ECN-RX--05-122

Hydrotalcite as CO2 sorbent for sorptionenhanced steam reforming of methane

H.Th.J. Reijers, S.E.A. Valster-Schiermeier, P.D. Cobden and R.W. van den Brink
to be published in the CO2 Capture special issue of Industrial & Engineering Chemistry Research

edited by name: H.Th.J. Reijers function: Project leader d.d.: 2005-07-28 verified by name: S.C.A. Kluiters function: Project leader d.d.: verified by name: J.W. Erisman function: Unit manager d.d.:

modified by name: H.Th.J. Reijers function: Project leader d.d.: authorized by name: R.W. van den Brink function: Group manager d.d.:

JULY 2005

Acknowledgement/Preface
This research has been carried out by ECN in the CATO-project (Carbon Capture, Transport and Storage), which is financed by the BSIK subsidy program of the Dutch Government agency SenterNovem. It is also part of ECN's DECAFF-program (Decarbonisation of Fossil Fuels), which is sponsored by the Dutch ministry of Economic Affairs.

Abstract
Sorption-enhanced reforming of methane is an attractive option for the combined production of electricity and capture of CO2. In this process, the steam reforming catalyst is mixed with a CO2 sorbent. During the reaction, CO2 is adsorbed implying an increase of the hydrogen production rate. Once the sorbent is saturated, it must be regenerated using purge gas, usually steam. The amount of steam needed for CO2 removal from the saturated sorbent determines the system efficiency of the process to a large extent. In this paper, several potassium-promoted hydrotalcite samples, both commercially obtained and in-house prepared, are tested for their suitability as CO2 sorbent for sorption-enhanced reforming of methane. In particular, the purge gas to adsorbed CO2 ratio is determined at various operation conditions. It is shown that this ratio is still too large at the investigated conditions. Mixed with steam reforming catalyst the hydrotalcite sorbent can adsorb sufficient CO2 to enhance the CH4 conversion to almost 100%. Keywords: adsorption, desorption, sorption-enhanced, CO2 capture, hydrotalcite, steam reforming

ECN-RX--05-122

Contents
List of tables List of figures Summary 1. 2. 3. 4. 5. 6. 7. Introduction Sorbent selection Hydrotalcites Experimental Data analysis Results and discussion Conclusions 4 4 5 7 8 9 10 10 11 13 13 13

List of symbols References

ECN-RX--05-122

List of tables
Table 1 Characteristics of CO2 sorbent materials (*** = good, ** = fair, * = poor, ?=unknown) ............................................................................................................ 15 Table 2 Specifications of HTCs according to the supplier ......................................................... 16 Table 3 Standard experimental conditions............................................................................ 17 Table 4 CO2 adsorption capacities at breakthrough and during the whole adsorption step at various conditions ................................................................................... 18

List of figures
Figure 1 Figure 2 Figure 3 Figure 4 Schematic representation of a SERP system for electricity production ....................... 19 Structure of hydrotalcite........................................................................................... 20 Schematic diagram of the lab-scale experimental apparatus ...................................... 21 Approximation of peak area given by the solid line (dashed line: curve fit, symbols:experimental data points) ........................................................................ 22 Figure 5 Adsorption and desorption profiles of various potassium- promoted HTC samples (the desorption profiles of ECN-HTC/22 wt% K2CO3 coincides with that of PURAL MG70/22 wt% K2CO3) .............................................................................. 23 Figure 6 Adsorption and desorption profiles of PURAL MG70 with different K2CO3 loadings 24 Figure 7 Desorption characteristics determined for different desorption flows (the dashed curve is a repetition of 100 ml/min at the end of the experiment) .......................... 25 Figure 8 Desorption characteristics determined for different desorption times (tads = 10 min, Fdes = 30 ml/min) .................................................................................................... 26 Figure 9 Effluent gas composition during reaction and regeneration using a mixture of steam reforming catalyst and HTC impregnated with 22 wt% K2CO3 ................... 27 Figure 10 CO2 flow during regeneration in 17.4% and 29% H2O containing N2 gas ............... 28

ECN-RX--05-122

Summary
Sorption-enhanced reforming of methane is an attractive option for the combined production of electricity and capture of CO2. In this process, the steam reforming catalyst is mixed with a CO2 sorbent. During the reaction, CO2 is adsorbed implying an increase of the hydrogen production rate. Once the sorbent is saturated, it must be regenerated using purge gas, usually steam. Compared with other purge gases (air, nitrogen, methane), steam can be easily separated, thus enabling the supply of a concentrated CO2 stream suitable for sequestration. The amount of steam needed for CO2 removal from the saturated sorbent determines the system efficiency of the process to a large extent. In this paper, several potassium-promoted Mg-Al based hydrotalcite samples, both commercially obtained and in-house prepared, are tested for their suitability as CO2 sorbent for sorption-enhanced reforming of methane. The effects of promoter loading, temperature, duration of the desorption step and purge flow on the adsorption and desorption properties are studied. In particular, the ratio of the amount of purge gas needed to remove all adsorbed CO2 and the amount of adsorbed CO2 ratio is used to find the optimum sorbent properties and experimental conditions. It is shown that this ratio is still too large at the investigated conditions. As a proof-of-principle experiment, it is shown that mixed with steam reforming catalyst, the hydrotalcite sorbent can adsorb sufficient CO2 to enhance the CH4 conversion to almost 100%.

