H₂S removal via reactive-absorption techniques (passage of biogas through alkaline solutions—NaOH, CaO) is not a feasible method. This is because it is not a selective process; CO₂ also reacts with alkaline solutions and would thus consume the costly alkalis. Moreover, as we shall see, CO₂ is itself an economically valuable biogas component that can be used for cultivation of agricultural plants or for the production of added-value products by means of appropriate upgrading.

Various types of activated carbon have been investigated for H₂S removal mainly by adsorption/oxidation to produce elemental sulfur and to a lesser extent by conversion to SO₂ (Bagreev and Bandosz, 2001; Bashkova et al., 2007; Xiao et al., 2008; Pipatmanomai et al., 2009; Kwansy and Balcerzak, 2016). Pore structure and surface characteristics as well as nitrogen content of the activated carbon affect both H₂S breakthrough capacity and selectivity toward elemental sulfur formation (Bashkova et al., 2007). On the other hand, Bagreev et al. (2001) concluded that the surface chemistry, specifically the surface acidity, of activated carbon seems to be the key factor that plays the dominant role in the H₂S breakthrough capacity, rather than surface area and pore volume characteristics. The relative humidity of the gas stream and alkali-impregnation were also found to be crucial to the performance of activated carbons used for H₂S removal (Xiao et al., 2008; Pipatmanomai et al., 2009). A reaction pathway for H₂S oxidation over activated carbon that involves the participation of all three vapor-liquid-solid phases has been proposed by Bagreev et al. (2001) and Xiao et al. (2008). The liquid phase consists of an extremely thin water film formed on the AC surface via water vapor condensation into which H₂S dissolves prior of its dissociation toward H⁺ and HS⁻ ions. Then a surface reaction occurs between HS⁻ and dissociatively adsorbed oxygen (O_ads*) on the activated carbon active sites (*) forming elemental sulfur and water at the edge of the liquid film. Some of the HS⁻ ions is are also oxidized to H₂SO₄. The net results is

According to Bagreev et al. (2001) and Xiao et al. (2008) the steps involved are as follows, where * is an active site on the AC surface.

The importance of water on the AC surface is clear from the above mechanism and explains why both relative humidity (RH) and the ability of the AC surfaces to adsorb water have a major influence on H₂S removal efficiency. Low RH values or hydrophobic AC surfaces indeed exhibit low H₂S removal efficiencies as shown by Xiao et al. (2008) who showed that impregnation of AC with Na₂CO₃ modifies the chemistry of the AC surface, enhancing water adsorption and promoting dissociation of H₂S and its subsequent removal.

Zeolite-based materials also possess high H₂S breakthrough capacities and therefore present an alternative strategy for H₂S removal (e.g., ion-exchanged zeolites, CuO or ZnO-modified zeolites, etc.; Cosoli et al., 2008; Micoli et al., 2014), mesoporous silica (Belmabkhout et al., 2009) and iron oxide/hydroxide systems (so-called iron sponges) (Abatzoglou and Boivin, 2009). The basicity of these materials induces acid-base or redox reactions that result in efficient removal of H₂S, for example:

Iron sponge can also remove mercaptans

and are easily regenerated by O₂

Besides the above physical and chemical methods, biological processes are also widely employed for H₂S removal by microorganisms. These can achieve a satisfactory degree of desulfurization without the disadvantages associated with chemical processes. They can transform H₂S into S⁰ or SO42- (depending on O₂ availability) generating readily separated by-products that could be used for other industrial processes. In addition, they require minimum nutrient input and display high robustness to temperature, pH and moisture fluctuations (Oyarzun et al., 2003; Syed et al., 2006; Abatzoglou and Boivin, 2009; Sun et al., 2015).

The most common biological technologies for H₂S removal include biofilters (i.e., Chung et al., 1996; Elias et al., 2002; Oyarzun et al., 2003), biotrickling filters (e.g., Kim and Deshusses, 2005; Fortuny et al., 2008; Rodriguez et al., 2014) and bioscrubbers (e.g., Sorokin et al., 2008; Van Den Bosch et al., 2008). These processes are effective and environmentally friendly for the removal of H₂S, in particular at low concentrations of the latter (Fortuny et al., 2008; Tang et al., 2009). Moreover, most biological processes for H₂S removal use sulfide-oxidizing bacteria (SOB), especially chemotropic species (mostly Thiobacillus sp., Thiotrix sp., Beggiato sp., Thermothrix sp.). A number of chemotrophic thiobacteria have been studied and found suitable for H₂S biodegradation and can be used in both aerobic conditions with O₂ and anaerobic conditions (Syed et al., 2006; Abatzoglou and Boivin, 2009). Three patented H₂S purification processes, namely Thiopaq®, Biopuric®, and H2SPLUS SYSTEM®, combining chemical scrubbers and bioreactors have been commercialized for large-scale biogas desulfurization (Fortuny et al., 2008; Sorokin et al., 2008; Abatzoglou and Boivin, 2009).

