized by FT/IR (Thermo Nicolet iS5, Swiss). A certain amount of potassium bromide (KBr) was taken and pressed to obtain the tablet, and about 0.1 mL of organic phase was added to the tablet. Then, the tablet was measured, and the wavenumber scan range was 4000–400 cm⁻¹ with a resolution of 0.1 cm⁻¹.

The molecular compositions of organic phases were detected by GC/MS (Thermo Trace DSQ II, USA). Conditions: Carrier gas of high purity helium (He), gas flow rate of 1 mL/min, temperature at injection port of 250 °C, sample injection volume of 1 μL. Ion source temperature of 230 °C, MS transfer line temperature of 250 °C, ionization mode of EI, ionization energy of 70 eV, mass scan range of 30–500 m/z, and scan time of 1 s. The column temperature was maintained at 30 °C for 2 min, then increased to 100 °C at 15 °C/min and then to 250 °C at 10 °C/min and kept 250 °C for 3 min. Solvent (CH₂Cl₂) delay time was 3 min. The identification of the peaks in the chromatogram was based on the comparison with the standard spectra of compounds in the NIST library. The contents of the products were also calculated, which corresponded to the respective relative peak area among all of the detected peaks.

Furthermore, the chemical compositions of organic phases were characterized and confirmed by NMR (Bruker 600M, Germany), a 5-mm tube was used, 80 mg of sample was taken and dissolved in 300 μl deuterium acetone reagent, and testing temperature was 298 °C. The ¹H spectrum: Scanning width was 6000 Hz, sampling points were 66,000, scanning times were 16, and delay time was 1.0 s. The ¹³C spectrum conditions: Scan width 36,000 Hz, sampling points were 66,000, scanning times were 512, delay time was 2.0 s, and using reverse-gated de-duplication method to eliminate NOE effect. A semi-quantitative analysis was made to compare the distribution of aliphatic hydrogen/carbon and aromatic hydrogen/carbon in different bio-oils based on the same amount of solvent (CH₂Cl₂).

Spent and Regenerated Catalyst Evaluation

The TGA was also used to determine the coke deposited on the spent catalysts. The air was used as the carrier gas, flow rate was 50 mL/min, and heat rate was 10 °C/min from 40 to 800 °C. The TG and DTG curves were recorded synchronously. In addition, the BET analysis was carried out for the regenerated catalyst and the conditions were also the same as above.

Thermogravimetric Analysis

The TG and DTG curves of rapeseed shell pyrolysis without catalyst are shown in Fig. 2a, b. The main pyrolysis transformation stage (TG curve) of biomass shifted to the low-temperature side with the increase of heating rate, and the corresponding DTG curve showed the same shift law and the peak value gradually increased. Generally speaking, heating rate has two positive and negative effects on biomass pyrolysis: With the increase of heating rate, the response time of biomass particles to the temperature required for pyrolysis reaction becomes shorter, which is beneficial to pyrolysis, but the reaction time becomes shorter and the reaction degree decreases. Besides, thermal hysteresis will occur when the temperature difference between inside and outside the particle increases, and the pyrolysis products in the outer layer of the particle will not have time to diffuse, which hinders the internal pyrolysis process [34]. Therefore, the migration direction of TG and DTG curves depends on the combined influence of two aspects. Maschio et al. [35] studied the effect of particle size on biomass pyrolysis characteristics, and they found that the pyrolysis process of biomass particles with particle size less than 1 mm was mainly controlled by intrinsic dynamic rate, and the effect of heat and mass transfer in the particles could be neglected, and the pyrolysis process was controlled by heat and mass transfer and chemical dynamic rate when the particle size was larger and the heating rate was higher. The particle size of biomass used in this study was 0.10–0.15 mm, so there was no heat and mass transfer hysteresis effect, and the pyrolysis process shifted to the low-temperature side. At the same time, it can be seen from Fig. 2a, b that the deviation of conversion curve was not directly proportional to the increase of heating rate, which further indicated that high heating rate did increase the effect of heat and mass transfer on pyrolysis process.