ECN-RX--05-122

ECN-RX--05-122

1.

Introduction

There is a growing awareness that energy must be produced at lower greenhouse gas emissions. Fossil fuels, though, will remain the most important energy source for the first half of this century. This has led to new technologies to reduce the emission of the CO2 produced from the burning of fossil fuels. This CO2 can be captured with end-of-pipe technologies from the flue gas of the energy production unit using a CO2 scrubber, e.g. an amine solution. Other possibilities include the application of nitrogen-free combustion of fossil fuels which facilitate CO2 capture, but require a costly nitrogen-oxygen separation step1. A third possibility is to use precombustion decarbonization. Here, the CO2 produced is captured prior to combustion, while transferring the energy content of the fuel to hydrogen. Various pre-combustion routes for electricity production have been investigated, using a membrane reactor2,3 or the sorption-enhanced reaction process (= SERP)4,5. The relevant reaction(s) are:

CO + H 2 O CO2 + H 2 ,
for water gas shift, and in addition

(1)

CH 4 + H 2 O CO + 3H 2 .

(2)

if steam reforming is included. Conventional steam reforming of methane is performed at high temperatures, typically between 850 and 950 C, and 80% to 90% of the methane is converted to hydrogen. When the steam reforming catalyst is mixed with a sorbent, as in the sorptionenhanced reaction process, the CO2 produced in the above reactions is simultaneously adsorbed, so that the hydrogen production rate is enhanced. Thus, to obtain the same CH4 conversion, the steam reforming process may be performed at lower temperatures compared with conventional reforming, typically between 450 and 550 C. Sorption-enhanced reforming is a batch process necessitating the regeneration of the sorbent which is saturated with CO2. The preferred regeneration or purge gas is steam if the captured CO2 is to be sequestrated. Compared with other purge gases (air, nitrogen or methane), steam can be easily separated from the purge stream, leaving a pure CO2 flow. An example of a sorption-enhanced reforming system for electricity production is shown in Figure 1. Both SERP reactors are filled with a mixture of steam reforming catalyst and CO2 sorbent. A CH4/H2O mixture is fed to reactor 1, purge gas to reactor 2 for regeneration. To improve the desorption of CO2, the temperature may be increased or the pressure may be decreased with respect to the temperature and pressure during adsorption. After the sorbent of reactor 1 is saturated with CO2, the gas flows to both reactors are interchanged and CO2 is desorbed from the bed of reactor 1, whereas steam reforming is performed in reactor 2. Part of the produced H2 is fed to a gas turbine. The other part is used as fuel for the burner, which in turn drives the endothermic steam reforming reaction and the regeneration process. Production of 1 mole of steam is energy-intensive requiring about 20% of the reaction enthalpy of the steam reforming of the same amount of methane. Thus, its consumption for regeneration must be limited in order to obtain an acceptable system efficiency6. None of the mentioned groups who investigated sorption-enhanced steam reforming of methane, has optimized the sorbent properties and the system configuration such that the lowest cost of CO2 removal is obtained. The purpose of this paper is to investigate the relation between the purge gas to adsorbed CO2 ratio and the operation conditions in more detail. The amount of purge steam needed for complete desorption per mol CO2 adsorbed by the bed, will be used as the parameter to characterize the efficiency of the desorption step. This ratio depends only on the material properties of the sorbent, not on the system configuration. It should be in the same range as the steam/methane ratio of the feed gas (i.e. 3 - 6) to obtain an acceptable system efficiency. Note,

ECN-RX--05-122

that the amount of purge steam is expressed with respect to the amount of adsorbed CO2, since it is proportional with the amount of adsorbed CO2 to first approximation. Clearly, this is a simplifying assumption since a real system is more complex. A smaller adsorption capacity implies a higher switch frequency between adsorption and desorption steps, and higher pressure drop losses due to the larger adsorbent bed. Though these will affect the efficiency as well, such factors will not be taken into account here. The purge steam/adsorbed CO2 ratio needed to desorb a certain amount of CO2, hereafter S/CO2, will be experimentally determined. The plot of the desorbed fraction of CO2 versus S/CO2, is called the desorption characteristic. The steeper the characteristic, the more easily the CO2 is desorbed. The relation between the purge steam/adsorbed CO2 ratio for complete desorption, hereafter (S/CO2)c and the properties of the chosen sorbent, promoted hydrotalcite, and operation conditions for sorption-enhanced steam reforming of methane is investigated in this paper.

2.

