Introduction

Fossil fuels are used to produce energy worldwide, but they come with major problems such as global warming (Zhang and Liu 2021). To reduce the use of fossil fuels, biomass-based fuels offer a great potential alternative since biomass is one of the most promising renewable and sustainable energy resources (Houshfar et al. 2014; Singh and Dhepe 2018; Wang et al. 2021) While biomass waste has increased markedly due to the intensification of agricultural and industrial activities to serve a growing population, its disposal, utilization, and management practices are not efficiently employed (Tripathi et al. 2019). Biomass waste is usually discarded to degrade or to be burned in open areas, resulting in haze that harms the local environment (Madhu et al. 2018). Biomass waste-to-energy conversion is a promising solution for the above-mentioned challenges.

In the past decades, biomass conversion processes have been extensively studied for converting biomass into liquid biofuel via gasification (Fan et al. 2020), pyrolysis (Onsree et al.2018a, 2018b, 2019; Onsree and Tippayawong 2020a), and hydrothermal processing (Kantakanit et al. 2018). Among these processes, only hydrothermal liquefaction (HTL) can be efficiently applied to convert high moisture content or wet biomass materials generally having a moisture content of more than 50% w/w, such as fresh biowaste, food waste, algal biomass, and wastewater sludge to biocrude oil without a drying requirement. HTL is a thermochemical process in the presence of a solvent at moderate temperatures (200–374 °C) and high pressures, between 5 and 20 MPa (Ponnusamy et al. 2020). The process undergoes reactions similar to the formation of underground fossil fuels, but with far shorter reaction times, requiring merely a few minutes or hours, compared to the millions of years of the fossilization process (Dimitriadis and Bezergianni 2017; Gollakota et al. 2018). By using HTL, over two-thirds of the carbon content in biomass could be recovered as either biochar or biocrude oil, with expected yields of more than 40% w/w and heating values ranging from 30 to 40 MJ/kg (Tekin et al. 2014; Yang et al. 2019). One of the challenges in utilizing the HTL process is that a high-pressure reactor is required, making the process complicated and costly (Zhang et al. 2010).

Rather than simply maintaining a high pressure, materials used to build the reactor have to withstand extremely corrosive environments during the hydrothermal reactions (Zhang et al. 2010). In general, there are two types of HTL reactors: batch and continuous (Déniel et al. 2016a; Peterson et al. 2008). Construction of a continuous system presents further technical challenges, such as feedstock loading, in which feeding the slurry into the system at a certain concentration is a complex task (Déniel et al. 2016a). In other words, the system requires the development of methods for feeding raw biomass to the process inlet as well as for pumping highly viscous biocrude oil at the outlet. Elliott et al. (2015) noted that continuous HTL reactor technology has significant commercialization potential, but several challenges remain that need to be addressed before this can occur. To gain insight into the mechanism of HTL, experiments in a batch reactor are still necessary before progressing to commercial scales. Small batch reactors (10–1000 mL) are mostly used in laboratories, as operating conditions and feedstock loadings are easily controlled. For example, Jensen et al (2017) used a 10 mL micro-batch reactor to probe the HTL of lignocellulosic biomass. They obtained biocrude oil yields as high as 45% w/w. Xu and Lad (2008) employed a 14 mL bomb reactor to study the HTL of Jack pine sawdust and reported that the maximum liquid biocrude oil obtained without using catalysts was over 50% w/w at 300 °C and 30 min. Jindal and Jha (2016) used a 600 mL batch reactor to investigate the HTL of waste furniture sawdust and found that the biocrude oils had improved HHVs up to between 17 and 30 MJ/kg. Arun et al. (2021), who investigated the HTL of Prosopis juliflora biomass, found that the optimum biocrude oil yield was 22.3% w/w at 300 °C, 60 min, and a 1:20 biomass to solvent ratio. Wang et al (2019) carried out HTL of waste Tetra Pak in sub- and supercritical water in a micro-batch reactor (100 mm long and 7.5 mm in diameter). They found that the maximum yield of biocrude oil was about 36% w/w at 360 °C, 22 MPa, 30 min, with a feed concentration of 20% w/w, and their biocrude oil mainly contained ketones, phenolics, esters, and alcohols.

