Adsorption

, Volume 20, Issue 1, pp 201–210

Microwave assisted vacuum regeneration for CO2 capture from wet flue gas

Authors

    • Department of Chemical and Biomolecular EngineeringThe University of Melbourne
  • Jun Zhang
    • Department of Chemical and Biomolecular EngineeringThe University of Melbourne
Article

DOI: 10.1007/s10450-013-9563-y

Cite this article as:
Webley, P.A. & Zhang, J. Adsorption (2014) 20: 201. doi:10.1007/s10450-013-9563-y
  • 334 Views

Abstract

Small scale processing of flue gas with the goal of enriching the stream in CO2 for sequestration or use is an interesting application area for adsorption technology. For example, boiler flue gas which may contain up to 10 % (v/v) CO2 in air can be readily enriched to a stream containing >70 % CO2 which may be ideal for use within a process such as acidification, precipitation, stripping, etc. The challenge in these applications is producing high purity CO2 without excessive energy use and handling high concentrations of water vapor without the added complication of a pre-drying stage. In this study we have examined the use of microwave assisted vacuum as a way of rapidly directing thermal energy to the adsorbent surface to liberate water and CO2. Preliminary “proof-of-concept” pump down experiments were conducted on a small transparent adsorption column of 13X zeolite pre-saturated with a 12 % CO2 in N2 gas mixture. Both wet and dry gas tests were conducted. The addition of microwave radiation improved the rapid desorption of CO2 and water and improved the integrated CO2 purity in the blowdown stream from 60 to 80 %. In the case of dry CO2 mixtures, the enhancement is due to microwave heating of the 13X zeolite facilitated by the high cation density in the faujasite structure. In the case of water and CO2 desorption, the temperature rise of the adsorbent upon microwave heating was much lower than that predicted by simple heating suggesting that the microwave radiation is absorbed primarily by the adsorbed water. A simplified energy analysis suggests that brief exposure of an adsorbent to microwave radiation will raise the required vacuum level for regeneration of high humidity flue gas streams and may lead to an overall lower energy penalty. The selective ability of microwave radiation to target different species provides scope for optimized, compact, flue gas treatment systems.

Keywords

CO2 captureMicrowave regenerationZeolite 13XVacuum swing adsorption

List of symbols

C

Total gas concentration (mol/cm3)

Ci

Concentration of species i (mol/cm3)

t

Time (s)

\(\varepsilon_{b}\)

Bed voidage

\(\rho_{\text{ads}}\)

Adsorbent density (g/cm3)

\(\bar{n}_{i}\)

Loading of component i (mol/g)

\(\bar{n}_{T}\)

Total loading (mol/g)

\(\bar{n}_{i}^{*}\)

Equilibrium loading of component i (mol/g)

u

Gas interstitial velocity (cm/s)

z

Axial distance (cm)

ki

Linear driving force constant (1/s)

Deff

Effective diffusion coefficient for i in the adsorbent (cm2/s)

Rp

Pellet radius (cm)

Cp, ads

Adsorbent specific heat (J/g K)

Cp,gas

Gas specific heat (J/mol K)

T

Temperature (K)

ΔHi

Isosteric heat of component i (J/mol)

\(\dot{Q}_{gen}\)

Rate of heat generation by external means (W/g)

hw

Bed-wall heat transfer coefficient (W/cm2 K)

Tw

Wall temperature (K)

D

Bed inner diameter (cm)

m1i, m2i

Isotherm parameters (mol/g)

b10i, b20i

Isotherm parameters (1/bar)

Q1i, Q2i

Isotherm parameters (J/mol)

P

Pressure (bar)

PMW

Microwave power per unit volume (W/m3)

\(\omega\)

Microwave frequency (rd/s)

\(\varepsilon_{\text{o}}\)

Dielectric permittivity of vacuum (F/m)

\(\varepsilon^ {\prime \prime}\)

Dielectric loss coefficient

E

Microwave electric field (V/m)

