1 Introduction

Currently, the unsustainable consumption of fossil fuels worldwide has encouraged research into the use of waste-to-energy processes. Recovering useful energy from solid wastes using thermochemical conversion is regarded as a feasible strategy for both waste management and energy production [1, 2]. In Brazil, about 27.9 million people are smokers, consuming 110 billion cigarettes per year [3]. In this context, about 44 thousand tons of smoked cigarette butts (one cigarette butt has an average weight of 0.4 g [4]) are generated as garbage annually in Brazil, consequently, causing several environmental problems due to their toxic and hazardous characteristics [4,5,6,7]. The smoked cigarette butts consist mostly of cellulose acetate fibers, and small fractions of harmful substances (such as tar and nicotine), unburned tobacco, ash, and paper [4, 8], which results in a high degree of carbon atoms in its composition [7]. The majority fraction, composed by cellulose acetate, is non-biodegradable [8, 9]; thus, cigarette butts may not be decomposed in a landfill. Public concern over substantial pollution caused by smoked cigarette butts has demanded the search for new recycling routes, which can convert smoked cigarette butts into valuable products and reduce the environmental problems due to their inadequate management. These characteristics indicate that smoked cigarette butts can potentially be used as carbonaceous raw material for energy production. Smoked cigarette butt’s disposal often involves inefficient landfilling and incineration; however, they are being discouraged since are not environmentally friendly, sustainable, and feasible disposal options [8]. Alternatively, the predominant effort considers the recycling of smoked cigarette butts based on the production of adsorbent materials for several applications [3, 5, 7, 9]; however, the smoked cigarette butts are still considered solid waste after their use in the form of absorbent materials (becoming saturated), generating secondary pollution, and requiring expensive forms of treatment [8]. In this context, the pyrolysis process appears as an environmentally friendly disposal way for smoked cigarette butts, which provides an opportunity of converting smoked cigarette butts into useful biofuels [8].

Pyrolysis is a promising possibility towards a safe circular economy, as it can handle harmful substances and provides an opportunity of valorization by converting carbon-based materials into energy carries such as biochar, bio-oil, and pyrolytic syngas [10,11,12]. There are several benefits of using pyrolysis as a disposal way for solid waste when compared to the landfilling and incineration. For instance, when compared to combustion (or incineration), pyrolysis has a lower process temperature and lower emissions of air pollutants, as well as zero waste generation and more flexible and compact equipment at plant scale [2, 13]. Smoked cigarette butts waste that contains cellulose acetate filters saturated with toxic compounds can be classified as municipal and hazardous waste [4], which requires a long period to decompose in the environment, but that also presents great polluting potential, making unfeasible the use of landfills. In addition, the decreasing amount of land available for solid waste disposal and high maintenance costs have constrained the choice for landfills [14]. For example, disposal in Brazilian landfills is costly, estimated at US$ 70 per ton [15]. Thus, the pyrolysis of smoked cigarette butts waste seems to be a more adequate disposal way over traditional disposal methods.

Solid residues, as coffee paper cup waste, polyvinylchloride cable sheath, medium density fiberboard, and plastic solid waste, have recently been proposed as feedstocks to the pyrolysis process [1, 16,17,18]. In this regard, few attempts have been carried out to study the feasibility of smoked cigarette butts in the production of biofuels through pyrolysis [8]. For now, to our knowledge, the present information is insufficient concerning an in-depth understanding of the kinetic and thermodynamic parameters of the pyrolysis of smoked cigarette butts. This detailed knowledge is fundamental to achieving efficient and competitive waste-to-biofuel conversion [16, 17, 19].

Thermogravimetric analysis (TGA) is a useful technique to investigate the solid-state pyrolysis patterns in the form of kinetic curves (conversion rate as a function of temperature or time) under an oxygen-free atmosphere [20,21,22,23]. With the combined use of non-isothermal TGA outputs and mathematical methods, thermal behavior, reaction kinetics, and heat transfer data can be identified and investigated and, together with thermodynamic parameters, they can help to provide a reliable description of the entire pyrolysis process [24,25,26,27]. The kinetic triplet, i.e., the apparent activation energy (Ea), the frequency factor (A), and the reaction model (f(α)), can be determined by TGA measurements associated to numerical kinetic approximations, which has practical importance to advance in large-scale reactor design [23, 28]. The fundamental concepts of the energy barrier, the frequency of vibrations of the activated complex, and the reaction mechanism have equal significance [29]. The isoconversional methods of Friedman (FR), Flynn-Wall-Ozawa (FWO), Kissinger-Akahira-Sunose (KAS), and Starink (STK) express the values of Ea with the conversion, which can be estimated without the assumption of the reaction model and frequency factor [21,22,23, 29] The compensation effect and the master plot methods have been frequently applied to obtain the A and f(α) parameters, respectively and these methods are enabled to attend this study due to their recognized accuracy. Thermodynamic favorability and energy requirements of the pyrolysis process can be evaluated by calculating the thermodynamic parameters such as the enthalpy change (ΔH), Gibb’s free energy change (ΔG), and the entropy change (ΔS) by the Eyring Theory [1, 17, 18].

Usually, pyrolysis of solid residues tends to be a complex process since it involves several physical and chemical pathways with the overlapping of numerous reactions [1]. The use of deconvolution has been successfully adopted as a reliable pathway to distinguish and analyze individual processes in a complex pyrolysis reaction [1, 23, 30]. Thus, the kinetic triplet estimated by the sum of the several individual analysis can be useful for the reconstruction of the entire experimental pyrolysis behavior with high accuracy [1, 23]. A reliable description of the entire pyrolysis process, in terms of kinetics and thermodynamic aspects, is essential for the design and scale-up of an effective reactor [26, 27]. In this context, a significant part of the published kinetic studies does not provide sufficiently reliable information on kinetics and thermodynamic parameters since the complex reactions are often interpreted as single-step kinetics, i.e., by a single kinetic triplet. In addition, the validation of the estimated kinetic data is frequently omitted in kinetics investigations, although the ICTAC Kinetics Committee recommends that to ensure data reliability [29]. Although the pyrolysis of smoked cigarette butts has been used for the production of ester-rich bio-oil [8], no research has been done on the pyrolysis characteristics of smoked cigarette butts by exploring its kinetic triplet and thermodynamic parameters, which motivates this study.