Tobacco is a major economic crop in many countries. In Thailand, its processing is mainly managed by the Thai Tobacco Authority with reported revenue in the range of US$1.65 billion annually. Prior to the current covid-19 pandemic period, its growth was about 2.4–4.2% a year (Thai Tobacco 2021). There are large amounts of wastes and residues generated from tobacco processing. Since they are usually considered worthless, the waste is either heaped on farmlands or burnt on-site for disposal. So far, only a few papers on converting tobacco residues to biocrude oil have been reported (Chumsawat and Tippayawong 2020; Khuenkaeo et al. 2020). Pyrolysis was used as the main conversion process in all these papers. Their results indicated that using pyrolysis, a high reaction temperature of about 600 °C was necessary to maximize the biocrude oil yield, compared to the HTL reaction which only needs temperatures between 280 and 340 °C. Moreover, published reports on HTL of agricultural residues, especially tobacco processing waste, remain scarce.

Therefore, the specific objective of this work was to investigate the thermal conversion of tobacco processing waste to biocrude oil by HTL in a multiple batch reactor. The operating conditions considered were reaction temperatures of 280, 310, and 340 °C, and residence times of 15, 30, and 45 min at a fixed biomass-to-solvent ratio of 1:3. The distribution and yields of liquid-, solid-, and gas-phase products were examined across those operating conditions. For each condition, at least three repeats were performed. The main components of the biocrude oil were also analyzed.

Materials and methods

Biomass sample and chemicals

Tobacco processing residues, provided by the Tobacco Authority of Thailand‘s Denchai redrying plant in Phrae, Thailand, were used as the biomass sample. Prior to the tests, all dirt and impurities were manually removed from the waste sample. It was visually inspected to ensure no contamination occurred. It was then ground to approximately 3–5 mm in diameter. The samples were dried in an oven at 105 °C for 24 h and cooled to room temperature in a controlled humidity chamber. The dried samples were stored in Ziplock bags prior to being used in any experiments. Deionized (DI) water was used as the solvent for the HTL reactions. Dichloromethane (DCM), acetone, and methanol with over 99% purity, purchased from RCI Labscan, were used in product extractions and chemical composition analysis processes.

HTL reactor setup

HTL reactions occur at high pressure (5–20 MPa) and moderate temperature (200–374 °C) (Ponnusamy et al. 2020). During the reactions, a mixture of solvent and biomass may produce corrosive substances inside the reactor (Zhang et al. 2010). In the present work, four custom-built identical batch container vessels were used. Each was designed with an internal volume of 200 mL for reaction pressures and temperatures up to 22 MPa and 400 °C, respectively. Each container vessel was made of high-quality stainless steel (SS-304), in which the container body was tightly sealed with a steel flange and six bolts, as shown in Fig. 1.

Fig. 1
figure 1

A multi-batch reactor setup

Full size image

The container vessels were heated by an external high-temperature muffle furnace (Dongguan LIYI Environmental Technology Co., Ltd., China), generating a maximum temperature of 900 °C. For safety, a rupture disk was installed to prevent equipment failure in the case of excessive internal pressure (25 MPa). For each container, the biomass sample was mixed with DI water at a fixed mass ratio of 1:3. All reaction vessels were tested at maximum temperature and pressure conditions: the batch reactor was filled with 20 ± 1 g of tobacco residues and 60 ± 2 g of DI water to simulate HTL reactions.

Monitoring the pressures and temperatures inside the batch containers during the reactions was complicated. In this work, a dummy batch container was deployed and installed inside the oven chamber. It was mounted and connected to a Huadong pressure gauge (< 1% full scale) and a digital Sensoheat K-type thermocouple (± 0.75% of reading). During each test, the dummy batch reactor was filled with a mixture of biomass and DI water similar to the tested samples. The pressure and temperature inside the dummy batch reactor were regularly measured and recorded as a function of time and considered representative for all batch containers. The uncertainty of the measurement and the results from other batches compared to the dummy were within 5%. The slurry mixture in the batch reactors was expected to be in a near-critical fluid state. When the temperature of the mixture was increased, the pressure increased monotonically, according to saturation. Therefore, in this study, only the temperature was controlled and varied, and the reaction time was considered to begin after the temperature of the dummy container reached the set point.

