3D-printed activated carbon for post-combustion CO 2 capture

The applicability of 3D-printed activated carbons for their use to CO 2 capture in post-combustion streams and the influence of activation conditions on CO 2 uptake and CO 2 to N 2 selectivity were studied. For two monoliths with the same open cellular foam geometry but low and high burnoff during activation, a series of fixed-bed break-through adsorption experiments under typical post-combustion conditions, in a wide range of temperature (313 and 373 K), and partial pressure of CO 2 up to 120 kPa were carried out. It is shown that the higher burnoff during activation of the 3D printed carbon enhances the adsorption capacity of CO 2 and N 2 due to the increased specific surface area with sorption uptakes that can reach 3.17 mol/kg at 313 K and 120 kPa. Nevertheless, the lower burnoff time on monolith 1 leads to higher selectivity of CO 2 over N 2 , up to 18 against 10 on monolith 2, considering a binary interaction to a mixture of CO 2 /N 2 (15/85 vol%) at 313 K. The single and multicomponent adsorption equilibrium is conveniently described through the dual-site Langmuir isotherm model, while the breakthrough curves simulated using a dynamic fixed-bed adsorption linear driving force model. Working capacities for the 3D printed carbon with lower burnoff time lead to the best results, varying of 0.15 – 1.1 mol/kg for the regeneration temperature 300 – 390 K. Finally, consecutive adsorption-desorption experiments show excellent stability and regenerability for both 3D printed activated carbon monoliths and the whole study underpins the high potential of these materials for CO 2 capture in post-combustion streams.


Introduction
CO 2 capture from post-combustion streams in coal-fired power plants can play an important role to mitigate global warming [1].This carbon source is responsible at least for 73% of annual energy sector emissions [2].Therefore, to achieve the International Panel on Climate Change (IPCC) target by keeping the global warming average less than 2 • C, great efforts have been made towards the decrease of penalty energy and increase CO 2 recovery on actual capture technologies [1].
The chemical absorption technology with monoethanolamine still is the most widely used for CO 2 capture [1].However, its regeneration step requires high energy consumption which makes the process expensive [3].Among other technologies, namely cryogenic [4] and membrane [5], adsorption processes are recognized as an attractive and costly-efficient alternative for CO 2 capture on post-combustion applications [1,6,7].Additionally, the availability of several solid-based materials and many regeneration steps (temperature, pressure, vacuum, and electrical swing adsorption) makes the optimization of the adsorption processes to ensure acceptable footprint and cost versatile.
Commonly solid-based materials used to adsorb CO 2 are zeolites (13X, 5A, and 4A) [8][9][10], activated carbon [11], and metal-organic frameworks [12].These materials are synthesized into a powder form and transformed into pellets or beads before being applied to adsorption systems.However, when using pellets or beads on a system that needs high gas velocities, e.g.systems with short cycle times, the process tends to suffer from higher pressure-drop [13].Besides, package-bed with pellets or beads exhibit an inhomogeneous temperature along the bed due to low electrical conductivity, as well as, barriers to heat transfer caused by contact between particles [14,15].The resulting maldistribution of temperature is a disadvantage on cycles that work with regeneration step based on temperatures, like Temperature or Electrical Swing Adsorption (TSA and ESA).
Alternatively to conventional beads/pellets configurations, novel adsorbent structures have been developed to overcome the main issues related to packaging beds [3,14,[16][17][18][19][20][21].Monolith presents an entirely non-particulate structure and open channels that can reduce the pressure drop on the system over higher flowrates [22].Additionally, the temperature distribution in a monolith is nearly uniform due to the ease heat dissipation of throughout its structure, as there is no thermal resistance between the particles [14,15].Among other advantages, the monolith can be designed with a high level of freedom to achieve better performance towards CO 2 capture.For example, Zhao et al. [23] synthesized a hybrid activated carbon (from phenolic resin) and zeolite NaUSY monolith with excellent electrical properties for ESA applications.They have shown a great reduction (from 73% to 18%) in energy loss in the contact between the hybrid adsorbent surface and the copper electrode.The reduction of energy loss allows a quick increase of temperature in the regeneration step by heating the adsorbent and, hence, operational cost reduction.Regufe et al. [24] have reported a 3D-printed honeycomb monolith composed of zeolite 13X, activated carbon, and binder especially design for the application in the ESA process.The material showed a fast increase in temperature, e.g. it achieved 377 K in 180 s from room temperature.A short cyclic adsorption process submitted at humidity conditions was evaluated by Verougstraete et al. [25] using a new honeycomb carbon monolith for CO 2 capture from flue gases.It was shown that the monolith can CO 2 capture at very high gas flow rates (up to 5 LPM which corresponds to an interstitial gas velocity equal to 3.23 m/s) and the presence of humidity did not strongly affect the capacity of adsorption as compared to 13X zeolite.
These results suggest that monolith has the potential to be a promising CO 2 capture adsorbent, counting on synergies between fast energy dissipation and low pressure-drop that reduce energy penalty on cyclic adsorption processes.Accordingly, the present work seeks to evaluate the applicability of a recently presented new route towards 3D-printed activated carbon combining micropores from activation and meso-and macropores from porogen templating [14], for the applicability in CO 2 capture from the post-combustion stream.Therefore, two monoliths with the same open cellular 3D printed structure and porogen templated meso-and macroporosity but varying microporosity due to differing burnoffs in the activation steps were employed in a series of fixed-bed adsorption experiments.Within these experiments, the adsorption equilibrium and the dynamic separation between CO 2 and N 2 at the temperature range between 313 and 373 K and partial pressure up to 120 kPa were studied from the respective breakthrough curves.Performance metrics such as selectivities and working capacities were also evaluated.A mathematical fixed-bed adsorption model was used to simulate the experimental data and to evaluate process performance parameters of both 3D printed activated carbons for the design of continuous cycling temperature swing adsorption process (TSA).