The current study represents the first attempt at exploring the pyrolysis kinetics and the physicochemical properties of smoked cigarette butts in order to allow an evaluation of the energy potential of this solid residue. Firstly, the physicochemical characterization was performed to identify the capacity of the smoked cigarette butts to be used as a feedstock for pyrolysis. The asymmetrical Fraser-Suzuki function was used to deconvolute the overall differential curves into several pseudo-components, while the apparent activation energy was determined, in a second step, using four isoconversional methods (FR, FWO, KAS, and STK). The compensation effect and the master plot methods were adopted for each separate pseudo-component to find the frequency factor and possible pyrolysis reaction models, respectively. The adequacy of the resulting kinetic triplets was tested by the reconstruction of the experimental weight-loss curves. Thermodynamic parameters were also estimated using the Eyring Theory to characterize the conversion of smoked cigarette butts to biofuels. Findings from this study on the smoked cigarette butts subjected to pyrolysis will be critical to transforming this solid waste into a potential energy feedstock on a sustainable basis.

2 Materials and methods

2.1 Collection and preparation of smoked cigarette butts

The smoked cigarette butts (SCB) were collected from the Reitor João David Ferreira Lima University Campus (Federal University of Santa Catarina, Florianópolis) in December 2019. The smoked cigarette butts were ground using a cutting mill (Solab, SL-31, Piracicaba, Brazil) and sieved using a mechanical sieve shaker (Bertel, series 1.0, São Paulo, Brazil) to obtain particles of < 250 μm in size (60 mesh). The powdered sample was stored in a sealed polyethylene bag for later use. Proximate analysis, ultimate analysis, and heating values were carried out to characterize the SCB sample. All the experiments were repeated at least twice, and the characterization results were expressed in terms of averages.

2.2 Physiochemical analyses

On an as-received basis, the percentage by weight (wt%) of moisture (MO), volatile matter (VM), and inorganic material (ASH) in SBC was determined by a thermogravimetric analyzer model TGA-Q50 (TA Instruments, New Castle, USA) following the standard method ASTM E1131-08 [31]. The percentage by weight of fixed carbon (FC) was estimated by difference on an as-received basis, as follows:

$$ \mathrm{FC}\left(\mathrm{wt}\%\right)=100\ \mathrm{wt}\%-\mathrm{MO}\left(\mathrm{wt}\%\right)-\mathrm{VM}\left(\mathrm{wt}\%\right)-\mathrm{ASH}\left(\mathrm{wt}\%\right) $$
(1)

The ultimate analysis is a procedure used to quantify the percentage by weight of carbon, hydrogen, nitrogen, sulfur, and oxygen in an organic compound. In this work, the standard protocol ASTM D5373-08 [32] was used to determine carbon, hydrogen, and nitrogen in an Elemental Analyzer 2400 CHN Perkin-Elmer Series II, while the protocol ASTM D4239-14 [21] was used to determine the sulfur in an ICP-OES Analyzer Spectro Arcos. The oxygen content was determined by the difference in dry weight basis, according to Eq. 2.

$$ \mathrm{O}\left(\mathrm{wt}\%\right)=100\ \mathrm{wt}\%-\mathrm{C}\left(\mathrm{wt}\%\right)-\mathrm{H}\left(\mathrm{wt}\%\right)-\mathrm{N}\left(\mathrm{wt}\%\right)-\mathrm{S}\left(\mathrm{wt}\%\right)-\mathrm{ASH}\left(\mathrm{wt}\%\right) $$
(2)

The Channiwala and Parikh correlation was used to estimate the higher heating value (HHV) on a dry weight basis, according to Eq. 3 [33].

$$ {\displaystyle \begin{array}{c}\mathrm{HHV}\left(\mathrm{MJ}\ {\mathrm{kg}}^{-1}\right)=0.3491\cdot \mathrm{C}\left(\mathrm{wt}\%\right)+1.1783\cdot \mathrm{H}\left(\mathrm{wt}\%\right)+0.1005\cdot \mathrm{S}\left(\mathrm{wt}\%\right)\\ {}-0.1034\cdot \mathrm{O}\left(\mathrm{wt}\%\right)-0.0151\cdot \mathrm{N}\left(\mathrm{wt}\%\right)-0.0211\cdot \mathrm{ASH}\left(\mathrm{wt}\%\right)\end{array}} $$
(3)

The lower heating value (LHV), on a dry weight basis, was calculated using an empirical correlation (Eq. 4), which deducts the value of the latent heat of vaporization of water from HHV [34].

$$ \mathrm{LHV}\left(\mathrm{MJ}\ {\mathrm{kg}}^{-1}\right)=\mathrm{HHV}\left(\mathrm{MJ}\ {\mathrm{kg}}^{-1}\right)-0.2183\cdot \mathrm{H}\left(\mathrm{wt}\%\right) $$
(4)

2.3 Thermogravimetric analysis

Thermogravimetric experiments were performed using a thermogravimetric analyzer model TGA-Q50 (TA Instruments, New Castle, USA) under an oxygen-free atmosphere for pyrolysis. A sample mass of 10 mg was placed into a standard platinum crucible and heated from room temperature to 800 K under controlled high-purity nitrogen (N2, 99.999%), with a gas flow rate of 50 mL min−1. This procedure was performed for the non-isothermal condition with heating rates of 5, 10, 20, and 30 K min−1. Slow heating rates (≤ 30 K min) and small particle size (< 250 μm) were applied to reduce heat and mass transfer restrictions, as reported by Van de Velden et al. [35]. A purge step with a high purity nitrogen atmosphere (99.999%) was performed by 30 min with a gas flow rate of 50 mL min−1 before each thermogravimetric analysis. This first step was necessary to avoid secondary oxidation reactions during heating. A blank TG run (using an empty sample crucible) was used to eliminate systematic errors from each experimental run of pyrolysis. The described procedures were adopted following the ICTAC Kinetics Committee recommendations [36].