HTL procedure and product extraction

In our HTL reactor, the biomass sample and DI water were mixed and stirred vigorously to make sure that the mixture was homogeneous. Then, the container vessels were heated to a temperature (280, 310, and 340 °C) in an oven for a specific reaction time of 15, 30, and 45 min. Next, the container vessels were rapidly cooled to room temperature in a cold-water bath, immediately halting the reactions, similar to previous work (Aierzhati et al. 2019).

Figure 2 shows the product extraction. After the HTL reactions ended, the gas-phase product in each batch container was vented to the atmosphere at room conditions, and the remaining reaction mixture was carefully collected. In line with the previous reports (Caprariis et al. 2017; Jiang and Savage 2017; Valdez et al. 2012; Xiu et al. 2010; Zhou et al. 2010), the solid and liquid products were separated using filter paper under vacuum. The filtrate was subsequently extracted with an equal quantity of DCM, and the resulting suspension consisted of two layers: (i) DCM-soluble and (ii) water-soluble fractions (Valdez et al. 2012). The DCM solution was evaporated at 38 °C (Jiang and Savage 2017), while the water-soluble fraction was evaporated at 65 °C for 12 h to completely remove the water (Xiu et al. 2010). The organic fractions from the filtrate which remained were labeled as light biocrude oil (LBO) and water-soluble (WS) fractions, respectively. The solid product remaining on the filter paper was washed with acetone until the solvent became colorless, and then it was dried at 105 °C for 24 h to obtain the solid product (Zhou et al. 2010). After removing the acetone from the washing liquid in an evaporator at 60 °C, this fraction was quantified and labeled as the heavy biocrude oil (HBO) (Caprariis et al. 2017). The total liquid biocrude oil yield was a sum of LBO, HBO, and WS. With the total initial biomass loading mass (\(M\)), the yield of each product can be calculated (on a dry basis) from Eqs. 1 and 2:

$${y}_{i}=\frac{{m}_{i}}{M}\times 100$$
(1)
$$\sum {y}_{i}=100$$
(2)

where \(y\) and \(m\) are the yield and mass of each product, respectively, the subscript \(i\) refers to the liquid and solid products.

Fig. 2
figure 2

Schematic of the product extraction procedure

Full size image

Product analyses

The moisture content, ultimate composition, and higher heating value of the tobacco residues and solid products were analyzed according to the ASTM D3173, D5373, and D5865 standards. The ultimate analysis of the organic liquid products (HBO, LBO, and WS) was performed using a ThermoScientific CHNS/O analyzer (Flash 2000, Thermo Scientific, Italy), while the chemical composition was analyzed using a gas chromatograph-mass spectrometer (GC: Agilent 7890B and MS: Agilent MSD 5977B, USA) with a DB5-MS column (30 m/0.25 mm/0.50 μm) and an automated liquid sampler (Agilent 7693A). The HBO and LBO fractions were diluted with acetone, and the WS product was diluted with methanol. Approximately 1 μL of the sample mixture was injected into the column with a split ratio of 10:1. Helium was used as the carrier gas at a flow rate of 10 mL/min. The oven temperature was initially set to 60 °C with a hold time of 5 min; then, it was increased at a constant rate of 5 °C/min until it reached 280 °C, where it was held for 5 min. The injector and detector temperatures were set at 250 and 280 °C, respectively. The compounds were identified by comparison with the NIST mass spectral database.

The higher heating value (HHV) of the liquid products was approximated by Eq. 3 (Rajagopal et al. 2021):

$$HHV=0.338C+1.428H-0.179O$$
(3)

where \(C\), \(H\), and \(O\) are the carbon, hydrogen, and oxygen contents of each liquid product (in w/w). The energy recovery (\(ER\)) of the products was calculated from Eq. 4 (Mishra and Mohanty 2020):

$${ER}_{i}=\frac{{{y}_{i}\times HHV}_{i}}{{HHV}_{tobacco}}$$
(4)

where \({HHV}_{tobacco}\) is the higher heating value of the tobacco waste.