Chung et al. (2007), based on their previous findings of the behavior of the Thiobmillus thiopurus biofilter (Chung et al., 1996), demonstrating for the first time a two-stage biofilter for sequential treatment of concentrated H₂S and diluted NH₃ mixtures. Their strategy of using a first biofilter of Thiobacillus thioparus for H₂S removal and a second biofilter of Nitrosomonas europaea for NH₃ removal was very effective. Kobayashi et al. (2012) studied microbial mats for desulfurization of biogas in a full-scale anaerobic digester and characterized them in terms of their structure and chemical and microbial properties. Their results indicated that the key players in sulfide oxidation and sulfur production in the bio-desulfurization in the headspace of the digester were likely to be two sulfide-oxidizing bacteria species related to H. neapolitanus and S. denitrificans: the microbial community, cell density, bacterial activity varied depending on the environmental conditions. They also showed that the habitat of the SOB should be confined to the lower part of the headspace so as to improve operating conditions. Lohwacharin and Annachhatre (2010) investigated the successful operation of the biological sulfide oxidation process in an airlift biological reactor under oxygen-limited conditions and showed that up to 90% of sulfide removed was converted to elemental sulfur (S2 O32- was the main by-product).

Fortuny et al. (2008) studied biotrickling filters with two different packing materials for the removal of ultra-high concentrations of H₂S from oxygen-poor gases as an interesting alternative for the treatment of off-gases containing high concentrations of H₂S. They found that optimization of packing and operating conditions could improve the process. Similarly, Fernandez et al. (2013) tested a biotrickling filter packed with polypropylene Pall rings to remove H₂S from biogas under anoxic conditions and achieved 99% sulfur removal for H₂S inlet loads lower than 120 gSm⁻³h⁻¹.

Ramos et al. (2014a,b) and Diaz et al. (2015) studied microaerobic conditions in order to control H₂S content and showed that the application of such conditions was an efficient method for H₂S control and removal from biogas. Moreover, Jeníček et al. (2017) confirmed the effectiveness of microaeration as a biochemical method of sulfide oxidation to elemental sulfur, obtaining H₂S removal efficiency more than 90% in most cases. An unusual approach is the combination of chemical and biological processes. Ho et al. (2013) proposed a chemical–biological process to remove a high concentration of H₂S in biogas, in which H₂S was first oxidized by ferric iron to generate S⁰ in a chemical reactor and the resulting ferrous iron was then oxidized in a biological reactor by iron-oxidizing bacteria. An H₂S removal efficiency of 98% was achieved indicating the feasibility of the method. Likewise, Lin et al. (2013) developed a pilot-scale chemical–biological H₂S removal process for biogas achieving H₂S removal efficiency up to 95%, further highlighting the chemical-biological approach feasibility for biogas desulfurization.

Organic compounds that contain Si-C bonds, called organosilicons, are classified into organosilanes and organosiloxanes. The former are polymeric compounds containing Si-Si bonds with organic side-chains, the latter consist of a backbone of alternating Si-O units with organic side-chains attached to each Si atom (de Arespacochaga et al., 2015). Siloxanes have exceptional properties including thermal stability, low flammability, low surface tension and toxicity, hydrophobicity and high compressibility. They are therefore widely used in industry as additives in many products, including pharmaceuticals, detergents, cosmetics, shampoos, shaving foams, textiles, and coatings. Depending on the raw biomass used for biogas production, and in particular from landfill wastes and composts, the resulting biogas can contain significant amount of siloxanes, volatile methyl siloxanes (VMSs) being the most common species found in digester and landfill biogas. The most significant ones are listed in Table 2.

When siloxanes-containing biogases are used as an energy vector, silica microparticulates formed at high temperatures can create fouling and abrasion effects that are highly detrimental to the engine components of natural gas-fueled vehicles (Nair et al., 2012, 2013; Jalali et al., 2013; de Arespacochaga et al., 2015), to the anodic side of biogas-fueled solid oxide fuel cells (Haga et al., 2008; Madi et al., 2015) and to the catalysts used for production of added-value chemicals from biogas. Accordingly, together with H₂S, siloxanes are considered as one of the most undesirable contaminant in biogas (Dewil et al., 2006; Ohannessian et al., 2008; Jalali et al., 2013). It is therefore of importance to remove siloxanes from biogas at the first stages of upgrading and also before its use as RNG in energy generation or catalytic production of valuable chemicals. The subject is attracting increasing attention (Abatzoglou and Boivin, 2009; de Arespacochaga et al., 2015) and undergoing rapid development.

