1. Introduction
The increase in energy consumption due to population growth and the dependence on fossil fuels have enlarged greenhouse gases emissions (GHG) with a major impact on environment and global warming [
1]. As a result, the use of renewable resources for sustainable energy production has been recently promoted [
2]. Lignocellulosic biomass, which includes agricultural and agroindustrial residues [
3], is considered as an interesting renewable resource since it has low cost, could be carbon neutral [
4], and its conversion implies low GHG emissions [
5]. Different processes have been proposed for the use of lignocellulosic biomass, such as pyrolysis [
6], gasification [
7], combustion [
8,
9], carbonization [
10] and liquefaction [
11]. A combination of processes has been proposed as a non-conventional energy technology to produce hydrogen (H
2) from biomass [
6,
12,
13]. H
2 has a high calorific value and can be used in fuel cells (FC), which convert chemical energy into power and heat [
14].
Colombia is the third Latin American country in biomass production [
15] and generates approximately 72 million tons of agricultural waste per year with a potential energy of at least 331,000 TJ/year [
16]. Sugarcane press-mud is a byproduct obtained from the clarification of sugarcane juice during the non-centrifugal sugar production [
17]. This residue is obtained with a yield of 3 to 5 wt% [
18], which represents about 1.36 Mton/year of sugarcane press-mud [
19]. Currently, this residue is used as a raw material for organic fertilizers [
17] or, more often, discarded in large quantities, generating pollution in sources of water. Sugarcane press-mud has been used to produce bioethanol through fermentation [
19]; nonetheless, it contains approximately 30 wt% lignocellulosic rich solid waste that is currently discarded [
19]. This solid waste will be hereinafter called sugarcane residual biomass and will be the focus of this study.
Among thermochemical processes, pyrolysis of biomass is a widely used technology to produce power or syngas in the absence of O
2 [
20,
21]. Moreover, it is the most studied process since it precedes other thermochemical processes as gasification and combustion [
22]. Some studies have shown that, due to the lignocellulosic composition of sugarcane residual biomass, pyrolysis is an alternative to convert it into valuable products, such as H
2-rich syngas, bio-oils and biochar [
3,
23,
24]. However, one of the main problems during thermochemical processes is the low quality of the produced gas due to the presence of higher organic and oxygenated compounds known as tars [
25]. These condensable compounds decrease gas yields and process efficiency [
26]. In order to improve the gas quality, some authors have proposed the integration of pyrolysis with reforming [
27] or gasification [
28] to reform the volatiles obtained during pyrolysis, therefore obtaining a H
2-rich gas stream [
27]. Nevertheless, these processes are more complex, since each one operates under different optimal conditions [
29]. Thus, costs increase because of the need of more than one thermochemical unit [
13,
27,
30]. Hence, catalytic pyrolysis has emerged as a feasible and economic alternative due to several reactions taking place, such as catalytic cracking, reforming and deoxygenation reactions of heavy compounds that allow for organic compounds degradation [
6] and carbon conversion [
31]. Consequently, less tars and a H
2-rich gas can be obtained in a single step.
Several catalysts have shown to improve the formation of gases during biomass pyrolysis [
22,
31,
32]. Among them, there is a trend in the use of Ni-based catalyst due to the higher activity and low cost [
25]. However, they can present deactivation due to the formation of coke on the catalyst surface [
33]. Moreover, catalysts with noble metals such as Pt/Al
2O
3 [
34], Rh-perovskite [
30] and Pt-Rh/MgAl(O) [
28] have been tested during integrated pyrolysis processes with steam reforming to improve the quality of condensable and non-condensable gas streams from pyrolysis [
35,
36]. In these studies, Pt and Rh have shown great activity promoting reforming reactions; Pt has a great selectivity to H
2, and Rh has a great capacity to break O–H bonds, which deliver an increase in the H
2 and CO yields [
36]. Although these catalysts have been used in steam reforming, it has been observed that during pyrolysis H
2O is present throughout the temperature range, allowing the reforming reactions to take place [
31]. The above is caused by the H
2O contained in the sample and the degradation of hemicellulose and lignin [
31]. Thus, studying low noble metal loading (<1% wt) catalysts in pyrolysis could improve the composition of the gas streams obtained from this step, reducing additional equipment requirements or subsequent high temperature conditions.