1 Introduction

The use of adsorption technology to capture CO2 from a variety of process streams had been extensively investigated particularly in the last 15 years (Casas et al. 2013; Xiao et al. 2012; Won et al. 2012; Wang et al. 2012; Su and Lu 2012; Liu et al. 2012; Kelley et al. 2010; Tlili et al. 2009). A very large number of promising adsorbents have been developed for this purpose with properties tailored for particular process gas streams (Sumida et al. 2012; Lin et al. 2012; Lee et al. 2012; Wang et al. 2011a; D’Alessandro et al. 2010; Choi et al. 2009). The economic assessment of an adsorption system for CO2 capture is often driven by energy costs and in adsorption technology the energy is needed for regenerating the beds. The suitability of the regeneration option is dependent on the application (feed conditions and product specifications) and even the specific site location for the process as well as integration options at the site. For example, a remote site without ready access to steam may favor electrical energy provided by on-site generators as an input (e.g. vacuum swing adsorption) rather than thermal energy. On the other hand, locations with available low quality waste heat (e.g. a power plant) make temperature swing operation more attractive. In the suite of regeneration options are pressure reduction (PSA or VSA), temperature swing processes (either heating through the vessel walls or internal heaters or hot gas purge), electrical swing processes (which are an alternative heating option and really fall within the TSA field), and purge desorption (e.g. moisture swing adsorption, Wang et al. 2011b). Combinations of these regeneration options have also been investigated (Su and Lu 2012; Wurzbacher et al. 2011).

As is well known, temperature swing processes provide a stronger driving force for desorption of CO2 giving a larger working capacity but are hindered by the large time scale involved in heating and cooling large vessels of adsorbent. In this respect, research towards rapid thermal processes is a logical next step in TSA development. In addition, the use of a hot purge gas (such as steam) dilutes the recovered CO2 and a separate downstream separation is required. On the other hand, pressure swing processes can operate extremely fast but require more valuable, high quality electrical energy input and, depending on how tenaciously the CO2 is held to the surface, may require very low pressures to regenerate. These low pressures (often less than 10 kPa) lead to large piping and valve sizes to minimize pressure drop and therefore restrict the scalability of the technology. One of the key challenges for pressure or vacuum swing adsorption separation of CO2 from N2 (e.g. flue gas applications) is to produce sufficiently pure CO2 product (often >95 %) without requiring excessively low vacuum levels. Past reported studies (e.g. Su and Lu 2012) suggest that for 13X zeolite, vacuum levels below 10 kPa are needed to produce CO2 product purities of greater than 95 %. Alternatively (or in addition), a CO2 product purge is often required to enhance the system purity. This step may lead to a reduction in CO2 recovery as the effluent from this step may contain CO2. It is therefore interesting to investigate alternative regeneration methods which may lessen the severity of the vacuum levels needed to achieve satisfactory CO2 purity.

In addition to the energy required for CO2 removal, the high water levels present in the system offer additional challenges. Often flue gas contains almost as much H2O as CO2. Our earlier studies attempted to deal with this high water level by a multilayered approach or process modifications (Li et al. 2011, 2008). While it is possible to control the location of the water loaded zone in the bed, high water levels require unreasonably deep vacuum levels for regeneration and not all zeolites can be regenerated in this way.

One interesting option for rapid adsorbent regeneration is the application of microwaves. This regeneration method has been investigated for over 50 years particularly for dehydration and VOC removal (Chandrasekaran et al. 2012). The interaction of microwaves with a material may show thermal and athermal effects, the latter particularly exploited in microwave chemistry. The presence of water in many CO2 capture applications makes the microwave technique particularly attractive. Zeolites and CO2 are transparent to microwave radiation whereas water shows substantial absorption. The dielectric loss coefficient (\(\varepsilon^ {\prime \prime}\)) of water (which directly relates to the conversion of radiation to heat) is substantially larger than for zeolite or CO2 (Turner et al. 2000), therefore adsorbed water is selectively heated rather than heating the entire adsorbent and gas system, resulting in a potential enhancement of desorption with lower energy consumption. Studies utilizing microwave radiation for regenerating desiccant zeolite rotors used in air dehumidification (Ohgushi and Nagae 2003; Kubota et al. 2011a, b) found that the desorption rate was about 5 times higher for microwave heating at 800 W than for hot-air heating. Also, the amount of water desorbed from the zeolite by microwave heating was 1.6–2.0 times larger than by hot-air heating.

Pure silica zeolites have low dielectric loss and do not heat appreciably under microwave radiation. However, as suggested by others (Turner et al. 2000), the silica bonds which end in silanols at the surface or defects may have substantially higher dielectric loss coefficients—in particular, dangling OH bonds can become rotationally excited producing heat which is transferred directly to the bulk of the crystal. This suggests that the surface can be deliberately targeted in a microwave desorption process. Dry NaX zeolite can be heated to greater than 1,200 °C by microwave radiation over 6 min while a wet NaX sample reached only 33 °C in 6 min (Whittington and Milestone 1992). This study speculated that the Na+ cations in the supercage are excellent microwave absorbers and substantially increased the thermal effect, in contrast to high silica materials such as de-aluminated Y zeolite (DAY). Zeolites with different water loading levels exhibit different overall dielectric properties and consequently respond differently to microwave radiation (Carreras et al. 1987).