Sorbent selection

For use in a steam reforming reactor, the sorbent material should fulfill the following requirements: 1. It should be able to remove sufficient CO2 in the temperature range 400 - 600 C and in the pressure range 1 - 40 bar. 2. It should be able to withstand the high psteam/pCO2 ratios at the steam reforming conditions in a SERP reactor (> 20), where psteam and pCO2 are the partial pressures of respectively steam and CO2 gas. 3. It should be mechanically and chemically stable for extended periods at these conditions. 4. The kinetics of adsorption and desorption should be sufficiently fast. From a literature search follows that the materials which are of interest for SERP applied to a fuel-processing step can be divided in the following 5 groups: 1. metal oxides7,8,9 2. hydrotalcites5,10,11,12 3. double salts13 4. lithium metal oxides14,15, and 5. supported sorbents. Metal oxides, like CaO and MgO, react chemically with CO2. Hydrotalcites are discussed in detail below. Potassium double salts were identified as the active phase in the decomposition products of K2CO3-impregnated hydrotalcite using high-temperature X-ray diffraction. Synthesizing double salts directly, rather than preparing them as a side product by decomposition of hydrotalcite, resulted in sorbents having higher adsorption capacity and improved kinetics compared with promoted hydrotalcites. Lithium metal oxides were discovered by Toshiba in their molten carbonate research when they added ZrO2 to the molten electrolyte. The product, Li2ZrO3, was found to be a good CO2 sorbent, as were other lithium metal oxides. Supported sorbents are combinations of a sorbent of group 1 4 and a suitable support material. An example is CaO supported by Cabot Superior Micropowder16. The properties of the above 5 groups of sorbents are summarised in Table 1. Those sorbent materials which are used for CO2 removal at other conditions are excluded. Materials like zeolites, amines, carbonates and active carbon are used at relatively low temperatures (mostly lower than 100 C) and are thus not taken into consideration here. The stability of double salts has not been investigated very well since a good recipe for a binder to produce pellets from the precipitates has not been found yet. For that reason, they were not used for our research. Trying to select a sorbent from the remaining 4 ones on the basis of this table proves hard since the overall performance of the various sorbent groups is rather similar. Therefore, system aspects need to be taken into account as well. Note that of these 5 groups, the materials of the 1st and 4th group are preferably regenerated by temperature swing since they react with CO2 in a strongly exothermic reaction (H = 170 kJ/mol for CaO at 800 C, H = 170 kJ/mol for Li4SiO4 at 700 C). Those of

ECN-RX--05-122

the 2nd and 3rd groups can be regenerated by pressure swing since the heat of adsorption is relatively low (17 kJ/mol17). Those of the last group may be regenerated using either temperature or pressure swing depending on the sorbent material. Clearly, the chosen regeneration method (temperature swing or pressure swing) has important consequences for the system configuration and efficiency. Without knowing the details of the system, we decided that pressure swing is to be preferred to temperature swing. In the case of temperature swing, the reaction step and regeneration take place in two different, fluidized bed reactors at different temperatures, between which the sorbent particles are continuously circulated18. This makes the system more complex and thus less attractive. Also, heat-resistant alloys must be used for vessel construction due to the higher temperatures involved (> 700 C). For that reason, the CaO and lithium metal oxide based adsorbents were dropped from our experimental work and attention was focused on hydrotalcites which could be obtained commercially as well as prepared relatively easily in our laboratory. It should be kept in mind that in the case of pressure swing operation, the parameter (S/CO2)c provides only a relative indication of the ease of regeneration of the CO2 sorbent. It is not suitable for process design purposes since the bed is never fully regenerated. Instead, a lower purge gas amount would be used with an associated penalty in the CO2 adsorption capacity. An optimization of energy and capital cost would determine which purge gas amount is most desirable.

3.

Hydrotalcites

HTCs (= Hydrotalcites), also called layered double hydroxides or Feitknecht compounds, belong to the family of anionic clays. Their general formula is
3+ M 12+x M x (OH ) 2 ( A n ) x / n .mH 2 O ,

where M2+ and M3+ are divalent and trivalent metal ions respectively, and An- is an anion. The value of x should be in the range 0.20 - 0.33. The metal ions and anions appear in different layers (Figure 2). The metal ion host layer has the brucite structure of Mg(OH)2, in which the metal ions are octahedrally coordinated by OH- ions. Part of the divalent metal ions is replaced by trivalent ions, leaving the brucite structure intact. Consequently, this layer has a net positive charge which is compensated by the charge of the anion layer. The empty sites of the anion layer are filled with water molecules. The most common HTC is

( Mg )1 x ( Al ) x (OH ) 2 (CO3 ) x / 2 .mH 2 O ,

occurring in nature as Mg0.75Al0.25(OH)2(CO3)0.125.0.5H2O (the mineral hydrotalcite) with which the above class of anionic clays is isostructural. HTCs can be easily synthesized by coprecipitation from a solution of soluble salts containing the metal ions, usually at slightly elevated temperature and constant pH19. By slow addition of a carbonate salt, a precipitate is formed. The precipitate is separated by filtration and dried. Thanks to the structure of HTC, the metals are well mixed and dispersed and thus, HTCs are ideal candidate catalyst precursors. When the HTC is calcined in air or N2, first it loses its interlayer water up to approximately 200 C. In the range 200 500 C, dehydroxylation and decarbonisation take place and the specific surface area increases strongly. The suitability of HTC as high temperature CO2 sorbent stems from the plurality of strong basic sites at the surface this structure offers, which favours the adsorption of the acidic CO2 according to the Lewis acidbase theory20. When no further heating occurs, the thus obtained mixture of metal oxides can be transformed back into the original HTC structure by contacting it with a carbonate solution. This is the so-called memory-effect of HTCs. If the metal oxide mixture is heated to temperatures above 500 C, the structure changes irreversibly and amongst other phases a spinel phase is formed.

