Chemical Engineering Journal 223 (2013) 795–805
Fluid bed adsorption of carbon dioxide on immobilized polyethylenimine (PEI): kinetic analysis and breakthrough behavior Esmail R. Monazam1,2, James Spenik1,2, and Lawrence J. Shadle1,∗ 1
National Energy Technology Laboratory U. S. Department of Energy 3610 Collins Ferry Rd. Morgantown, West Virginia 26507-0880 2
REM Engineering Services, PLLC 3537 Collins Ferry Rd. Morgantown, West Virginia 26505
Abstract The adsorption of carbon dioxide (CO2) by immobilized polyethylenimine (PEI) on mesoporous silica was investigated in a fluid bed. The tests were performed to determine breakthrough behavior with varying bed temperature, flow rates and feed concentrations. Experimental breakthrough curves were analyzed using a theoretical 1D model developed by Bohart and Adams. The results showed that Bohart-Adams model was suitable for the normal description of breakthrough curve for the temperature ranges of 40-90oC. The maximum capacity increased with temperature up to 70 C and then decreased. The adsorption rate constant exhibited a negative temperature dependence decreasing as the temperature increased. Parameters characteristic of a fluid bed adsorber were inferred from these breakthrough curves including the breakthrough time, saturation time, critical reactor length, and length of mass transfer zone LMTZ. These parameters can be used to design fluid bed adsorption system without resolving the mechanistic contributions of dispersion, mixing, and intraparticle diffusion. 1. Introduction Carbon dioxide (CO2) is an omnipresent species in the flue gas of fossil fuel combustion that is thought to be a main contributor to the global warming. Fossil fuel-fired power plants produce and release CO2 and account for one-third of the global CO2 emissions to the atmosphere [1]. This has resulted in a global research effort to mitigate the emissions of CO2 and examine ways to capture CO2 directly from power plant flue gases [2]. Adsorption of gases onto solid adsorbents has great environmental significance since it can effectively remove pollutants from gaseous streams. Low-temperature solid adsorbents reported in the literature [3, 4] include physical adsorbents (carbon-based materials [5–9], zeolites [10–16] and metal organic frameworks [17–21]), amine-loaded carbons [22–26] and amine-loaded silica’s [27–36]. ∗
Corresponding author: Tel. 304-285-4647; fax – 304-285-4403; email –
[email protected]
Chemical Engineering Journal 223 (2013) 795–805
For the purpose of reducing CO2 emissions to the atmosphere, researchers at The Department of Energy’s National Energy Technology Laboratory (NETL) have developed a high adsorption capacity sorbent by immobilizing amine functional group(s), such as those found in polyethylenimine (PEI). These are grafted on high surface area supports such as mesoporous silicas [37]. This high surface area support can immobilize a large number of active amine sites per gram of support. The nature of the amine functional groups determine the amount of CO2 adsorbed and the energy required for regenerating the sorbent. However, there are few publications in the literature relating to the adsorption of gaseous pollutants onto immobilizing amine / mesoporous silica support, particularly using fluid bed configurations [38]. Fluid-bed experiments are usually done to measure the effectiveness of adsorption for removing specific adsorbates as well as to determine the maximum adsorption capacity. In this adsorption mode, there is a fluid bed of adsorbents through which the stream under treatment is passed. As the polluted stream travels through the bed, adsorption of the contaminants occurs and a purified effluent exits the column until breakthrough, which is when the outlet CO2 concentration reaches a quality limit set by emission regulations. The time for breakthrough and the shape of the breakthrough curve are very important characteristics for determining the operation and the dynamic response of an adsorption column. During breakthrough CO2 is removed from flue gas by the sorbent. The adsorption performance can be characterized by measuring the inlet CO2 concentration, outlet CO2 concentration and determining the adsorbed CO2 concentration from the ratio of effluent CO2 concentration to inlet CO2 concentration as a function of time for a given bed height. In this present work, a systematic study on the adsorption of CO2 onto the immobilizing amine /mesoporous silica support in a fluid bed was carried out to provide a deeper insight into the physics of the dynamic adsorption processes. 2. Experimental Methods 2.1 Sample preparation and characterization The sorbent used, designated sorbent AX, was manufactured by Pressure Chemicals under DOE specifications of nominal 40% PEI (BASF, Lupasol PR8515) on a silica substrate (PQ Inc. 2129). Particle density was 0.88 g/cc as measured from the bulk density and the particle size. The particle size was measured using QICPIC particle analyzer (Sympatec GmbH, Model-QP0104). The size distribution for the raw sorbent was Gaussian as displayed in Figure 1. The cumulative volume distribution at 10 and 90 % were 114 and 188 μm, respectively. The Sauter mean size was 142 μm. The batch of prepared sorbent came in six 35 kg barrels from testing conducted at ADA Inc. and exhibited some variability such that the cumulative distribution at 10 and 90% ranged from 53 up to 188 μm, respectively. The active ingredient in the sorbent is PEI which is a polymer of 1, 2-ethanediamine and aziridine with a number average molecular weight of 2000. It contains a mixture of primary, secondary and tertiary amines, and in dry conditions the main reaction between amine and CO2 is the formation of carbamate. These are grafted on the high surface mesoporous silica which was measured prior to loading PEI to have 312 m2/g using N2
Chemical Engineering Journal 223 (2013) 795–805
via BET method and a porosity of 2.49 ml/g using isopropanol (IPA). This high surface area support immobilizes a large number of active amine sites per gram of support. The nature of the amine functional groups determines the amount of CO2 adsorbed and the energy required for regenerating the sorbent [39, 40]. The amines were dissolved in methanol and combined with the silica substrate at a 0.40 amine: 2.0 methanol: 1.0 silica weight ratio. This substrate slurry was then placed in a rotary evaporator, and the methanol was removed, physically adsorbing the amine into the silica. A nominal total of 40% PEI by weight was immobilized onto the substrate, and then the CO2 capture sorbent was tested using a thermogravimetric analyzer (TGA). This impregnation procedure was reported to produce a well-distributed layer of PEI on mesoporous silica substrate [30]. Xu et al. describe a synergistic influence of the PEI with the mesoporous substrate such that the adsorption capacity in the “molecular basket” increases over that observed for either individually when PEI loadings exceed 20% by weight [30]. The synergistic effect maximizes at 50% PEI loading and this is described as the optimal loading of PEI on the mesoporous surface. The PEI in the sorbent, Lupasol PR8515, was approximately 25, 50, and 25% wt. primary, secondary, and tertiary amine, respectively. The stoichiometry of CO2 uptake reactions is 2:1 for primary and secondary amines, but is more complex with tertiary amines. In the presence of water the tertiary amine reacts with CO2 producing the tertiary ammonium bicarbonate, but in dry environment the tertiary amine reacts in concert with a primary or secondary amine. The theoretical reaction order for CO2 can be calculated to be 0.625 for the PEI mixture. It was found to be 0.69 in TGA testing in a dry environment [41]. An SEM photomicrograph of the sorbent after testing at elevated temperatures up to 120 °C is shown in Figure 2. These used sorbent particles have some micro-cracks eventually resulting in particle attrition and fragmentation. The bulk density was measured to be 0.63 g/cc and the packed bed voidage was calculated to be 30%. The aspect ratio for the fresh sorbent ranged from 0.6 to 0.75. The aspect ratio is defined as the ratio of the minimum to maximum FERET diameter and extracted from the QICPIC particle analyzer.