To take advantage of the excellent microwave selectivity of water in the adsorbed phase, a novel approach in flue gas CO2 capture, for both dry and wet systems, is put forward in this study. Microwave energy is introduced into the CO2 separation system during the desorption step to enhance the overall performance. Preliminary proof of concept experiments were conducted to quantify the effect and a simple mass and energy balance were developed and validated with the data to provide insight into the underlying mechanisms. A system energy consumption analyses was also performed to assess this idea. Our goal in this study was not to utilize microwave radiation alone but to examine the features of a hybrid vacuum-microwave system. Our intention is to retain the vacuum as the primary source of regeneration energy while using microwave assist to mitigate the deep vacuum levels needed for conventional vacuum desorption—we denote this process microwave assisted vacuum swing adsorption (MVSA). We are particularly interested in using microwave assist for wet gas systems with high gas humidity. For flue gas streams with high water content (>5 % water) very low vacuum levels are often required (<3 kPa) to provide sufficient purge to remove the water (Li 2010). It may be possible to elevate these vacuum levels to more practical values (e.g. ~20–30 kPa) by a short exposure of part of the bed to microwave radiation.

The experiments presented in this study are preliminary and were conducted to assess the merits of MVSA in the first instance.

2 Experimental work

2.1 Materials and isotherms

In this study, 13X zeolite (UOP APG I, (8 × 12 beads) was chosen as the adsorbent because of its high CO2/water adsorption capability, the presence of numerous studies on CO2 capture using 13X, and stability under repeated cycling. Adsorption isotherms for CO2, N2 and water were measured over the range 20–90 °C and are reported in our earlier work (Zhang et al. 2009). Isotherms for CO2 and N2 were measured by a Micromeritics ASAP2010 and water isotherms were measured by an Intelligent Gravimetric Analyser (IGA), Hiden Isochema, UK.

2.2 Microwave desorption experiments

A transparent plexiglass adsorption column (ID 20 mm, OD 30 mm, Packed Length 100 mm) was packed with 20 g of dry, activated UOP APG I adsorbent The column was placed inside a microwave chamber (2.45 GHz) in a vertical orientation. A schematic of the column-microwave system is provided in Fig. 1. The temperature of the adsorbent was not measured in the column since conventional thermocouples are microwave absorbing. The temperature of the gas entering and leaving the column was measured outside the microwave chamber as shown by the location of the thermocouples in Fig. 1. To control the rate of pump down, we placed a valve in the suction line between the bed and the vacuum pump. The bed pressure was measured between the valve and the suction of the vacuum pump during the pump down experiments. A premixed gas mixture of 12 % CO2 in N2 was used for the experiments.
https://static-content.springer.com/image/art%3A10.1007%2Fs10450-013-9563-y/MediaObjects/10450_2013_9563_Fig1_HTML.gif
Fig. 1

Schematic of the microwave-adsorption experimental apparatus

Pump down experiments were performed by first saturating the column with the premixed gas by doing a full breakthrough experiment and ensuring that the temperature and composition were uniform. Thereafter, the exit of the column was closed and a vacuum applied (BOC Edwards E2M1.5 vacuum pump) to the feed end (countercurrent to the feed) for a fixed duration.

Pump down experiments were conducted for (a) the column pre-saturated with a dry CO2/N2 (12 % CO2 in N2) gas mixture at 120 kPa and 20 °C; and (b) a column pre-saturated with a wet CO2/N2 (12 % CO2 in N2) gas mixture in which the relative humidity of the mixed gas was 18 % at 120 kPa and 20 °C (H2O mole fraction of 0.0048). The wet gas was prepared by passing the dry gas through a water filled bubbler controlled at a fixed temperature and blending the wet gas with a dry gas stream of 12 % CO2. The dew point of the blended feed gas was measured with an in situ water sensor (HMT330, VAISALA, Finland, ±0.05 % full scale).

In each case the pump down was conducted with and without the presence of microwave radiation. The response parameters included integrated CO2 purity in the pump down gas, temperature history, pressure history, gas flow history and energy consumption, either measured or calculated.

It was found that the microwave oven has an uneven distribution of microwave radiation, as expected. Careful calorimetric experiments were conducted to “map” the energy profile within the oven. After mapping the energy field, the column was placed in a uniform radiation field to achieve more homogeneous heating. These same calorimetric measurements showed that the power received at the column when filled with water was approximately 840 W when the microwave setting was 1,100 W. Furthermore, only short durations of microwave radiation were used. It was found that excessive radiation time (>40 s) could lead to damage of the column and its contents.

