Comparison of amine-impregnated mesoporous carbon with microporous activated carbon and 13X zeolite for biogas purification

Three materials are directly compared for their potential for biogas purification: 13X zeolite, microporous activated carbon and mesoporous activated carbon impregnated with polyethyleneimine. The amine-impregnated material shows the highest selectivity for CO2 over CH4 but this should be balanced by the higher operating temperature required. All three materials could be used for biogas purification with the advantages and diasadvantages clearly presented.


Introduction
With the world's ever increasing requirement for green energy, there is great potential to reduce carbon emissions through the optimisation of current energy generation methods. One such green technology is the production of biogas via the fermentation of plant material or waste to produce a mixture of predominantly CO2 and CH4.
Depending on the process used during production, along with the type of fermented material, the composition of the produced gas can vary significantly. However, from an anaerobic digester, a significant portion of the produced gas will always be CO2.
In order to enhance the gas stream for energy production processes, an adsorption process can be used to purify the individual components.
[1] Purification of the gas mixtures to produce two high purity gas streams has the added benefit of producing a higher value product of close to pure methane along with a CO2 stream that could potentially be sequestered, preventing the release of CO2 into the atmosphere and hence reducing the environmental impact of the process. This is referred to as biogas upgrading and, as a result of its green power generation credentials, the optimisation of the upgrading process has recently begun to attract interest as an area of research.[1-3] The optimal technology for biogas upgrading is highly dependent on the specific biogas process/plant. The biogas feedstock, the scale of the plant and the acceptable concentration of impurities in the product streams are all important factors in selecting an upgrading technology. A variety of technologies have been investigated and, in certain cases, implemented such as water scrubbing and pressure swing adsorption (PSA) [4], cryogenic separation, chemical absorption, physical absorption and membrane separation [5,6]. A review comparing the cost and investigating the appropriate utilisation of the various approaches was recently published by Sun et al. [7]. From this review it is clear that further work is required to establish the potential of the different technologies if biogas upgrading is to find more widespread application. There are also several recent reports that propose systems to lower the cost of gas separation. In 2015 Kim et al. [8] proposed a four column PSA process using a carbon molecular sieve as adsorbent that only had a selectivity for CO2 over CH4 of 1.9. Grande et al. proposed a layered pressure swing adsorption system where a kinetic adsorbent such as a carbon molecular sieve was layered with an equilibrium adsorbent [9]. This combination improved the productivity of the setup and resulted in a potential size reduction of the separation unit by up to 60%. The selection of an appropriate, novel adsorbent could significantly enhance the efficiency of an adsorption separation process. However, there are only few reports in the literature regarding the development of optimised adsorbent material for biogas upgrading.
The main materials used in PSA are zeolites and activated carbons. Alonso-Vicario et al. compared commercial zeolites 13X, 5A and natural clinoptilolite using breakthrough experiments and concluded that despite its lower CO2 capacity, clinoptilolite was preferred as it was able to separate both the CO2 and H2S that was present in their biogas stream, from CH4 [10]. Various activated carbons have been investigated for their ability to separate CO2 from CH4 with a selectivity of 2-4, depending on the material and the process conditions [10,11]. Triamine grafted pore expanded silica was investigated by Belmabkhout et al. who proposed, on the basis of single component adsorption data, that it had great potential to separate acidic gases from CH4 with high selectivity [12].
In this paper, we compare the selectivity for CO2 over CH4 of three different adsorbents: commercial zeolite (13X), commercial microporous activated carbon (micro-AC) and an amine-impregnated activated carbon (meso-AC2-PEI). The first two materials provide a benchmark and direct comparison between well-characterised and studied materials while the third material is, to our knowledge, the first report of the study of an amine-impregnated activated carbon for biogas upgrading. The three materials allow a direct comparison of the advantages and disadvantages of using physical adsorption (13X, micro-AC) or chemical adsorption (meso-AC-PEI) to separate CO2 from CH4. We show that the impregnated AC material has the highest selectivity (→ ∞) that, together with its insensitivity to water but relatively high operating temperature, could make this a very suitable class of material for integration into temperature swing adsorption processes.