2.2 Experimental procedure and apparatus A schematic diagram for the fluid-bed column systems is shown in Figure 3. The fluidbed columns were made of polyethylene tubes 14 cm internal diameter and 101 cm in height. Thermocouples were located at three levels within the bed, (bottom – 6 cm above distributor, middle – 18 cm, top – 36 cm). An average of these three thermocouples was taken as the bed temperature. Standard conditions were taken as 20 °C and 1 atmosphere. An amount of sorbent (2.4 kg) was introduced to the reactor to yield a 25.4 cm deep packed bed. Gas flows to the plenum were maintained using Alicat mass flow controllers and an additional flow of air was supplied to the exiting gas to dilute the effluent
Chemical Engineering Journal 223 (2013) 795–805
sufficiently to maintain CO2 concentration below 10% as necessary. The CO2 exit concentration was monitored using a PP Systems WMA-4 CO2 analyzer (Range – 0 – 10,000 ppm). Bed temperatures were maintained using an oil flow through coiled heat exchange tubes immersed in the bed. The heat transfer oil could be either heated or cooled to maintain the desired bed temperature. Data regarding gas flows, temperatures, CO2 concentrations and other pertinent conditions were acquired at one second intervals throughout the duration of the experiments. The sorbent was initially pre-treated by heating to 110°C under a fluidizing gas flow of N2 with 2% humidification (by volume) until the CO2 exit concentration was negligible. To begin the adsorption phase, the bed temperature was reduced to the desired temperature under a flow of humidified nitrogen. The reactor exit was open to ambient conditions. When the bed was at the desired thermal condition, gas flows were adjusted to the desired overall inlet flow rate and CO2 concentration. During the adsorption phase, effort was made to maintain the bed at a constant temperature. This condition was maintained until the CO2 outlet concentration was maximum and invariant. After maximum adsorption, the bed was transitioned to the desorption phase by raising the bed temperature to the regeneration temperature of 110°C. The gas flows employed during this period were usually identical to the adsorption phase flows. Once the 110°C temperature was reached, CO2 flow was discontinued and the CO2 outlet concentration was monitored until negligible. The bed was again cooled under humidified nitrogen flow to the desired temperature of the next adsorption condition. The selection of the regeneration temperature of 110 °C was a compromise between sorbent stability and expected working capacity. This sorbent exhibited thermal and chemical instabilities such that elevated temperatures, particularly in the presence of O2 and the absence of moisture, can lead to the decomposition, oxidation, desiccation, and breakdown of the active surface sites in PEI [39]. Tests were conducted to evaluate the severity of these effects by cycling the sorbent through adsorption and regeneration in the presence of oxygen and the absence of moisture. In one case the sorbent was visibly discolored, producing yellow, red, and black particles when heat exchange fluid temperatures reached 140 °C, corresponding to bed temperatures of 130 °C. Subsequent tests on this same sorbent demonstrated reduced capacity for CO2. The capacity of a similar sorbent was sufficiently reduced at temperatures above 100 °C to provide working capacities of about 1 mole/kg [41]. The working capacity is defined as the difference between capacities under adsorption and regeneration conditions. Adsorption temperatures were varied between expected flue gas temperatures of 40 °C up to regeneration temperatures to evaluate the influence of temperatures on both capacity and rates of CO2 uptake. The capacity of PEI sorbent is a relatively insensitive to temperature over this range, but the rate of adsorption decreases with increasing temperature [41]. There are advantages to maximizing the adsorber temperature to reduce the process heat requirements. The CO2 concentrations were tested at nominally 15 and 30 vol% based upon the fact that a 500 MW power plant produces between 14 and 21% CO2 in the flue gas [42, 43]. The
Chemical Engineering Journal 223 (2013) 795–805
higher level was selected to evaluate the process sensitivity to a change in the CO2 concentration. The experiment instrumental control and accuracy were better at higher rather than going to lower concentrations. Typical temperature profiles across the fluidized sorbent bed during adsorption are presented in Figure 4 when fluidization velocity was 20 times minimum fluidization velocity. Thermal deviations during CO2 adsorption elevated the temperature above setpoint, but these deviations peaked prior to breakthrough and became small as the reaction front approached the outlet. The adsorption temperature was found to increase by 15 to 20 °C at 33.3% CO2, but was less at 16.7% CO2 concentration. The heat release extended over a longer time period for the lower CO2 concentration thereby reducing the deviation from nominal temperature. By the time the CO2 concentrations began to breakthrough the sorbent bed the thermal deviation returned to the nominal operating temperature. A thermal swing process, such as being investigated here, is more efficient if the temperature difference is relatively small between the adsorber and regenerator. Thus, there is an advantage to being able to operate the adsorber at elevated temperatures. The lack of temperature dependence in this type of sorbent [30, 39, 41] allow the temperature of the sorbent to increase during CO2 uptake without sacrificing capacity. Thus, there is a smaller temperature swing required and the heat required during regeneration can be minimized. From these experiments the heat transfer coefficients and heats of reaction were extracted by conducting an energy balance on each experiment considering the heat absorbed from the heat exchange fluid; sensible heat lost to fluidizing gases, through the walls to the environment; and the sensible heat of the reactor walls, copper heat transfer coils, and sorbent bed. The heat of reaction is the difference between the heat adsorbed and the sum of all the sensible heat losses. The heat transfer coefficient is the total heat adsorbed over the duration of the experiment, Q, divided by the tube surface area, At, times difference between the average oil temperature, Toil , and the average bed temperature, Tbed . Q hc = At (Toil − Tbed ) The values for the heat of reaction and heat transfer coefficients are presented in Table 1 for the test matrix. The heat transfer coefficient did not vary appreciably with temperature or CO2 concentration and was found to be about 139 ± 41 W/m2-K. The heat transfer coefficient was lower when the fluidizing gas velocity was decreased from 20 to 10 · umf, where hc=87.7±8.2 W/m2-K. This was due to the fact that the test period was longer for the lower flow rates. These values are consistent with following correlations for horizontal tubes in a fluidized bed developed by Vreedenburg in 1958 [44]:
ρ µ g2 hc Dt = 420 s Prg ρg kg gρ s2 d 3p
0.3
Re 0D.3
for
ρs Re p ≥ 2550 ρg
Chemical Engineering Journal 223 (2013) 795–805
ρ (1 − ε ) hc Dt = 0.66 Prg0.3 s ρε kg g
0.44
Re0D.44
for
ρs Re p ≤ 2050 ρg
where Re D =
Dt ρ g U g
µg
,
Re p =
d p ρ gU g
µg
The heat of reaction for sorbent AX was found to be exothermic at 63 ± 16 kJ/mole. There was a slight decrease in heat of reaction with increased CO2 concentration, but this can be attributed to the higher thermal deviation from the nominal temperatures for these conditions. 3. Results and Discussion The effect of reaction temperatures on the experimental breakthrough curves for a feed containing 33.3% CO2 for immobilized PEI /silica sorbent particle of 142 µm diameter is illustrated in Figure 5. The breakthrough point is the point that CO2 was initially detected in the effluent. The breakthrough curve obtained at the lower temperature (40 oC) shows a much shorter time with undetectable CO2 effluent before the breakthrough and a much sharper shape. Figure 5 also shows that when the temperature increased from 40 oC to 70 o C that the time required for complete saturation increased from 560 sec to about 1160 sec. In general rate of uptake is higher at higher temperature; however, the slope of the 90 °C breakthrough curve was less sharp than other breakthrough curves. The shape of the breakthrough curve indicates that the rate decreased as the temperature increased. Nucleation growth mechanisms have been proposed to explain this behavior [41]. The broadness of the trailing edge of the breakthrough curves could perhaps result from slow intraparticle diffusion within the pores of the sorbent particle. The CO2 must first diffuse into the porous particles before interacting with the functional groups of PEI. The “tailing” of a breakthrough curve (i.e., a slow approach of Ct/C0 towards 1) is usually due to the dominant mass transfer process. In addition the sorbent capacity increased from 87.6 to 121.5 mg/g over 40 oC to 70 oC temperature range. This increase in capacity is consistent with findings for PEI based sorbents [30, 41] and reflects mass transfer restrictions to the deeper layers in the PEI at low temperatures. Mathematical models are useful for understanding fluid bed dynamics and for design and optimization studies since they help reduce time-consuming and repetitive experiments. For adsorption in a fluid bed column, the mass transfer in the gas phase can be modeled as a dispersed plug flow accounting for mixing in the axial direction [45]. As a general rule, when the ratio of the column-to particle radius is greater than 10, the difference of concentration in the radial direction is negligible. In this study, we consider an isothermal fluid bed with an immobilized PEI /silica sorbent through which a gaseous containing