Results and discussion

Temperature and pressure evolution in the container vessel

Figure 3 shows the temperature and pressure evolution profile obtained from a representative HTL of tobacco residues and saturated water in a batch container. The x-axis represents the temperatures of the ingredients inside the container. The y-axis is the pressure inside the container at any temperature. As expected, the mixture of tobacco residue and DI water solvent experienced a near saturated state. But the thermodynamic state of the slurry mixture appeared to have slightly higher pressures for temperatures above 200 C°. For example, at 230 °C, the mixture was at 3.0 MPa, whereas pure water saturation at this temperature is about 2.7 MPa. The difference was clearer for temperatures above 260 °C, and an average difference of approximately 1.3 MPa was maintained. In the mixtures, this may have been due to the initiation of the thermal decomposition of the lignocellulosic biomass at temperatures of about 200 °C, which generated many different substances in the container (Tippayawong et al. 2019). Higher temperatures increased the thermal decomposition of biomass (Onsree and Tippayawong 2020b), leading to a more complicated mixture in the batch container. The resulting pressure subsequently rose according to the saturation of some of those new substances (Déniel et al. 2016b).

Fig. 3
figure 3

Temperature and pressure evolution inside an HTL batch container

Full size image

Distribution and yields of HTL products

Figure 4 shows the major products in this study were liquid-, solid-, and gas-phase, which were calculated on a dry basis. Both temperature and residence time were found to affect the distribution and yields of HTL products. Increasing the temperature from 280 to 310 °C increased the liquid product yields but lowered the solid and gas product yields. When the reaction temperature increased to 340 °C, liquid biocrude oil yields decreased markedly, leading to a considerable increase in gas formation, while the solid yields were only slightly changed. The negative effect of temperature on HTL liquid products was consistent with previous publications on different biomass resources, such as (Zhu et al. 2015; Yin et al. 2010; Sugano et al. 2008), which established that the reaction temperature range of 300–315 °C was suitable for efficient production of biocrude oil. The effect of temperature on the variability of liquid product yields was due to the complex reaction mechanisms of biomass composition in terms of cellulose, hemicellulose, and lignin (Peterson et al. 2008; Zhang 2010). The decomposition of these three components in DI water at high temperatures was reported to give different products (Toor et al. 2011). Hemicellulose and cellulose decompose at about 220 and 280 °C, respectively, while lignin decomposes over a wide range, 200–500 °C. Hemicellulose and cellulose are beneficial for liquid products, while a high lignin content leads to a high amount of solid products (Zhong and Wei 2004). The decomposition mechanism during the HTL reactions consisted mainly of the following three steps: depolymerization, decomposition, and recombination (Toor et al. 2011). At the initial stage, the biomass was decomposed and depolymerized to form small compounds. These compounds may rearrange through condensation, cyclization, and polymerization to form new compounds (Sun et al. 2010). In other words, an increase in temperature triggered the liquid yields. After reaching the maximum yield of the liquid product, a further increase in temperature inhibited the HTL of the biomass, promoted the decomposition of all the compounds into gaseous products and possibly allowed the repolymerization or condensation of the intermediates into chars (Akhtar and Amin 2011). The reaction temperature to maximize the liquid product yield has previously been found to be slightly above 300 °C (Yin et al. 2010). In this work, the maximum yield of the liquid products was obtained (52.1% w/w) at 310 °C, the optimal conversion temperature for biocrude oil production from HTL of tobacco processing residues.