Materials
In this work, two CO 2 -activated carbon monoliths (M1 and M2) with a tetragonal unit cell structure and a diameter of 0.008 m were studied.Two pieces of each monolith were placed in front of each other to obtain a total length of 0.084 m.The synthesis of 3D-printed activated carbon monoliths was carried out according to the method described by Steldinger et al. [14].Briefly, a resin mixture of two monomers, 35% pentaerythritol tetraacrylate (PETA) and 35% divinylbenzene (DVB), and a porogen 30% bis(2-ethylhexyl) phthalate (DOctP) was prepared.Then, 0.4 mg mL − 1 sudan1 (dye) and 10 mg mL − 1 phenylbis(2,4, 6-trimethylbenzoyl) phosphine oxide (BAPO; initiator) were added before the porous polymer open cell structure was produced by stereolithographic 3D printing.After that, the porogen phase and the dye were extracted by soxhlet extraction using acetone.In the final step, the resultant polymer open cell structure was stabilized by a thermal treatment in the air (at 573 K for 6 h), pyrolyzed in nitrogen (at 1173 K for 0.3 h) and activated in CO 2 (at 1133 K).The only difference between monoliths M1 and M2 is the activation time, being 6 h for M1 and 12 h for M2.The adsorbates and inert gases were supplied by Air Liquide with the following purities: He (99.9998%),CO 2 (99.998%), and N (99.999%).

Characterization
The textural properties were determined using physisorption analysis with nitrogen at 77.4 K.The measurements were performed employing a BELSORP-maxII apparatus (Microtrac Retsch GmbH).Beforehand, the samples were degassed at 250 • C for 4 h under fine vacuum.The characteristic surface area was calculated applying the Brunauer-Emmett-Teller (BET) multi-point method considering five appropriate data points.Pore size distributions, the QSDFT surface area and the pore volume were deduced by applying the QSDFT adsorption model for carbon with a combined slit-and cylindrical pore geometry (fitting errors <0.2%, VersaWin Vers.1.0).Mercury porosimetry was performed using a combined Belpore LP/HP (Microtrac Retsch GmbH) in the pressure range from 0.013 to 400 MPa.The samples were outgassed at 200 • C overnight at atmospheric pressure.Pore size distributions were calculated based on the intrusion data from 1 to 400 MPa using a cylindrical pore model after cut-off of inter particle voids at pressures below 1 MPa and compressibility correction.