Chemical Engineering Journal 223 (2013) 795–805
CO2 flows with a constant linear velocity. The general mass balance for an absorption in a column is given by;
∂CCO2 ∂t
= DL
∂ 2CCO2 ∂z 2
−
U ∂CCO2 (1 − ε ) ∂q − ε ∂z ε ∂t
(1)
where CCO2 is the gas phase CO2 concentration, DL is the axial dispersion of the adsorbate in the gas stream, q is the CO2 concentration in the sorbent averaged over the sorbent volume, z is the length in the axial direction, U is the superficial velocity of the gas mixture, ε is the bed void fraction, and ∂q/∂t is the rate of adsorption. Although such models are useful for a fundamental understanding of the dynamics of fluid-bed, their practical utility is rather limited owing to the complicated solution methods. On the other hand, simplified models are useful when it is desired to develop a practical tool that can capture quantitatively the effects of main system variables on the column dynamics [46]. Therefore, the ability of a relatively simple model, known as the Bohart-Adams equation [47], to describe the breakthrough behavior of immobilized amine (PEI) on mesoporous silica was examined. It should be noted that the Bohart-Adams model assumes that adsorption rate described by a quasi-chemical kinetics rate expression as:
∂q = kc (qs − q)CCO2 ∂t
(2)
where kc is rate constant and qs is saturation capacity. Neglecting axial dispersion and assume t>>L*ε/U, The analytical solution to Eqs. (1) and (2) results in Bohart-Adams model. This model is capable of generating symmetrical, sigmoidal breakthrough curves. The development of the Bohart-Adams model is described in detailed by Chu and Hashim [46] and the solution can be summarized as:
Ct 1 = C0 1 + exp(k (q L − C t )) c s 0 U
(3)
Where kc (cm3/g.s) is rate constant and qs is the saturation capacity per unit volume of packed bed. Upon rearrangement, Eq. (3) becomes [46];
C ln 0 − 1 1 Ct L = t qs − UC0 kc C0
(4)
According to Eq.(4), plotting experimental data in terms of t versus ln(C0/Ct–1)/C0 should yield a straight line and the two parameters kc and qs can be estimated from the intercept
Chemical Engineering Journal 223 (2013) 795–805
and slope, respectively. Therefore, Eq. (4) implies that the service time is a linear function of the bed height and hence it can be written as: = t mL − b
(5)
where m is the slope and b is the intercept of the line. The slope, m, is equal to qs/UC0. Thus, determining the slope allows one to evaluate the sorbent capacity when known values of the inlet concentration of CO2 and gas velocity are used. Likewise the intercept, b, is equal to ln(C0/Ct–1)/C0kc such that once it is determined the value of kc can be extracted at given level of CO2 capture, C0/Ct. The simplicity and advantage of using the Bohart-Adams model is that it can applied for the prediction of the slope for any unknown flow rate with a known slope at a given flow rate. Thus, once the reactor design parameters are calculated at a given flow rate, the fluid-bed column can be designed for different flow rates, by modifying the equation (Eq. (5)) and utilizing the new slope, however, the intercept remains unchanged because it depends on the inlet gas concentration, C0. The Eq. (5) is modified as follows:
U mnew = mold old (6) U new Thus, when values for the velocity and slope are measured at a given condition, Uold and mold, respectively, then a new slope parameter can be obtained for a new gas velocity of interest. From this new slope the estimated sorbent CO2 uptake, qs, can be obtained. Similarly, the change in inlet concentration (C0) usually results in changes in slope and intercept of Eq. (5). The new slope and intercept values can be determined from: C0 (7) mnew = mold old C0 new Likewise, when values for the CO2 concentration and slope are measured at a given condition, Co old and mold, respectively, then a new slope parameter can be obtained for a new gas concentration of interest and the new value for qs can be obtained.
C0 bnew = bold old C0new
C0new − 1 ln Ct C ln 0old − 1 Ct
(8)
Similarly, when values for the concentration and intercept are measured at a given condition, Co old and bold, respectively, then a new intercept parameter can be obtained for a new concentration of interest. From this new intercept the estimated CO2 uptake rate parameter, kc, can be obtained. For example, Eq. (7) was tested by comparing the response at different CO2 concentration with experimental measurements reported in
Chemical Engineering Journal 223 (2013) 795–805
Table 2. By taking the values obtained from Eq. (4) the sorption capacity, qs, and rate constants, kc, were estimated from concentration data taken at 33.3% CO2 for different temperatures. For these four test cases the new capacities were found to be within 2.7% of the measured capacities, and the rate constants were within 3.6% of the measured rates. There was no observed bias or systematic deviation for the values estimated at different temperatures. At time t=0, the Eq. (4) can also be solved for bed length L0 as:
= L0
C U ln 0 − 1 qs kc Cb
(9)
where L0 is the minimum or critical bed length of the adsorbent sufficient to produce an outlet concentration Ct=Cb, where Cb is the concentration at the breakthrough defined as a limit concentration or a fixed percent of initial concentration. Note that the sorbent near the inlet is not necessarily fully saturated when the reactor length is equal to this critical bed length. In Fig. 6 the breakthrough data of Figure 5 is plotted in terms of Eq. (4). A linear relationship was found between t and ln(C0/Ct–1)/C0 for temperatures of 70 and 90 oC, indicating that the experimental results conformed to the linearized Bohart-Adams model. The implication is that kc and qs can be estimated from these linear curves. However, Figure 6 also display that the relationship between t and ln(C0/Ct–1)/C0 was non-linear for temperatures of 40 and 55 oC, indicating the experimental results do not conform to the Bohart-Adams model. The implication is that kc and qs cannot be estimated accurately from these nonlinear curves. An alternative method of estimating kc and qs is to fit Eq. (3) directly to the breakthrough data of Figure 5 by a nonlinear least-square fit. Therefore, for a given temperature, values of kc and qs were determined by curve fitting the data of Figure 5 with the parameters in eq. (3) using TABLECURVE available from Statistical Package for the Social Sciences. The experimental Ct/C0 data was compared to the Bohart-Adams model when the kinetic parameters were determined in this manner. This is illustrated in Figure 5 for different temperatures with the measurement represented by discrete symbols and the model estimates by the solids lines. The model data and experimental data agree over the entire Ct/C0 time with overall variance explained (R2) greater than 99.9%. However, the model overestimated the sharpness of the leading and trailing edges of the experimental breakthrough curves at the low temperatures. When fitting non-linear equations with a large number of parameters, it is prudent to consider how sensitive the fit is to the choice of input parameters, and whether equally good fits could be achieved with different sets of parameters. Indeed, it was found out the unknown parameters did not vary with different initial choices. The effect of reaction temperatures on the experimental breakthrough curves for a feed containing 16.7% CO2 for immobilized PEI /silica sorbent particle is also illustrated in Figure 7. Overall, the agreement between the experimental curves and calculated ones is
Chemical Engineering Journal 223 (2013) 795–805
very good. Similar to the test at higher CO2 concentration, the lowest temperature exhibited the largest deviation from the experimental data. It is apparent that prior to breakthrough at low temperature the first order reaction model overestimates the uptake rate. This is not unexpected since a more rigorous model including nucleation and growth mechanisms was needed to capture the reduced rate of uptake due to the sporadic nature of the nucleation at these low temperatures [41].