Besides, in order to obtain a rich gas outlet stream and to avoid catalyst deactivation, the use of multifunctional catalyst that combine different supports has been recently proposed [
33]. The presence of CeO
2 in catalysts such as Ni-Ce/Al
2O
3 and Ce/HZSM-5 avoids deactivation, since its redox properties prevent coke formation [
33]. Furthermore, noble metal catalysts have shown resistance to deactivation and higher gas yields during biomass pyrolysis [
28,
30] and other thermochemical processes such as combustion [
34] and gasification [
37]. CeO
2 used as catalyst support can improve the thermal stability and basicity of the catalyst, which increases CO
2 adsorption by inhibiting coke formation and reducing its deactivation [
38]. Additionally, supports such as SiO
2 offer high surface area, increasing the availability of active sites in the catalytic structure [
38]. In this sense, a Rh-Pt/CeO
2-SiO
2 catalyst designed by Cifuentes et al. [
39] for ethanol steam reforming has shown elevated activity and selectivity to H
2. Consequently, it is proposed to evaluate this catalyst during the biomass pyrolysis for H
2-rich syngas production.
In that sense, understanding the thermal degradation of biomass under different atmospheres is an important step in the design of biomass conversion process to obtain H
2. Thereby, differences in product distribution and kinetic parameters under catalytic and non-catalytic conditions must be addressed for different degradation atmospheres. Thus, pyrolysis (N
2) is commonly compared to combustion (O
2) [
40,
41,
42,
43], because the latter is the traditional thermal degradation process employed to handle lignocellulosic solid wastes. For that purpose, TG analysis have been widely used in the characterization of thermal degradation of different types of biomass, such as nutshell, pine sawdust [
1], other lignocellulosic biomass [
40] and plastics [
41]. TG analysis provides real-time information on the thermal degradation of the sample as a function of time and temperature [
44,
45].
Once thermal degradation of the sample is studied, kinetic studies could be performed. Degradation kinetics are an important tool to understand the progress of decomposition reactions [
46]. Besides, kinetic study provides kinetic parameters as activation energy (E
i), pre-exponential factor (k) and reaction models that describe the thermal degradation and allow for the design of thermochemical units suitable for this type of residues [
41]. Kinetic models of biomass pyrolysis are determined based on the correlation between thermal degradation analysis and information about the released products. This could be done by integrating TG analysis with FTIR (TG-FTIR) and mass spectrometry (TG-MS) [
8,
47].
Kinetic modelling is usually performed by numerical methods like model fitting methods, which estimate kinetic parameters of the thermal decomposition process using an integral approach, hence the correlation with experimental data is easy and precise [
9,
48,
49]. Gangavati et al. [
24] reported the kinetic parameters found through TGA of a press-mud obtained from a sugar mill in India. Parameters were calculated using different relations from literature such as Coast and Redfern, Agrawall and Sivasubramanian methods in order to compare the values obtained [
24]. Meanwhile, Garrido et al. [
41] studied the thermal decomposition of viscoelastic memory foam by TG Analysis under different atmospheres and proposed a model with three consecutive reactions and the kinetic parameters using integral methods that involve all the heating rates evaluated, which gives more accurate parameters [
49,
50]. The above agree with Anca-Couce et al. [
51], who compared kinetic parameters obtained by model free methods and model fitting methods during beechwood pyrolysis and concluded that model fitting methods are more reliable and show a better fit. However, for catalytic processes, the kinetic parameters have been only obtained by model free methods. For instance, Yang et al. [
22] evaluated the effect of the multifunctional Ni-CaO-Ca
2SiO
4 catalyst on the kinetics of catalytic pyrolysis of straw, sawdust and cellulose finding an increase in the intensity of H
2 and CO observed by TG-MS and the reduction of activation energies for all biomasses. Moreover, Loy et al. [
5] reported a kinetic parameter during non-catalytic and catalytic pyrolysis of rice husk, using rice hull ash catalyst, obtaining
Ei values in the range of 190–186 kJ mol
−1 and 154–150 kJ mol
−1, respectively. The parameters were obtained by model free methods [
5].