Several methods are effective for removal of siloxanes and may be considered as possible strategies for this purpose:

selective absorption in organic solvents,

To this end, Schweigkofler and Niessner (2001) showed that nitric acid and sulfuric acid are especially potent agents for siloxane removal (efficiencies >95%) at moderately elevated temperatures (ca. 60°C) and in concentrated solutions, 65 and 97 wt% respectively, whereas phosphoric acid was ineffective. Countercurrent absorption towers are necessary to ensure sufficient contact and therefore fast mass transfer between gas and liquid, although the high acidity involved is a significant techno-economic drawback for application of the method. The same authors have also researched siloxanes abatement via adsorption on a variety of solids. The adsorption capacity was found to depend strongly on the siloxane type (L2 and D5 siloxanes were tested), relative humidity of the biogas (the higher the humidity the lower the siloxane removal capacity), and of course on the nature of the adsorbent itself (activated charcoal, carbopack B, Tenax TA, XAD II resins, molecular sieve 13X and silica gel were tested). Activated charcoal and silica gel were found to be exceptionally effective sorbents and silica gel showed excellent thermal regeneration properties as well (ca. 250°C).

Regenerable, activated alumina beds operated in continuous mode (double-bed, cyclic; alternately fused or trapping and regeneration) have been proposed for the efficient removal of siloxanes by Higgins (2007).

Montanari et al. (2010) studied D3 adsorption and adsorbent regeneration over three solids (silica gel, faujasite NaX zeolite and pure activated carbon) by means of FT-IR spectroscopy. Only partial regeneration was achieved with all these adsorbents in the temperature range of 20–200°C with either N₂ flow or by applying vacuum. Hydrogen bonding to surface silanol groups is involved in the adsorption of D3 on silica, while with NaX zeolite molecular adsorption as well as chemical adsorption occurs. The authors discussed the advantages of choosing various types of activated carbon as preferred adsorbents for removal of D3 contamination from biogas.

Cabrera-Codony et al. (2014) investigated 12 commercial types of activated carbon as D4 siloxane adsorbents. They found a strong correlation between the textural properties of the ACs (in particular the total pore volume) and their D4 adsorption capacities, with a wood-based H₃PO₄-activated carbon offering the optimum adsorption capacity (D4 ~1750 mg/g) with dry N₂ as carrier. This value was reduced by >50% under typical biogas concentrations of D4 and in the presence of the major biogas components CH₄, CO₂ and water vapor. However, polymerization of siloxane on the adsorbent surface, promoted by oxygeated functional groups (phenolic and carboxylic) that occur on these wood-based types of AC, inhibits their thermal regeneration. The authors concluded that the activated carbons with high pore volume and low carboxylic and phenolic content may be very promising materials with both high siloxane capacities and good thermal regeneration characteristics.

Sigot et al. (2014) investigated the D4 adsorbing capacity of three materials: a coconut-based activated carbon (930 m²/g BET surface area), a 13X zeolite (700 m²/g BET surface area) and a Chameleon® silica gel (690 m²/g BET surface area). The silica gel exhibited the highest D4 adsorbing capacity: ~250 mg/g at room temperature and 0% relative humidity (RH). At a temperature ~20°C higher, only a 15% decrease in capacity was found; however RH of the order of 70% catastrophically decreased D4 adsorption at both temperatures. The surface chemistry of silica gel is dominated by siloxane Si-O-Si and silanol Si-O-H groups which show an affinity for compounds similar to D4: these are considered to play a key role in the superior capacity of silica gel compared to the other materials tested.

Jiang et al. (2016) explored D4 siloxane adsorption over mesoporous aluminosilicate (UCT-15), a zeolite-type material developed at the University of Connecticut, which offers tuneable textural properties via variations in aluminum content and calcination ramp rate. The best D4 adsorption capacity (105 mg/g) was shown by UCT-15 with Si:Al = 5 and a 10°C/min ramp rate, a value that is almost twice that of commercial ZSM-5, which has similar BET surface area and total pore volume. The as-prepared UCT-15 had a larger BET surface area, external surface area, mesopore volume and total pore volume. External surface area and mesopore volume were the key parameters governing the adsorption capacity. Hydroxyl groups on the surface of the aluminosilicates were found to promote the undesirable polymerization of D4, which is detrimental to adsorbent regeneration. In a more recent report Jafari et al. (2016) modified the textural properties of the mesoporous silica UCT-14, by tuning the temperature of gelation, calcination temperature and the heating rate to produce a material with high D4 adsorbing capacity (686 mg/g). This high value is comparable to that achieved with commercial silica gel under both dry and humid conditions. The modified UCT-14 (designated Si-Syn120) was more stable under consecutive use-regeneration cycles and somewhat more resistant to humidity in regard to performance deterioration, compared to commercial silica gel.