2.3 Adiabatic model for MVSA

To help interpret the experimental results and provide a basis for comparison with alternate regeneration strategies, a simple pump down model of the MVSA process was developed. The model is a conventional set of mass and energy balance equations written for a single desorption step for a three component system: H2O, CO2 and N2. The model only captures first order effects and omits features such as axial dispersion of mass and energy, thermal conductivity in the axial direction, and heat loss to the wall. The major equations are:

Species i balance where i is CO2 and H2O (N2 determined by difference: CN2=C–CH2O–CCO2):
$$\frac{{\partial C_{i} }}{\partial t} + \left( {\frac{{1 - \varepsilon_{b} }}{{\varepsilon_{b} }}} \right)\rho_{ads} \frac{{\partial \bar{n}_{i} }}{\partial t} = - \frac{{\partial uC_{i} }}{\partial z}$$
(1.1)
Total mass balance:
$$\frac{\partial C}{\partial t} + \left( {\frac{{1 - \varepsilon_{b} }}{{\varepsilon {}_{b}}}} \right)\rho_{ads} \frac{{\partial \bar{n}_{T} }}{\partial t} = - \frac{\partial uC}{\partial z}$$
(1.2)
Rate of adsorption for each species:
$$\frac{{{\text{d}}\bar{n}_{i} }}{{{\text{d}}t}} = k_{i} \left( {\bar{n}_{i}^{*} - \bar{n}_{i} } \right), \quad k_{i} = \, \frac{{15D_{eff} }}{{R_{p}^{2} }} \; i = {\text{ CO}}_{ 2} , {\text{ H}}_{ 2} {\text{O, N}}_{ 2} \varepsilon^ {\prime \prime}$$
(1.3)
Energy balance
$$\rho_{ads} \left( {\frac{{1 - \varepsilon_{b} }}{{\varepsilon_{b} }}} \right)C_{p,ads} \frac{\partial T}{\partial t} - \rho_{ads} \left( {\frac{{1 - \varepsilon_{b} }}{{\varepsilon_{b} }}} \right)\sum\limits_{i = 1}^{3} {\Updelta H_{i} \frac{{\partial \bar{n}_{i} }}{\partial t}} = \rho_{ads} \left( {\frac{{1 - \varepsilon_{b} }}{{\varepsilon_{b} }}} \right)\dot{Q}_{gen} - C_{\text{p,gas}} \frac{\partial uCT}{\partial z} - \frac{{4h_{w} }}{{\varepsilon_{b} D}}\left( {T - T_{w} } \right)$$
(1.4)
Dual Site Langmuir Equilibrium Model
$$n_{i} = \frac{{m_{1i} b_{1i} y_{i} P}}{{1 + b_{1i} y_{1i} P}} + \frac{{m_{2i} b_{2i} y_{i} P}}{{1 + b_{2i} y_{i} P}};{\text{ i = H}}_{ 2} {\text{O,CO}}_{ 2} , {\text{N}}_{ 2}$$
(1.5)
$$b_{1i} = b_{10i} \exp \left( {\frac{{Q_{1i} }}{RT}} \right); \, b_{2i} = b_{20i} \exp \left( {\frac{{Q_{2i} }}{RT}} \right)$$
The velocity at the exit of the column, uexit, was calculated from the actual volumetric flow rate as supplied by the vendor pump curve which was a function of the suction pressure. A valve placed between the column and the pump also changed the profile of the pump down curve and this was modeled using a valve equation (Chou and Huang 1994). It should be noted that the molar flow rate is not constant over the pump down period. The model is run with a constant actual volumetric flow rate of the Edwards vacuum pump (which, over the pressure range is very flat and constant) together with a valve equation between the vacuum pump and the bed. The boundary condition on the bed can also be quite well represented as a constant velocity. The instantaneous and integrated molar flow rate is calculated and the integrated CO2 concentration is then calculated from the integrated moles which have been accumulated at each time. The isotherm data were fit to a Dual Site Langmuir Equation for each individual species and the isosteric enthalpies, ΔHi were determined as a function of loading by the Clausius–Clapyron Equation. For the dry gas simulations, the competitive adsorption of CO2 and N2 was modeled using the extended Langmuir Equation with the parameters shown in Table 1. For the wet gas cases, the adsorption of CO2 and N2 were first calculated as indicated in the dry case and then reduced in line with the experimental observation of Wang and Levan (2010) according to the water loading. The rate constants for CO2, N2 and H2O were determined by matching the breakthrough curves to the simulation data to get the effective diffusivity values, Deff. The instantaneous bed pressure was calculated from the ideal gas equation once the gas concentration C was solved for. The set of PDE’s described in Eqs. (1.11.4) were discretized spatially using the finite volume method with Sweby flux limiters and solved using the MATLAB routine ode15 s to provide the dynamic and integrated information. The program was also used in scale up calculations to estimate the total energy required. All of the model parameters are provided in Table 1.
Table 1