Open image in new window

The TG and DTG curves of rapeseed shell pyrolysis with catalyst (0%PW/HZSM-5) are shown in Fig. 2c, d. They show that when 0%PW/HZSM-5 was added to in the catalytic pyrolysis of biomass, the effect of heating rate on catalytic conversion process was similar with the pyrolysis process. The difference is that the addition of zeolite catalyst can reduce the pyrolysis temperature of biomass and the temperature corresponding to the maximum weight-loss rate. HZSM-5 is a typical acidic zeolite catalyst with special pore structure and acid distribution. Hence, there are two aspects of the influence of zeolite catalysts on biomass pyrolysis: The strong acidity of the catalyst can easily decompose the compounds in biomass and advance the pyrolysis conversion process [36], and the larger specific surface area of zeolite has stronger adsorption on pyrolysis products, and the special pore structure has shape-selective catalysis on pyrolysis products, these two factors can slow down the release of products [37]. After adding 0%PW/HZSM-5 zeolite alone, the effect of acidity on the pyrolysis process was more obvious, and the TG and DTG curves shifted to the side of low temperature. Simultaneously, the strong acidity catalysis accelerated the conversion process and caused the maximum weight-loss rate to increase further. When PW-modified zeolites with different loadings were added to the conversion process, the specific TG and DTG curves are shown in Fig. 2e, f. With the increase of PW loading, the acidity distribution of zeolite was adjusted (mainly reflected in the distribution of B acid and L acid). This adjustment could not be simply evaluated as the increase or decrease of acidity but played a role in increasing or decreasing the selectivity of a certain reaction. At this time, the effect of catalyst acidity on biomass pyrolysis characteristics should be much smaller than that of specific surface area and pore structure. PW modification increased the specific surface area of zeolite from 66.031 to 245.733 m²/g and the pore volume from 0.107 to 0.220 cm³/g (in which the micropore volume increased from 0.025 to 0.117 cm³/g). This obvious change enhanced the absorption and catalysis of pyrolysis products and decreased the volatilization and precipitation rate of products. Therefore, the conversion process was slightly delayed compared with 0%PW/HZSM-5 catalysis, and the TG and DTG curves shifted back to the high-temperature side to some extent. At the same time, the maximum weight-loss rate was reduced due to the inhibition of product diffusion.

Kinetics

Product Distribution

FT/IR Analysis

The FT/IR technique was employed to investigate the chemical structure of the obtained organic phases. The FT/IR spectrums of the organic phases and the identified band positions, functional group absorbance type, and corresponding chemical species are summarized in Supplementary Material. The functional groups of each oil phase were similar, but there are still some differences. The first broad band around 3380 cm⁻¹ corresponded to the O–H stretching vibration mainly caused by phenolic and alcoholic functional groups, and another band indicating O–H bending vibration was all detected in obtained organic phases, which further showed the inevitable existence of alcohols. Differently, when PW was introduced, the intensity of the band decreased slightly, whose change rule was more consistent with the change of oxygen content, indicating that the hydroxyl functional groups also showed the law of decreasing first and then increasing. Then, the bands between 2864 and 2850cm⁻¹ caused by the stretching vibration of the saturated alkyl hydrogen C–H bonds, simultaneously, accompanied by the bending vibration (bands between 1470 and 1350cm⁻¹), indicating that there are many –CH₃ functional groups in all organic phases. In addition, an absorption band around 3056 cm⁻¹ was examined in OP-III, which was caused by =C–H stretching, indicating that the alkenyl functional group had a more existence. Thirdly, an obvious band around 1700 cm⁻¹ was caused by C=O stretching vibration, which showed the existence of carbonyl groups from ketones or carboxylic acids, so the stability of organic phase needed further improvement. Fourthly, the vibration absorption bands within the range of 1600–1500cm⁻¹ were caused by the skeleton structure of aromatic C=C bonds; meanwhile, the absorption bands located within the range of 870–625cm⁻¹ was caused by the bending vibration of aromatic =C–H hydrogen bonds, and interestingly, the C=C and =C–H bond types of benzene rings in OP-II, III were relatively few, but the band intensities were significantly enhanced (especially for OP-III), indicating that the benzene ring structures in organic phases may be different, which was mainly reflected in the monocyclic or polycyclic structure. It can be inferred that there are more aromatic compounds with monocyclic structure in OP-III. Last point, the absorption bands between 1300 and 1000cm⁻¹ were mainly caused by the stretching vibration of C–O bond in the alkyl aryl ethers, alcohols, esters, or aliphatic ethers, and generally, the C–O bond stretching vibration of bio-oil after catalytic upgrading is mainly caused by methoxy.