ECN-RX--05-122

4.

Experimental

For our work, four commercially available HTCs (PURAL MG70, PURAL MG61 HT, PURAL MG50 and PURAL MG30) were obtained from SASOL. They are aluminum magnesium hydroxide carbonates of general formula (Mg)1-x(Al)x(OH)2(CO3)x/2.mH2O. Their specifications are given in Table 2. We also prepared several HTC samples in-house, called ECN-HTC. To a well-stirred NaHCO3 solution of 65 C, a solution containing the metal nitrates in the ratio Mg/Al = 3 was added dropwise resulting in precipitation of the HTC precursor. The pH was kept at a constant value of 8.00 by adding a NaOH solution when needed. After that, the solution was allowed to cool down to room temperature overnight under vigorous stirring. Next day, the precipitate was separated from the suspension by filtration and dried for 24 hours at 120 C. The HTC samples, both commercial and in-house prepared, were activated by heating them in air to 400 C and by keeping them at this temperature for 4 hours. After that, the obtained powders were loaded with 22 wt% K2CO3 using dry impregnation and dried overnight21. The impregnated powders were compacted at a pressure of 275 atm and at room temperature. A sieve fraction of these particles was obtained with sizes between 0.212 and 0.425 mm. The experiments were performed using a glass reactor tube of 16 mm inner diameter and 150 mm length. The reactor was filled with 3 g of particles. The sample was heated under N2 to the temperature of the experiment, usually 400 C. Humidified N2 (29% H2O) was passed along the sample for 75 minutes to remove any calcination products, the so-called pre-desorption step. After that the sample was subjected to a series of adsorption/desorption cycles. The conditions of the adsorption and desorption steps of a standard experiment are given in Table 3. The duration of both steps is 75 minutes and 60 gas samples per step from the dry effluent are analysed by a MicroGC (Hewlett Packard M200H), using a sample interval of 75 s. A permapure is used for drying the effluent. At the end of the experiment, the sample is cooled down to room temperature under dry N2. The water is evaporated and mixed with the feed gas flow in a CEM (Controlled Evaporation Mixer) unit. The mass flow controllers, CEM unit and valves are computercontrolled. A schematic diagram is shown in Figure 3. To determine the amount of CO2 adsorbed and desorbed by the HTC samples only during respectively the adsorption and desorption steps, a so-called blank experiment was performed to correct for instrumental effects. For the blank experiment, a non-adsorbing material (SiC) of the same particle size range and bed volume as the adsorbing samples was used. The experimental conditions were identical to those applied in the adsorption and desorption steps of the experiments using an adsorbing sample. The following parameters were varied: K2CO3 loading, H2O fraction of the feed gas, promoter, desorption flow and desorption time. Finally, the HTC sorbent was mixed with a low-temperature steam reforming catalyst and a CH4/H2O/N2 gas mixture was fed to the reactor to investigate sorption-enhanced reforming of methane.

5.

Data analysis

The measured CO2 concentrations were converted into CO2 flows. Curve fitting was applied to the resulting adsorption and desorption profiles using

a 0 (1 exp((t a1 ) / a 2 ) ) + b0 (1 exp((t b1 ) / b2 ) ) + c0 and a 0 exp((t a1 ) / a 2 ) + b0 exp((t b1 ) / b2 ) + c0 + c1t ,

for respectively the adsorption and desorption profiles. The parameters a0, a1, a2, b0, b1, b2, c0 and c1 are determined by a least-squares fit. The R-square value was better than 0.999 and 0.99 for respectively the adsorption and desorption profiles, showing that the quality of the fitted

10

ECN-RX--05-122

curves was good. The area under the curves was subsequently calculated by integrating the above equations up to the appropriate time 't'. The same procedure is applied to both the adsorbing sample and the blank. The amounts of CO2 adsorbed and desorbed at the time 't' elapsed since the start of the step, respectively Vads and Vdes, follow from:

Vads (t ) = Area under ' blank ' curve Area under ' CO2 adsorbent ' curve
and

Vdes (t ) = Area under ' CO2 adsorbent ' curve Area under ' blank ' curve ,
where blank and CO2 adsorbent curve refer to the adsorption or desorption profiles of the experiments using respectively the blank and the CO2 adsorbent. The height of the peak, which arises shortly after switching from adsorption to desorption conditions, cannot be accurately described due to lack of data points. Therefore, for the calculation of the curve area this peak is approximated by a straight line going from the vertical axis to the second data point, see Figure 4.