Model parameters used in simulation

Item

Value

 

Bed properties

 Bed voidage

0.37

 

 Adsorbent internal voidage

0.50

 

 Adsorbent density (g cm3)

1.011

 

 Deff, H2O

6 × 10−6

 

 Deff, CO2

5 × 10−5

 

 Deff, N2

7 × 10−5

 

 Particle diameter

8 × 12 mesh

 

 Bed internal diameter (cm)

2.0

 

 Bed length (cm)

10.0

 

Operating conditions

 Initial column pressure (bar)

1.2

 

 Initial column temperature (K)

300

 

 Initial column mole fraction

Dry

Wet

 

 H2O

0.00

0.0048

 

 CO2

0.12

0.119

 

Isotherm parameters

H2O

CO2

N2

 m1 (mmol/g)

5.699

2.610

2.456

 b10 (1/bar)

1.169 × 10−6

1.510 × 10−6

2.89 × 10−4

 Q1 (J/mol)

47,965

30,180

16,000

 m2 (mmol/g)

9.262

2.081

0.000

 b20 (1/bar)

3.088 × 10−6

7.090 × 10−5

0.000

 Q2 (J/mol)

58,488

35,181

0.000

3 Results and discussion

3.1 Dry gas experiments with and without microwave

Figure 2 shows the integrated concentration of CO2 in the suction gas from the bed during the pump down step for the case of the adsorbent bed pre-saturated with dry, 12 % CO2 in N2 gas. The curves are continuous output from the Servomex CO2 analyzer attached to the discharge of the vacuum pump. The concentration is therefore not instantaneous CO2 concentration of the gas leaving the bed since accumulation within the vacuum pump, analyzer line etc. will cause integration of the CO2 profile. From 0 to 360 s there is no application of microwave radiation and the concentration of CO2 in the vacuum line increases as expected. The model is able to capture the rise quite well. The simulation predicts a rise in CO2 concentration to a final integrated CO2 purity of 63 %, in good agreement with the experimental data. The microwave is applied at 360 s for 25 s. The model shows a rapid rise in CO2 concentration in the integrated pump down gas to 80 %, in good agreement with the data. The additional desorption of CO2 can be interpreted entirely as a thermal effect since CO2 is transparent to microwave radiation. Thus, heating of the adsorbent is rapid and, unlike using a hot purge gas which would dilute the CO2 product, the effluent CO2 concentration is significantly improved. The microwave heating in our model is represented as a simple Joule heating effect (inclusion of Qgen in the energy balance) which appears to be adequate for this case.
https://static-content.springer.com/image/art%3A10.1007%2Fs10450-013-9563-y/MediaObjects/10450_2013_9563_Fig2_HTML.gif
Fig. 2

Integrated mole fraction of CO2 in the pump down gas during the pump down process—dry gas case

The pressure and temperature history of the gas leaving the bed entering the vacuum pump is shown in Fig. 3a, b. The pressure profile matches the data well.
https://static-content.springer.com/image/art%3A10.1007%2Fs10450-013-9563-y/MediaObjects/10450_2013_9563_Fig3_HTML.gif
Fig. 3

a Bed pressure and b exit gas temperature during pump down of the bed (dry gas)

The simulated temperature history shows the instantaneous bed temperature and the temperature of the integrated gas leaving the bed, assuming no heat transfer to the piping leaving the bed. In practice, the rather small scale of the experiment and large surface to volume ratio means that ambient heat transfer will mitigate temperature changes, as shown by the experimental data in Fig. 3c. Nevertheless, it is clear that there has been a considerable temperature rise on application of the microwave radiation which has more than offset the drop in temperature due to desorption of the CO2 in the bed.

This is consistent with a pure thermal effect as zeolite heating under microwave radiation is well known (Whittington and Milestone 1992).