Breakthrough apparatus and experimental procedure
The single-component adsorption of CO 2 and N 2 on monoliths M1 and M2 were studied in a fixed-bed apparatus [10].The apparatus consists of three sections, namely the gas preparation section, adsorption section, and analytical section.In the gas preparation section, CO or N 2 partial pressure is balanced with inert helium (He) up to 120 kPa to be introduced in the adsorption column.In the adsorption section, an adsorption column (stainless steel) dimension of 0.01 m internal diameter and 0.1 m length is located inside the oven of SRI 8610C (Gas Chromatograph equipment -USA) with controlled temperature.After measurement of partial pressure (CO 2 or N 2 balanced with inert He) the gas is sent to the adsorption column into a given constant temperature (313, 343, or 373 K).The outlet gas goes directly to the Thermal Conductivity Detector (TCD), which will evaluate the specie conductivity.Helium is used as a reference gas in the TCD analyses.
To maintain the material fixed inside the column the whole monolith is covered with glass wool and the void volume inside the column is filled with glass spheres.The monolith diameter is somewhat lower than the column inner diameter, being necessary a thin layer of glass wool to fix the monolith in the column.Before the first run, the adsorption column is activated for 12 h at 475 K under pure He flows (10 NmL/ min).Between the runs, the column is heated up to 455 K and cooled up to the experiment temperature (±5 K/min) under pure He flows (20 NmL/min) to keep the column clean for the next run.
The equilibrium loading is obtained by integrating the molar flow profiles of the breakthrough curves by the following equation, where m ads is the adsorbent mass, F fi is the feed molar flow rate of component i at the inlet of the bed, F i is the molar flow rate of component i at the outlet of the bed, t ∞ is the saturation time, ε b is the bed porosity, V c is the column adsorption volume, and C 0i is the feed gasphase concentration at the inlet of the fixed bed.
The apparatus is designed to perform single or multi-component experiments, thus the dynamic separation of CO 2 and N 2 is also studied by adsorption-desorption experiments.In the gas preparation section, two synthetic gas mixture, namely (i) CO 2 /N 2 (15/85 vol%), and (ii) CO 2 /N 2 (50/50 vol%), at 100 kPa was set to goes through the adsorption column at three temperatures 313, 343, and 373 K. Before each experiment, the adsorption column is saturated with pure N 2 at 100 kPa.During the experiment, the outlet gas goes to a VICI Valco 6way valve, which in determined time change its position to send a sample, trapped in the loop (1 mL), to a chromatography column (HayeSep N packed column, 24 g, and mesh 80/100) at 373 K before goes to TCD to be analyzed.The chromatographic peaks are quantified via a conductivity measurement (TCD), and the final loading is calculated by Eq. ( 1).After the adsorption step is finished, the desorption step is started by closing the flow rate of CO 2 .The flow rate of N 2 on the desorption step is equal to the adsorption step.The outlet gas is analyzed with the same procedure described above until the column is completely cleared.
To evaluate the regenerability of adsorbent materials a set of six adsorption-desorption experiments was carried out.The adsorption experiments start at the same condition, namely a synthetic mixture of 15% CO 2 and 85% N 2 at 100 kPa and 313 K.At the end of the adsorption step, the column is completely cleared under constant N 2 flow and heating, to ensure that there is no permanent CO 2 in the column to the next adsorption cycle.

Adsorption equilibrium model
To describe the adsorption equilibrium behavior of CO 2 and N 2 on the monoliths, the dual-site Langmuir model (DSL) was used to fit the experimental data.This model (Eq.( 2)) is commonly used for adsorbents where a certain degree of energy surface heterogeneity interaction adsorbate-adsorbent is expected [26], where q m1,i and q m2,i are the maximum adsorbed concentration of component i in each site, b 1i and b 2i are the adsorption affinity constant of component i in each site.
The effect of temperature on the adsorption affinity constant, b i is taken into account by the van't Hoff expression, where b ∞,i is the pre-exponential factor of the affinity constant at infinite temperature, R is the ideal gas constant, (− ΔH i ) is the heat of adsorption, and T is the temperature.
The extended DSL model isotherm was used to predict the interaction of binary mixture CO 2 /N 2 systems by the following equation,