6.

Results and discussion

Figure 5 shows the adsorption and desorption profiles corresponding with the 20th cycle of the experiments up to t = 40 min using the various PURAL samples and ECN-HTC, all loaded with 22 wt% K2CO3. The CO2 adsorption decreases during the first 20 cycles for all samples. After the 20th cycle, it becomes approximately constant. The results of the blank experiments are indicated by dashed lines. The adsorption profiles of PURAL MG70, MG61 HT and ECN-HTC are similar (breakthrough times respectively 7.5 min, 8.8 min, 9.0 min), that of PURAL MG50 starts to break through at the same time as the before-mentioned ones (7.9 min) though it rises less strongly, whereas breakthrough of PURAL MG30 clearly occurs at a later time (12.5 min). For all samples we find that after breakthrough, the profile does not show a sharp rise towards the inlet CO2 flow (1.5 ml/min), but rather slowly creeps to this value. Table 4 shows the calculated amount of CO2 adsorbed up to the breakthrough time tbt and during the whole adsorption step (here 75 min), respectively qads(tbt) and qads(t=75min). About two thirds of all CO2 is adsorbed before breakthrough. The desorption profiles show a rapid initial CO2 release, which considerably slows down during deeper desorption and never becomes zero. At the end of the desorption step, the CO2 flow is in the range 0.05 - 0.1 ml/min. Apart from the desorption peak, which is highest for PURAL MG30, the profiles of the various samples are quite similar. The desorbed amount exceeds the adsorbed amount of CO2. During impregnation of calcined HTC with K2CO3, the original HTC structure is partly restored due to the memory effect19. When the impregnated HTC is heated again, calcination of the restored structure occurs which is apparently not complete after the pre-desorption step. We believe that the mismatch between adsorbed and desorbed CO2 is an inherent property of the HTC, not the result of experimental errors in the MicroGC data or the analysis procedure. First, because the mismatch gets smaller when more adsorption/desorption cycles are performed (up to 40). Second, because in only one case (the experiment performed at 500 C, see below), the mismatch (almost) vanished, although it did not even here disappear completely. Third, because we have checked the MicroGC by using a CO2 analyzer allowing a CO2 concentration measurement per 15 seconds which is a factor 5 more than the GC allows. This did not change the mismatch between adsorbed and desorbed CO2. Using an activation temperature of 500 C as a measure to reduce the mismatch, is undesirable because of the decrease of the CO2 adsorption capacity. Other measures like increasing the duration of the activation step or pre-desorption step, while keeping the temperature at 400 C, could not reduce the mismatch, though.

ECN-RX--05-122

11

Figure 6 shows the adsorption and desorption profiles corresponding with the 20th cycle of experiments using PURAL MG70 with different K2CO3 loadings. For comparison, also the unloaded sample is included. It is seen that the CO2 adsorption is enhanced by a factor 3 when loading the HTC with K2CO3, implying that the original HTC structure is not simply restored upon contacting the calcined product with K2CO3. The CO2 adsorption is not very sensitive to the loading percentage (Table 4). The desorption profiles of the various samples are quite similar. The shift of the peak position is due to the extent of sorbent loading, not to differences in desorption properties of the various samples. The presence of water during adsorption is important. Without water, the CO2 adsorption capacity is smaller than in the presence of water. At H2O fractions in the feed gas of 15% or higher, the CO2 adsorption does not increase (Table 4). The temperature of the experiment, which refers to both the temperature of activation and the temperature at which the series of adsorption and desorption cycles is performed, does not have a strong effect on the CO2 adsorption capacity. From 400 to 450 C, it hardly varies, while it has slightly decreased at 500 C (Table 4). The effect of water on CO2 desorption is discussed below in the sorption-enhanced steam reforming experiment. As pointed out in the introduction, S/CO2, the purge gas to adsorbed CO2 ratio, is an important parameter. The desorption characteristic yields the variation of the fraction of desorbed CO2 with S/CO2. Figures 7 and 8 show the desorption characteristics of PURAL MG70 loaded with 22 wt% K2CO3, determined at different desorption flows (tads = tdes = 75 min, Fads = 30 ml/min, Fdes = 10, 20, 30, 60, 100 ml/min) and desorption periods (tads = 10 min, tdes = 10, 25, 50, 75 min, Fads = Fdes = 30 ml/min). All characteristics rise above values of 100% implying that more CO2 is desorbed than adsorbed. As pointed out above, this is a consequence of the release of CO2 from the sorbent structure occurring during the desorption steps. The value of S/CO2 for complete desorption, (S/CO2)c, is in the range 66 336, which is clearly larger than the desired range 3 - 6. Decreasing the desorption flow results in a smaller (S/CO2)c, though the CO2 adsorption capacity is also reduced such that (S/CO2)c shows a net decrease. When the desorption flow is decreased from 100 to 10 ml/min, (S/CO2)c drops by approximately a factor 5. Apparently, mass transfer limitation occurs during CO2 transport from the particle interior to the purge gas, because the amount of purge gas can be reduced by a factor 5 to remove the same amount of CO2. The desorption characteristics shown in Figure 8 are getting steeper and (S/CO2)c decreases by a about factor 3 when tdes is reduced from 75 to 10 minutes. This indicates that the smaller part of the bed which is used, the more easily the adsorbed CO2 can be removed from the bed. Finally, to deliver a proof-of-the-principle test for sorption-enhanced steam reforming of methane, an experiment has been performed using a mixture of 1.5 g low-temperature steam reforming catalyst and 3.0 g PURAL MG70 impregnated with 22 wt% K2CO3. As before, the sample was subjected to a series of reaction/regeneration cycles at 400 C. Feed gas containing 2.9 % CH4, 17.4% H2O and balance N2 at a flow rate of 25 ml/min was fed to the sample during the reaction step, while the standard feed gas was used for regeneration (Table 3). The results of the last (100th) cycle are shown in Figure 9. More H2 is produced during reaction than predicted by thermodynamic equilibrium without sorbent. An average CH4 conversion of 95% is obtained (thermodynamic equilibrium conversion of CH4 without sorbent: 53%), demonstrating the suitability of HTC for sorption-enhanced reforming of methane. In the same experiment, the effect of water on CO2 desorption was investigated. An extra cycle was added with the same reaction conditions but with regeneration performed in 17.4% H2O containing N2 gas. After the sample had been completely desorbed at these conditions (114 min), the H2O concentration was increased to 29%. A clear increase of the CO2 flow is observed.