2.4 Kinetic study

The pyrolysis kinetics of a residue is evaluated through a process of converting solids into volatiles, which occurs by a sequence of chemical reactions. The complex structure of the residue is broken into different small structures by the temperature increase; consequently, the process results in a weight loss of solid structure. This behavior can be mathematically expressed by the following equation [29]:

$$ {\left(\frac{d\alpha}{d t}\right)}_{\mathrm{Biomass}}={A}_{\mathrm{biomass}}\cdot {e}^{Ea_{\mathrm{Biomass}}/R\cdot T}\cdot f{\left(\alpha \right)}_{\mathrm{Biomass}} $$
(5)

where R (kJ mol−1 K−1) represents the gas constant, α, t, and T are the physical parameters of conversion, time (min), and temperature (K), respectively, while the kinetic parameters of A, Ea, and f(α) are the frequency factor (min−1), apparent activation energy (kJ mol−1), and reaction model, respectively. The physical parameters are detected by thermogravimetric analysis, in which the conversion is represented by the relationship between the initial and final mass of the solid, with the mass for each pyrolysis time, or temperature. Equation 5 (kinetic parameters constants) returns a single thermal behavior, i.e., the parameters from Eq. 5 are suitable for solids with just one thermal behavior [23]. The thermogravimetric data report a mass loss of the solid as an overall behavior; thus, Eq. 5 is unsuitable for a complete description of the thermal behavior since different compounds constitute the structure of residue. The deconvolution method can be useful to predict the thermal behavior of every single possible component decomposed during pyrolysis. In addition, the estimation of the kinetic parameters in a kinetic study is commonly performed by non-isothermal programming at different heating rates. In this context, Eq. 5 is transformed into Eq. 6, which describes the sum of each thermal decomposition of the components of pyrolysis.

$$ {\left(\frac{d\alpha}{d t}\right)}_{\mathrm{Biomass}}\equiv \sum \limits_{i=1}^N{c}_i\cdot {\left(\frac{d\alpha}{d T}\right)}_i=\frac{1}{\beta}\cdot \sum \limits_{i=1}^N\left[{c}_i\cdot {A}_i\cdot {e}^{Ea_i/R\cdot T}\cdot f{\left(\alpha \right)}_i\right] $$
(6)

where β (K min−1) is the heating rate, c is the contributed fraction in conversion, the subscript i represents each possible component during the pyrolysis of the residue, and N is the total of components proposed for the residue.

2.4.1 Determination of pseudo-components by deconvolution

The complex composition of residues results in the overlapping of each component. The description of each component can be obtained by the deconvolution method, which offers a reliable way to obtain the curves of the pseudo-components for posterior kinetic evaluation [23, 37]. Perejón et al. [38] compared several functions for deconvolution curves and obtained the best results using the Fraser-Suzuki function. Accordingly, the pseudo-component curves in thermogravimetric data in this work were obtained from the Fraser-Suzuki function.

$$ \frac{dm}{dT}=\theta \cdot \exp \left\{-\frac{\ln (2)}{s^2}\cdot {\left\{\ln \left[1+2\cdot s\cdot \left(\frac{T-{T}_p}{w}\right)\right]\right\}}^2\right\} $$
(7)

where θ is the amplitude, Tp is the peak temperature, w is the half-width in the curve, and s is the asymmetry shape parameter (s ≠ 0). The sum of each curve is represented by the overall thermal behavior, as shown in Eq. 8.

$$ \frac{d m}{d T}=\sum \limits^N{\eta}_i\cdot {\left(\frac{d m}{d T}\right)}_i=\sum \limits^N\frac{\eta_i}{m_0-{m}_{\infty }}\cdot {\left(\frac{d\alpha}{d T}\right)}_i\equiv \sum \limits^N{c}_i\cdot {\left(\frac{d\alpha}{d T}\right)}_i $$
(8)

where η is the contribution of the pseudo-component i in unit of mass, m0 is the initial mass, m∞ is the final mass, and the subscript i represents the pseudo-components in SCB waste. The parameters in Eq. 7 are obtained by minimizing the residual sum of squares (RSS).

$$ RSS=\sum \limits^M{\left[{\left(\frac{dm}{dT}\right)}_{\mathrm{exp}}-{\left(\frac{dm}{dT}\right)}_{\mathrm{dec}}\right]}^2 $$
(9)

where M is the total number of points used for the deconvolution, (dm/dT)exp is the experimental values measured, and (dm/dT)dec represents the values from deconvolution.