The calculated breakthrough rate (d(Ct/C0)/dt vs. t) data is illustrated by Figure 8 for different temperatures. The experimental breakthrough curves are symmetrical. The maximum rate decreased with the increasing temperatures. The calculated breakthrough rate is a function of Ct/C0 as presented in Figure 9 for isothermal operating temperatures between 40 and 90oC. The rate increased with fractional Ct/C0 up to 50% Ct/C0 before dropping again as the bed became saturated. This rate dropped systematically with temperature. Wang et al. [48] determined that the rate controlling process for CO2 uptake must be changing with temperature. The analysis by Monazam et al. [41] supported this conclusion stating that while the uptake of CO2 is a low-energy process, it is not a barrierless process and so the rate controlling process must be changing with temperature. They demonstrated that this process is described using a nucleation and growth mechanism in which the rate of platelet-shaped (2 dimensional) nuclei decreases with increasing temperature, while growth of the nuclei increases in rate with increasing temperature. The nuclei growth step is limited by thermodynamic constraints and controlled by mass transport of CO2 into the PEI. The values of kc, qs, maximum breakthrough rate, and its time are also tabulated in Table 2. The values of kc decreased as temperature increased for a given inlet CO2 concentration. Assuming an exponential temperature dependence (Figure 10), k=Ae(-d/T), where A is a pre-exponential factor, d is the exponential constant, and T is temperature in K. The temperature dependence can be rewritten as
d (10) T From Eq. 10, it is apparent that the plot of ln k vs. 1/T should result in a straight line with a slope of d and intercept of ln A. A plot of ln(k) vs. 1/T for amine/silicate sorbent with particle size of 142 µm according to particle-CO2 uptake is shown in Figure 10 for each reaction temperature and two CO2 concentrations of 33.3% and 16.7%. The values for pre-exponential factor, A, and exponential constant, d, for the CO2 removal were obtained from the intercept and slope of straight line in Figure 10. ln= k ln A −
The values for A varied substantially as the CO2 level changed, increasing with increasing CO2. The exponential constant, d, was indeed fairly constant for different CO2 levels as evidenced by the nearly parallel lines. Overall, the rate of reaction decreased as the temperature increased. This apparent negative temperature dependence is consistent with observations by Monazam et al. [41] and Ebner et al. [49] for similar sorbent.
Chemical Engineering Journal 223 (2013) 795–805
The sorbent capacities for CO2 uptake measured in the fluid bed were between 86 and 122 mg/g and depended strongly on temperature, but less so on CO2 concentration (Table 2). TGA capacities measured at 60 °C in 100% CO2 were comparable with values between 114 to 126 and an average of 124 mg/g. There was only a slight increase in CO2 uptake at the higher CO2 concentration in the fluid bed. This is consistent with the adsorption isotherms for a similar sorbent reported by Monazam et al. [41] in which there is only small increase in uptake between 10 and 100% CO2 . For the temperature range studied in the fluid bed Monazam et al. reported that CO2 adsorption follow the isotherms formulated by Freundlich and Temkin for heterogeneous and mixed homogeneous heterogeneous surfaces as might be expected for a complex polymer such as PEI on a mesoporous substrate. That study demonstrated that the uptake only begins to drop off below 10% CO2. Increasing the velocity through the fluid bed increased the rate for CO2 breakthrough; however, the capacity did not change appreciably as discussed in additional detail below. Figure 11 shows the effect of temperature on the CO2 sorption capacities. The saturated CO2 capacity increases with temperature up to 70 oC and then decreases at higher temperature. As the temperature is increased, PEI becomes more flexible and hence, more CO2-affinity sites are exposed to the CO2 leading to higher capacities [30]. As temperature increased above 70 oC, the equilibrium shifts to the reverse direction and desorption become more favored [50]. This same trend was observed in an earlier study on the same type of sorbent, designated 196C using TGA apparatus to measure the sorbent capacity [41]. That sorbent was 50% PEI loaded on a Cariact mesoporous substrate and when this difference of loading was taken into account the capacities agreed closely (Figure 12). It is interesting to note the primary discrepancy between the two sorbents was that sorbent AX had slightly lower capacity at lower temperature, even though this sorbent had the lower PEI loading. It has been argued that the CO2 capacities for PEI based sorbent are reduced at lower temperatures where transport through the PEI is thought to restrict CO2 access to the deeper PEI multilayers [30, 39]. Figure 12 provides evidence that the capacity was less dependent on the thickness of the PEI layer than other factors such as the substrate or the molecular weight of the PEI. The PEI number average molecular weight was 2000 for AX while it was only 400 for the 196C. The characteristic parameters, including breakthrough time, saturation time and length of the mass transfer zone (MTZ), derived from the breakthrough curves of a 25.4 cm length bed for various bed are listed in Table 3. The length of the mass transfer zone (MTZ) is the portion of the adsorbate bed where adsorbate is actually being adsorbed on the adsorbent. The MTZ typically moves from the inlet towards the outlet during operation. That is, as the adsorbent near the inlet becomes saturated (spent) with adsorbate, the zone of active adsorption moves toward the effluent end of the bed where the adsorbate is not yet saturated (Figure 13). The MTZ is sometimes called the adsorption zone or critical bed depth. For constant pattern adsorption, the MTZ length was approximated as follows [51]:
Chemical Engineering Journal 223 (2013) 795–805
LMTZ =
2 L(ts − tb ) (ts + tb )
(11)
where L is the length of the bed (cm); and tb and ts are the breakthrough and saturation time (min) respectively, which are normally defined as the times when the outlet concentrations are 5% and 95% of the inlet concentration, respectively. The breakthrough time increased and the saturation time increased with increasing temperature, resulting in longer mass transfer zones (Table 3). In general, the lower temperature, the faster is the nucleation of product species on the surface and thus, the shorter the mass transfer zone. In contrast to Lo, the length of the mass transfer zone, LMTZ, is the length from a fully saturated gas inlet to where the CO2 reaches breakthrough at the top. This is always equal to or larger than Lo. The overall bed capacity utilization for a symmetric breakthrough curve can be given as follows:
0.5 LMTZ (12) L The utilization factor decreased as the temperature increased (Table 3). Likewise, the capacity utilization decreased with inlet CO2 concentration due to the reduction in the driving force for adsorption. The utilization factor approaches 1 when the reactor length is large. The design engineer must balance the cost of excess reactor size with inefficient utilization of the reactor sorbent inventory. f = 1−
The carbon dioxide loading in the bed was determined by a dynamic mass balance, which requires numerical integration of the data. Using the breakthrough curve data from an adsorption breakthrough run, the time equivalent to the total or stoichiometric capacity of the column for CO2 is calculated by numerically integrating the following equation:
Ct (13) 1 − dt 0 C0 where tq is the time equivalent to the total or stoichiometric capacity, t is time, Ct is the concentration of CO2 at time t, and C0the feed concentration of CO2. With tq determined, the dynamic adsorption capacity, q, of the bed was calculated using following equation [45]; = tq
q=
∫
∞
Q f C0tq mads
(14)
where q is the CO2 loading, Qf is the volumetric feed flow rate at standard conditions, and mads is the mass of adsorbent loaded in the bed. It can be seen from Figure 11 that the experimental (using Eq. 14) and predicted values (using Eq. 3) of CO2 loading showed excellent agreement.