3.2 Pumpdown of bed pre-saturated with wet gas

Figure 4 shows the integrated CO2 concentration (dry basis) in the gas leaving the bed for the case where we have pre-saturated the column with a wet CO2/N2 mixture also containing 12 % CO2. The relative humidity of the stream was 18 % as measured by the Vaisala humidity meter corresponding to a water mole fraction of 0.0048. Microwave radiation is applied at 400 s for 20 s. Figure 4 also shows the prediction from the simulation which is reasonably close but under predicts the final CO2 level. We see that in a wet gas system, the peak integrated CO2 purity increase upon microwave addition (from ~55 to ~68 %) is less than that in a dry system (~60 to ~80 %). This is due to the lower CO2 loading on the bed initially in the presence of water vapor as well as the lower temperature increase of the system. Figure 5 shows the temperature of the gas leaving the bed. In this case, the desorption of water by microwave radiation has attenuated the temperature rise in the bed far below that predicted by the model. It is clear that the absorption of microwave radiation by water is not satisfactorily represented by a simple heating term as in the case of a dry gas. Comparison with Fig. 3b shows that the maximum experimentally measured temperature rise for the dry case (320 K) is considerably higher than that observed for the wet case (305 K). This trend is consistent with that observed by Turner (2000) in their study of cyclohexane and methanol desorption from silicalite and DAY. In that study, it was found that virtually all of the heating came from the microwave’s interaction with the adsorbent when the material with low dielectric loss (such as cyclohexane) was adsorbed.
https://static-content.springer.com/image/art%3A10.1007%2Fs10450-013-9563-y/MediaObjects/10450_2013_9563_Fig4_HTML.gif
Fig. 4

Integrated mole fraction of CO2 in the pump down gas—wet gas case

https://static-content.springer.com/image/art%3A10.1007%2Fs10450-013-9563-y/MediaObjects/10450_2013_9563_Fig5_HTML.gif
Fig. 5

Temperature of the gas leaving the bed for the case in which the bed is initially saturated with wet gas

In summary, the adsorbent temperature in the dry gas cases increased considerably more than the wet gas system, which indicates that in the wet gas case, the radiation has been directed at liberating water from the surface of the adsorbent with little increase in adsorbent temperature. This is a useful athermal feature of microwave radiation for wet adsorbents and deserves closer study to understand the role of zeolite surface chemistry such as cation density (the cations acting as microwave absorbers) and silanols groups on water desorption.

3.3 Model analysis

To provide some theoretical insights and to assess the merits of the microwave assisted proposal, numerical simulations were performed using the model described above. In this instance we have assumed a dry feed (no water). Figure 6a shows the effects of heat generation on the integrated CO2 purity obtained by desorption from the bed for different heat inputs. As before, we have represented the heat input solely as \(\dot{Q}_{gen}\) in Eq. (1.4). In practice, the absorbed power per unit volume of material is:
$$P_{MW} = \varepsilon^{\prime \prime} \varepsilon_{o} \omega E^{2}$$
(1.6)
where \(\omega\) is the angular microwave frequency, \(\varepsilon^ {\prime \prime}\) the dielectric loss coefficient (imaginary part of the complex relative dielectric permittivity), \(\varepsilon_{o}\) the dielectric permittivity of vacuum, and E the microwave electric field. The absorption of the microwave radiation by the system therefore depends on \(\varepsilon^ {\prime \prime}\) and the relative amount (volume) of the components in the system. Of the components, the adsorbent (NaX) has the largest relative volume but low \(\varepsilon^ {\prime \prime}\). The adsorbed water has high \(\varepsilon^ {\prime \prime}\) but a low volume and the CO2 and N2 have both low volume and \(\varepsilon^ {\prime \prime}\). Therefore we expect an interesting change in P during the desorption process as the water is first removed and then, depending on its adsorbed amount (and isotherm shape), the zeolite is thermally heated. Regardless, this level of detail is not yet available and our analysis simply uses a single \(\dot{Q}_{gen}\) term regardless of specificity. In addition, we have not accounted for losses (i.e. <100 % effectiveness of heating)—the power scale in Fig. 6 should therefore be regarded as indicative only. As shown in Fig. 6(a), with the increase of microwave power level, the CO2 purity steadily increases with the marginal gain decreasing with continued increase of microwave power level. We expect this as the non-linear isotherm for CO2 would make initial temperature increases advantageous but after some temperature, additional desorption would be negligible given the drop in CO2 capacity.
https://static-content.springer.com/image/art%3A10.1007%2Fs10450-013-9563-y/MediaObjects/10450_2013_9563_Fig6_HTML.gif
Fig. 6