2.4.2 Apparent activation energy determination

The apparent activation energy for each pseudo-component can be estimated after data extraction through the deconvolution method. The determination of Ea values is performed using the integral form and differential form of Eq. 6, which are known as isoconversional methods. The differential form is expressed by Eq. 10, while the integral form is expressed by Eq. 11.

$$ \ln {\left(\frac{d\alpha}{d t}\right)}_i=\ln {\left(\beta \cdot \frac{d\alpha}{d T}\right)}_i=\ln \left[{A}_i\cdot f{\left(\alpha \right)}_i\right]-\frac{E_{ai}}{R\cdot T} $$
(10)
$$ g{\left(\alpha \right)}_i=\underset{0}{\overset{1}{\int }}\frac{d\alpha}{f{\left(\alpha \right)}_i}=\frac{A_i}{\beta}\cdot \underset{T_0}{\overset{T}{\int }}{e}^{\raisebox{1ex}{$-{E}_{ai}$}\!\left/ \!\raisebox{-1ex}{$R\cdot T$}\right.} dT\equiv \frac{A_i}{\beta}\cdot I\left({E}_{ai},T\right)=\frac{A_i\cdot {E}_{ai}}{R\cdot \beta}\cdot p(x) $$
(11)

where g(α) is the integral form of the reaction model, and p(x) represents an approximation for temperature integral equation with x = Ea/RT. The subscript “i” represents each possible component during the pyrolysis of the SCB sample. Unlike the differential expression (Friedman method), the estimation of Ea values from Eq. 11 is not reached since the integral expression does not provide an analytical solution. Thus, approximate equations are alternative solutions to provide expressions with acceptable accuracy and also recommended by the ICTAC Kinetics Committee [29]. The most common integral expressions solved by approximated equations are the methods of Flynn-Wall-Ozawa (FWO), Kissinger-Akahira-Sunose (KAS), and Starink (STK) represented by Eq. 12, Eq. 13, and Eq. 14, respectively [21,22,23].

$$ \log \beta =\log \left(\frac{A_i\cdot {E}_{ai}}{R\cdot g{\left(\alpha \right)}_i}\right)-2.315-0.4567\cdot \frac{E_{ai}}{R\cdot T} $$
(12)
$$ \ln \left(\frac{\beta }{T^2}\right)=\ln \left(\frac{A_i\cdot R}{E_{ai}\cdot g{\left(\alpha \right)}_i}\right)-\frac{E_{ai}}{R\cdot T} $$
(13)
$$ \ln \left(\frac{\beta }{T^{1.92}}\right)=\mathrm{Constant}-1.0008\cdot \frac{E_{ai}}{R\cdot T} $$
(14)

A linear regression analysis between the left side of Eq. 12, Eq. 13, and Eq. 14, and 1/T, is used to obtain the Ea values, i.e., logβ versus 1/T for the FWO method, ln(β/T2) versus 1/T for the KAS method, and ln(β/T1.92) versus 1/T for STK method.

2.4.3 Frequency factor determination

The isoconversional methods are not suitable for the determination of the frequency factor due to a lack of knowledge of the reaction model expression. In this case, it is necessary to use alternative methods to determine the values of A, although the presence of A in the expression of isoconversional methods. The compensation effect (Eq. 15) is generally used for the estimation of A and has the advantage of offering a simultaneous evaluation of both lnA and Ea [29]. The use of the compensation effect method for A determination is in accordance with the ICTAC Kinetics Committee recommendations [29].

$$ \ln {(A)}_i={a}_i\cdot {E}_{ai}+{b}_i $$
(15)
$$ \ln \left(\frac{g{\left(\alpha \right)}_{ij}}{T^2}\right)=\ln \left[\frac{A_{ij}\cdot R}{\beta \cdot {E}_{aij}}\cdot \left(1-2\cdot \frac{R\cdot T}{E_{aij}}\right)\right]-\frac{E_{aij}}{R\cdot T} $$
(16)

where a and b are the compensation parameters, and the terms i and j are the residue component and the reaction model, respectively. The method consists of obtaining the compensation parameters from a model-fitting, such as Coats-Redfern method (Eq. 16), at a single heating rate and several reaction models found in the literature, as initial step [39]. After, the plotting provides a straight line, where a is obtained from the slope and b is obtained from the y-intercept. Finally, the values of A are estimated using the compensation parameters and Ea values previously determined by an isoconversional method.

2.4.4 Reaction model determination

The master plot is often used for the determination of the reaction model in a pyrolysis process [1, 34]. This method is a well-established mathematical procedure, which is based on the comparison between the curves plotted as g(α)/g(0.5) versus α for theoretical and experimental data. The theoretical curves are plotted for several reaction models found in the literature [1, 40], while the experimental curves are obtained from the ratio between the temperature integral equations for each conversion point and the conversion when the reaction reaches 50% (Eq. 17).

$$ \frac{g{\left(\alpha \right)}_i}{g{(0.5)}_i}=\frac{p{(x)}_i}{p{\left({x}_{0.5}\right)}_i} $$
(17)

Equation 17 indicates that the ratio of the integral function can be associated with the ratio of the integral approximation and the overlapping of Eq. 17 in theoretical curves, providing information about the reaction model.

2.4.5 Reconstruction of the curves using kinetic triplet

The reconstruction of the thermal behavior for each heating rate (5, 10, 20, and 30 K min−1) was performed using the kinetic triplet (determined by isoconversional methods, compensation effect, and master plot) in Eq. 6. Subsequently, the new expression was solved using the numerical method of Runge-Kutta 4th-order, from which are obtained curves in terms of dα/dt versus T. Each experimental condition was overlapped with its respective reconstructed curve to compare both curves and to confirm the obtained kinetic triplet. The curves were mathematically compared using the coefficient of determination (R2), quality of fit (QOF), residual (y), and mean residual (My), as shown in Eq. 18, Eq. 19, Eq. 20, and Eq. 21, respectively.

$$ {R}^2=1-\frac{\sum {\left[{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{exp}}-{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{rec}}\right]}^2}{\sum {\left[{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{exp}}-{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{av}}\right]}^2} $$
(18)
$$ QOF\left(\%\right)=\left(1-\frac{\sqrt{\sum {\left[{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{exp}}-{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{rec}}\right]}^2/N}}{{\left[{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{exp}}\right]}_{\mathrm{max}}}\right)\cdot 100 $$
(19)
$$ y={\left(\frac{d\alpha}{d t}\right)}_{\mathrm{exp}}-{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{rec}} $$
(20)
$$ My=\frac{\sum \left[{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{exp}}-{\left(\frac{d\alpha}{d t}\right)}_{\mathrm{rec}}\right]}{N} $$
(21)

where (dα/dt)exp represents the experimental values of dα/dt, [(dα/dt)exp]max represents the maximum conversion rate within the experimental values of dα/dt, (dα/dt)av is the average of experimental values of dα/dt, (dα/dt)rec represents the values of dα/dt calculated by the numerical method of Runge-Kutta 4th-order, and N is the number of experimental points used.