12

ECN-RX--05-122

7.

Conclusions

The selected high temperature CO2 sorbent, Mg-Al HTC, is able to adsorb CO2 at temperatures between 400 and 500 C. The adsorption is considerably enhanced by impregnating this HTC with K2CO3. Loadings between 11 and 44 wt% have been tested, but the adsorption proved not to be very sensitive to the loading percentage. The adsorption, at least the part after breakthrough, and desorption profiles of the potassium-promoted HTC consist roughly of two parts: a part characterized by fast adsorption (desorption) and a part characterized by slow adsorption (desorption). In the performed experiments, desorption was never complete. The end of the desorption profiles did not go to zero and the desorbed amounts of CO2 exceeded the adsorbed amounts of CO2. This is most probably due to CO2 released from the HTC structure. Water assists the CO2 adsorption and desorption process. In the investigated temperature range, 400 500 C, the CO2 loading capacity hardly varies. Lowering the desorption flow or the duration of the desorption step, results in smaller values of (S/CO2)c. A proof-of-the-principle test has shown the workability of SERP for the steam reforming of methane.

List of symbols
Fads = Fdes = qads(t) = S/CO2 = (S/CO2)c= t tads tbt tdes Tads Tdes Vads Vdes = = = = = = = = molar gas flow during the adsorption step (ml/min) molar gas flow during the desorption step (ml/min) specific adsorbed amount of CO2 (mmol/g) purge steam to adsorbed CO2 ratio (mol purge gas/mol adsorbed CO2) purge steam to adsorbed CO2 ratio for complete CO2 removal (mol purge gas/mol adsorbed CO2) time (min) duration of the adsorption step (min) breakthrough time (min) duration of the desorption step (min) temperature during the adsorption step (C) temperature during the desorption step (C) amount of adsorbed CO2 (ml) amount of desorbed CO2 (ml)

References
(1) Matisson, T.;Zafar, Q.;Lyngfelt, A.; Gevert, B. Integrated hydrogen and power production from natural gas with CO2 capture, in Proceedings of the 15th International Hydrogen Conference, Yokomhama, 27 June- 2 July 2004. (2) Bracht, M. Water gas shift membrane reactor for CO2 control in IGCC systems: technoeconomic feasibility study. Energy Conv. Management, 1997, 38 (supplement), 159-162. (3) Dijkstra, J.W.; Jansen, D.; Leendertse, G.P.; Pex, P.P.A.C.; Hydrogen membrane reactor as key technology for sustainable use of fossil fuels, in Proceedings of the 1st European Hydrogen Energy Conference, Grenoble, 2 - 5 September, 2003. (4) Hufton, J.R.; Allam, R.J.; Chiang, R.; Middleton, P.; Weist, E.L.; White, V. Development of a process for CO2 capture from gas turbines using a sorption enhanced water gas shift reactor system, in Proceedings of the 7th International Conference on Greenhouse Gas Technologies, Vancouver, September 2004. (5) Hufton, J.R.; Mayorga, S.; Sircar, S. Sorption-enhanced reaction process for hydrogen production. AIChE Journal, 1999, 45, 248-256.