Performance metrics
Two important performance metrics used to rank adsorbents are the selectivity and working capacity.Although these performance metrics do not take into account mass and heat transfer effects, cycle configurations, and pressure drop effects, as addressed by Rajagopalan et al. [7], they still are a simple way to have a first impression of the adsorbent behavior.
The selectivity (S) of CO 2 over N 2 can be evaluated by Eq. ( 5) [7,27], where q CO2 * and q N2 * are the equilibrium loading of CO 2 and N 2 , respectively; y CO2 and y N2 are the molar fraction of CO 2 and N 2 , respectively.Eq. ( 5) can be used to predict pure selectivity (S P ) or competitivity selectivity (S C ), in which the pure or competitive loading is calculated from Eqs. (2) or (4).
The working capacity (WC) can be estimated according to the isotherm of CO 2 in two different conditions (high and low temperature for the case of TSA or high and low pressure for the case of Pressure Swing Adsorption (PSA)) by the following equation [7,27,28], where q CO2 (T 1 , P 1 ) * is the CO 2 equilibrium loading at the temperature T 1 and CO 2 partial pressure P 1 in the adsorption step, and q CO2 (T 2 , P 2 ) * is the CO 2 equilibrium loading at the temperature T 2 and CO 2 partial pressure P 2 in the desorption step.

Mathematical modeling
The dynamic behavior of adsorption processes can be numerically predicted according to the mass and energy conservation laws.The mathematical model employed for the simulation of the breakthrough curves in this work includes both the effect of axial dispersion and mass transfer resistances.The sorbates adsorption kinetics is taken into account by the linear rate driving force model (LDF) [29][30][31].The set of coupled partial and algebraic differential equations that represent the fixed bed adsorption model are summarized in Table 1 and the respective initial and boundary conditions are summarized in Table S3 in Supporting Information (SI).
For the solution of the model, the set of coupled partial and algebraic differential equations are reduced into a system of ordinary and algebraic differential equations by the method of the lines [32].Orthogonal Collocation method was applied to discretize the spatial coordinate [33].The computation of the collocation points was determined by the position on the spatial coordinate using Jacobi polynomial, P N (α,β) (x), with α = 0 and β = 0.The approximation of the first and second derivatives was made by collocation matrices routines A i,j and B i,j ,

Table 1
Mathematical model equations for fixed-bed adsorption simulation.

Phenomenon model Equations
ideal gas C = P RT (7) overall mass balance mass transfer rate L.F.A.S. Zafanelli et al. respectively.Thus, the system of ordinary differential equations has been solved using ode15s, a stiff ODE solver available in the MATLAB library [34], and the algebraic differential equations were solved by Gauss elimination.Twenty-five special collocation points appeared to give satisfactory accuracy and stability to the numerical solution.The solution has been implemented on a personal computer.

Adsorbent properties and characterization
The synthesis method of the 3D-printed activated carbons allows to control the structure ranging from the mm to the nm scale, which was characterized by nitrogen physisorption analysis and Hg porosimetry.The results are summarized in Fig. 1 and Table 2.As indicated by Fig. 1a for a piece of monolith M1, the macroscopic size and shape of the activated carbon is determined by the computer designed structure (Fig. S1), while on the nm level, the porosity can be varied via the CO 2 activation conditions.By increasing the activation time at a temperature of 1133 K from 6 to 12 h, the burnoff almost linearly scales from 14 wt% to 26 wt%.The partial oxidation of the monoliths leads to a distinct rise of the nitrogen uptake for monolith M2 (12 h activated monolith) in the low-pressure region of the adsorption branch (p/p 0 < 0.1) compared to M1 (6 h activated monolith), indicating the creation of additional microporosity (Fig. 1b).The prolonged activation equally increases the specific surface area and pore volume for monolith M2 (QSDFT: 1307 m 2 g − 1 ; 0.52 cm 3 g − 1 ) compared to M1 (QSDFT: 1048 m 2 g − 1 ; 0.39 cm 3 g − 1 ) by ca.30%.
According to the pore size distributions shown in Fig. 1c, the largest gain can be attributed to micropores with a diameter less than ca.1.5 nm.In accordance, the Hg porosimetry data reveal a similar pore size range but a ca.20% higher intruded volume for M2 (0.18 cm 3 g − 1 ) compared to M1 (0.15 cm 3 g − 1 ).In other words, the CO 2 activation can be utilized to introduce additional micropores without leading excessive altering of the meso-to macroporosity.Apart from this, it should be mentioned that both activated monoliths were stable and could be normally handled in the lab.