2.5 Estimation of thermodynamic parameters

The thermodynamic parameters in pyrolysis process provide valuable information about energy balances. The thermodynamic parameters, i.e., the change of enthalpy (ΔH), change of Gibbs free energy (ΔG), and change of entropy (ΔS), were determined using the activated complex theory (Eyring Theory) [41,42,43], as follows:

$$ \Delta G={E}_a+R\cdot {T}_m\cdot \ln \left(\frac{k_B\cdot {T}_m}{h\cdot A}\right) $$
(22)
$$ \Delta H={E}_a-R\cdot T $$
(23)
$$ \Delta S=\frac{\Delta H-\Delta G}{T_m} $$
(24)

where kB is the Boltzmann constant (1.381 × 10−23 J K−1), h is the Plank constant (6.626 × 10−34 J s−1), and Tm is the maximum temperature peak observed from the differential thermogravimetric (DTG) curve.

3 Results and discussion

3.1 Smoked cigarette butts characteristics

The results for proximate analysis, ultimate analysis, and heating values are frequently used to obtain useful information about the potential use of a feedstock in pyrolytic conversion. After a detailed search in related literature, it was observed that the physicochemical characteristics of the SCB waste are slightly similar to the cashew nutshell, as observed in Table 1. The proximate and ultimate analysis data found in the study performed by Mythili et al. [44] were converted on a dry basis by Eq. 1 and Eq. 2, while the heating values were calculated by Eq. 3 and Eq. 4, to compare with the SCB data. The SCB sample is characterized by higher volatile matter (76.72 wt%) and lower ash content (8.34 wt%) when compared with typical Brazilian coals [45]. This characteristic is attractive for a solid fuel due to a high volatile matter and a low ash content, which are desirable in a feedstock for energy generation. Ash content is critical for biomass conversion since a high ash content can cause undesirable operational issues, such as fouling and slagging, that limit mass and heat transfer [46]. The SCB sample contains an amount of fixed carbon content (14.94 wt%) close to those reported for lignocellulosic residues, such as cashew nutshell (15.83 wt%), sugarcane bagasse (14.69 wt%), and sawdust (15.51 wt%) [44]. The solid residue has a high VM/FC ratio (5.14), which is comparable with the typical VM/FC ratios for other biomasses (greater than 4.00). Thus, the SCB can be classified as an adequate energy feedstock in terms of reactivity with easy ignition and a high degree of combustibility [46].

Table 1 Comparative physicochemical characterization between SCB and cashew nutshell
Full size table

The ultimate analysis showed that the carbon, hydrogen, and oxygen contents were 44.46 wt%, 6.13 wt%, and 39.08 wt%, respectively, for the SCB sample. Typical values of carbon, hydrogen, and oxygen contents found for biowastes are in the following ranges 41.62–48.88 wt%, 5.39–6.22 wt%, and 41.71–46.34 wt%, respectively [44]. Thus, the SCB shows characteristics of a valuable carbonaceous resource for energy recovery or chemical product generation. The SCB sample contains minimal amounts of sulfur (< 0.01 wt%) and nitrogen (1.99 wt%), which minimizes SOx and NOx emissions and transform this residue as an environmentally friendly resource for energy. Typical nitrogen amounts found in commercial Brazilian coals are between 0.3 and 1.0 wt% [45]. Despite pyrolysis has been placed as a promising way of disposal for smoked cigarette butts waste, studies on the combustion behavior and potential emissions also seem interesting goals for future research.

The molar ratios of H/C and O/C for cashew nutshell (1.52 and 0.56, respectively) were lower than the ratios of the SCB sample (1.65 and 0.66, respectively), which is directly correlated to a higher value of HHV for cashew nutshell. The H/C and O/C molar ratios for the SCB sample, as well for cashew nutshell, are comparable with those of lignocellulosic biomasses and vary from 1.52 to 2.05 and from 0.56 to 0.90 for H/C and O/C, respectively [46]. The high HHV of SCB (18.50 MJ kg−1) represents an important advantage of this residue considering its application as an energy feedstock. The HHV value of SCB was superior to that of several lignocellulosic residues such as plum pits (13.71 MJ kg−1), olive pits (17.02 MJ kg−1), sawdust (12.19 MJ kg−1), and jackfruit peel (16.27 MJ kg−1), which are suggested as alternative feedstocks for energy purposes [28, 46].

Slight deviations in physicochemical properties of smoked cigarette butts can occur due to the different cigarette brands and the presence of impurities and humidity in the garbage dump. In addition, it is reported that smoked cigarette butt waste has bulk density ranging from 89.8 to 159.0 kg m−3 [47], which is substantially low as compared to the well-established lignocellulosic biomasses for bioenergy, such as coconut shell (293.90 kg m−3), sugarcane bagasse (293.00 kg m−3), wood sawdust (792.80 kg m−3), and rice husk (510.10 kg m−3) [44]. The low bulk density of this waste can result in high transportation and handling costs, which can potentially cause a not economically viable processing logistics. In this regard, aiming large-scale processing, low bulk density, and the possible variability in the physicochemical properties of smoked cigarette butt waste can be circumvented by torrefaction combined with densification. The torrefaction is considered a well-established thermochemical pre-treatment to upgrade a low-grade feedstock into a high-grade feedstock with uniform, enhanced physicochemical properties, including reduced moisture content, increased calorific value, and improved energy density [48]. On its turn, the densification processes (such as briquetting) allowed significantly improving the bulk density and energy content per unit volume from different bioenergy feedstocks [49, 50]. Therefore, the use of torrefaction combined with densification can be prospectively explored to make smoked cigarette butts more interesting to be used in waste-to-energy processes.