ECN-RX--05-122

13

(6) Reijers, H.Th.J.; Roskam-Bakker, D.F.; Dijkstra, J.W.; Smit, R.; Groot, A. de; Brink, R.W. van den. Hydrogen production through sorption-enhanced reforming, in Proceedings of the 1st European Hydrogen Energy Conference, Grenoble, 2 - 5 September, 2003. (7) Balasubramanian, B.; Ortiz, A.L.; Kaytakoglu, S.; Harrison., D.P. Hydrogen from methane in a single-step process. Chem. Eng. Science, 1999, 54, 3543-3552. (8) Harrison, D.P.; Peng, Z. Low-carbon monoxide hydrogen by sorption-enhanced reaction. Int. Journal Chem. Reactor Eng. 2003, 1, 1-9. (9) Meyer, J.; Eriksen, D.O.; Glckner, R.; rjaster, B.E. Hydrogen production by integrated reforming and CO2 capture, in Proceedings of the 1st European Hydrogen Energy Conference, Grenoble, 2 - 5 September, 2003. (10) Waldron, W.F.; Hufton, J.R.; Sircar, S. Production of Hydrogen by Cyclic Sorption Enhanced Reaction Process. AIChE Journal, 2001, 47, 1477-1479. (11) Ding,Y.; Alpay, E. Adsorption-enhanced steam-methane-reforming. Chem. Eng. Science, 2000, 55, 3461-3474. (12) Ding, Y.; Alpay, E. Equilibria and kinetics of CO2 adsorption on hydrotalcite adsorbent. Chem. Eng. Science, 2000, 55, 3929-3940. (13) Mayorga S.G. Carbon dioxide adsorbents containing magnesium oxide suitable for use at high temperatures. Air Products and Chemicals Inc. Allentown (PA). US Patent no. 6280503 B1, 2001. (14) Kato, M.; Yoshikawa, S.; Nakagawa, K. Carbon dioxide absorption by lithium orthosilicate in a wide range of temperature and carbon dioxide concentrations. J. Mat. Sci. Letters. 2002, 21, 485-487. (15) Nakagawa, K.; Kato, M.; Yoshikawa, S.; Essaki, K. A novel CO2 absorbent using lithiumcontaining oxides, in Proceedings of the 2nd Annual Conference on Carbon Sequestration, Alexandria VA, 5 - 8 May 2003. (16) Stevens, J.F.; Atannassova, P.; Shen, J.P. Development of a 50 kW fuel processor for stationary fuel cell applications using revolutionary materials for absorption-enhanced natural gas reforming, in DOE Hydrogen Program FY2004 Progress Report, http://www.eere.energy.gov, 2004. (17) Ding, Y.; Alpay, E. Equilibria and kinetics of high temperature CO2 adsorption on hydrotalcite adsorbent, Chem. Eng. Science, 2000, 55, 3461-3474. (18) Meyer, J.; Aaberg, R.J.; Andresen, B. Chapter 12 Generation of hydrogen fuels for the thermal power plant with integrated CO2 capture using a CaO-CaCO3 cycle, in Carbon dioxide capture for storage in deep geologic formations, volume 1. Elsevier: Oxford, 2005. (19) Cavani, F.; Trifir, F.; Vaccari, A. Hydrotalcite-type anionic clays: preparation, properties and applications. Cat. Today, 1991, 11, 173-301. (20) Di Cosimo J.I.; Diez, V.K.; Xu, M.; Iglesia, E.; Apesteguia, C.R. Structure and surface and catalytic properties of Mg-Al based basic oxides. J. Cat. 1998, 178, 499-510. (21) Nataraj, S.; Carvill, B.T.; Hufton, J.R.; Mayorga, S.G.; Gaffney, T.R.; Brzozowski, J.R. Materials selectively adsorbing CO2 from CO2 containing streams. Air Products and Chemicals Inc., EP Patent no. 1006079A1, 2000.

14

ECN-RX--05-122

Table 1 Characteristics of CO2 sorbent materials (*** = good, ** = fair, * = poor, ?=unknown) group representative member ads.cap. stability kinetics metal oxides CaO *** * *** hydrotalcites Mg6Al2(OH)16[CO3].4H2O/K2CO3 * *** * double salts (K2CO3)(2KHCO3)(MgCO3)(MgO).xH2O ** ? ** Li metal oxides Li4SiO4 ** ** *** supported sorbents CaO on Cabot Superior Micropowder16 ** *** ***

ECN-RX--05-122

15

Table 2 Specifications of HTCs according to the supplier spec. surface pore volume (m/g) (g/cm) PURAL MG70 > 1801) > 0.21) PURAL MG61 HT 16 not specified PURAL MG50 > 2001) > 0.21) 2) 1) PURAL MG30 > 250 > 0.51)
1) 2)

bulk density (g/cm) 0.35 - 0.55 0.15 0.45 - 0.65 0.35 - 0.55

MgO:Al2O3 (wt%:wt%) 70:30 61:39 50:50 30:70

After 3 hours of activation at 550 C PURAL MG30 contains a significant amount of boehmite

16

ECN-RX--05-122

Table 3 Standard experimental conditions flow (ml/min) composition T (C) duration (min) adsorption 30 5% CO2/29% H2O/66% N2 400 75 desorption 100 29% H2O/71% N2 400 75