GC/MS Analysis

Then, organic phases were analyzed by GC/MS technique, and more than 75% of hydrocarbons were detected in all organic phases, so hydrocarbons were the main products, and aromatic hydrocarbons were the main components of hydrocarbon products, which became the most important factor affecting the physicochemical properties of organic phase. Therefore, the detected aromatic hydrocarbons were summarized in Supplementary Material. Generally, aromatic hydrocarbons can be classified as MAH and PAH, in order to better analyze the hydrocarbon composition of organic phase. MAH is further divided into MAH-I (contains only one benzene ring without any other ring structure) and MAH-II (contains one benzene ring with one or more hydrogenated benzene ring) in this study, from the perspective of fuel grade and environmental protection; MAH-II and PAH should belong to undesirable products. The change of catalyst made the selection of products inconsistent, but some products belonged to similar compounds, and there are 14 kinds of MAH-I compounds, five kinds of MAH-II compounds and 14 kinds of PAH compounds in all detected aromatics. Vinyl-benzene as a typical high content (6.52%) of MAH-I was only obtained in OP-III, which verified the FT/IR results of =C–H stretching peak around 3056 cm⁻¹, and from this point, the quantitative analysis of organic phase chemical constituents using GC/MS technique compromised with the FT/IR qualitative results. The critical diameter (width), maximum diameter (length), and kinetic diameter of the oxygenates, olefins, and aromatics were reported in literature [47], and it was found that benzene, toluene, naphthalene, indene, ethylbenzene, and p-xylene are sufficiently small to diffuse into the zeolite micro-pores, while the larger aromatic products are most likely formed in the particle gaps.

In order to make the difference of hydrocarbon composition in organic phase clearer, the data of hydrocarbon composition were analyzed statistically, and the results are shown in Fig. 3. As can be seen in Fig. 3, the content of LAH in organic products was low, and only a small amount of LAH was detected in OP-III and IV, further indicating that the catalytic performance of zeolite catalyst was mainly reflected in aromatization. Specifically, the undesired PAH presented a trend of decreasing first and then increasing, while the MAH-II basically showed a trend of gradual increase, and the variation of desired MAH-I was just opposite to that of PAH. The explanation of change laws should be divided into two parts: (a) from OP-I to OP-III, actually, the reduction of PAH contributed to the increase of MAH, which due to the enhancement of absorbance performance (including specific surface area and pore volume) and the increase of B acid and decrease of L acid (especially for L₁₆₁₆), on the one hand, the conversion process was more complete, and on the other hand, the aromatization performance was obviously reduced; (b) from OP-III to OP-IV, despite that the weakening of acidity might be helpful to inhibit the aromatization, the increase of specific surface area and pore volume made the catalytic reaction more sufficient. Therefore, the combined effect of the two factors made the aromatization performance not be inhibited, but obviously strengthened, which led to the increase of the undesired MAH-II and PAH. Another notable phenomenon is that the increase of MAH-II essentially meant the increase of hydro-saturated benzene ring, while the slight decrease of (H/C)_eff indicated the reduction of hydrogen retention, so the increasing PW load could promote hydrogenation reaction (or hydrogen transfer reaction) to a certain extent.