Therefore, this study aimed to evaluate the effect of the Rh-Pt/CeO2-SiO2 catalyst during the pyrolysis and combustion of sugarcane residual biomass. Thermal degradation kinetic models were proposed, and their parameters calculated by model fitting methods based on the released products obtained from TG, TG-FTIR and TG-MS analysis. Obtaining the accurate kinetic parameters of catalytic conditions under different atmospheres and understanding the products evolution of the biomass catalytic pyrolysis will help us with a rigorous reactor design of the thermal degradation of sugarcane residual biomass.
3. Materials and Methods
3.1. Biomass Recolection, Pretreatment and Characerization
The liquid sugarcane press-mud residue was collected from Tolima, Colombia. Initially, the press-mud residue was hydrolyzed at 130 °C for 1 h in an autoclave (TOMY Digital Biology, Tokyo, Japan) for subsequent fermentation. Then, the sugarcane press-mud was filtered using a sieve (70-mesh, 212 µm) to remove the solid phase. Afterwards, the solid residue containing the lignocellulosic material was dried at 60 °C for 72 h, grounded and sieved in a AS200 sieve (Retsch, Haan, Germany). Finally, the dry solid fraction (sugarcane residual biomass) with particle sizes <212 μm was the biomass used in this study.
Samples were characterized by elemental analysis using a CHNS analyzer FlashEA 1112 Series (Thermo Fisher Scientific, Waltham, MA, USA). Oxygen content was determined by difference on a dry ash basis. The proximate analysis was performed by thermogravimetry in a TGA/DSC1 (Mettler Toledo, Columbus, OH, USA), following the method described by García et al. [
65] The enthalpy of combustion was measured in a calorimetric pump AC-350 (LECO, St. Joseph, MI, USA); this was used to determine the lower and higher heating value on dry basis (
LHVdb and
HHVdb), according to Equations (9) and (10).
where %
N,
S and
H are the weight percentages from elemental analysis of the sample,
is the enthalpy of combustion of the biomass, and the numbers represent the different formation enthalpies in cal g
−1.
Moreover, the quantitative analysis of the composition of the ashes was carried out by inductive coupling plasma mass spectrometry (ICP-MS 7700x) (Algilent Technologies, Santa Clara, CA, USA). Samples were prepared following the EPA 3051A method (acid digestion with microwaves for sediments, sludges, soils and solids). For this, the digestion of 0.1 g of biomass was performed using 4 mL of HNO3 and 1 mL of H2O2, then the digestion was completed by microwave with a maximum power of 950 W. Finally, the digested sample was filtered with glass fiber and diluted into water to a volume of 25 mL and analyzed in the ICP-MS. To obtain information of the functional groups in the biomass, the sample was characterized by Attenuated Total Reflection Fourier Transform Spectroscopy (ATR-FTIR, IFS 66S) (Bruker, Billerica, MA, USA). Each IR spectrum was obtained in a scanning range of 4000 and 600 cm−1 with 4 cm−1 of resolution.