Adsorption equilibrium
To set-up the adsorption isotherms of CO 2 and N 2 in the 3D-printed activated carbons a series of breakthrough curves on the respective    2a that the moles fed to reach the breakthrough of CO 2 are much higher than in the case of N 2 (Fig. 2b) on monolith M1.Consequently, the equilibrium loading for CO 2 and N 2 is 1.74 and 0.23 mol/ kg, respectively, which means that the sorption uptake of CO 2 is almost 7.56 times higher in CO 2 than in N 2 (Run 1.3).The same trend was observed on monolith M2 (Fig. 2c and d) and also at the other experimental conditions.Moreover, as the partial pressure increases the moles fed to reach breakthrough increases which indicate that data is thermodynamically consistent (the loadings of CO 2 and N 2 in both monoliths were obtained by applying Eq. ( 1)).Fig. 3 shows the experimental (symbols) and theoretical (lines) isotherms measured for CO 2 and N 2 on monoliths M1 (closed symbols) and M2 (open symbols) at (a) 313, (b) 343, and (c) 373 K from the breakthrough curves.As can be seen from Fig. 3a, CO 2 loading on monolith M1 (3.0 mol/kg) is slightly lower than monolith M2 (3.18 mol/kg), especially at the higher partial pressure (>50 kPa) and lower temperature (313 K).For the other temperatures, the loading of CO 2 is practically the same on both monoliths (Fig. 3b and c).In the case of N 2 (Fig. 3a-c), the loading reached on monolith M2 is higher than in monolith M1 at all temperatures studied.Monoliths M1 and M2 were activated under a CO 2 atmosphere at 1133 K for 6 and 12 h, respectively.The specific surface area and pore volume increased with the increase of CO 2 activation time (Table 2).Additionally, the monolith narrow micropores are widened, as activation proceeds, being more accessible for N 2 which can justify the increased N 2 loading on monolith M2.In contrast, the widened monolith M2 narrow micropores have a negative effect to "trap" CO 2 at low partial pressure (as can be seen in Fig. 3a).This effect has been also observed by Querejeta et al. [22] who have studied CO 2 , N 2 , O 2 , and H 2 O on two carbon honeycomb monoliths with different burn-off degrees.
Table 3 shows the dual-and single-site Langmuir parameters calculated from the fitting of the adsorption equilibrium data for CO 2 and N 2 shown in Fig. 3, respectively.The lines in Fig. 3 clearly show that the isotherm model describes well the equilibrium data.The heat of adsorption for CO 2 varies between − 40 and − 20 kJ/mol on M1, and -24.7 to − 22.5 kJ/mol on M2.For N 2 , the heat of adsorption obtained on M1 is equal to − 16 kJ/mol, and on M2 is equal to − 13 kJ/mol.The heats of adsorption obtained in this work are within the range reported to similar materials, namely, AC monolith [22,25] and hybrid (AC/zeolite) [27] monolith.