Chemical Engineering Journal 223 (2013) 795–805
The advantage of the using Eq. 3 (Bohart-Adams model) is that any experimental test can be reliably scaled up to other flow rates and inlet CO2 concentrations without further experimental testing. Figure 14 gives the concentration breakthrough curves obtained for CO2 adsorption for two different flow rates to see the effect of the feed flow rate at a constant adsorption pressure of 1 atm. It was observed that a higher feed flow rate of 30 slpm contributed to an early breakthrough since the bed got saturated faster with more CO2 going through the column compared to a lower feed flow rate of 15slpm. The minimum fluidization velocity for the sorbent was 0.15 cm/s corresponding to nominally 1.5 slpm flow in the fluid bed. Thus, changing the flow from 10 to 20 times the minimum fluidization velocity did not affect the measured capacity indicating that there was little or no leak through the fluid bed. LMTZ was close to the length of the fluid bed for the higher flow rate (30 slpm), but the critical length (11 cm) was still well below the bed height of 25 cm (Table 3). Flows higher than this would be expected to leak CO2 through the fluid bed. Figure 14 also shows that when the flow rate increased from 15 to 30 slpm the time required for breakthrough decreased from 700 s to about 460 s. For 30 slpm, the mass transfer zone moved quicker and mass transfer coefficient was higher because of the higher Reynolds number. Tailing was seen towards the approach to sorbent saturation for both flow rates. This is an indication of the slow intra-particle diffusion within micropores of the adsorbent, presence of non homogeneous particles and varying flow patterns. With higher flow the degree of mixing and turbulence is generally higher. At the end of the breakthrough curve where CO2 approached the inlet concentration, there was a somewhat slower approach to the final outlet concentration for the high flow case as compared to lower flow case (Figure 14). In addition, there was a difference in the leak through rates prior to breakthrough, such that the CO2 in the outlet gas was greater for the longer duration tests. At low flows or lower concentrations the concentration of CO2 in the outlet gas was found to be slightly higher than the higher gas flows or CO2 concentrations. Prior to breakthrough the outlet concentrations were essentially the same whether or not the test conditions took a long or short time. The outlet concentration increased gradually over time. Thus, for the longer test cases the outlet concentration continued to increase with time until breakthrough occurred. In general, the cumulative leak rate was higher for the longer duration tests, irrespective of the gas velocity and bed voidage. The effect of a variation of the inlet CO2 concentration from 16.6 to 33.3% used with the same adsorbent bed height of 25.4 cm and flow rate of 30 slpm is shown by the breakthrough curve in Figure 15. It is illustrated that the breakthrough time slightly decreased with increasing inlet CO2 concentration. When the inlet concentration was increased from 16.6 to 33.3% the time required for the breakthrough decreased from 570 s to about 490 s. At lower inlet CO2 concentrations, gases were dispersed and breakthrough occurred slower. As influent concentration increased, sharper breakthrough curves were obtained. This can be explained by the fact that a lower concentration
Chemical Engineering Journal 223 (2013) 795–805
gradient resulted in slower transport due to a decrease in the diffusion coefficient or mass transfer coefficient. The larger the inlet concentration, the steeper was the slope of breakthrough curve and shorter was the breakthrough time. These results demonstrated that the change of concentration gradient affected the saturation rate and breakthrough time, or in other words, the adsorption process was concentration dependent. As the influent concentration increased, CO2 loading rate increased, because the driving force increased for mass transfer. Fixed bed tests are by nature less isothermal than a fluid bed because the heat transfer coefficients in a fluid bed are an order of magnitude larger than those in a fixed bed. Since the CO2 loading on the sorbent was found to be relatively insensitive to temperature over the range between 40 and 80 °C (90-120 mg CO2/g sorbent), it is not expected that the breakthrough curves would change significantly as long as the heat rise is maintained within this temperature range in the fixed bed. Meeting that constraint, however, depends upon the gas flow rates and adsorbate concentration as well as the dimensions of the adsorber bed. The worst case scenario might be thought to be high CO2 concentrations with low gas flows rates through a relatively thick bed; these conditions would tend to produce hot spots endothermicly releasing CO2 to moderate the temperature. A fixed bed adsorber would be expected to adsorb less CO2 per unit weight sorbent in this situation. On the other hand, the rate of CO2 uptake was found to be somewhat slower at higher temperatures, and thus an adiabatic fixed bed would have a tendency to exhibit a shallower breakthrough curve extending over a longer period of time. These kinetic considerations can impact the required contact time, the size of the adsorption bed, and ultimately the capital costs. Fixed beds are also plagued with relatively high pressure drops which create a greater potential for gas channeling thereby requiring much larger particles than those tested in this study. Fines less than 43 µm are known to exacerbate the tendency to channel and are typically avoided in fixed bed units.
4. Conclusions The adsorption of carbon dioxide (CO2) by immobilized amine (PEI) on mesoporous silica was investigated using fluid bed. The kinetic, equilibrium and breakthrough curves were discussed. The kinetic process was better described by the pseudo-first-order kinetic model. The breakthrough curves were significantly affected by flow rate and inlet CO2 concentration. As the flow rate increased, the breakthrough curve became steeper, the break point time and the amount of adsorbed CO2 decreased, probably due to an insufficient residence time of the CO2 gas in the bed. The breakthrough capacity of the sorbent was found to increase with inlet CO2 concentration. The breakthrough data were fitted by the Bohart-Adams equation at experimental conditions using nonlinear regressive analysis. The calculated theoretical breakthrough curves were in good agreement with the corresponding experimental data.
Chemical Engineering Journal 223 (2013) 795–805
Acknowledgments The authors acknowledge the Department of Energy for funding the research through the Fossil Energy’s Carbon Sequestration/CO2 Capture Research program.