Materials
The zeolite 13X and the microporous activated carbon (SRD 10061) are commercially available materials from UOP (Honeywell) and Calgon Carbon, respectively. The microporous-AC had a BET surface area of 1336 m 2 g -1 with a total pore volume of 0.68 cm 3 g -1 of which 0.59 cm 3 g -1 consisted of micro-pores with dimensions < 2nm [13]. The meso-porous-AC material was synthesised by a templating method using sucrose and a silicagel with an average pore size of 150 Å, following the procedure described previously [13]. It had a BET surface area of 1254 m 2 g -1 and a total pore volume of 3.1 cm 3 g -1 . In this case, ~2.9 cm 3 g -1 consisted of meso-pores with dimensions in the range 2nm -50 nm (the pore size distribution is provided in the Electronic Supplementary Information). As published previously, impregnation of mesoporous-AC with amines was shown to significantly increase the CO2 uptake capacity at 0.1 bar, changing the mechanism from physisorption on the empty material to chemisorption on the impregnated material [13]. Large molecular weight amines were found to be more suitable due to their higher thermal stability and recyclability in spite of the slightly less efficient use of the amino groups.
Microporous-AC was shown to be unsuitable for impregnation due to the tendency for pore blocking. In the present study, the mesoporous AC was impregnated with polyethyleneimine (PEI, MW 1200) at a ratio of ca. two parts polymer to one part carbon, meso-AC-PEI (65.7 wt.%, corresponding to approximately ¾ pore filling), following the procedure detailed previously [13].

Extended Zero-Length Column Breakthrough Technique
The extended zero-length column technique (E-ZLC) is similar to the more traditional ZLC which is a powerful method for providing an initial ranking of adsorbents, requiring only small amounts of sample (5 -15 mg) [14]. The E-ZLC makes use of a larger column, ca. three times the length of the ZLC, housed in a 1/8" Swagelok bulkhead connector. This allows more sample to be packed in the adsorption column to achieve a clear separation of components in a binary mixture and determine the binary adsorption selectivity [15]. The advantage of E-ZLC over a traditional breakthrough column is that a relatively small amount of sample is required (ca. 50 mg, compared to ca. 5 g for a standard column) and that the column can be considered to be isothermal, as experimentally tested and discussed previously [15]. In a typical experiment, the sample is packed in the column and regenerated at high temperature under inert gas flow. The sample is cooled to the temperature of interest and then equilibrated with a gas stream containing a known partial pressure of sorbates, in the present case 45% CO2, 55% CH4. The gas stream is then switched to a stream of pure purge gas (N2) and the desorption profile is monitored by a mass spectrometer.
Three E-ZLC were packed with the commercial zeolite 13X (63.8 mg), the micro-AC (37.9 mg) and the meso-AC-PEI (25.2 mg). The different masses used for the experiments are a consequence of the different densities of the adsorbents. The breakthrough experiments were run at 35 o C for 13X and micro-AC, which as physisorbents have a higher CO2 capacity at lower working temperatures, and at 75 o C for meso-AC-PEI, due to the slower reaction kinetics of the impregnated chemisorption material [13]. The desorption profiles were determined for different gas flow rates and modelled using the Cysim simulator [15,16].
The simulation parameters needed to reproduce the experimental breakthrough response of all three samples are reported in Table 1. Comparison with experimentally determined volumetric isotherms, measured with an iQ1 volumetric system (Quantachrome), was used to determine the parameters used in the simulations for 13X and micro-AC.

Volumetric Isotherms
The volumetric isotherms measured for 13X and micro-AC and fitted to obtain the Langmuir isotherm parameters used in the breakthrough simulations are shown in  Table 1. Volumetric isotherms could not be measured for meso-AC-PEI since the amine could potentially damage the iQ1 system. In this case the simulation parameters were solely established by fitting the E-ZLC breakthrough curves.
A dual-site Langmuir isotherm was used to fit the 13X data with the same saturation capacity (qs) used for each gas on each site, to ensure thermodynamic consistency. A single-site Langmuir expression was used to fit the micro-AC sample, and also assumed for the breakthrough measurements on the meso-AC-PEI sample.