3.2 Thermal behavior

Figure 1a and Fig. 1b show respectively the TG and DTG profiles obtained for the pyrolysis of smoked cigarette butts under heating rates of 5, 10, 20, and 30 K min−1. The first mass-loss region, at temperatures below 350 K, is attributed to the evaporation of adsorbed moisture. The second and third regions of mass-loss at temperatures between 500 and 700 K are the active pyrolytic zone, where the maximum devolatilization degree occurs (70.29 ± 2.53 wt% of mass loss). The second mass-loss region can be associated with the decomposition of a small amount of organic volatiles in saturated filters that include tar and nicotine (from 350 to 500 K). The third mass-loss region can be associated with the decomposition of unburned tobacco and cellulose acetate fibers (from 500 to 710 K) [51, 52].

Fig. 1
figure 1

(a) TG and (b) DTG profiles obtained for the pyrolysis of smoked cigarette butts at 5, 10, 20, and 30 K min−1; and (c) three isolated peaks proposed by the Fraser-Suzuki deconvolution function (at 5 K min−1) and fitting parameters for each pseudo-component for the pyrolysis of smoked cigarette butts at 5, 10, 20, and 30 K min−1

Full size image

The fitting curve perfectly overlapped with the complex pyrolysis behavior of SCB with a high coefficient of determination (R2 = 0.972) using the Fraser-Suzuki function, as demonstrated at 5 K min−1 (Fig. 1c). This tendency was also verified for all selected heating rates with R2 close to 1. The active pyrolytic zone can be divided into three pseudo-events: PS-1 (from 350 to 492 K), PS-2 (from 415 to 709 K), and PS-3 (from 559 to 643 K), which correspond to the devolatilization overlapping of the three respective pseudo-components: organic volatiles, unburned tobacco, and cellulose acetate fibers. The application of the Fraser-Suzuki deconvolution function led to the separation of three defined peaks, where the temperature characteristics for the three isolated peaks were slightly shifted to higher values with the increasing heating rate (Fig. 1c). The heat transfer is more effective at higher heating rates than at lower heating rates leading to a devolatilization process at higher temperatures [53]. The isoconversional methods, the compensation effect, and the master plot were applied in the deconvoluted data, where the reaction was considered as three parallel reactions in the active pyrolytic zone to get reliable three kinetic triplets.

3.3 Kinetic triplet examination

3.3.1 Isoconversional estimation of activation energy

The isoconversional methods represented by Eq.10 (FR method), Eq.12 (FWO method), Eq.13 (KAS method), and Eq.14 (STK method) are solved by a linear relationship; thus, Fig. 2 shows the linear fit over conversion range of 0.1 ≤ α ≤ 0.9 for the FR, FWO, KAS, and STK methods for each pseudo-component proposed from deconvolution.

Fig. 2
figure 2

The linear fit curves obtained from the FR (ln(dα/dt) versus 1/T), FWO (log(β) versus 1/T), KAS (ln(β/T2) versus 1/T), and STK (ln(β/T1.92) versus 1/T) methods for the pseudo-components in SCB pyrolysis

Full size image

The straight lines with a high coefficient of determination (R2 > 0.99) in Fig. 2 indicate good applicability of all isoconversional methods to determine the values of Ea. In Fig. 2, the parallel straight lines for PS-1 and PS-2 at conversion below 0.2 show a greater gap than those observed for PS-3. This gap is associated with the energy required to breaking the chemical bonds, where the conversion takes place at a lower rate and high-temperature range in the initial conversion. A greater amount of energy is required to solid conversion beginning (α ≤ 0.2) for PS-1 and PS-2 when compared to PS-3. This behavior can be seen in Fig. 1c; a larger temperature range is required for the initial conversion of PS-1 and PS-2 that results in a flattening of the curve, while the initial conversion of PS-3 occurs under a short temperature range. For conversion ranges between 0.2 and 0.8, parallel straight lines are closer than at the beginning of conversion, which has the following order: PS-2 > PS-1 > PS-3. This behavior indicates that the variation of activation energy as a function of temperature, and conversion, for PS-2 is significantly greater than for PS-3. Thus, there is a relationship between the flattening of the deconvoluted curves (Fig. 1c), where a greater temperature range for the complete conversion of a component results in a greater gap between the straight lines; consequently, the variation of activation energy as a function of conversion is associated to the temperature range of deconvoluted curves. A gap is observed at conversion above 0.8 which is similar to conversions below 0.2. In this conversion step, the activation energy increases significantly due to the greater thermal stability of the chemical bonds in the solid structure.

Fig. 3a shows values of Ea obtained for each pseudo-component that was calculated for FR, FWO, KAS, and STK methods at the conversion range of 0.05 ≤ α ≤ 0.95. The values of Ea for PS-1 and PS-3 follow a slight increase with conversion (0.05 ≤ α ≤ 0.95), which indicates a slight change in the energy required to the breakdown of the chemical bonds for these pseudo-components. The difference between the observed Ea values for the beginning (α = 0.05) and ending (α = 0.95) of the thermal decomposition are 50.24–54.26 kJ mol−1 and 36.56–38.47 kJ mol−1 for PS-1 and PS-3, respectively. The difference of the values of Ea between the beginning (α = 0.05) and ending (α = 0.95) for PS-2 was 120.81–127.10 kJ mol−1, which is greater than those observed for PS-1 and PS-3. The low Ea values for PS-1 in Fig. 3a indicate that this component can be composed mainly by semi-volatile (harmful substances such as tar and nicotine), which remain impregnated in the SCB residue after the smoking. This behavior of Ea explains the low temperature at which this component is released (101.87–108.77 kJ mol−1).