Kinetics of sorption
For each experimental condition (feed composition, gas velocities, temperatures, and pressures), the values of the transport parameters were obtained by commonly used correlations as summarized in Table S4 in the SI.As an exception, the mass transfer coefficient (K LDF ) was obtained from a "trial-and-error" procedure by applying the linear driving force model (LDF) to the mass transfer rate equation to obtain the best fit of the breakthrough curves.The sum squares errors between the experimental breakthrough and the simulated breakthrough, ∑ t∞ i=0 (y exp (t i ) − y sim (t i )) 2 , was used as an objective function to optimize the    the single and binary experiments are summarized in Tables S5 and S6 (single), and S8 (binary) on the SI.In Fig. 4a and b, the dashed line indicates the CO 2 stoichiometric time of each run, being 16.44 min in monolith M1 (a) and 12.50 min on monolith M2 (b).The stoichiometric time is derived through an overall mass balance of an adsorption column in equilibrium with the adsorbate [29,35], where v i is the interstitial velocity, L is the column length, ε b is the bed porosity, ρ b is the bulk density, q 0 is the adsorbed concentration in equilibrium with the gas phase concentration c 0 at the column feed condition.
As can be seen in Fig. 4a and b, the experimental and simulated breakthrough curves cross the stoichiometric time at the same point, indicated by an arrow in Fig. 4a and b, which is around 50-60% of the breakthrough curves.This effect is expected since the resistance to mass transfer must respect the stoichiometric time.In other words, for values of K LDF , in which the simulator provides consistent results, the breakthrough curves must always pass through the stoichiometric point.Thus, when the simulator is submitted to different values of K LDF , the expected response is more dispersed curves for lower K LDF values and less dispersed curves for higher K LDF values.These effects are similar to the results presented by Lapidus and Amundson [36], which evaluate for the first time the effect of longitudinal diffusion in chromatography and ion exchange fixed-bed columns.
Concerning a comparison with other works, Zhao et al. have reported kinetic coefficients (K LDF ) of 0.100 s − 1 at 293 K for a carbon monolith [37], and 0.120 s − 1 at 298 K for a hybrid monolith (H-ZSM5/activated carbon) [27], for a feed gas mixture of 15% CO 2 balanced with N 2 , which are in the same range of the values reported in this work.

Binary adsorption
For the binary separation studies, it was considered two different ratio gas mixtures: (i) CO 2 /N 2 (15/85 vol%), and (ii) CO 2 /N 2 (50/50 vol %), at a constant total pressure of 100 kPa and three temperatures 313, 343, and 373 K.All the runs studied, started from a column saturated with pure N 2 at 100 kPa.The desorption of the fixed bed was also carried out with pure N 2 at the same total flow rate value used in the adsorption step.The experimental conditions are summarized in Table S7 in the SI.As can be seen in Fig. 5a, for the experiment with the mixture ratio CO 2 /N 2 (15/85 vol%), at 313 K the CO 2 breaks the column at around 10 min (black symbols) and reaches saturation at around 25 min.Interestingly, the breakthrough time of CO 2 in the bed decreases from 10 to 2 min, as the temperature increases from 313 K (Fig. 5a) to 373 K (Fig. 5e), which indicates a strong impact of temperature on the CO 2 sorption uptake.Regarding the desorption curves (Fig. 5b), the CO 2 concentration decreases slowly, and hence, the complete cleaning of the bed from CO 2 takes around 45 min.This effect is expected since the isotherm of CO 2 on monolith M1 is of favorable nature for adsorption (Type I is observed in the measurement range) being unfavorable for desorption (see Fig. 3).Thus, the CO 2 desorption curves concentration waves are dispersive which requires a long time for the regeneration of the column.As the temperature increases, the time to regenerate the bed decreases from 45 min at 313 K (Fig. 5b) to 12 min at 373 K (Fig. 5f), since the equilibrium loading of CO 2 is lower as the temperature increases.For the mixture ratio 50/50 (Fig. 6), the breakthrough curves become steeper due to the higher concentration of CO 2 in the feed leading to a higher compression of the concentration wave fronts.Regarding the adsorption and desorption of CO 2 on monolith M2 (Figs. 7 and 8), the same trends were observed.
Overall, the lines shown in Figs.5-8, which represent the simulation results, are in good agreement with the experimental data.The simulation parameters are summarized in Table S8.It is worth mentioning that the extended multi-component DSL model was used to predict the multicomponent adsorption equilibria on both monoliths.