Nomenclature b CCO2 C0 Cb Ct DL f kc L L0 LMTZ m mads q qs Qf t tb tq ts U z ε
intercept in eq. 5, (s) concentration of CO2 (g/cm3) concentration of CO2 at the inlet (g/cm3) concentration of CO2 at the breakthrough, 5% C0 (g/cm3) concentration of CO2 at the time t (g/cm3) axial dispersion of the adsorbate in the gas stream (cm2/s) utilization factor rate constant (cm3/g.s) bed length (cm) critical bed length (cm) Length of mass transfer zone (cm) slope in eq. 5, (cm/s) mass of adsorbent (g) CO2 uptake over sorbent volume (g CO2/cm3 of bed) CO2 uptake at saturation per unit volume (g CO2/cm3 of bed) volumetric feed flow rate (slpm) time (s) breakthrough time (s) time for total capacity (s) saturation time (s) superficial gas velocity (cm/s) axial length (cm) bed void fraction
Chemical Engineering Journal 223 (2013) 795–805
References [1] International Energy Agency Working Party on Fossil Fuels, CO2 capture at power stations and other major point sources, OECD/IEA, Paris, FR, 1–12, (2003). [2] P.A. Zhang, P. Xiao Webley, Effect of process parameters on power requirements of vacuum swing adsorption technology for CO2 capture from flue gas, Energy Convers. Manag. 49 (2008) 346–356. [3] A. Samanta, A. Zhao, G.K.H. Shimizu, P. Sarkar, R. Gupta, Post-Combustion CO2 Capture Using Solid Sorbents: A Review, Ind. Eng. Chem. Res. 51(2012) 1438–1463. [4] A. Sayari, Y. Belmabkhout, R. Serna-Guerrero, Flue gas treatment via CO2 adsorption, Chem. Eng. J. 171 (2011) 760–774. [5] P. Davini, Flue gas treatment by activated carbon obtained from oil-fired fly ash, Carbon 40 (2002) 1973–1979. [6] J. Przepiórski, M. Skrodzewicz, A.W. Morawski, High temperature ammonia treatment of activated carbon for enhancement of CO2 adsorption, Appl. Surf. Sci. 225 (2004) 235–242. [7] C. Pevida, M.G. Plaza, B. Arias, J. Fermoso, F. Rubiera, J.J. Pis, Surface modification of activated carbons for CO2 capture, Appl. Surf. Sci. 254 (2008) 7165–7172. [8] R.V. Siriwardane, M.S. Shen, E.P. Fisher, J.A. Posten, Adsorption of CO2 on Molecular Sieves and Activated Carbon, Energy & Fuels 15 (2001) 279–284. [9] C. Lu, H. Bai, B. Wu, F. Su, J.F. Hwang, Comparative Study of CO2 Capture by Carbon Nanotubes, Activated Carbons, and Zeolites, Energy Fuels 22 (2008) 3050–3056. [10] P.J.E. Harlick, F.H. Tezel, Adsorption of carbon dioxide, methane, and nitrogen: pure and binary mixture adsorption by ZSM-5 with SiO2/Al2O3 ratio of 30, Sep. Purif. Technol. 37 (2002) 33–60. [11] J.S. Lee, J.H. Kim, J.T. Kim, J.K. Suh, J.M. Lee, C.H. Lee, Adsorption Equilibria of CO2 on Zeolite 13X and Zeolite X/Activated Carbon Composite, J. Chem. Eng. Data 47 (2002) 1237–1242. [12] K.S. Walton, M.B. Abney, M.D. Levan, CO2 adsorption in Y and X zeolites modified by alkali metal cation exchange, Micropor. Mesopor. Mater. 91 (2006) 78. [13] W. Gao, D. Butler, D.L. Tomasko, High-Pressure Adsorption of CO2 on NaY Zeolite and Model Prediction of Adsorption Isotherms, Langmuir 20 (2004) 8083–8089. [14] F. Su, C. Lu, S.C. Kuo, W. Zeng, Adsorption of CO2 on Amine-Functionalized YType Zeolites, Energy Fuels 24 (2010) 1441–1448.
Chemical Engineering Journal 223 (2013) 795–805
[15] E. Diaz, E. Munoz, A. Vega, S. Ordonez, Enhancement of the CO2 Retention Capacity of Y Zeolites by Na and Cs Treatments: Effect of Adsorption Temperature and Water Treatment, Ind. Eng. Chem. Res. 47 (2008) 412–418. [16] R. Chatti, A.K. Bansiwal, J.A. Thote, V. Kumar, P. Jadhav, S.K. Lokhande, R.B.Biniwale, N.K. Labhsetwar, S.S. Rayalu, Amine loaded zeolites for carbon dioxide capture: Amine loading and adsorption studies, Micropor. Mesopor. Mater. 121 (2009) 84–89. [17] Y.S. Bae, K.L. Mulfort, H. Frost, P. Ryan, S. Punnathanam, L.J. Broadbelt, J.T. Hupp,R.Q. Snurr, Separation of CO2 from CH4 Using Mixed-Ligand Metal−Organic Frameworks, Langmuir 24 (2008) 8592–8598. [18] Z. Zhao, Z. Li, Y.S. Lin, Adsorption and Diffusion of Carbon Dioxide on Metal−Organic Framework (MOF-5), Ind. Eng. Chem. Res. 48 (2009) 10015– 10020. [19] A.R. Millward, O.M. Yaghi, Metal−Organic Frameworks with Exceptionally High Capacity for Storage of Carbon Dioxide at Room Temperature, J. Am. Chem. Soc. 127 (2005) 17998–17999. [20] V. Finsy, L. Ma, L. Alaert, D.E. de Vos, C.V. Baron, J.F.M. Denayer, Separation of CO2/CH4 mixtures with the MIL-53(Al) metal–organic Framework, Micropor. Mesopor. Mater. 120 (2009) 221–227. [21] J.R. Li, R.J. Kuppler, H.C. Zhou, Selective gas adsorption and separation in metal–organic frameworks, Chem. Soc. Rev. 38 (2009) 1477–1504. [22] F. Su, C. Lu, W. Chen, H. Bai, J.F. Hwang, Capture of CO2 from flue gas via multiwalled carbon nanotubes, Sci. Total Environ. 407 (2009) 3017–3023. [23] M. Cinke, J. Li, C.W. Bauschlicher Jr., A. Ricca, M. Meyyappan, CO2 adsorption in single-walled carbon nanotubes,Chem. Phys. Lett. 376 (2003) 761–766. [24] F. Su, C. Lu, H.S. Chen, Adsorption, Desorption, and Thermodynamic Studies of CO2 with High-Amine-Loaded Multiwalled Carbon Nanotubes, Langmuir 27 (2011) 8090–8098. [25] S. Hsu, C. Lu, F. Su, W. Zeng, W. Chen, Thermodynamics and regeneration studies of CO2 adsorption on multi walled carbon nanotubes, Chem. Eng. Sci. 65 (2010) 1354–1361. [26] A.K. Mishra, S. Ramaprabhu, Nanostructured polyaniline decorated graphene sheets for reversible CO2 capture, J. Mater. Chem. 22 (2012) 3708–3712. [27] H.Y. Huang, R.T. Yang, D. Chinn, C.L. Munson, Amine-Grafted MCM-48 and Silica Xerogel as Superior Sorbents for Acidic Gas Removal from Natural Gas, Ind. Eng. Chem. Res. 42 (2003) 2427–2433. [28] M.L. Gray, Y. Soong, K.J. Champagne, H. Pennline, J.P. Baltrus, R.W. Stevens Jr., R. Khatri, S.S.C. Chuang, T. Filburn, Improved immobilized carbon dioxide capture sorbents, Fuel Proc. Technol. 86 (2005) 1449–1455. [29] N. Hiyoshi, K. Yogo, T. Yashima, Adsorption characteristics of carbon dioxide on organically functionalized SBA-15, Micropor. Mesopor. Mater. 84 (2005) 357–365.