Fig. 3
figure 3

(a) Ea versus α curves obtained from FR, FWO, KAS, and STK, (b) master plot curves between (symbol) experimental and (line) theoretical data, and (c) linear fitted obtained from compensation effect for PS-1, PS2, and PS-3

Full size image

For comparison purposes, the activation energy values of other organic materials under pyrolysis conditions are given in Table 2. Kibet et al. [54] evaluated the kinetics of nicotine in mainstream cigarette smoking that was observed to have an Ea value of 108.85 kJ mol−1 for a commercial cigarette. The significant increase of Ea values for PS-2 indicates that this component is composed of a complex matrix of compounds. Wang et al. [55] evaluated the Ea values for the pyrolysis of tobacco leaf and was observed an average value of 131.74 kJ mol−1. Zhao et al. [56] studied the effect of inorganic and organic salts on the pyrolysis of cigarette paper, in which a decrease of Ea values was observed when inorganic and organic salts were added (225.4–251.8 kJ mol−1 to 134.2–203.4 kJ mol−1). Based on these data, it can be assumed that the values of Ea for PS-2 represent the complex and majority decomposition of unburned tobacco, carbonized tobacco, and paper. The values of Ea for PS-3 can be associated with the energy required for cellulose acetate decomposition, where similar Ea values were observed by Henrique et al. [57] during the thermal decomposition of cellulose nanocrystals. In general, the values of Ea were lower than those obtained for sub-bituminous coal (277.3–287.4 kJ mol−1) [40], which can indicate suitable applicability in the pyrolysis process. A lower activation energy is often associated with a fast reaction and high reactivity [18].

Table 2 Comparison of apparent activation energy for the pyrolysis of smoked cigarette butts with literature values for other organic materials under pyrolysis conditions
Full size table

3.3.2 Reaction model determination

The reaction model (f(α)) estimated by the master plot method appears as nth-order reaction mechanisms of the considered pseudo-components, as shown in Fig. 3b (F3 for PS-1, F5 for PS-2, and F1 for PS-3). It is observed that the pyrolysis of smoked cigarette butts occur from several parallel nth-order reactions, where the devolatilization starts with a third-order reaction (f(α)PS-1 = (1-α)3), followed by a fifth-order reaction (f(α)PS-2 = (1-α)5), and ending as first-order reaction (f(α)PS-3 = 1-α). Higher order reaction observed for PS-1 and PS-2 can be explained by the blocking of surface pores by low-temperature decomposition products of smoked cigarette butts, causing the hindrance of volatile releases [40]. The presence of several parallel nth-order reactions also was reported for the pyrolysis of plastic solid waste [1].

3.3.3 Frequency factor estimation

Satisfactory values of R2 were obtained from the compensation effect for PS-1 (R2 = 0.9904), PS-2 (R2 = 0.9771), and PS-3 (R2 = 0.9998), which provided values of frequency factor (A) of 1.77 × 1012 min−1 (lnAPS-1 = 28.20) for PS-1, 9.44 × 1016 min−1 (lnAPS-2 = 39.09) for PS-2, and 9.62 × 1020 min−1 (lnAPS-3 = 48.32) for PS-3 (Fig. 3c). These different values of A can be attributed to the different composition of the pseudo-components. High values of the frequency factor imply in an easier decomposition [18], where values of A for PS-3 (~ 1020 min−1) appeared to be quicker and easier to start devolatilization than PS-1 and PS-2 (~ 1012–1016 min−1). The frequency factor with magnitude lower than 109 indicates a surface reaction, while the frequency factor with an order of magnitude higher or equal to 109 indicates simpler reaction chemistry [17]. The values of Ea higher than 100 kJ mol−1 and A with a magnitude higher than 1012 min−1 imply that the pyrolysis of smoked cigarette butts is chemically controlled [16].

3.3.4 Simulated

The kinetic triplet estimated can be applied in Eq. 6 to provide an expression that represents the thermal decomposition of SCB. The contributed fractions for PS-1, PS-2, and PS-3 were 0.1445, 0.3865, and 0.3268, respectively. When compared to the isoconversional methods of FR, FWO, and KAS, the STK method improves the accuracy of the activation energy satisfactorily [23, 34, 58]. That is the reason why the activation energies obtained with the STK method were used into the overall kinetic expression (Eq. 25) and also to derive the thermodynamic parameters.

$$ \left(\frac{d\alpha}{d t}\right)=0.1445\cdot \left(1.77\times {10}^{12}\right)\cdot {e}^{-102240/ RT}\cdot {\left(1-\alpha \right)}^3+0.3865\cdot \left(9.44\times {10}^{16}\right)\cdot {e}^{-181630/ RT}\cdot {\left(1-\alpha \right)}^5+0.3268\cdot \left(9.62\times {10}^{20}\right)\cdot {e}^{-246880/ RT}\cdot \left(1-\alpha \right) $$
(25)

