Valorization of spent disposable wooden chopstick as the CO2 adsorbent for a CO2/H2 mixed gas purification

A series of activated carbons (ACs) derived from spent disposable wooden chopsticks was prepared via steam activation and used to separate carbon dioxide (CO2) from a CO2/hydrogen (H2) mixed gas at atmospheric pressure. A factorial design was employed to investigate the effects of the activation temperature and time as well as their interactions on the production yield of ACs and their CO2 adsorption capacity. The activation temperature exhibited a much higher impact on both the production yield and the CO2 adsorption capacity of ACs than the activation time. The interaction of both parameters did not significantly affect the yield of ACs, but did affect the CO2 adsorption capacity. The optimal preparation condition provided ACs with a desirable yield of around 23.18% and a CO2 adsorption capacity of 85.19 mg/g at 25 °C and 1 atm and consumed the total energy of 225.28 MJ/kg AC or 116.4 MJ/g-mol CO2. A H2 purity of greater than 96.8 mol% was achieved from a mixed gas with low CO2 concentration (< 20 mol%) during the first 3 min of adsorption and likewise around 90 mol% from a mixed gas with a high CO2 concentration (> 30 mol%) during the first 2 min. The CO2 adsorption on the as-prepared ACs proceeded dominantly via multilayer physical adsorption and was affected by both the surface area and micropore volume of the ACs. The adsorption capacity was diminished by around 18% after six adsorption/desorption cycles. The regeneration of the as-prepared chopstick-derived ACs can be easily performed via heating at a low temperature and ambient pressure, suggesting their potential application in the temperature swing adsorption process.

www.nature.com/scientificreports/ q CO 2 adsorption capacity of adsorbent (mg/g) q D Maximum adsorption of D-R model (mg/g) q e Equilibrium CO 2 adsorption from experiment (mg/g) q e Average value of equilibrium CO 2 adsorption from experiment (mg/g) q e,mod Equilibrium CO 2 adsorption from model (mg/g) q m Maximum monolayer adsorption of CO 2 (mg/g) R Universal gas constant (8.314 J/mol K) R 2 Determination coefficient (-) S Normalized standard deviation (%) S BET Total surface area (m 2 /g) S t-plot Micropore surface area (m 2 /g) t Adsorption time (min) T Absolute temperature (K) V mes Mesopore volume (cm 3 /g) V mic Micropore volume (cm 3 /g) w Amount of the adsorbent (g) w AC Weight of obtained activated carbon (g) w m Moisture content in chopstick (kg) w SC Weight of spent chopstick (g) Y Yield of activated carbon (%) Carbon dioxide (CO 2 ) is currently recognized as the most prominent contributor to global warming 1,2 . The main sectors of CO 2 emission are energy production (~ 40%), industry (~ 23%), buildings (~ 10%), transport (~ 23%), and others (agriculture, forestry, and other land uses) (~ 5%) 3,4 . Most of the CO 2 emissions from energy production are derived from the burning of fossil fuels, like coal (~ 72.5%) and oil and natural gas (~ 27.5%) 4 . The technology currently used to produce energy in power plants is the Integrated Gasification Combined Cycle (IGCC). With this technology, the energy carriers, such as coals, are gasified with oxygen and steam to form syngas, a mixture of carbon monoxide (CO) and hydrogen (H 2 ). The obtained syngas is then further processed through the water-gas shift reaction to convert CO to CO 2 by the reaction with H 2 O, yielding a CO 2 /H 2 mixed gas [5][6][7] . To meet the target of the Paris Agreement (2015), which urged to reduce the greenhouse gas emissions by 45% by 2030 compared to 2010 and then to zero emission by 2050 8 , various strategies have been developed and applied to separate CO 2 from a mixed gas stream. These include membrane separation, physical and chemical absorptions, cryogenic separation, and adsorption. Among these developed CO 2 separation technologies, adsorption is the most promising and versatile technology due to its low energy consumption and operating cost, high separation efficiency, and high possibility of adsorbent regeneration [9][10][11] . Based on the literature, activated carbons (ACs) are one of the most appropriate adsorbents for CO 2 capture based on their high performance and stability 12,13 . Typically, a high CO 2 adsorption capacity is achieved at a low temperature and high pressure [14][15][16] . Besides, the properties of ACs also play an important role on their CO 2 uptake. The volume of micropores in the range of 0.33-0.82 nm of bamboo-derived ACs was the main factor responsible for the CO 2 adsorption at 273 K and 1 bar, while neither the surface area nor the total pore volume were significant factors 13 . In contrast, both the surface area and micropore volume played a crucial role on the CO 2 uptake by coconut shell-derived ACs 17 , local coal-derived AC activated by KOH 18 and corncob-derived AC activated by KOH 19 . A high CO 2 uptake of Mesua ferrea seed cake-derived AC was obtained when the AC had a high micropore quantity and surface functionality 