Regenerability and performance metrics
To assess the monolith's feasibility to be renewed and reused over several adsorption/desorption cycles, six cycles were carried out for a mixture of CO 2 /N 2 (15/85 vol %) at 100 kPa and 313 K. Fig. S6 in the SI shows the loading of CO 2 in both monoliths is stable.It is worth mentioning that the initial condition from one experiment to another shows a tiny difference (see Table S9 on SI) which can explain the slight difference between the loadings for the same monolith.In general, both monoliths show excellent stability and regenerability over consecutive cycles performed in this work.
The CO 2 working capacity (Eq.( 6)) was calculated for several combinations of regeneration temperature and desorption pressure.The loadings of CO 2 were obtained from the extended DSL isotherm taking into account the competitive interaction with N 2 .The adsorption capacity, q * ads = q(293 K, 15kPa), was calculated at a fixed condition of 15% CO 2 and 85% N 2 at 293 K and 100 kPa.The desorption capacity, q * des = q(T 2 , P 2 ), was calculated taking into account a range of CO 2 pressure between 1 and 100 kPa and a range of temperature from 293 to 473 K. Fig. 9 shows the contours of the CO 2 working capacity for monoliths M1 and M2 as a function of different regeneration temperatures and pressures.As expected, the working capacity increase as desorption pressure decrease and regeneration temperature increase for both monoliths.For monolith M1 (Fig. 9a) keeping the desorption pressure at 15 kPa the working capacity increase from 0.15 to 0.90 mol/kg as regeneration temperature increase from 300 to 354 K (see the gray dashed lines in Fig. 9a).This result shows that the 3D-printed activated carbon can be explored in a temperature swing adsorption process due to a great increase in the WC with a small change in the regeneration temperature.In the case of monolith M2, considering the same conditions, the WC does not reach 0.75 mol/kg, as seen in Fig. 9b.This can be explained due to the N 2 being more adsorbed on M2 than M1, which results in a negative effect on the performance metrics of M2.As shown by Rajagopalan et al. [7], the affinity of N 2 can play a critical role in deciding the process performance for post-combustion CO 2 capture.In our study, N 2 has more affinity with monolith M2 due to a higher burnoff during activation, which increase the specific area and pore volume accessible to N 2 molecules and thus harms the performance.In contrast, the monolith M1 has a better performance WC as the affinity of N 2 is lower.Besides, two other advantages come from the lower activation time, namely, the reduction of the manufacturing cost and increase of the product yield.
The CO 2 selectivities over N 2 calculated from Eq. ( 5), considering the binary interaction, are 18 and 10 for monoliths M1 and M2, respectively.These values were calculated from isotherm parameters considering a mixture of CO 2 15% and N 2 85% at 100 kPa and 313 K.For similar materials, the selectivity values obtained in this work are within the range reported in the literature [22,38].

Conclusions
In this work, the potential of two 3D-printed activated carbons with different activation times (M1<M2) has been investigated for their use in CO 2 capture in post-combustion streams.A series of fixed-bed breakthrough adsorption experiments were performed in a wide range of temperature and pressure of interest for post-combustion applications, namely: between 313 and 373 K and the partial pressure up to 120 kPa.The adsorption equilibrium and heat of adsorption in both monoliths follow the order CO 2 ≫N 2 .The isotherm Langmuir model was used to predict the N 2 adsorption equilibrium and dual-site Langmuir model for the CO 2 , both being in good agreement with the experimental data.The extended multi-component dual-site Langmuir model was used and provided a good prediction of the competitive interaction between the compounds in multicomponent breakthrough experiments.It was shown that a higher burnoff rate during the activation results in a higher specific area and pore volume.As a result of the higher specific area and pore volume, it is expected higher adsorption capacity for both components.However, the adsorption capacity of CO 2 at higher temperatures (343 and 373 K) and higher pressures (>50 kPa) is similar in both monoliths which means that the increase in the specific surface area of monolith M2 is related to microporosity only accessible to N 2 molecules.Nevertheless, as the N 2 affinity increases, a negative effect is observed on CO 2 selectivity and working capacity on monolith M2.For example, considering a typical post-combustion condition, the selectivity of CO 2 over N 2 is 18 for M1 and 10 for M2.In the case of CO 2 working capacities, the monolith M1 has a better performance than M2 at the same range of regeneration temperature.Both materials presented excellent stability and regenerability over consecutive adsorption-desorption experiments.The mathematical model used in this work describes well the adsorption and desorption breakthrough curves data.Finally, the present work indicates that the 3D-printed activated carbon can be used in a temperature swing adsorption process for CO 2 capture.Bulk density (kg m − 3 ) θ = q/q m ) gas-phase energy balance − Kax ∂ 2 T ∂z 2 + Fcpg ∂T ∂z + Cε b Cpg ∂T ∂t + (1 − ε b )aphp(T − Ts) + achw(T − Tw) = 0 (11) solid-phase energy balance Cps ∂Ts ∂t = aphp(T − Ts) + ρ p ∑ n i=1 ( − ΔHst,i)