Fig. 4
figure 4

Yields and distribution of a liquid-, b solid-, c gas-phase products from the HTL of tobacco residues

Full size image

The residence time was also observed to have a considerable effect on the distribution and yields of HTL products. In the present work, it was defined as the time which the biomass spent in the reactor after reaching the set point temperature. The times for heating up and cooling down the reactor were not included. From Fig. 4, at a reaction temperature of 280 °C, the maximum yield of liquid products (51.7% w/w) was obtained at a residence time of 30 min. However, when the temperature was raised to 310 °C, the maximum yield increased to 52.1% w/w at a residence time of only 15 min. This appeared to be the optimal condition for producing the highest yield of HTL liquid products from tobacco residues based on a combination of reaction temperature and time. At 340 °C, the liquid product yields decreased to 37% w/w.

In contrast, the gas-phase product yields increased slightly with an extended period of residence time. The highest gas product yield was found at 45 min and 340 °C. The solid yield was not significantly affected by the residence time. This indicated that the liquid products reached their maximum before decreasing at longer residence times, whereas the gas yields continuously increased until the saturation point was reached, similar to previous work (Jindal and Jha 2016). There appeared to be an interrelationship between the temperature and residence time needed to maximize the yield of the liquid products. A higher temperature required a lower residence time. However, too short a duration could lead to incomplete degradation and polymerization reactions, while too long a duration caused polymerization of intermediates (Cao et al. 2017). Under extended residence times, the lower yield of the liquid products could be due to the cracking reactions of liquid products to gas-phase products and the formation of char products by condensation (Karagöz et al. 2005). When considering the residence time, the yields of the liquid products in our study were comparable to those reported by Yang et al (2016) and Jena et al (2011).

In order to gain insight into the combined effects of temperature and residence time, Fig. 5 shows the proportions of HBO, LBO, and WS in the liquid products obtained at varying conditions of temperature and residence time. With an increase in temperature from 280 to 310 °C, the HBO fraction increased substantially from 18.9 to 29.6% w/w, whereas the LBO and WS fractions had slight changes with average values of approximately 9.7 and 12.8% w/w, respectively. Compared to a previous study (Rajagopal et al. 2021), the trend of increasing biocrude oil production was the same, except for the WS fraction, which was reported to decrease at a higher temperature. By increasing the residence time from 15 to 45 min, both biocrude oil fractions increased slightly. The WS fraction decreased to 8.7% w/w. When only liquid products were considered, tobacco processing residues hydrothermally decomposed to generate LBO and WS fractions at a low temperature and generate more of the HBO fraction at higher temperatures. An extended residence time did not increase either of the biocrude oil yields, but caused the depolymerization of the light WS components, lowering the liquid yields (Aierzhati et al. 2019; Mishra and Mohanty 2020).

Fig. 5
figure 5

HBO, LBO, and WS in the liquid products at a 280 °C for 15 min, as well as 310 °C for b 15 min and c 45 min

Full size image

Characterization of HTL products

Table 1 shows the ultimate and HHV analyses of the tobacco waste and HTL products and energy recovery of each product. Compared to the raw biomass material, both of the biocrude oils were rich in C (> 65% w/w) and H (> 7% w/w), as well as low in O (< 26% w/w). The solid product had a C content of up to 70% w/w and an O content of about 24% w/w. Atomic O/C and H/C ratios of these HTL products were observed to be 0.8–1.4 and 0.2–0.3, respectively, which were significantly lower than those of tobacco processing residues. The decreased O/C and H/C ratios of the HTL products were due to deoxygenation and decarboxylation of the biomass feedstock during the liquefaction process (Chen et al. 2019). This also resulted in an increased HHV of the HTL products, which was over 28 MJ/kg for the biocrude oils and about 26 MJ/kg for the solid products. The superior HHV indicated an energy recovery of almost 90% from the raw material. In this case, the HBO and LBO products accounted for more than 65% of the energy recovered. This intensification of energy density in the biofuels is an essential factor for sustainable energy conversion processes. Meanwhile, the WS fraction was found to have a high O content, 80% w/w, with only 11% w/w C and an HHV of only 1.4 MJ/kg HHV. This result was similar to that reported in previous work (Aierzhati et al. 2019).