20 . However, different results were observed with corn stalk-derived ACs, in which the mesopore volume played the key role in the CO 2 adsorption capacity at a low BET surface area (< 500 m 2 /g), while the micropore area played the main role at a high BET surface area (> 500 m 2 /g) 11 . The CO 2 uptake of water caltrop shell-derived nitrogen-doped porous carbons was enforced by the synergetic effect of N content and narrow microporous volume 21 , similar to the hazelnut shell derived N and S co-doped porous carbons, in which its CO 2 uptake was dictated by the joint effect of narrow microporosity and N and S content 22 . The amine-impregnated AC exhibited considerably low BET surface area but importantly high CO 2 uptake compared with the virgin AC 23 . This is because the impregnated amines acted as the active sites to adsorb the CO 2 molecules through the chemical adsorption mechanism. The BaO-impregnated AC exhibited a considerably higher CO 2 uptake than the unimpregnated one due to its high surface basicity 24 . The MgO-impregnated AC nanofiber can promote the CO 2 uptake of the virgin material due to the generation of chemical bindings between the acidic CO 2 molecules and existing basic functional groups 25 . According to above results, it seems to be that both the textural property and surface chemistry affect the quantity of the CO 2 uptake. Nevertheless, it is still controversial to conclude which textural properties of ACs affect the CO 2 adsorption capacity, probably due to the differences in the utilized raw materials and conditions used to prepare the ACs as well as the condition used to test the CO 2 uptake. Nevertheless, the preparation of ACs with good textural properties and high surface might benefit for the CO 2 adsorption.
Typically, there are two sequential steps that are involved in the production process of ACs, including carbonization and activation 26 . Carbonization (or pyrolysis) is the thermal decomposition of the raw material at high temperatures (400-1200 °C) in an inert atmosphere, such as nitrogen (N 2 ) or argon (Ar), in order to eliminate volatile compounds, getting the carbonized carbonaceous material with high fixed carbon (or biochar) 12,27 . For activation, there are two established processes including physical and chemical activations. Physical activation involves thermal elimination of carbon oxides from the carbon surface using activating gases, such as CO 2 , steam, ammonia, or a combination of them 20,28,29 , while the chemical activation involves the impregnation of dehydrating agents or oxidants, such as potassium hydroxide (KOH), sodium hydroxide (NaOH), potassium carbonate (K 2 CO 3 ), or zinc chloride (ZnCl 2 ), and heating the mixture in an inert atmosphere 18 www.nature.com/scientificreports/ with physical activation, chemical activation can be achieved at a lower temperature (< 600 °C) 20 with higher yields 26 and higher surface areas 28 . However, it is energy consuming process due to the required severe condition to proceed the reaction 20 and requires chemical reagents that can contaminate the obtained ACs as well as the environment 29 . Thus, physical activation is more preferable than chemical activation when considered in terms of environmental safety. The frequently used gases for physical activation are CO 2 and steam because both gases provided ACs with a comparable BET surface area 12 . The CO 2 activation usually facilitates the development of new micropores that are responsible for the CO 2 adsorption 31,32 . However, it exhibits a four-fold slower reaction rate than that of steam activation 31 , leading to a long production time and high energy consumption. To conform to the need of economic feasibility, the steam activation seems more favorable than the CO 2 activation. Nevertheless, the steam activation still faces the weakness of that the excessively high steam activation temperatures and/ or times promote the creation of new micro-pores and/or widen the existing micropores, which consequently decrease the surface area and total pore volume 29,33 . Moreover, they induce a high burn-off, resulting in a low yield of ACs. Thus, knowing which preparation parameters (temperature and time) or their interaction affect the textural property of ACs and yield might help the sustainable production of ACs. Previously, the typically precursors used to produce ACs were coal, peat, lignite, and petroleum residues 34 . However, the production of ACs from these finite resources is expensive, requires intensive regeneration, and cannot serve a high and increasing demand for global AC consumption 12 Thus, a plethora of research have focused on the synthesis of ACs from sustainable resources, such as biomass/agricultural wastes 14,16,24,[35][36][37][38] , municipal wastes 30,32,39 , and industrial wastes [40][41][42] . The production of AC from wastes is not only a sustainable process but also an environmentally friendly and a cost-effective strategy based on the reduction of waste disposal and the low production cost of AC 43 .