ECN-RX--05-122

17

Table 4 CO2 adsorption capacities at breakthrough and during the whole adsorption step at various conditions qads(tbt) qads(t=75 min) tbt (min) (mmol/g) (mmol/g) HTC sample PURAL MG70 7.5 0.14 0.33 PURAL MG61 HT 8.8 0.18 0.28 PURAL MG50 7.9 0.17 0.33 PURAL MG30 12.5 0.29 0.44 ECN-HTC 9.0 0.18 0.35 K2CO3 loading (wt%) 0 2.5 0.02 0.06 11 6.4 0.13 0.18 22 6.9 0.14 0.18 33 6.4 0.12 0.16 44 5.2 0.09 0.13 H2O content of feed gas (%) 0 5.0 0.10 0.21 7.5 6.1 0.13 0.29 15 8.8 0.20 0.33 29 8.8 0.21 0.33 temperature (C) 400 8.8 0.18 0.33 450 8.7 0.18 0.32 500 7.5 0.14 0.27

18

ECN-RX--05-122

H2 + steam
1

gas turbine SERP generator

steam natural gas

in adsorption

air
2

exhaust steam

SERP in regeneration water knock out

CO2

Figure 1 Schematic representation of a SERP system for electricity production

ECN-RX--05-122

19

Al(OH)6 - octahedron Mg(OH)6 - octahedron

H2O

CO32-

Figure 2 Structure of hydrotalcite

20

ECN-RX--05-122

He

liquid water

through reactor

valve CH4 water flow controller CO2 CEM N2 Mass flow controller permapure bypass

reactor

MicroGC vent

vent flow meter

Figure 3 Schematic diagram of the lab-scale experimental apparatus

ECN-RX--05-122

21

CO2 flow (ml/min)

0 0 10 20 30 40 50 60 70 80

Elapsed time (min)

Figure 4 Approximation of peak area given by the solid line (dashed line: curve fit, symbols:experimental data points)

22

ECN-RX--05-122

1.6
blank

1.2
PURAL MG70/22 wt% K2CO3 PURAL MG61 HT/22 wt% K2CO3

0.8

PURAL MG50/22 wt% K2CO3

ECN-HTC/22 wt% K2CO3 PURAL MG30/22 wt% K2CO3

0.4

adsorption

0.0 4
PURAL MG30/22 wt% K2CO3

desorption

CO2 flow (ml/min)

2
PURAL MG50/22 wt% K2CO3

PURAL MG70/22 wt% K2CO3

1
PURAL MG61 HT/22 wt% K2CO3

blank

0 0 10 20 30 40

Elapsed time (min)

Figure 5 Adsorption and desorption profiles of various potassium- promoted HTC samples (the desorption profiles of ECN-HTC/22 wt% K2CO3 coincides with that of PURAL MG70/22 wt% K2CO3)

ECN-RX--05-122

23

1.2
44%

adsorption
non-impregnated

0.8

33%

11%

0.4

22%

1.6 0.0

non-impregnated 44%

1.2

desorption

CO2 flow (ml/min)

33% 11%

0.8

22% 0.4

0.0 0 10 20 30 40

Elapsed time (min)

Figure 6 Adsorption and desorption profiles of PURAL MG70 with different K2CO3 loadings

24

ECN-RX--05-122

140

Desorption percentage (%)

10 ml/min

20 ml/min 60 ml/min 30 ml/min 100 ml/min, beginning

100 100 ml/min, end

60

20 0 100 200 300 400

Purge gas (mol gas/mol adsorbed CO2)

Figure 7 Desorption characteristics determined for different desorption flows (the dashed curve is a repetition of 100 ml/min at the end of the experiment)

ECN-RX--05-122

25

140 10 ml/min 20 ml/min 30 ml/min 60 ml/min

Desorptiepercentage (%)

100 100 ml/min, beginning

100 ml/min, end

60

20 0 100 200 300 400

Purge gas (mol gas/mol adsorbed CO2)

Figure 8 Desorption characteristics determined for different desorption times (tads = 10 min, Fdes = 30 ml/min)

26

ECN-RX--05-122

1.6 reaction regeneration

16

CO2

CH4,CO2 concentration (%)

1.2

solid line: experiment

12

H2

dashed line: calculated from thermodynamics without adsorbent present

0.8

8 H2

0.4

4 CH4 CO2 0 0 10 20 30 40

0.0

Elapsed time (min)

Figure 9 Effluent gas composition during reaction and regeneration using a mixture of steam reforming catalyst and HTC impregnated with 22 wt% K2CO3

ECN-RX--05-122

27

H2 concentration (%)

CH4

0.4

0.3

17.4% H2O in N2

29% H2O in N2

CO2 flow (ml/min)

0.2

0.1

0
0 50 100 150 200 250

Elapsed time (min)

Figure 10 CO2 flow during regeneration in 17.4% and 29% H2O containing N2 gas

28

ECN-RX--05-122