Chemical Engineering Journal 223 (2013) 795–805
[30] X. Xu, C. Song, J.M. Andrésen, B.G. Miller, A.W. Scaroni, Preparation and characterization of novel CO2‘‘molecular basket’’ adsorbents based on polymermodified mesoporous molecular sieve MCM-4. Microporous Mesoporous Mater. 62 (2003) 29–45. [31] G.P. Knowles, S.W. Delaney, A.L. Chaffee, Diethylenetriamine[propyl(silyl)]Functionalized (DT) Mesoporous Silicas as CO2 Adsorbents, Ind. Eng. Chem. Res. 45 (2006)2626–2633. [32] M.B. Yue, L.B. Sun, Y. Cao, Z.J. Wang, Y. Wang, Q. Yu, J.H. Zhu, Promoting the CO 2 adsorption in the amine-containing SBA-15 by hydroxyl group, Micropor. Mesopor.Mater. 114 (2008) 74–81. [33] R. Serna-Guerrero, E. Da’na, A. Sayari, New Insights into the Interactions of CO2 with Amine-Functionalized Silica, Ind. Eng. Chem. Res. 47 (2008) 9406–9412. [34] C. Lu, H. Bai, F. Su, W. Chen, J.F. Hwang, H.H. Lee, Adsorption of Carbon Dioxide from Gas Streams via Mesoporous Spherical-Silica Particles, J. Air Waste Manage. Assoc. 60(2010) 489–496. [35] C. Lu, F. Su, S. Hsu, W. Chen, H. Bai, J.F. Hwang, H.H. Lee, Thermodynamics and regeneration of CO2 adsorption on mesoporous spherical-silica particles, Fuel Process. Technol.90 (2009) 1543–1549. [36] S.H. Liu, Y.C. Lin, H.R. Hyu, Adsorption of CO2 from Flue Gas Streams by a Highly Efficient and Stable Aminosilica Adsorbent, J. Air Waste Manage. Assoc. 61 (2011) 226–233. [37] R.A. Khatri, S.S.C. Chuang, Y. Soong, M.Gray ,Carbon dioxide capture by diamine- grafted SBA-15: a combined Fourier transform infrared and mass spectrometry study, Ind. Eng. Chem. Res. 44 (2005) 3702–3708. [38] Sharon Sjostrom, Holly Krutka, Travis Starns, Tom Campbell, Pilot test results of post-combustion CO2 capture using solid sorbents, Energy Procedia 4 (2011) 1584-1592. [39] A. Heydari-Gorji, Y. Belmabkhout, and A. Sayari, Polyethylenimine
impregnated mesoporous silica: effect of amine loading and surface alkyl chains on CO2 adsorption, Langmuir 27 (2011) 12411–12416. [40]A. Samanta, A. Zhao, G.K.H. Shimizu, P. Sarkar, R. Gupta, Post-Combustion CO2 Capture Using Solid Sorbents: A Review, Ind. Eng. Chem. Res. 51(2012) 1438– 1463. [41] E.R. Monazam, L.J. Shadle, D.C. Miller, H.W. Pennline, D.J. Fauth, J.S. Hoffman, M.L. Gray, Equilibrium and kinetics analysis of carbon dioxide capture using immobilized amine on a mesoporous silica, AIChE J. 2012, DOI: 10.1002/aic.13870. [42] Eslick, J. and D.C.Miller. "A multi-objective analysis for the retrofit of a pulverized coal power plant with a CO2 capture and compression process". Computers and Chemical Engineering. 35 (2011) 1488– 1500. [43] Woods.C.,Capicotto,P.J.,Haslbeck,J.L.,Kuehn,N.J.,Matuszewski,M.,Pinkerton, L. L.,etal. (2007). Bituminous coal and natural gas to electricity final report. Cost and Performance Baseline for Fossil Energy Plants, US Department of Energy. Report number DOE/NETL-2007/1281. [44] Yang, W.C. Handbook of Fluidization and Fluid Particle Systems, Marcel Dekker, Inc., New York (2003) 263.[45] R.Serna-Guerrero, A. Sayari, Modeling
Chemical Engineering Journal 223 (2013) 795–805
adsorption of CO2 on amine-functionalized mesoporous silica. 2: Kinetics and breakthrough curves, Chem. Eng. J. 161 (2010) 182-190. [46] K. H. Chu, M. A.Hashim, Copper biosorption on immobilized seaweed biomass: Column breakthrough characteristics, Journal of Environmental Sciences 19(2007) 928–932. [47] G. S. Bohart, E. Q. Adams, Some aspects of the behavior of charcoal with respect to chlorine[J], J. Am. Chem. Soc. 42 (1920) 523–544. [48] X. Wang, V. Schwartz, J.C. Clark, X. Ma, S.H. Overbury, X. Xu, C. Song, Infrared study of CO2 sorption over ‘‘molecular basket’’ sorbent consisting of polyethylenimine-modified mesoporous molecular sieve, J. Phys. Chem. C. 113 (2009) 7260–7268. [49] A.D. Ebner, M.L. Gray, N.G. Chisholm, Q.T. Black, D.D. Mumford, M.A. Nicholson, J.A. Ritter, Suitability of a solid amine sorbent for CO2 capture by pressure swing adsorption. Ind. Eng. Chem. Res. 50 (2011) 5634–5641. [50] C.Chen, W.-J.Son, K.-S.You, J.-W. Ahn, W.-S Ahn, Carbon dioxide capture using amine-impregnated HMS having textural Mesoporosity, Chemical Engineering Journal 161 (2010) 46–52. [51] A. A. Pota, A. P. Mathews, Effects of particle stratification on fixed bed adsorber performance. J. Envir. Engrg. 8 (1999) 705-711.
Chemical Engineering Journal 223 (2013) 795–805
LIST of TABLES Table 1. The effect of temperature and inlet CO2 concentration on heat of reaction, Hr, and overall fluid bed heat transfer coefficient, hc. Table 2. Parameters obtained by fitting nonlinear Eq. (3) to experimental breakthrough curves using total flow of 30 slpm with balance N2 and 2% H2O Table 3. Effects of temperatures and inlet concentration on the characteristic parameters using total flow of 30 slpm with balance N2 and 2% H2O Table 1. The effect of temperature and inlet CO2 concentration on heat of reaction, ∆Hr, and overall fluid bed heat transfer coefficient, hc. Tbed °C 40 55 70 90 40 55 70 90
Inlet CCO2 ∆Hr hc slpm kJ/mole W/m2-C 10 -84 185 10 -58 132 10 -53 143 10 -39 65 5 -85 196 5 -74 148 5 -59 124 5 -53 123 Average -63 139 std dev 16 41
Chemical Engineering Journal 223 (2013) 795–805
Table 2. Parameters obtained by fitting nonlinear Eq. (3) to experimental breakthrough curves using total flow of 30 slpm with balance N2 and 2% H2O Inlet CCO2 (slpm) 10 10 10 10 5 5 5 5
T(oC) 40 55 70 90 40 55 70 90
kc (cm3/g.s) 31.7 24.2 18.4 15.9 33.6 27.4 16.1 13.6
qs (mg/g) 87.6 105.6 121.5 97.4 86.6 110.7 120.7 105.7
d(Ct/C0))/dt,max (s-1) 0.00453 0.00330 0.00240 0.00195 0.00240 0.00187 0.00105 0.00089
time d(Ct/C0))/dt,max (s) 505 608 700 561 997 1275 1389 1283
Table 3. Effects of temperatures and inlet concentration on the characteristic parameters using total flow of 30 slpm with balance N2 and 2% H2O
T(oC) 40 55 70 90 40 55 70 90
Inlet CCO2 (slpm) 10 10 10 10 5 5 5 5
tb t0.05 (s) 368 375 372 217 1170 958 417 243
ts t0.95 (s) 562 695 1163 986 1853 1738 1736 2041
LMTZ (cm) 10.6 15.2 26.2 32.5 11.5 14.7 31.1 40.0
L0 (cm) 8.2 9.3 11.1 17.1 7.8 7.9 12.8 16.5
f 0.79 0.70 0.48 0.36 0.77 0.71 0.39 0.21
Chemical Engineering Journal 223 (2013) 795–805
LIST OF FIGURES Figure 1. Particle size distribution of sorbent AX using QICPIC particle analyzer. Figure 2. SEM photomicrograph of sorbent AX after testing under temperature excursion Figure 3. Schematic diagram of fluid-bed column system Figure 4. Typical temperature profiles across the fluid bed and raw exit CO2 concentrations (with dilution) during breakthrough testing of sorbent AX at a) 70 °C and 33.3% CO2 , and b) 40 °C and 16.6% CO2. Figure 5. Effect of temperature on breakthrough curved at feed 33.3% CO2 and flow rate of 30 slpm. Figure 6. Experimental breakthrough data plotted according to Eq. 4. Figure 7. Effect of temperature on breakthrough curved at feed concentration of 16. 7% CO2 and flow rate of 30 slpm including insert of the initial breakthrough. Figure 8. Effect of temperature on breakthrough rate at feed concentration of 10slpm and flow rate of 30 slpm. Figure 9. Effect of temperature on breakthrough rate at 33.3% CO2 and flow rate of 30 slpm. Figure 10. Temperature dependence of the reaction rate, k (1/s). Figure 11. Comparison of the experimental (symbol) and predicted values (solid line) of CO2 loading for two different inlet CO2 concentrations at different temperatures. Figure 12. Measured CO2 uptake values for sorbent AX at different nominal temperatures as compared to TGA measurements for sorbent 196C normalized to 40% PEI loading. Figure 13. Movement of the mass transfer zone and the resulting breakthrough curve Figure 14. Effect of flow rate on breakthrough curves using feed CO2 concentration of 33.3% Figure 15. Effect of feed concentration (C0) on breakthrough curves at flow rate of 30 slpm.