E-ZLC Breakthrough Measurements
The adsorption breakthrough profiles of 13X, micro-AC and meso-AC-PEI are shown in Fig. 3. Time 0 indicates the change from the pure purge gas (N2) to the mixture of 45% CO2 and 55% CH4. A clear separation of the CO2 and CH4 is seen for the 13X sample, Fig. 3(a). At a flow rate of 10 cm 3 min -1 the breakthrough times are approximately 9 s and 84 s for CH4 and CO2, respectively, after subtraction of the breakthrough time of the blank response (14 s). "Roll-up" of the CH4 (C/C0 > 1) is observed. This is due to all the CO2 being adsorbed by the 13X, with the consequence that the gas at the outlet, prior to the breakthrough of CO2 is pure CH4. The magnitude of the roll-up is larger than expected due to the over-response of the mass spectrometer to the large step change in the gas phase concentration of CH4 as it breaks through. Several methods have been suggested to correct for the flow rate but generally are only valid for small step changes [17]. The Cysim simulation corrects for the flow rate and ensures that the mass balance closes [16]. By calculating the selectivity from the desorption curves, the true binary selectivity is established because the integration of the area under the curve (accounting for the variable flowrate) directly yields the adsorbed amount of the binary mixture.
The desorption curves for 13X are shown in Fig. 4 along with the model prediction on a semi-log plot vs. time. The parameters determined by the volumetric isotherm measurements, were used to simulate the CO2 desorption curves and the methane parameters were carefully fitted to match the experimental data. The adsorbed amounts of each component, extracted from the simulations are provided in Table 2 along with the selectivity of the adsorbents with respect to CO2, defined as (1) where = 0.45 bar and = 0.55 bar.  The total uptake of CO2 was less for the impregnated sample than for 13X, Table 2, however, the presence of water does not significantly hinder the uptake of CO2 by amine-impregnated samples [18,19], unlike the situation for 13X [20]. This is advantageous as biogas often has a high water content. The meso-AC-PEI adsorbs more CO2 per unit mass than the micro-AC. CO2 binds strongly to the amine, as can be seen from the shape of the desorption curve and also from the high value extracted for the heat of adsorption, ∆H, of approximately 90 kJ mol -1 , Table 1. The CO2 is therefore very favourably adsorbed compared to the CH4 and the majority of the CO2 starts to desorb from the sample at a lower CO2 partial pressure (low C/Co) than is the case for 13X and meso-AC. The strong chemisorption between the amine and the CO2 provides high selectivity at low partial pressure.  Figure 3). This allowed the selectivity corresponding to the best fit to the experimental data to be reported taking into account any necessary flow rate corrections. To achieve the best fit, the b1,0 parameter for CO2 was adjusted from 2.17 x 10 -5 bar -1 to 1.87 x 10 -5 bar -1 . Gil et al. [11] reported a selectivity factor of 3.2 for CO2 over CH4 on a comparable microporous activated carbon, in good agreement with the selectivity of 3.0 (2.59) reported here.
Although the unmodified activated carbons may have a disadvantage over zeolites in terms of selectivity, activated carbons are relatively inexpensive and stable over many cycles. As shown here, the selectivity can be significantly enhanced by modifying the adsorbent through impregnation with polyamine. The basic amine groups preferentially chemisorb the CO2 and, additionally, loading the pores with amine through a wet impregnation method has the added benefit of filling the pore volume of the activated carbon, greatly reducing the number of sites available for physisorption of CH4. To facilitate the chemisorption and increase the reaction kinetics the process must be carried out at elevated temperature, again reducing the volume of adsorbed CH4 and further enhancing the selectivity of the impregnated activated carbon.

Conclusions
All three investigated materials in this study, 13X, micro-AC and meso-AC-PEI, can be used to separate CO2 from CH4 in a biogas upgrading adsorption process. Both meso-AC-PEI and 13X have high selectivity, adsorbing only small (in the case of meso-AC-PEI undetectable) amounts of CH4. Depending on the type of process to be developed, the biogas feedstock and the purity requirements of the product streams, all three adsorbents could potentially be utilized to upgrade biogas.
Commercial zeolite 13X has a high selectivity of up to 66, however, in the presence of water vapour, the total uptake of CO2 is significantly reduced [20] and it would therefore be desirable to ensure dry feed gas.
The required operation temperature for the highly selective amine-impregnated material would make it suitable for integration into a temperature swing adsorption process, using excess heat from the biogas plant for regeneration. However, due to the high input partial pressure of CO2, it may not always be necessary to incorporate the strong amine-CO2 chemisorption sites. In some cases, the high regeneration costs may outweigh the advantages of the high selectivity of the amine impregnated material. Process simulations would be required in each case to fully assess the suitability and viability of each material.
As a larger number of biogas plants are introduced to the energy mix, novel materials will be required to upgrade the fuel to the required purity in the most economical manner possible. It is likely that no single material will be suitable for all situations and it is therefore important to understand the parameters influencing the performance and directly compare different classes of material.