Open image in new window

In the composition of MAH-I, BTEX is the basic chemical raw materials in great demand and can be blended into gasoline with higher octane number. Hence, it is necessary to investigate the selectivity of PW/HZSM-5 to BTEX individually, the total content of BTEX was counted, and the proportion of BTEX in each sample was calculated, as shown in Fig. 4. As can be seen from Fig. 4 that only 6.15% in the OP-I, and more than half (52.37%) in the OP-III, and the OP-II, IV accounted for 20.16% and 21.32% respectively. As discussed above, the increase of BTEX selectivity mainly attributed to the release of more micro-pores after introducing PW modification; of course, the promotion of aromatization should also be considered. Then, the decrease of BTEX selectivity after 20%PW modification owed to the increase in pore structures, and although the increasing micro-pores might further increase the BTEX production, the increasing meso-pores or macro-pores could result in a great increase in the possibility of product polymerization to generate PAH [48].

Open image in new window

Furthermore, the oxygenate compounds detected in the organic phase were classified and analyzed as shown in Fig. 5. As can be seen from Fig. 5, the oxygenate compounds mainly involved in four categories: ketones, phenols, alcohols and others (mainly ethers), which were much less than those obtained by other catalytic methods, such as Stefanidis et al. [43] examined 10 types of oxygenated organics during the in situ upgrading of biomass vapors using various modified ZSM-5 on a fixed bed reactor, and Asadieraghi et al. [49] identified nine kinds of oxygenate compounds by employing meso-HZSM-5 and Ga/meso-HZSM-5 in a multi-zone fixed-bed reactor. Firstly, the ketones decreased first and then increased with the increase of PW loading and the content of ketones in OP-I was the highest (4.97%) and that in OP-III was the lowest (2.47%). The decreasing amplitude reached 50%, but the change was relatively small, because the content was low so that the incompleteness of the transformation was relatively high. Secondly, phenols showed a significant decrease trend, and even no phenols were detected in OP-III, IV. This was due to that PW modification enhanced catalytic alkylation and dehydroxylation, and on the other hand, because hydrogen transfer or hydrogenation enhanced the conversion of some phenols to alcohols, which also explained the increase of alcohols. Thirdly, the decrease of alcohols and the change of other organics were mainly attributed to incomplete deoxygenation and alkylation reactions so that some methoxy-containing compounds (ethers) were produced. In addition, the oxygen atom distributions in the organic phases were also classified as shown in Fig. 6. As can be seen from Fig. 6, compounds containing two oxygen atoms were substantially reduced until they were undetectable, and some were not completely deoxygenated, increasing the number of compounds containing one oxygen atom. In terms of deoxygenation degree, 10%PW/HZSM-5 had reached the highest catalytic performance, and the increase of PW loading could not further enhance the deoxygenation capacity, and even rebounded slightly, which was basically consistent with the results of above elemental analysis.