Fig. 1 .
Fig. 1.A) Single piece of monolith M1; Low-temperature N 2 physisorption analysis: B) ad-(closed symbols) and desorption (open symbols) isotherm for M1 and M2 in linear and log-scale (inset); C) Cumulative pore volume derived from the nitrogen sorption data assuming a combined slit-and cylindrical pore model (QSDFT adsorption kernel); D) Cumulative and relative intruded Hg volume.
monoliths M1 and M2 were measured at three temperatures, 313, 343, and 373 K, and partial pressures up to 120 kPa.The experimental conditions are summarized in TablesS1 and S2in the Supporting Information (SI).Fig.2a-dshows the observed experimental breakthrough curves (symbols) at 313 K plotted in terms of normalized molar fraction y i /y i0 against moles fed per unit mass of monolith M1, for components CO 2 (a), and N 2 (b), and of monolith M2 for components CO 2 (c), and N 2 (d).All the breakthrough curves observed for CO 2 and N 2 on monoliths M1 and M2 at 343 and 373 K are shown in Figs.S2-S5 of the SI.It is clear from Fig.

a
The reference temperature used is 313 K.

Fig. 4 .
Fig. 4. Kinetic coefficient effects on numerical breakthrough curves of CO 2 on (a) M1, and (b) M2 at 313 K. Dashed lines indicate the CO 2 stoichiometric time; Experimental = symbols; Numerical = lines.Sum squares of error were shown in brackets.

Figs. 5
and 6 show the adsorption and desorption breakthrough curves on monolith M1 and Figs.7 and 8 on monolith M2.
Intracrystalline diffusivity (m 2 s − 1 ) F Total molar flux (mol m − 2 s − 1 ) h p Film heat transfer coefficient (W m − 2 K − 1 ) h w Wall heat transfer coefficient (W m − 2 K − 1 ) K LDF Linear Driving Force coefficient (s − 1 ) K ax Effective axial bed thermal conductivity (W m − 1 K − 1 ) K Dimensionless Henry's law equilibrium constant (− ) k ′ Overall effective rate coefficient (s − 1 ) k f Film mass transfer coefficient (m s − 1 ) L Length of column (m) M i Molecular mass of component i (kg kmol − 1 ) m ads Mass of adsorbent (kg) p i Partial pressure of component i (kPa) P Total pressure of column (kPa) P em Peclet number (− ) q i Adsorbed phase concentration of component i (mol kg − 1 ) q i Average adsorbed phase concentration of component i (mol kg − 1 ) q * Equilibrium adsorbed concentration of component (mol kg − 1 ) q m Maximum adsorbed phase concentration (mol kg − 1 ) r c Crystal radius (m) R p Particle radius (m) R Universal gas constant (J mol − 1 K − 1 ) S ij Selectivity of component i over j (− ) t Time (s) t st Stoichiometric time (s) T Temperature in bulk gas phase (K) T s Temperature in solid phase (K) T w Column wall temperature (K) v i Interstitial velocity (m s − 1 ) y i Molar fraction of component i (− ) z Axial coordinate in bed (m) Greek Letters ΔH i Heat adsorption of species i (J mol − 1 ) ΔH st Isosteric heat adsorption (J mol − 1 ) Apparent adsorbent density (kg m − 3 ) ρ b

Table 3
Adsorption equilibrium model parameters for sorption of CO 2 and N 2 on monoliths M1 and M2.