Effect of microwave power level on a CO2 purity, b CO2 working capacity

A similar pattern was also observed with the working capacity of the adsorbent (Fig. 6b). We have defined working capacity as the net amount of moles of CO2 removed during the desorption step, per unit mass of adsorbent. With the help of microwave energy, the purity exceeds 92 % and the productivity reaches 2.66 mol/kg adsorbent (a typical CO2VSA system working capacity is usually less than 1 mol/kg). This represents a large decrease in bed volume with corresponding capital savings. Figures 7 and 8 show the effect of the microwave radiation on the vacuum level needed to achieve the required purity. This was important to our study and shows that with the assistance of microwave, we may be able to save pumping energy to achieve the same purity and working capacity as that achievable only under deep vacuum. The addition of microwave energy has changed the desorption path and provided a more isothermal pump down process. For example, in the absence of microwave-assist, a vacuum level of 3 kPa is needed to achieve 70 % purity. This vacuum level of 3 kPa is frequently needed to achieve high CO2 purities in CO2VSA systems (Li et al. 2011, 2008) and is difficult to achieve at large scale. The addition of microwave energy raises this to 8 kPa for the same purity. The microwave power only plays a minor role in this process compared with the additional vacuum pump energy needed. Considering the large improvement of working capacity and adsorbent regeneration achieved after microwave desorption, it may be worthwhile to spend this extra energy on the separation process depending on individual applications. An optimized combination of microwave power and vacuum power may be able to reduce the overall energy requirements.
https://static-content.springer.com/image/art%3A10.1007%2Fs10450-013-9563-y/MediaObjects/10450_2013_9563_Fig7_HTML.gif
Fig. 7

Effect of desorption pressure on CO2 purity with and without microwave radiation. Isothermal case included as reference

https://static-content.springer.com/image/art%3A10.1007%2Fs10450-013-9563-y/MediaObjects/10450_2013_9563_Fig8_HTML.gif
Fig. 8

Effect of desorption pressure on CO2 working capacity with and without microwave radiation. Isothermal case included as reference

These simple calculations are based on a bed uniformly saturated with a gas of a fixed composition (12 % CO2 in N2, dry). During an actual vacuum swing cycle, the composition profile in the bed at the start of the pump down step will not be uniform. It is likely that there will be far less CO2 adsorbed near the exit of the bed. In addition, if a CO2 product purge is used, the entrance region of the bed will contain a much higher concentration of CO2 (close to the product concentration). In this case we expect the results shown in Figs. 5, 6, 7 and 8 to change. It is not possible to predict these without a full cycle simulation since the presence of microwave radiation during a full cycle and its effect on the extent of require product purge is not known. Given the promise suggested by our initial proof of concept experiments, we are currently undertaking an analysis of this situation.

3.4 Calculation of energy requirements

To assess the effectiveness of microwave energy and vacuum energy in achieving CO2 desorption, we have used our model to calculate the desorption energy needed and purity attainable for 4 cases of pump down of a bed: (a) a bed saturated with dry gas (12 % CO2/N2) employing vacuum only down to 30 kPa (b) a bed saturated with dry gas (12 % CO2/N2) employing vacuum and microwave energy (c) a bed saturated with wet gas (12 % CO2/N2) employing vacuum only down to 30 kPa and (d) a bed saturated with wet gas (12 % CO2/N2) employing vacuum and microwave energy. We have chosen 30 kPa as a reasonably achievable vacuum pressure using a single stage blower. The results are shown in Table 2. It is seen that with a small amount of microwave energy, a dry gas system yields much higher purity and productivity. The addition of microwave energy helps attenuate the temperature drop [which was from 20 to 7 °C in case (a)] providing higher productivity and purity. The additional recovery has lowered the vacuum energy costs by more than the required additional microwave energy resulting in an overall lower total energy.
Table 2

Comparison of the energy requirement for four pump down scenarios

 

Dry gas, no MW

Dry gas, MW

Wet gas, no MW

Wet gas, MW

Integrated CO2 purity (%)

28.33

39.60

24.04

31.63

Productivity (mol/kg)

0.618

1.108

0.30

1.046

Vacuum pump energy (MJ/kg CO2)

1.47

0.93

1.96

0.82

Microwave energy (MJ/kg CO2)

0

0.10

0

1.07

Total energy (MJ/kg CO2)

1.47

1.03

1.96

1.90

Final desorption temperature (°C)

7.0

12.0

5.0

56.3

For a wet system, since water is a very strong energy absorber (high \(\varepsilon^ {\prime \prime}\)), we expect much of the microwave power to be absorbed by the adsorbed water (Eq. 1.6). The wet gas case (d) shows lower purity and productivity due to reduction in CO2 capacity in the presence of competing water. As mentioned earlier, we have found experimentally that under the same microwave irradiation conditions, a wet system temperature rise is much smaller than a dry system. This contrasts with our calculation in Table 2 which shows a temperature increase to 56.3 °C. This reason for this discrepancy is the way in which we have represented the addition of microwave power through a simple \(\dot{Q}_{gen}\) term. Clearly, the effect of microwave radiation on removal of water is not a simple thermal effect and we will need to understand and improve our representation of this effect in our model to match our experimental data. As suggested by Carreras et al. (1987), zeolites with different water loading levels exhibit different overall dielectric properties and respond differently to microwave radiation. The addition of a component-specific “absorption” fraction to \(\dot{Q}_{gen}\) will be needed based on Eq. (1.6). The total energy for the wet case may therefore be lower than that calculated in Table 2.