Figure 4a shows a satisfactory equivalence between the experimental and reconstructed curves, which obtained high values of the coefficient of determination (R2 > 0.93). The coefficient of determination indicates that the set of kinetic triplets obtained from deconvolution can describe adequately the decomposition process of SCB. The value of QOF was above 94.30% and shows a deviation between the experimental and reconstructed curves of less than 5.70%. These values of QOF are acceptable since most of the thermal behavior could be described. Based on this information, Eq. 25 can describe acceptably the thermal behavior of the SCB under an oxygen-free atmosphere. On the other hand, some temperature regions were described with a low adjustment, as seen in Fig. 4b. The low adjustment was observed in some temperature regions for the heating rate of 30 K min−1. These regions comprise the temperature range between the end of PS-1 decomposition and the beginning of PS-2 decomposition (483–530 K), and some experimental points at a high decomposition rate. The high relative error at high decomposition rate can be associated with some noise in experimental data; experimental data can influence the high deviations between the calculated data, which leads to the difference of these values. This slight deviation can also be explained by a significant delay in the heat transfer, which is usually reported for the pyrolysis of solid waste under high decomposition rates [22]. On the other hand, the average value of the residual for the heating rate of 30 K min−1 was 0.00795, which represents a low value when compared to the regions with the high values of relative error. In general, it is noted that the data of kinetic triplet can describe the pyrolysis process of the SCB and the data can assist in the optimization and modeling of a pyrolysis reactor for this type of waste.

Fig. 4
figure 4

(a) Reconstructed (line) and experimental (symbol) curves of dα/dt versus T for pyrolysis of SCB at heating rate of 5, 10, 20, and 30 K min−1 and (b) respective residuals

Full size image

3.4 Thermodynamic analysis

The thermodynamic parameters (ΔH, ΔG, and ΔS) calculated for the three pseudo-components by using the Ea values are shown in Table 3. Average values of ΔH were calculated as 98.67 kJ mol−1, 176.84 kJ mol−1, and 241.79 kJ mol−1 for PS-1, PS-2, and PS-3, respectively.

Table 3 Thermodynamic parameter calculated for the pseudo-components (PS-1, PS-2, and PS-3) of SCB
Full size table

The positive values of ΔH indicate an endothermic aspect of the pyrolysis of smoked cigarette butts, which confirm the requirement of an external energy source to produce biofuels under an oxygen-free atmosphere. Furthermore, the Ea-ΔH values were lower than 5.09 kJ mol−1 that indicates a facility to overcome the potential energy barrier, which is related to an energy-efficient manner to product formation [1, 17].

Changes of the Gibbs free energy (ΔG) were estimated to be 122.63 kJ mol−1 for PS-1, 158.29 kJ mol−1 for PS-2, and 175.43 kJ mol−1 for PS-3. The ΔH and ΔG values were positive, and suggest a non-spontaneous decomposition reaction that needs heat to occur [16]. The high value of ΔG denotes a lower favorability for pyrolytic conversion [18]. The ΔG values found during the pyrolysis of smoked cigarette butts (from 122.63 to 175.43 kJ mol−1) are in agreement with the range found for pyrolysis of plastic solid waste (from 160.96 to 215.56 kJ mol−1) [1], PVC cable sheath (from 158.95 to 162.19 kJ mol−1) [17], medium density fiberboard (from 147.87 to 169.20 kJ mol−1) [18], and coffee paper cup waste (from 138.50 to 141.90 kJ mol−1) [16]. This insight indicated that the conversion of smoked cigarette butts to energy is comparable with the abovementioned typical solid wastes in terms of favorability.

It is well known that the values of ΔS manifest the degree of disorder of a pyrolysis reaction. The negative value of ΔS for PS-1 (− 55.79 J mol−1 K−1) indicates that the first devolatilization event has a lower reactive characteristic associated with a higher degree of arrangement [16, 28]. This insight can be interpreted as a thermally stable product that has been produced, i.e., the activated complex produced presents a more well-organized structure than that in the initial state [1, 17]. This indicates that the decomposition of the complex matrix of compounds in SCB is close to its thermodynamic equilibrium. The ΔS values for PS-2 (32.25 J mol−1 K−1) and PS-3 (108.46 J mol−1 K−1) were positive, which is associated with a higher reactive characteristic, indicating that less time is required for the activated complex generation [1]. Thus, in terms of the activated complex theory [41,42,43], the pyrolysis of smoked cigarette butts can be described as an initial “slow” devolatilization event for PS-1 (tar and nicotine), followed by “fast” devolatilization events for PS-2 (unburned tobacco) and PS-3 (cellulose acetate fibers). The negative and positive ΔS values during the pyrolysis of smoked cigarette butts can be interpreted as a result of the complex reactions involved. This tendency is consistent with other kinds of solid waste proposed as feedstocks for pyrolysis, such as PVC cable sheath and plastic solid waste [1, 17]. The thermodynamic analysis suggests the smoked cigarette butts as a suitable feedstock for pyrolysis with excellent prospects to produce energy, illustrating the opportunity of making good use of this solid residue with toxic characteristics. Future research directions will include investigating the pyrolysis of smoked cigarette butts on a bench-scale reactor to gain insight into the product yields and their characterization.

4 Conclusions

For the first time, it is revealed how smoked cigarette butt waste may be a suitable energy feedstock under pyrolysis kinetics and thermodynamics aspects. This paper addressed the pyrolysis of smoked cigarette butts in the form of multiple-step kinetics, with the contribution of three pseudo-components distinguished and quantified using deconvolution with the asymmetrical Fraser-Suzuki function. The order of magnitude of the frequency factor indicates a simpler reaction pathway for the decomposition of smoked cigarette butts. All pseudo-components obey nth-order reaction mechanisms according to the master plot method. The kinetic triplets were satisfactorily able to reconstruct original experimental curves. It is worth highlighting that the deconvolution combined with isoconversional analysis used here can potentially be extended to other solid wastes with multiple-step kinetic behaviors. Findings related to thermodynamic investigations attested to the endothermic and non-spontaneous nature of the pyrolysis of smoked cigarette butts. The value of Ea-ΔH smaller than 5.09 kJ mol−1 implies that pyrolysis is potentially able to convert smoked cigarette butts in energy in a viable manner. The reported results confirm that smoked cigarette butts are a suitable feedstock toward renewable energy production.