Open image in new window

Open image in new window

¹H/¹³C NMR Analysis

The larger compounds in the organic phases cannot be validated and recognized in GC-MS, and this defect can be remedied by NMR technique through the functional groups’ confirmation. The ¹H NMR and ¹³C NMR experiments (C₃D₆O, 600 MHz) were conducted, and the specific results are shown in Fig. 7. Figure 7a shows the ¹H NMR spectra of organic phases, which provided the detailed information about organic components. According to the spectra, it was indicated that the main chemical shift (δ) regions for ¹H NMR were 0.5–3.0 ppm and 6.5–8.0 ppm. The former region mainly is related to aliphatic protons and the latter is attributed to aromatic protons [50]. Specifically, the region from 0.5 to 1.5 ppm corresponds to protons on aliphatic carbon atoms at least two bonds away from C=C or heteroatom, the region from 1.6 to 2.5 ppm corresponds to alkyl protons, and the peaks located in the range of 2.5–2.8ppm are caused by protons on a-position of aromatic rings. Besides, the weakened peaks in the 3.0–4.5 ppm and 4.5–6.0 ppm ranges respectively represented methoxyl protons and ether protons. Lastly, the peaks located in the range of 6.0–8.5ppm corresponded to the aromatic protons. To judge specific changes, the RP (%) of the main chemical shift region was calculated by integral normalization based on the proton of CH₂Cl₂ occurred (peak at 5.63 ppm) in each sample, and the results are also listed in Fig. 7a. The ratios of aliphatic proton to aromatic proton in OP-I, II, III, and IV were 3.0, 3.6, 8.3, and 2.0, respectively. This change confirmed the cracking effect of appropriate amount of PW, and the chemical shift for aliphatic proton of OP-III was mainly located in the 1.6–2.5 ppm, which further indicated that Dawson-type PW had strong alkylation performance. Besides, as shown in Fig. 7b, the main chemical shift regions of ¹³C NMR were 10–40 ppm and 120–150 ppm, mainly corresponding to aliphatic carbon and aromatic carbon respectively [51]. The peaks in the range of 22–29ppm were attributed to carbon branched methyl carbon. The peaks, in the range of 120–150ppm, were attributed to the aromatic carbon. Also, the RC (%) of the main chemical shift region was calculated according to the carbon of CH₂Cl₂ occurred (peak at 55 ppm) in each sample, and the calculation results are listed in Fig. 7b. The change rule of the RC was in accordance with the RP change rule. Generally, the change rules of organic groups demonstrated from NMR results were basically consistent with the discussion in GC/MS analysis.

Open image in new window

Spent and Regenerated Catalyst Analysis

The TG and DTG curves of the spent catalysts are shown in Fig. 8. Figure 8 shows that the total weight losses of 0%PW/HZSM-5(s), 5%PW/HZSM-5(s), 10%PW/HZSM-5(s), and 20%PW/HZSM-5(s) were 18.62%, 16.58%, 11.56%, and 14.81%, respectively. The weight loss processes could be roughly divided into three stages. The first stage below about 250 °C was caused by the release of water and light species, the second stage between 25 and 700 °C was attributed to the burning of deposited coke, and the third stage above about 700 °C belonged to the decomposition of heavy carbonaceous species and the slow stabilization process. To identify the characteristics of cokes more accurately, the second stage needed to be analyzed. Based on the DTG curves, the specific temperature ranges of main weight loss in the second stage for the four spent catalysts were 350–700 °C, 265–635 °C, 250–510 °C, and 300–650 °C respectively, which showed that PW modification could reduce the activation energy of the catalytic reaction, while the coking reaction also shifted to the low-temperature direction, and the shift degree varied with the amount of PW. For the corresponding coke contents, 12.13%, 11.27%, 5.25%, and 8.47% were respectively determined, which showed that the 10%PW modification not only obtained the higher conversion proficiency, but it could also enhance the catalyst stability. Generally, the in situ pyrolysis-catalysis method could produce two types of coke on the catalyst including thermal coke and catalytic coke, and the decomposition temperature of catalytic coke was relatively high [52], which could be verified through the acromion on the DTG curve. The thermal coke usually deposited on the catalyst surface and its formation was induced by thermal and weak acid sites (or inert sites), and the catalytic coke was formed in the inner pores during the aromatization and reformation [53]. In this study, 10%PW modification dramatically increased the B acid sites and the pore volume which could obviously decrease the coke contents, while 20%PW modification led to the decrease of B acid sites and further increase of specific surface area (the diffusion performance of products was deteriorated), and both factors finally resulted in the recovery of coke content.

Open image in new window

To evaluate the regeneration performance of the catalysts, the optimal catalyst 10%PW/HZSM-5 was subjected to the BET analysis after roasting regeneration. The specific surface area and pore volume of the fresh catalyst were 148.936 m²/g and 0.220 cm³/g, and those of the regenerated catalyst were 145.241 m²/g and 0.214 cm³/g. The properties recovered 97.52% and 97.27%, which showed that the catalyst had high regeneration activity, and it could be continuously regenerated and reused.