Table 1 Properties of tobacco processing waste, HBO, LBO, WS, and solid products, as well as the energy recovery of each product
Full size table

The HBO, LBO, and WS products were analyzed by GC–MS. The peak values from this analysis do not represent the actual concentrations of compounds, they indicate the product distributions associated with the solvent (Khuenkaeo et al. 2020). Figure 6 depicts the main composition of each liquid product at the optimum reaction condition (310 °C and 15 min). Other compounds with low boiling points may have been present but were covered by solvent peaks or may have been lost during acetone evaporation. Based on the relative percentage of the peak areas, the HTL products mainly contain phenols, ketones, and nitrogenous compounds. The HBO product had the highest proportion of phenols at about 30%, while the LBO fraction showed the highest proportion in ketones at over 32%. Nitrogenous compounds were found to be the most abundant in the WS product at over 35%. The compounds of the liquid products found here were in agreement with those from previous research (Jindal and Jha 2016; Kim and Um 2020), where phenols, carboxylic acids, aromatic hydrocarbons, ketones, aldehydes, and nitrogenous ring structures were reported to be the main components of liquid products from HTL of lignocellulosic biomass. However, the produced HTL biocrude oil in the present work differed from the biocrude oils obtained from food waste and algae with high protein and carbohydrate contents (Chen et al. 2019; Aierzhati et al. 2019; Toor et al. 2013), in which their components consisted mainly of aromatics, phenolic derivatives, carboxylic acids, esters, and nitrogenous ring structures. It was likely that the compounds of HTL biocrude oil products were also influenced by the raw material compositions, apart from the HTL conditions. Attempts have been made to propose possible formation pathways of biocrude oils from HTL of lignocellulosic biomass (Yang et al. 2015). During HTL, polysaccharides, basic components of lignocellulose, are decomposed to phenols and ketones (Kumar et al. 2018), as shown in Fig. 7a. Proteins which could be present in the plant cells can be decomposed into low molecular weight compounds such as pyrroles, pyrazines, and amines. Shown in Fig. 7b for N-containing compounds, proteins are decomposed to amino acids, and then the amino acids further decompose either to carbonic acids and amines by decarboxylation or to ammonia and organic acids by deamination reaction (Peterson et al. 2008). Phenolic compounds are likely derived from the degradation of lignin, while the ketones are probably formed from further reactions of the sugars generated by the degradation of hemicellulose and cellulose (Barbier et al. 2012). Large amount of nitrogenous compounds observed in the WS product may be due to the fact that the tobacco processing residues are initially rich in N content (Khuenkaeo et al. 2020), compared to other types of lignocellulosic biomass (Tanyaket et al. 2020).

Fig. 6
figure 6

Main chemical compounds extracted from the liquid biocrude oil products

Full size image
Fig. 7
figure 7

Possible formation pathways of biocrude oils from a polysaccharides and b proteins as proposed by Yang et al. (2015). The dotted boxes refer to the compound products in our biocrude oil products

Full size image

Practical application of HTL as a clean technology

HTL has been shown to be a promising clean technology in converting organic materials into value added oils. The technology can be adopted to support Thailand’s waste management system and its goal to achieve zero waste by 2030, as well as the country’s bio-circular economic policy in reusing, recycling and utilizing waste and converting these resources into value added products. Further research and development in HTL will help propel the future demonstration of this clean technology in managing wet waste. Rather than discarding about 2500 tons of tobacco processing waste annually, the HTL technology may be adopted to convert this waste to biofuels and biochemicals. Apart from tobacco processing waste, the technology can also be applied to other wet organic wastes such as other agricultural residues, food waste, waste from fresh markets, and the organic fraction of municipal solid waste.

Conclusion

Hydrothermal liquefaction (HTL) can be used to convert tobacco processing waste to biomass-based biocrude oil, effectively. By using a batch reactor, the maximum biocrude oil yields over 52% w/w with about 90% of energy recovery at conditions of 310 °C, 15 min and 1:3 biomass to DI water ratio. The main components of the biocrude oils were phenols and ketones. These findings provide a positive outlook for utilizing wet agricultural waste for energy and chemical productions via HTL as a clean technology. Further studies on separation, purification and refining of highly valuable components or upgrading of the biocrude oils to biofuels should be considered and undertaken.