The different effects of microwave radiation on the dry case and wet case provide scope for innovative design of the microwave system. In usual practice, the wet and dry zones in an adsorbent bed may be spatially separated with the wet zone coincident with the feed and may even use a different adsorbent. On the other hand, the dry zone is usually located towards the exit of the bed. Application of the microwaves to the wet zone only (or perhaps differing amounts to the two zones) can provide an optimal irradiation profile to minimize total energy in concert with an appropriate vacuum level. In addition, the use of intermittent microwave radiation during the pump down process can be optimized to correspond to changes in the isosteric heat of water as the loading decreases.

3.5 Practical aspects and implementation

The practical application of microwave technology to adsorption systems is restricted to small systems. Microwave compressed air desiccant dryers are already industrially available (Arrow Pneumatics 2011). The penetration distance of the microwaves dictates the largest diameter which can be used. This penetration distance depends on the system characteristics, especially the dielectric constants and must usually be determined experimentally. Low Si:Al ratio zeolites are expected to have relatively short penetration depths. A rough estimate for NaX is 15 cm (Bathen 2003) whereas more transparent zeolites such as DAY may provide >100 cm penetration depth. This raises an interesting option of using high silica adsorbents in the water guard layer of the system to achieve high penetration depths and assist in water desorption without large temperature increases. A penetration depth of 100 cm translates to a bed diameter of at least 200 cm which is useful for the scale of CO2 capture from flue gas systems of small boilers. The ability to raise the vacuum level from <5 to >30 kPa is a large benefit because of the associated reduction in gas velocity in the suction lines and correspondingly lower pipe and valve sizes. Far more study is needed to understand the impact of microwave radiation on the cyclic stability of an adsorption process, especially the integration of a microwave system into a complex cycle and the quantification of microwave energy needs for different adsorbent systems. The need to cool the bed before re-admitting feed gas for the next cycle remains but the rapid heating and smaller temperature rise we have observed for water desorption suggests that this cooling may be accomplished more rapidly than in conventional TSA.

The MVSA process appears at first glance to be similar to electrical swing adsorption (ESA). However, the latter relies entirely on Joule heating and the impact on the system is independent of the molecules—all are heated together with the heating of the adsorbent. Furthermore, ESA is restricted to conducting adsorbents (e.g. carbon monoliths) and the risk of fire in systems containing high voltage, carbon and oxygen is significant. In contrast, MVSA requires transparent adsorbents (such as high silica zeolites), is specific to adsorbed species with high dielectric loss coefficients, and can be rapidly applied and removed during a cycle.

Finally, it should be recalled that we are converting high quality electrical energy into thermal energy for desorption of molecules and comparison with true TSA processes which use low quality heat should always include the Carnot factor associated with the heat-work conversion. Based on the energy penalty at a typical power plant, this factor is approximately 3.2 (DOE/NETL-2012/1557 2012). Therefore, an electrical specific energy of 1 MJ/kg CO2 for adsorption systems which use vacuum only to regenerate the adsorbent translates to 3.2 MJ/kg of thermal energy which is currently comparable to or slightly greater than than that required for conventional liquid-scrubbing MEA based systems.

4 Conclusions

We have examined the action of microwave radiation on the CO2 capture characteristics of the pump down process for adsorbent regeneration. Experiments conducted on wet and dry CO2/N2 mixtures with and without microwave radiation indicated that in the dry case, the microwaves provided simple heating of the zeolite which led to an improvement in CO2 desorption and an improvement in product purity. Unlike the usual temperature drop accompanying desorption, the action of microwave radiation could keep the bed more nearly isothermal improving purity and capacity. In the case of wet CO2/N2 mixtures, the desorption of water (a strong microwave absorber) attenuated the temperature rise and indicated that microwave assisted vacuum could provide improved CO2 desorption and column regeneration compared to the case with no microwave radiation. A simple model was developed to describe the effect of heating on the process. Representing the heating as a simple uniform heat generation process was not successful in duplicating the data for the wet case. A more detailed microwave power model is needed which captures the targeted nature of the radiation. The analysis indicates that there is scope for integration of a microwave process into a vacuum desorption system and it may be possible to elevate the vacuum levels and provide capital and energy savings. Far more work is needed though to understand the complex relationship between the microwave radiation and the interaction of the zeolite chemistry and guest molecules CO2, H2O and N2.

Acknowledgments

We acknowledge funding for this CO2CRC project provided by the Australian Government through its CRC program.

Copyright information

© Springer Science+Business Media New York 2013