3.2. Catalyst Synthesis
The Rh-Pt/CeO
2-SiO
2 catalyst was prepared following the methodology proposed by Cifuentes et al. [
38] For this, the mixed support was obtained by dissolving the Ce(NO
3)
3·6H
2O (99.9%, Merck, Darmstadt, Germany), as CeO
2 precursor, in distilled water and added slowly to the SiO
2 (Merck, Darmstadt, Germany). Subsequently, the support was dried for 24 h in an oven at 80 °C and calcined at 500 °C for 4 h. Rhodium (III) chloride hydrate (RhCl
3·H
2O) (Merck, Darmstadt, Germany) and hexachloroplatinic acid hexahydrate (H
2PtCl
6·6H
2O) (Merck, Darmstadt, Germany) were used as precursor salts of the metals and were added by the incipient wet impregnation method [
39] up to a total load of 0.4 wt% of each metal. These loads of rhodium (Rh) and platinum (Pt) were selected considering their reported activity in reforming reactions [
39]. The final solid was dried at 80 °C for 24 h, then calcined at 700 °C for 2 h and reduced with 8% H
2/He at a flowrate of 200 mL min
−1. Finally, to ensure a particle size <177 μm, the final solid obtained was sieved on an 80-mesh sieve. The effect of the Rh-Pt/CeO
2-SiO
2 catalyst was evaluated using a 1:1 biomass/catalyst ratio. This ratio was selected based on the results reported by [
6,
22,
59]. A complete characterization of the catalyst has been previously reported [
38], with a surface area of 104 m
2 g
cat−1.
3.3. TG Analysis
Thermal degradation of biomass was evaluated at three different heating rates (5, 10 and 20 °C min−1) up to 900 °C in two reaction atmospheres, pyrolysis (N2) and combustion (synthetic air). These conditions were applied to the sugarcane residual biomass samples with and without catalyst, for a total of 12 experiments. The analysis was carried out in a thermobalance model STA6000 (Perkin Elmer, Waltham, MA, USA). For all the experiments, 2–5 mg of dried samples was used, with a total flow rate of 100 mL min−1. To ensure the reproducibility of the experiments, duplicates of experiments were carried out randomly, ensuring a difference <5%. Weight loss was defined as the ratio between the mass of the solid at any time (m) and the initial mass of the solid (m0). Moreover, the DTG curves represent the weight change with time.
3.4. TG-FTIR Analysis
Volatile compounds obtained during the thermal degradation were analyzed by TG-FTIR analysis, using a TGA/DSC1 (Mettler Toledo, Columbus, OH, USA), coupled to a Nicolet 6700 FT-IR spectrometer (Thermo Scientific, Waltham, MA, USA). The experiments were carried out in the two reaction atmospheres, pyrolysis and combustion, with a flow rate of 100 mL min−1, heating up to 900 °C at 10 °C min−1 with approximately 5 mg of the samples. The absorbance was measured with a resolution of 4 cm−1 in a range of 3600–600 cm−1.
3.5. TG-MS Analysis
To identify diatomic molecules such as H2, which cannot be identified by TG-FTIR, and to associate the identified functional groups with specific compounds, TGA-MS analysis was performed. Tests were carried out in a thermobalance TGA/SDTA851e/LF/1600 model (Mettler Toledo, Columbus, OH, USA), coupled to a mass spectrometer Thermostar GSD301T model (Pfeiffer vacuum, Asslar, Germany), which works on Square-Input Response (SIR) mode with ionization of 70 eV. In these experiments, the gases used were He (pyrolysis) and He:O2 = 4:1 (combustion), both with a flow rate of 100 mL min−1 and approximately 5 mg of sample, heating up to 900 °C at 30 °C min−1. To track a broad spectrum of compounds, two different runs were performed. In the first, the mass/charge ratios (m/z) were followed in the range of 2–46 and the next, in the range of 50–106. The response of the different ions was divided by the He response (m/z = 4). Finally, to obtain the proportions of the species, the areas of the followed ions were calculated integrating the TG-MS results.
3.6. Kinetic Model
A model fitting method was used for the kinetic modeling. For that purpose, a model explaining thermal decomposition in both atmospheres (pyrolysis and combustion) was proposed. This methodology has been used in kinetic models for biomass [
44] and other types of materials [
41]. The kinetic model proposed for the pyrolysis of biomasses could be interpreted considering the materials formed by three different fractions, as shown in Equation (1). Note that, at first, none of the components is related to a particular chemical structure; i.e., Component
i do not correspond to celullose, hemicellulose or lignin fractions [
64].
All raw and processed Excel data from TG, TG-FTIR, TG-MS analysis and the fitting method of the estimated kinetic parameters can be downloaded from [dataset] [
66].