In this work, spent disposable wooden chopsticks were used as a raw material to prepare ACs by steam activation and then used to capture CO 2 from a mixed CO 2 /H 2 gas. A 2 k factorial design was carried out to investigate the effect of the activation temperature and time as well as their interactions on the yield and CO 2 adsorption capacity of ACs. The benefit of this work is the utilization of spent disposable wooden chopsticks, one of the large scales generated municipal wastes coming from the sharp growth of food delivery services in Thailand as a sustainable carbon source to prepare ACs. This can reduce the environmental and economic burden of waste management by the government and related agencies as well as achieve the cost-effective production of ACs.

Methods
Spent disposable wooden chopsticks were collected from an urban area in Thailand and employed as the raw material to prepare ACs. Prior to utilizing, they were cleaned, naturally dried, crushed in a knife mill and sieved to get a particle size in the range of 0.21-4.76 mm. The dry-basis proximate analysis displayed the presence of volatile matter, fixed carbon, and ash of 80.15 ± 0.38, 18.74 ± 0.37, and 1.12 ± 0.01 wt%, respectively. The ultimate analysis showed the existence of C, H, N, O, and S contents of around 54.05 ± 7.40, 6.86 ± 0.92, 0.21 ± 0.04, 38.75 ± 8.47, and 0.14 ± 0.11 wt%, respectively, and also trace minerals, such as potassium, magnesium, silicon, calcium, or iron.

Preparation of disposable wooden chopstick-derived AC.
A two-step process (carbonization and physical activation) was performed to prepare the disposable wooden chopstick-derived ACs. The carbonization was carried out at 500 °C for 15 min in a N 2 atmosphere. In each experiment, the raw material was pre-dried at 105 °C for 3 h to eliminate the free moisture. Then, approximately 100 g of dried raw material was placed in a cylindrical stainless-steel reactor. Gaseous N 2 (99.999%, Alternative Chem) was continuously supplied throughout the reactor at a constant flow rate of 1,000 mL/min for 30 min to build up an inert environment. Next, the reactor was slowly heated at a constant heating rate of 10 °C/min from room temperature (~ 30 °C) to 500 °C and maintained at this final temperature for 15 min. Afterwards, the reactor was left to cool down slowly to below 105 °C and the carbonized wooden material or biochar was withdrawn. For the steam activation, a 2 k factorial design was performed to explore the effect of the activation temperature (A: 700-900 °C) and activation time (B: 1-2 h) on the production yield and CO 2 adsorption capacity of the obtained ACs. In each batch, approximately 40 g of the obtaining biochar was physically activated by steam in a horizontal fixed bed reactor. The steam generated from the deionized water was continuously supplied at a flow rate of 8 mL/min, while N 2 (protecting gas) was supplied into the reactor at a rate of 1000 mL/min. After completion of the processing time, the reactor was left to cool down overnight and the resulting ACs were kept in desiccator for further characterization and utilization. Samples were coded as ACx-y, where x represents the activation temperature (in 100 °C units) and y represents the time (h). For example, AC7-1 indicates the AC which was activated at 700 °C for 1 h. The yield of ACs was computed from the weight ratio between the obtained AC and the utilized raw material as Eq. (1).
Characterization. The micromorphological characteristics of the biochar and ACs were determined by scanning electron microscopy and energy dispersive X-ray spectrometry (SEM-EDX; IT-500HR JEOL) and high-resolution transmission electron microscopy (HRTEM; JEOL-JEM-3100F) with an accelerating voltage of 300 kV. The qualitative functional groups presented on the surface of all ACs were characterized by Fouriertransform infrared spectroscopy (FTIR; FT/IT-6800 JASCO). The textural properties of the ACs, including the specific surface area and pore size distribution, were computed by N 2 adsorption/desorption isotherms at 77 K using a Multipoint Surface Area Analyzer (Micromeritics, Tristar II3020) coupled with the classical adsorption theories of Brunauer-Emmett-Teller (BET) methods. from a CO 2 /H 2 mixed gas in a horizontal glass tube reactor having an inside diameter of 8 mm ID and 600 mm length at constant temperature of 25 °C and 1 atm. Prior to conducting the experiment, the AC was dried at 105 °C for 5 h to eliminate free moisture and then approximately 2 g of AC was carefully packed in the glass column, providing an effective adsorption length of around 230-250 mm. Afterwards, a CO 2 /H 2 mixed gas was supplied at a constant flow rate of 100 mL/min into the reactor. The concentration of CO 2 in the mixed gas stream was varied over the range of 10 to 50 mol%, controlled by mass flow controller (S48-2-HMT, Horiba).
As the adsorption proceeded, the outlet gas stream was sampled to analyze the gas concentration using gas chromatography (GC; Shimadzu GC-8A) with a thermal conductivity detector (TCD) and an INJ/DET temperature of 120 °C, column temperature of 100 °C, and current of 100 mA. The amount of CO 2 adsorption (mg CO 2 per gram of bulk adsorbent) was obtained from integration of the transient CO 2 concentration from the breakthrough curves using Eq.
(2). The average value of at least three experimental data was reported to reduce the relative errors (3%).
Modelling of adsorption isotherms. Three adsorption isotherm models were used to fit the experimental adsorption data, including Langmuir, Freundlich, and Dubinin-Radushkevich (D-R) models. The Langmuir model describes a monolayer adsorption of adsorbates onto a homogeneous surface with a constant adsorption energy in the absence of interaction between the adsorbates and neighboring sites 44,45 . The nonlinear-and linearized equations of the Langmuir model are shown in Table 1. A plot of P e against P e /q e provides the slope and y-intercept, which can be used to estimate q m and k L , respectively. The Freundlich model explains a multilayer adsorption of adsorbates on the heterogeneous surface of adsorbents 46 . The adsorption energy is initially high and exponentially decreases as the degree of occupied sites increases 47,48 . Both nonlinear-and linearized forms of the Freundlich model are given in Table 1. A plot of log P e versus log q e allows the estimation of n and k F from the slope and y-intercept, respectively. Lastly, the D-R model is appropriate to describe the equilibrium adsorption of gases and vapor on the heterogeneous surface of carbonaceous materials with a wide distribution range of pore sizes 44 . The adsorption occurs via pore volume filling rather than film formation on the perforated walls 44,49 .
The original nonlinear form of the D-R model together with its linear form are also tabulated in Table 1. A plot of (lnB) 2 versus lnq D gives the slope and intercept, which can be used to compute the energy parameter (E) and q D , respectively. The goodness of fit between each isotherm model and the experimental data was determined via the determination coefficient (R 2 ) and the normalized standard deviation (S), as expressed by Eqs. (3) and (4), respectively:

Results and discussion
Effect of the activation temperature and activation time. Representative SEM and HRTEM images showing the microstructures of AC7-2 and AC9-2 together with the original biochar are illustrated in Fig. 1. The biochar showed the development of some large pores in a longitudinal direction due to the opening of vascular bundles of the wooden material during the carbonization (Fig. 1a). The HRTEM images revealed a concentric arrangement of small packets of carbon layers. Compared with the original biochar, the steam activation induced the generation of well-developed pores as well as a surface roughness due to the formation of more gasified components during the activation process. This is because the reactions between carbon and steam are endothermic, and so well-developed carbons form efficiently under elevated temperatures 33 . That is, a high temperature can effectively remove the disordered carbon coming from the deposition and decomposition of the generated tar Freundlich q e = k F P 1/n e log q e = log k F + 1 n log P e log P e vs log q e  51 . Analysis of the AC structures by HRTEM revealed the defective graphene-like layers (dark area) of different sizes and shapes, which were bonded with the neighboring layers to create the spaces or porosity (grey area) on the surface of ACs (Fig. 1b,c). Figure 2 shows the FTIR spectra of the parental biochar and all ACs prepared by steam activation at 700-900 °C for 1-2 h. The FTIR spectrum of the biochar that appeared at a wavenumber lower than 920 cm −1 indicated the presence of aromatic C-H out-of-plane bending 52 . Bands of intensity between wavenumbers of 920-1300 cm −1 are the overlapping C-O stretchings of various surface groups, including the C-O vibration of ethers (942-1300 cm −1 ), esters (1100-1250 cm −1 ), cyclic ethers (1140 cm −1 ), lactonic groups (1160-1370 cm −1 ), phenolic groups (1180-1220 cm −1 ), and also carboxylic acids and cyclic anhydrides (1180-1300 cm −1 ) 53 . The bands at 1480-1650 cm −1 indicated the presence of polyaromatic C=C stretching vibration of sp 2 hybridized Intense spectra appeared a wavenumber of 2300-2400 cm −1 due to atmospheric CO 2 54,55 . After steam activation, qualitative changes were observed in all six ACs, from which some bands were diminished. That is, the peak intensities of the aromatic C-H out-of-plane bending mode, C-O stretching vibration of different surface groups, C=C vibration of sp 2 hybridized carbon, and C=O stretching vibration were all reduced. This was attributed to the loss of more volatile compounds that were released due to the gasification and the reaction between biochar and steam during the steam activation.
The physical adsorption/desorption isotherms and pore size distribution of the ACs are shown in Fig. 3. The isotherm of AC7-1 and AC7-2 (Fig. 3a) conformed to the Type I isotherm according to the IUPAC classification 56 , indicating the presence of a predominately microporous structure with a narrow pore size distribution and a well-developed mesoporous structure 29 . The isotherms of the four ACs prepared at higher activation temperatures and times displayed a hysteresis loop; the usual characteristics of some mesopore-dominant porous materials associated from the capillary condensation in their mesopores 29 . This suggested the emergence of a mesoporous structure in the different distributions. As also displayed as inset of Fig. 3b, the generation of medium-size mesopores was initially observed for AC8-1 and was more pronounced for AC8-2, AC9-1, and AC9-2. This is attributed to the widening of the original micropores to mesopores in the presence of a high activation temperature and long activation time as well as the generation of new mesopores.
The quantitative values of the textural properties of all six ACs are tabulated in Table 2. It can be seen that, at an activation time of 1 h, all the monitored textural properties, including the S BET , S t-plot , V mic and V mes , increased as the activation temperature increased from 700 to 900 °C. The AC8-1 exhibited the highest micropore volume ratio (V mic /V mic + V mes ) of 81.57%. At an activation time of 2 h, the S BET , S t-plot , and V mic also increased as the activation temperature increased from 700 to 800 °C but then slightly decreased at 900 °C. This was not the case for V mes , in which it continuously increased over the whole range of activation temperatures. The micropore volume ratio decreased slightly from 79.83 to 77.18%, indicating the relatively low existence of a microporous structure at a high activation temperature and long activation time.
The CO 2 adsorption capacity of all six ACs prepared by steam activation from a CO 2 /H 2 mixed gas is also summarized in Table 2. The steam activation significantly improved the adsorption capacity of the biochar from around 19.20 mg/g to greater than 74.46 mg/g, a greater than 3.88-fold improvement. Increasing the activation time from 1 to 2 h increased the CO 2 adsorption capacity of the ACs prepared at 700 °C, but a longer activation time at 800 °C was not significant. However, it negatively affected the CO 2 adsorption capacity of ACs prepared at 900 °C. The AC9-1 exhibited the maximum adsorption capacity (around 89.85 mg/g), while AC9-2 displayed a remarkably lower CO 2 adsorption capacity (81.13 mg/g).
The relationship between the CO 2 adsorption capacity and textural properties of ACs prepared at different activation temperatures and activation times is plotted in Fig. 4. The CO 2 adsorption exhibited a direct  Table 2. Textural property and CO 2 adsorption capacity at 25 °C and 1 atm of the six ACs and the parental biochar.

Sample
S BET (m 2 /g) S t-plot (m 2 /g) V mic (cm 3 /g) V mes (cm 3 /g) V mic /V mic + V mes (%) q (mg/g) www.nature.com/scientificreports/ relationship to the S BET , S t-plot , and V mic of the ACs and a relatively fluctuating trend with respect to the V mes . This suggested that both the surface area and micropore volume were the dominant factors promoting the CO 2 adsorption, in accord with previous studies that mentioned that micropores play a crucial rule in CO 2 adsorption 11,57 . This is because a large quantity of CO 2 molecules can diffuse throughout a high surface area of ACs and strongly adsorb at their micropores via van der Waal's forces 24,58 . In comparison, the CO 2 uptake of the ACs prepared from various types of biomass-waste by physical activation reported in the literature are displayed in Table 3. The adsorption capacity of the chopstick-derived AC (AC7-2) was on par with those reported in the literature. The difference in adsorption capacity might be due to the differences in biomass-waste type and properties, the condition used to prepare the AC and to test the adsorption capacity (ex. gas composition and gas flow rate), as well as the equipment or reactors used to test the adsorption capacity.

Optimization of the AC preparation condition.
To further understand the impact of the AC preparation condition, in terms of the activation temperature (A) and activation time (B), on the production yield (Y) and CO 2 adsorption capacity (q) of ACs, a collection of statistical models and associated estimation procedures known as analysis of variance (ANOVA) was performed. Table 4 tabulates each experimental condition in terms of coded variables and response values. It can be seen that case 1 exhibited the highest production yield, but the lowest CO 2 adsorption, whilst case 5 showed the maximum adsorption capacity with an extremely low production yield but still in the acceptable range of the dry biomass-derived ACs, of 5-40 wt% 53 , while the lowest production yield was obtained at case 6. Table 5 illustrates the ANOVA analysis of these two response variables. The manipulated variables that had a p-value of less than 0.05 were considered as a statistically significant effect  www.nature.com/scientificreports/ at the 95% confidential interval level. Conspicuously, the activation temperature played an important role on the production yield of ACs, while the interaction between the activation temperature and the activation time exhibited an important effect on the CO 2 adsorption capacity. Plots of the main and interaction effects of the manipulated variables on the production yield of ACs are shown in Fig. 5. Both a high activation temperature and long activation time exhibited a negative effect on the AC production yield. The activation temperature exhibited a much steeper plot than that of the activation time, indicating its greater impact on the AC production yield than the activation time (Fig. 5a), while the interaction effect of both manipulated variables was not pronounced in this study range, as can be seen by the parallel graph lines (Fig. 5b). This is because the steam activation induced the decomposition of cellulose and hemicellulose leading to the formation of a high porosity in the structure of AC, which allowed the diffusion of the oxidizing agent into the carbon structures and consequently reacted with the lignin 62,63 . Upon increasing the temperature and time, more volatiles were released due to the gasification and the reaction between biochar (C f ) and steam, according to reactions (R1) and (R2) 64 , thus resulting in the decreasing yield of ACs 65 .
(R1)   www.nature.com/scientificreports/ Figure 6 depicts the plots of the main and interaction effects for both manipulated variables on the CO 2 adsorption capacity. Both the activation temperature and time exhibited a slight impact on the CO 2 adsorption capacity. The maximum adsorption was observed at a suitable activation temperature (Fig. 6a), indicating a nonlinear relationship between the activation temperature and CO 2 adsorption capacity. According to the interaction effect plot (Fig. 6b), the intersect of the two linear-curves was observed, indicting a significant interaction effect between the activation temperature and activation time on the CO 2 adsorption capacity. A high CO 2 adsorption was achieved for the ACs prepared at a lower activation temperature and a longer activation time (700 °C, 2 h), or those prepared at a high activation temperature and short activation time (900 °C, 1 h). The AC prepared at an elevated activation temperature and long activation time exhibited a markedly low CO 2 adsorption capacity (ex. case 6), because a long activation time at an elevated temperature can induce a high degree of widening of the existing pores instead of pore-deepening and/or new pore generation 33,66,67 . This pore-widening effect was experimentally confirmed by the decreased micropore volume ratio from 79.20 to 77.18% as the reaction time increased from 1 to 2 h at 900 °C (Table 2). Based on the statistical analysis, the regression models used to predict the production yield and adsorption capacity can be written as Eqs. (5) and (6), respectively.
where A and B are the coded activation temperature and time, respectively. Figure 7 shows the contour plots of both manipulated variables against both response variables. A high AC production yield was obtained at a low activation temperature and short activation time (Fig. 7a), while a high CO 2 adsorption was achieved at a high activation temperature and short activation time (Fig. 7b). From an economical point of view and adsorption performance, the optimal activation temperature and activation time for the preparation of AC was found to be at 700 °C for 2 h, respectively. At this predicted condition,  www.nature.com/scientificreports/ the approximated values of production yield and CO 2 adsorption capacity were around 20.79 and 86.20 mg/g, respectively, which were closed to those obtained from the bench-scale experiment. Figure 8a-e shows the breakthrough curve of an equimolar CO 2 /H 2 mixture at 25 °C and 1 atm and the concentration profile of the exit gas stream at different inlet CO 2 concentrations of all the adsorbents. The breakthrough curve exhibited a huge roll-up of H 2 (C/C 0 > 1) at the early adsorption period (< 2 min). This indicted a fast exit of H 2 in the exhaust stream or its lower adsorption compared with CO 2 68,69 . In other words, the as-prepared AC7-2 exhibited a strong CO 2 adsorption and a weak H 2 adsorption. This roll-up behavior was observed over the entire investigated range of CO 2 concentrations (10-50 mol%; data not shown), supporting that the separation of CO 2 from H 2 was due to thermodynamic separation 70,71 . This is because the weaker adsorbed H 2 exhibited a fast diffusion in the porous structure of AC compared with a stronger adsorbed CO 2 , leading to a transient H 2 rich adsorbed phase 71 . Besides, as CO 2 is heavier than H 2 it exhibited a higher adsorption affinity towards the AC adsorbents 9,72 , which has previously been ranked in the order of CO 2 > > CH 4 > CO > > H 2 72 . Due to the low adsorption affinity of H 2 from the CO 2 /H 2 mixed gas via the AC7-2, the calculation of either the CO 2 or H 2 selectivity using the thermodynamic analysis by Ideal adsorbed solution theory (IAST) is not applicable 71,73 . Thus, the composition of the outlet gas stream (M) or the exit gas concentration profile was calculated according to Eqs. (7) and (8) and depicted in Fig. 8f. www.nature.com/scientificreports/ According to the plot, the obtained profiles can be categorized into two distinct regions; (i) the breakthrough of H 2 at the early adsorption period and the production of a H 2 -rich gas stream, and (ii) the breakthrough of CO 2 and transient to reach the feed composition. The gas stream with the high CO 2 concentration exhibited a breakthrough faster than that with the low CO 2 concentration. The mixed gas stream with a low CO 2 concentration (< 20 mol%) gave an exit stream with a high H 2 purity (> 96.8 mol%) during the first 3 min of adsorption and then lessened afterwards to reach the feed composition, while the mixed gas stream with a CO 2 concentration of more than 30 mol% provided an exit gas stream with a H 2 purity greater than 90% during the first 2 min of adsorption time. This information will help engineers to design an industrial scale CO 2 capture system using the adsorption-based separation process from a CO 2 /H 2 mixed gas over a wide range of CO 2 concentrations. Adsorption isotherm. Figure 9 depicts the fitting curves between the experimental data (marker point) and isotherm results (dashed line) for the CO 2 adsorption by three ACs at 25 °C and 1 atm in the presence of different CO 2 concentrations. The obtained coefficients and fitting quality were considered in terms of the determination coefficient (R 2 ) and the normalized standard deviation (S), with the results summarized in Table 6.
The adsorption capacity increased as the increased CO 2 concentration, which was attributed to the high driving force of the CO 2 concentration between the bulk phase and the surface of AC that can promote a high mass transfer rate 74,75 . For all the explored ACs, the Freundlich model provided a better fit with the experimental results over the entire range of CO 2 concentrations than the Langmuir model, considered in terms of the higher R 2 and S values. This suggests that the CO 2 adsorption on the spent chopstick-derived AC occurred predominantly via a multilayer adsorption with a heterogeneous surface binding 44,49 . The value of n was higher than 1 (n > 1), confirming a favorable adsorption 75 as well as its high degree of heterogeneity and good adsorption intensity 49,76 . The adsorption energy parameters (E) obtained from the D-R isotherm model varied in the range of 11.3 to 11.7 kJ/mol, which were between 8 and 20 kJ/mol, and so were neither purely physical adsorption (< 8 kJ/mol) nor chemical adsorption (> 20 kJ/mol) 77 . Nevertheless, a deviation of energy parameters from a value of 8 kJ/mol of around 25% indicated a predominately physical adsorption. In other words, the CO 2 molecules were dominantly adsorbed via the intermolecular cohesion forces at the pore surface and small part of them were adsorbed via the surface functionalities that originated from pyrolysis as well as inorganic matters 20 .  The recyclability of an adsorbent plays an essential role in the economics of a commercial scale operation, where a high cyclic stability of any employed adsorbent is required. In this work, the cyclic stability of the AC7-2 sample was tested by repetitive adsorption of CO 2 from a CO 2 /H 2 mixed gas (50 mol% CO 2 ) at 25 °C and 1 atm. After each particular adsorption, the adsorbed CO 2 on the surface of the AC7-2 was simply desorbed in air at 105 °C at ambient pressure and then subjected to re-adsorb CO 2 from the mixed gas stream at the same CO 2 concentration. As shown in Fig. 10, the fresh AC exhibited a CO 2 adsorption capacity of 85.19 mg/g and this dropped slightly with increasing regenerative cycles. This suggested that the regeneration of AC via the heating process is effective to remove the weak adsorbed CO 2 at the outer layer of a multi-layer adsorption. At the even after six adsorption/desorption cycles, the employed adsorbent depicted a minimal loss of CO 2 adsorption capacity of around 18% of its initial capacity without any significant loss of ACs. Based on the regeneration results, the as-prepared chopstick-derived AC7-2 is recommended as the candidate adsorbent for CO 2 adsorption in the temperature swing adsorption process in air at low temperature (ex. 105 °C) and ambient pressure. Otherwise, to enhance a higher recyclability, other regeneration procedures such as depressurization or chemical regeneration should be investigated 78 .
Energy consumption analysis. The analysis of required energy to produce the spent disposable wooden chopstick-derived AC by steam activation was adopted from the energy model of Maski et al. 79 and Chen et al. 80 . The energy demand in the production process was separated into two sub-production processes, including the carbonization for biochar production and the steam activation. To perform the energy balance, the heat capacitis of chopstick (C p,sc ) and biochar (C p,char ) were estimated from Eqs. (9) and (10), respectively 81 .
For the basis of 1 kg of chopstick with 7.73 wt% moisture content, the energy consumption used to produce the chopstick-derived AC was determined as shown below.
For the carbonization, there are two types of energy required to prepare biochar; (i) energy required to dry the moisture-bearing chopstick and (ii) energy required to heat chopstick (E CS ) from room temperature (30 °C) to 500 °C and holding at this temperature for 15 min.   (13) and (14), respectively. Due to the complexity of reaction pathways during drying and carbonization, the energy consumptions during both periods were computed from the voltage-current-time profile of the oven and furnace, which were 0.0009 and 0.0011 MW, respectively. Based on the utilized processing time, the values of E cs,hold-1 and E cs,hold-2 were 9.979 and 0.993 MJ, respectively. Therefore, the total energy required to produce biochar per kg of chopstick was 11.7977 MJ.
For the steam activation, the energy requirements are mainly for (i) heating the biochar from 500 to 700 °C (E char,heat ), (ii) holding the biochar at 700 °C for 2 h (E char,hold ) and (iii) producing steam (E boiler ).
(i) Energy required to heat biochar was estimated from the sensible heat according to Eq. (15).
(ii) Energy required for holding the biochar at 700 °C for 2 h was computed from the voltage-current-time profile of the furnace, which was equal to 40.3200 MW. (iii) Energy required in boiler was estimated from the property of generated steam, which was at 1.2 barg.
Based on the quantity of required steam, the amount of required energy for boiler was around 0.0016 MJ.
In summary, the total energy required to produce AC from spent disposable wooden chopsticks was the summation of the energy required for carbonization (11.7977 MJ/kg chopstick) and activation (40.4229 MJ/kg chopstick) with a total of 52.2206 MJ/kg chopstick. Therefore, the total energies required to produce AC and to adsorb CO 2 based on the production yield of AC (23.18%) were 225.28 MJ/kg AC and 116.4 MJ/g-mol CO 2 , respectively. The estimated energy consumption was seemed to be high but was still in the range of biomassderived AC, of 43.4 -277 MJ/kg AC 81 .

Conclusion
In this work, a series of ACs was prepared from spent disposable wooden chopsticks by steam activation for CO 2 separation from a CO 2 /H 2 mixed gas. From ANOVA analysis, it was found that a high activation temperature and long activation time (900 °C, 2 h) negatively affected the production yield and properties of ACs. From an economical point of view and adsorption performance, the optimal activation temperature and activation time for the preparation of AC was found to be 700 °C for 2 h, providing an experimental production yield of 23.18% and CO 2 adsorption of 85.19 mg/g at 25 °C and 1 atm, respectively with the total energies required to produce AC carbon and to adsorb CO 2 of 225.28 MJ/kg AC and 116.4 MJ/g-mol CO 2 , respectively. A fast breakthrough of H 2 was observed via the as-prepared ACs over an inlet CO 2 concentration in a mixed gas of 10-50 mol%, with the release of an almost pure H 2 gas stream during the first 2 min of adsorption. The experimental data of CO 2 adsorption was adequately described by the Freundlich isotherm model, where the physical adsorption played a predominate role on the interaction between the AC-CO 2 molecules. The optimal AC exhibited a 18% loss of CO 2 adsorption after six adsorption/desorption cycles. To further increase the CO 2 absorption capacity, future research should be carried out to develop a large number of basic sites on the surface of AC such as adding basic metal oxides or alkali metals.

Data availability
All data generated or analyzed during this study are included in this published article.  (15) E char,heat = w char C p,char T = 0.1013 MJ.