Cryogenic methods may also be considered for biogas purification from siloxanes. Although siloxanes can be fully removed (>99%) at very low temperatures, ca. −70°C, Abatzoglou and Boivin (2009) pointed out the energy-intensive character of the method and the need for relevant techno-economic analysis to demonstrate its sustainability.

Comprehensive reviews of biogas purification processes have been provided by Abatzoglou and Boivin (2009) and Ryckebosch et al. (2011).

CO₂ separation from a gas stream via absorption is a classical method, based either on physical or preferably on chemically-driven absorption of CO₂ in liquids or liquid solutions, taking place in bubble cap trays or in randomly packed towers (containing inert solid elements) where a countercurrent flow of gas mixture and liquid absorbent is applied; spray contactors of absorbent and gas mixture may also be possible in practical applications (Kohl and Nielsen, 1997). Countercurrent flow in randomly packed columns is well suited to absorption applications due to their more reliable design and generally superior performance. Based on the significantly higher solubility of CO₂ in water compared to methane, particularly at lower temperatures (Figure 1), so-called water or physical scrubbing may be used for CH₄-CO₂ separation of biogas. The CO₂-rich water leaving the absorption tower is regenerated by flashing followed by recycling. Biogas upgrading plants using water scrubbing with yield capacities of the order of 30–100 m³/h are in current use.

On the other hand, with the most effective chemically-based absorption of CO₂ (often called chemical scrubbing), efficiencies up to 99.5% can be reached (Ryckebosch et al., 2011), with alkalis (e.g., NaOH; Ca(OH)₂) or alkanolamines (RNH₂; R is the organic component of the amine whose identity is not critical to the absorption reaction). Aqueous solutions of these bases are generally used, where CO₂ trapping occurs according to the following reactions:

The most common alkalolamines used include monoethanolamine (MEA), diethanolamine (DEA), triethanolamine (TEA), di-methylethanolamine (DMEA), methyldiethanolamine (MDEA), mixtures of glycol and monomethylamine, diglycolamine (DGA), diisopropanolamine (DIPA) and amine mixtures (Cebula, 2009; Petersson and Wellinger, 2009; Ryckebosch et al., 2011). After absorption, the consumed alkalolamine is regenerated by heating, and then recycled. During this process the following reaction takes place:

It is possible for chemical scrubbing to be applied for the simultaneous removal of CO₂ and H₂S, in which case higher temperature is required for amine regeneration (Petersson and Wellinger, 2009). It is therefore preferable to remove H₂S from biogas before CO₂ absorption as this is cheaper overall and generates clean CO₂ for possible subsequent use.

Adsorption, often referred to as chemisorption, the driving force behind heterogeneous catalytic reactions, is also widely applied for the selective separation of molecules from a fluid phase (Dabrowski, 2001). It is the spontaneous exothermic chemical reaction that occurs when a molecule (referred to as adsorbate), initially present in fluid phase, encounters the surface of an active solid (referred to as adsorbent). Adsorbents are porous solids with large surface areas per unit mass (typically 100–2,000 m²/g), such as activated carbons, molecular carbon sieves, fullerenes, carbonaceous nanomaterials, silica gels, activated alumina and other metal oxides, metal hydroxides, zeolites, clay minerals and pillared clays, etc. They are effective for the selective adsorption of specific species from a fluid (gas or liquid) phase, thus removing them from the mixture. The adsorption isotherms of adsorbate species, often described by the Langmuir equilibrium equation, are of critical importance for the design of an adsorption-based separation process and enable determination of the solid surface capacity of the adsorbent for a specific adsorbed molecule. On the other hand, the kinetics and dynamics of adsorption, described by the more general term “adsorption dynamics,” is affected by external, internal and surface diffusion characteristics of the adsorbed molecule, and, as described by Fick's second law, give the evolution with time of industrial adsorption processes used in separations (Kohl and Nielsen, 1997; Dabrowski, 2001).

Regeneration of the adsorbent, a key step in any separation process based on adsorption, can be performed by reducing the total pressure or by applying vacuum conditions on the saturated adsorbent: these processes are therefore termed pressure swing adsorption (PSA) or vacuum swing adsorption (VSA) respectively. Alternatively, regeneration may be achieved by increasing the temperature of the saturated adsorbent so as to desorb the adsorbate (temperature swing adsorption, (TSA)).

Adsoprtion-based removal of CO₂ from biogas is widely practiced and commercially applied at both pilot and demonstration plant levels and is still an active research subject (Cavenati et al., 2008; Cebula, 2009; Petersson and Wellinger, 2009; