Non-Energy Valorization of Residual Biomasses via HTC: CO 2 Capture onto Activated Hydrochars

: This study aims to investigate the CO 2 sorption capacity of hydrochar, obtained via hydrothermal carbonization (HTC). Silver ﬁr sawdust was used as a model material. The batch runs went at 200 ◦ C and up to 120 min. The hydrochar was activated with potassium hydroxide impregnation and subsequent thermal treatment (600 ◦ C, 1 h). CO 2 capture was assayed using a pressure swing adsorption (PSA) process. The morphology and porosity of hydrochar, characterized through Brunauer-Emmett-Teller, Barrett-Joyner-Halenda (BET-BJH) and scanning electron microscopy (SEM) analyses, were reported and the sorbent capacity was compared with traditional sorbents. The hydrochar recovered immediately after the warm-up of the HTC reactor had better performances. The Langmuir equilibrium isotherm ﬁts the experimental data satisfactorily. Selectivity tests performed with a model biogas mixture indicated a possible use of hydrochar for sustainable upgrading of biogas to bio-methane. It is conceivably a new, feasible, and promising option for CO 2 capture with low cost, environmentally friendly materials.


Introduction
Growing population and economy race are the two main concerns regarding the increase in emissions of greenhouse gases (GHG) and, subsequently, the accelerating climate change and global warming. Carbon dioxide is the first responsible for these unwanted effects [1]. Nowadays, reducing CO 2 emissions is one of the most challenging issues facing humanity [2]. Global climate reports show that the average world temperature is steadily increasing with respect to 20th century data [3]. Besides, worldwide researchers recognize the need for a sustainable development and more equity, including poverty eradication, and provide an ethical foundation for limiting the effects of climate changes. The rate of atmospheric accumulation of GHGs and the capacity to address its mitigation and adaptation differ for each nation. Often, many of the countries most vulnerable to climate change contribute little to GHG emissions.
Intergovernmental panel on climate change (IPCC) recently reported on the impacts of global warming of 1.5 • C above pre-industrial levels and the related global greenhouse gas emission pathways [4]. The attention is on the decision to adopt the Paris Agreement [5,6].
Comprehensive and sustainable strategies in response to climate change should consider the co-benefits, adverse side-effects, and risks inherent to both adaptation and mitigation options. European policy options emerge in the European Union (EU) Emission Trading System (ETS) that involve the

Experimental Apparatus
The HTC apparatus was designed and realized on purpose. Details of the piping and control instrumentation are reported elsewhere [34]. The 316-stainless steel batch reactor has an internal volume of about 200 mL and is equipped with a valve for the air initial evacuation and gaseous product withdrawal. A thermocouple and a pressure-meter allow monitoring and controlling the process temperature and pressure, respectively.

Experimental Procedure
The HTC experiments were performed at 200 • C, with a constant water biomass ratio 7:1, and residence time equal to 0 and 120 min. The substrate was ground in a mortar and sieved. The fraction below 1 mm was recovered and used for the reaction after drying until constant weight in an oven, at 105 • C. Demineralized water and dry biomass were loaded into the reactor in the chosen weight ratio. Then, the reactor was sealed, evacuated for about 15 min, and heated up to the process temperature. The warm-up lasted for about 20-25 min. Residence time was measured once it reached the temperature set point. At the end of the run, the reactor was quenched with air until 160 • C, and then with water until room temperature.
After the quenching step, the gas phase was recovered and weighed. Solid and liquid products were recovered and separated by filtration. The filter and solid phases were put in an oven at 105 • C for at least 24 h for dry weight determination. After drying, the hydrochar was milled and sieved (certified ISO 3310-1 & 2 and ASTM E11, Endecott sieves). The fraction between 106 and 355 µm was recovered and stored in vials before the CHNS characterization (data reported in Supplementary Material) and the activation step. Then, 15 grams of 1:2 solid mixture of hydrochar and KOH was put in an oven under nitrogen gas flow and warmed up to 600 • C (3 • C/min). The dwell step lasted 1 h [40]. The samples were washed with 10 wt% HCl to remove any inorganic salts, and then with demineralized water until neutral pH. Finally, the samples were dried in an oven at 105 • C for 24 h.

Characterization Methods
The nitrogen sorption isotherms of activated and non-activated hydrochars were determined with a high-speed surface area and pore size analyzer NOVA 1200e Alfates Quartachrome. Brunauer-Emmett-Teller (BET) [42] and Barrett-Joyner-Halenda (BJH) [43] methods were utilized for a standard determination of surface area, pore volume, and pore size distribution and the average pore diameter [44]. Each sample was outgassed for 3 h at 100 • C.
Textural properties were investigated by means of scanning electron microscopy and energy dispersive X-ray spectroscopy (SEM-EDS). The analyses were performed with a Field Emission Zeiss Gemini500SEM analyser (acceleration voltage from 0.02 to 30 KV-store resolution up to 32 k × 24 k pixels-0.6 nm at 15 KV, in back-scattering electron mode-pressure from 5 to 500 Pa), equipped with EDS (energy dispersive spectroscopy) microanalyzer OXFORD Aztec Energy with INCA X-ACT PELTIER COOLED ADD detector. All samples were coated with 5 nm of Cr in an automatic sputter coater Q 150T ES.

Experimental Apparatus
The experimental apparatus is shown in Figure 1. Details of the piping and control instrumentation are reported in a previous paper [22].

Experimental Procedure
Bed length was kept constant packing the column with the same HC volume same of the inert material volume used for the blank test.
Adsorption experiments were performed with two different feed mixtures (CO2 + N2 and CO2 + CH4). The PSA operating pressure varied up to 5 bars.
Tests with CO2 + N2 were carried out at 2, 3, 4, and 5 bar with 100 NmL/min of CO2 and 80 NmL/min of N2 (carrier gas). The regeneration step was performed with N2 at 280 NmL/min. A mass flow meter (MFM) quantified the total amount of the gas mixture leaving the column. During the adsorption tests, V-A, V-B, and V-C valves were kept open. The configuration for regeneration after each run was as follows: V-A and V-G closed, V-H switched for a quick depressurization and then repositioned, V-G re-opened, and N2 injected until no CO2 was detected. The blank tests were carried out in the same way.
Tests with CO2 + CH4 mixture were carried out at 2 and 5 bar with 76 NmL/min of CO2 and 76 NmL/min of CH4. The flow rate of N2 during the regeneration step was 280 NmL/min. The experiments were carried out as previously described, with CH4 in the inlet flow rate instead of N2. V-D and V-A operated simultaneously. The regeneration step lasted until no CH4 was detected at the column outlet. Details of the piping and control instrumentation are reported in a previous paper [22].

Experimental Procedure
Bed length was kept constant packing the column with the same HC volume same of the inert material volume used for the blank test.
Adsorption experiments were performed with two different feed mixtures (CO 2 + N 2 and CO 2 + CH 4 ). The PSA operating pressure varied up to 5 bars.
Tests with CO 2 + N 2 were carried out at 2, 3, 4, and 5 bar with 100 NmL/min of CO 2 and 80 NmL/min of N 2 (carrier gas). The regeneration step was performed with N 2 at 280 NmL/min. A mass flow meter (MFM) quantified the total amount of the gas mixture leaving the column. During the adsorption tests, V-A, V-B, and V-C valves were kept open. The configuration for regeneration after each run was as follows: V-A and V-G closed, V-H switched for a quick depressurization and then repositioned, V-G re-opened, and N 2 injected until no CO 2 was detected. The blank tests were carried out in the same way.
Tests with CO 2 + CH 4 mixture were carried out at 2 and 5 bar with 76 NmL/min of CO 2 and 76 NmL/min of CH 4 . The flow rate of N 2 during the regeneration step was 280 NmL/min. The experiments were carried out as previously described, with CH 4 in the inlet flow rate instead of N 2 . V-D and V-A operated simultaneously. The regeneration step lasted until no CH 4 was detected at the column outlet.
All tests were repeated almost five times. The synopsis is reported in Table 1. The mathematical model for determining the amount of the adsorbed CO 2 was extensively described in Di Felice et al. [45]. The blank test highlighted the domination in the response curve of a flow-mixing in the ancillary equipment and provided a means of confronting the gas outlet responses to the CO 2 capture characteristics of the bed itself. A simple descriptive FOPDT model (first order plus dead time model) of the gas phase allows extracting the response of the particle phase from the measured gas phase response of the entire system under CO 2 capture conditions. The response curve gives the molar total gas holdup, viz, the product of the total molar throughput of gas and the area enclosed by the response curve and normalized ordinate 1 (Figures 2 and 3).
The response curve may be used to evaluate the total amount of CO 2 present in the whole system as a function of time (i.e., its holdup: the moles of CO 2 that entered the system at time t minus those that left). The CO 2 holdup in the solid phase of the bed (i.e., the CO 2 captured by the hydrochar) is merely the total holdup of the entire system minus the holdup of the gas phase of the entire system [45]. Adsorption performance in CO 2 /CH 4 tests was evaluated estimating recovery, purity, and selectivity [22]. The gas recovery rate is the ratio of the quantity of gas recovered after the column to the fed quantity: The purity is defined by the quantity of the gas during the saturation phase divided by total gas leaving the column at the same time.
where y i is the molar fraction of the specific gas and q is the total gas flow. The selectivity is expressed as follows:

Results and Discussions
The assessment of new porous carbon material should be based on its adsorption properties, evaluated through the well-established procedures. Table 2 lists the results of the BET analysis. The surface area and the average diameter of the pore D av,BJH (obtained as 4·V BJHa /A BET , where V BJHa is the BJH desorption pore volume and A BET is the BET surface area) are critical parameters for physical adsorption. In general, the higher the superficial area, the better the sorbent capacity.
The increase of porosity is also highlighted by SEM-EDS images ( Figure 2), where, in the left-hand side (Figure 2a,c,e), the original wood structure is always present after hydrothermal treatment with several typical pits [46] and, at higher magnification (3000 and 10,000×), 10 µm macropores are also visible. On the right-hand side (Figure 2b,d,f), after the activation procedure, as expected, a well-developed sponge structure is evident at increasing magnification (400, 2000, and 10,000×).

Results and Discussions
The assessment of new porous carbon material should be based on its adsorption properties, evaluated through the well-established procedures. Table 2 lists the results of the BET analysis. The surface area and the average diameter of the pore Dav,BJH (obtained as 4·VBJHa/ABET, where VBJHa is the BJH desorption pore volume and ABET is the BET surface area) are critical parameters for physical adsorption. In general, the higher the superficial area, the better the sorbent capacity.
The increase of porosity is also highlighted by SEM-EDS images (Figure 2), where, in the lefthand side (Figure 2a,c,e), the original wood structure is always present after hydrothermal treatment with several typical pits [46] and, at higher magnification (3000 and 10,000×), 10 µ m macropores are also visible. On the right-hand side (Figure 2b,d,f), after the activation procedure, as expected, a welldeveloped sponge structure is evident at increasing magnification (400, 2000, and 10,000×).
The EDS analysis, not reported here, shows the carbon as the main element with a distributed trace of chlorine, owing to the HCl washing step of the activation method.  The EDS analysis, not reported here, shows the carbon as the main element with a distributed trace of chlorine, owing to the HCl washing step of the activation method.
As far as the time course of the outflowing CO 2 and CH 4 concentration is concerned, a sigmoidal behavior was observed for all tests and at all values of the operating pressure, as shown in Figure 3. The recorded signals are reported as a normalized value, that is, divided by the inlet concentration.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 14 As far as the time course of the outflowing CO2 and CH4 concentration is concerned, a sigmoidal behavior was observed for all tests and at all values of the operating pressure, as shown in Figure 3. The recorded signals are reported as a normalized value, that is, divided by the inlet concentration. By inspection of Figure 3, the time of first detection (arrows) depends evidently on the operating pressure. The translation of blank curves is inherent in the operating modality of the apparatus and depends linearly on the pressure (data not reported for the sake of brevity). The translation of the response curves suffers from the superimposition owing to the presence of the adsorbent bed. The difference between the two delays increases steadily. Table 3 is a synopsis of all the results.  By inspection of Figure 3, the time of first detection (arrows) depends evidently on the operating pressure. The translation of blank curves is inherent in the operating modality of the apparatus and depends linearly on the pressure (data not reported for the sake of brevity). The translation of the response curves suffers from the superimposition owing to the presence of the adsorbent bed. The difference between the two delays increases steadily. Table 3 is a synopsis of all the results.
The sorbent capacity of the activated hydrochar obtained at 200 • C and 0 min is higher than that of the corresponding material recovered after 120 min of retention in the HTC reactor. The results of the BET analysis confirm this finding, where the 0 minute sample shows a specific area as high as 210% (3.1 time bigger) and an average pore diameter decreased by 27% in comparison with sample HCA_200_120. These results warn that the HTC reaction time is a crucial parameter and that the existence of a possible optimal retention time will be worth study, and in the further developments, it will be coupled with a cost optimization aimed at industrial exploitation of the results.
On the other hand, the performances of non-activated hydrochar denounce of an adsorbent capacity of an order of magnitude lower, and are thus not acceptable for industrial applications.  Figure 4 shows typical adsorption curves obtained with the activated hydrochar at 2 and 5 bar (a and b, respectively) and the mixture CH 4 /CO 2 . Blank runs are reported as a reference. Arrows signal the time of the first detection. Both diagrams show that, in the blank runs, the CO 2 and CH 4 signals are indistinguishable. On the contrary, in the presence of a hydrochar bed, a selectivity appears evident, as proven by the temporal separation of the arrows. Methane appears first in the column outlet regardless of the operating pressure. The delay between the two signals is an increasing function of the operating pressure, and in any case, it is sufficiently broad for envisaging the development of an industrial process. All of this evidence proves that hydrochar is a suitable medium for separating the mixture by selective adsorption.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 14 The sorbent capacity of the activated hydrochar obtained at 200 °C and 0 min is higher than that of the corresponding material recovered after 120 min of retention in the HTC reactor. The results of the BET analysis confirm this finding, where the 0 minute sample shows a specific area as high as 210% (3.1 time bigger) and an average pore diameter decreased by 27% in comparison with sample HCA_200_120. These results warn that the HTC reaction time is a crucial parameter and that the existence of a possible optimal retention time will be worth study, and in the further developments, it will be coupled with a cost optimization aimed at industrial exploitation of the results.
On the other hand, the performances of non-activated hydrochar denounce of an adsorbent capacity of an order of magnitude lower, and are thus not acceptable for industrial applications. Figure 4 shows typical adsorption curves obtained with the activated hydrochar at 2 and 5 bar (a and b, respectively) and the mixture CH4/CO2. Blank runs are reported as a reference. Arrows signal the time of the first detection. Both diagrams show that, in the blank runs, the CO2 and CH4 signals are indistinguishable. On the contrary, in the presence of a hydrochar bed, a selectivity appears evident, as proven by the temporal separation of the arrows. Methane appears first in the column outlet regardless of the operating pressure. The delay between the two signals is an increasing function of the operating pressure, and in any case, it is sufficiently broad for envisaging the development of an industrial process. All of this evidence proves that hydrochar is a suitable medium for separating the mixture by selective adsorption.  Table 4 quotes the obtained selectivities and recoveries calculated using Equations (1)-(3). The obtained selectivities confirm that CO2 is preferred to CH4. The performance of hydrochar is worse than that obtainable with commercial porous sorbents such as zeolites or activated carbons [22]. As expected, the methane recovery is relatively low with a purity of 95% (Italian regulation for network injection) regardless of the investigated operating pressure. The doubling of recovery obtained with a purity of 70% is a valuable result because of energetic applications.   Table 4 quotes the obtained selectivities and recoveries calculated using Equations (1)-(3). The obtained selectivities confirm that CO 2 is preferred to CH 4 . The performance of hydrochar is worse than that obtainable with commercial porous sorbents such as zeolites or activated carbons [22]. As expected, the methane recovery is relatively low with a purity of 95% (Italian regulation for network injection) regardless of the investigated operating pressure. The doubling of recovery obtained with a purity of 70% is a valuable result because of energetic applications.  Figure 5 reports the averaged CO 2 sorbent capacity for the two samples of activated hydrochar as a function of the corresponding partial pressure in the gas phase. Data fit well to the Langmuir equation [47].
CO 2Sorbent capacity = C Max ·p CO 2 / K + p CO 2 (4) Appl. Sci. 2020, 10, x FOR PEER REVIEW 9 of 14 The regressions are reported as solid lines in the explored range and as dotted lines in the extrapolated ones. A wider partial pressure range should be explored to ascertain the correct equilibrium law. For the present paper, this preliminary investigation gives valuable information for steering future studies. The regression parameters are reported in Table 5.  [22,40,[48][49][50][51][52][53][54]. As a general finding, data on PSA at 1 bar appear in the literature sparingly, even for traditional sorption media. Figure 6 reports a possible comparison of the hydrochar sorption capacity. It appears that, despite the different capture techniques, the data of "HCA_200_0_calc" at 1 bar (calculated by Equation (4)) are comparable with those of the literature. On the other hand, the comparison with PSA experiments conducted on zeolites, activated carbon, and fly ash shows that, at the same operating conditions, the HCA exhibits much better results in terms of CO2 capture capacity. The regressions are reported as solid lines in the explored range and as dotted lines in the extrapolated ones. A wider partial pressure range should be explored to ascertain the correct equilibrium law. For the present paper, this preliminary investigation gives valuable information for steering future studies.
The regression parameters are reported in Table 5.  Figures 6 and 7 compare the performances of the hydrochars to those of traditional solid materials and those of some innovative materials, as reported in the literature [22,40,[48][49][50][51][52][53][54]. As a general finding, data on PSA at 1 bar appear in the literature sparingly, even for traditional sorption media. Figure 6 reports a possible comparison of the hydrochar sorption capacity. It appears that, despite the different capture techniques, the data of "HCA_200_0_calc" at 1 bar (calculated by Equation (4)) are comparable with those of the literature. On the other hand, the comparison with PSA experiments conducted on zeolites, activated carbon, and fly ash shows that, at the same operating conditions, the HCA exhibits much better results in terms of CO 2 capture capacity.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 10 of 14 Figure 6. Comparison of CO2 sorbent capacities with literature data obtained with batch equilibrium method (BPL: commercial activated carbon [48]; C35N400: ammonia-treated activated carbon [49]; G-900: activated graphite fibers [50]; MFB-600: N-doped activated carbon [51]; RN-800: ammonia-treated activated carbon [52]; RFL-500: N-doped porous carbon [53]; DO-88-M: activated carbon from petroleum pitch [54]; AS-2-600: sawdust-based porous carbon [40]). Figure 7 shows that the HCA_200_0 has the best CO2 sorbent capacity: 6.569 mmol/g, threefold concerning the best performance of traditional sorbents [22]. Another significant result appears in Figure 7. Hydrochars prepared after 120 min of HTC reaction halve their performance, even though remaining well above commercial zeolites and similar materials. This suggests investing in further research aimed to ascertain if a reaction time exists, which maximizes the sorption capacity. This more-in-depth investigation is of the utmost importance for industrial-scale process optimization.  Comparison of CO 2 sorbent capacities with literature data obtained with the dynamic experimental method (pressure swing adsorption (PSA)) (A1, A2, A3, A4, B1, B2: zeolites from fly ash; SG: commercial silica gel; AC: commercial activated carbon; 13X: commercial zeolite [22]). Figure 7 shows that the HCA_200_0 has the best CO 2 sorbent capacity: 6.569 mmol/g, threefold concerning the best performance of traditional sorbents [22]. Another significant result appears in Figure 7. Hydrochars prepared after 120 min of HTC reaction halve their performance, even though remaining well above commercial zeolites and similar materials. This suggests investing in further research aimed to ascertain if a reaction time exists, which maximizes the sorption capacity. This more-in-depth investigation is of the utmost importance for industrial-scale process optimization.
All of the results here reported highlight the concrete possibility of exploiting the residual biomass as an adsorption medium for biogas upgrading and encourages continuing research in this way.

Conclusions
Hydrochar from lignocellulosic residual biomass furnishes porous carbon materials via usual activation. The best reaction conditions were 200 • C and 0 min. The best results were BET area of 881 m 2 /g, average pore diameter of 11 Å, and sorbent capacity of 6.569 mmol/g at 5 bar. The hydrochar exhibits higher CO 2 adsorption than that of some traditional sorbents. The material possesses adequate capacity to adsorb CO 2 from mixtures with CH 4 selectively. This residual material could be exploited in a sustainable biogas upgrading process, contextually reducing CO 2 emissions and the related environmental impact.