Chemical Engineering Journal 223 (2013) 795–805
Figure 1. Particle size distribution of sorbent AX using QICPIC particle analyzer.
Figure 2. SEM photomicrograph of sorbent AX after testing under temperature excursion
Chemical Engineering Journal 223 (2013) 795–805
Effluent exit CO2 Analyzer
Sorbent slumped
N2 flow control
Dilution air
Cooling/ heating coils
Thermocouples
CO2 flow control Humidifier Flue gas inlet
Distributor Plenum
Figure 3. Schematic diagram of fluid-bed column system
Chemical Engineering Journal 223 (2013) 795–805
90
100000 a
85
T (o C)
75
60000
70
CO2 (ppm)
80000
80
40000
65 top
60
middle
20000
bottom
55
CO2 out
50 0
500
1000
0 1500
Time (s)
70
100000 b
top
65 middle
60
CO2 out
60000
50 40000
45
CO2 (ppm)
bottom
55 T (o C)
80000
40 20000 35 30 0
500
1000
1500
0 2000
Time (s)
Figure 4. Typical temperature profiles across the fluid bed and raw exit CO2 concentrations (with dilution) during breakthrough testing of sorbent AX at a) 70 °C and 33.3% CO2 , and b) 40 °C and 16.6% CO2.
Chemical Engineering Journal 223 (2013) 795–805
1.0 CO2 =33.3% 0.8
Ct/C0
0.6 o 40 C
0.4
55 70
0.2
90
0.0 0
200
400
600 800 Time (s)
1000
1200
1400
Figure 5. Effect of temperature on breakthrough curved at feed 33.3% CO2 and flow rate of 30 slpm.
Chemical Engineering Journal 223 (2013) 795–805
1200 CO2 =33.3%
y = -0.074x + 580.79 R² = 0.9797
1000
y = -0.0686x + 698.44 R² = 0.9961 y = -0.0527x + 644.34 R² = 0.9791 y = -0.0462x + 567.9 R² = 0.9473
Time (s)
800 600 90 o C
400
70 55
200
40 0 -1.E+04
-5.E+03
0.E+00
5.E+03
1.E+04
2.E+04
ln[C0 /(Ct-1]/C0
Figure 6. Experimental breakthrough data plotted according to Eq. 4. 1.0 CO2 =16.7%
0.8
0.15 0.6 Ct/C0
0.10 40 oC
0.4
0.05 55
0.2
70
0.00 0
500
0.0 0
500
1000 1500 Time (s)
2000
2500
Figure 7. Effect of temperature on breakthrough curved at feed concentration of 16. 7% CO2 and flow rate of 30 slpm including insert of the initial breakthrough.
1000
Chemical Engineering Journal 223 (2013) 795–805
0.005 CO2 =33.3%
0.004
40oC
d(Ct/C0)/dt,max ↙
55 70
d(Ct/C0 )/dt
0.003
90 0.002 0.001 0.000 0
200
400
600 800 Time (s)
1000
1200
1400
Figure 8. Effect of temperature on breakthrough rate at feed concentration of 10slpm and flow rate of 30 slpm. 0.0050 CO2 =33.3%
0.0040
d(Ct/C0 )/dt
0.0030 0.0020 40 55 70 90
0.0010 0.0000 0.0
0.2
0.4
oC
0.6
0.8
1.0
Ct/C0
Figure 9. Effect of temperature on breakthrough rate at 33.3% CO2 and flow rate of 30 slpm.
Chemical Engineering Journal 223 (2013) 795–805
-3
y = 1948.9x - 10.264 R² = 0.9836 33.3% CO2
lnk,s-1
-4
16.6% -5 y = 2557.3x - 12.798 R² = 0.9566 -6 0.0027 0.0028 0.0029
0.003 0.0031 0.0032 0.0033 1/T ,K-1
Figure 10. Temperature dependence of the reaction rate, k (1/s).
130
qs (mg-CO2 /g-sorbent)
120 110 100 90 80
16.67%CO2 33.30%
70 60 0
20
40 60 Temperature, o C
80
100
Figure 11. Comparison of the experimental (symbol) and predicted values (solid line) of CO2 loading for two different inlet CO2 concentrations at different temperatures.
Chemical Engineering Journal 223 (2013) 795–805
140
qs (mg CO2/g sorbent)
120 100 80 60 40
TGA on 196C
20
FB uptake on AX
0 20
40
60
80
100
120
T (°C) Figure 12. Measured CO2 uptake values for sorbent AX at different nominal temperatures as compared to TGA measurements for sorbent 196C normalized to 40% PEI loading.
Chemical Engineering Journal 223 (2013) 795–805
Saturation time,ts,t.95
Saturation
Effluent Concentration
C0
Breakthrough Breakthrough time,tb,t.05 Volume treated C1
Clean Sorbent
Time C2
C3
C4
C0
C0
C0
MTZ
MTZ
C0
Figure 13. Movement of the mass transfer zone and the resulting breakthrough curve.
Chemical Engineering Journal 223 (2013) 795–805
1.0 T=70 o C
0.8
Ct/C0
0.6 15 slpm
0.4
30 0.2 0.0 0
500
1000 Time (s)
1500
2000
Figure 14. Effect of flow rate on breakthrough curves using feed CO2 concentration of 33.3% 1.0 T=70 o C
Ct/C0
0.8
0.6
0.4
16.6% CO2 33.3%
0.2
0.0 0
500
1000 Time (s)
1500
2000
Figure 15. Effect of feed concentration (C0) on breakthrough curves at flow rate of 30 slpm.
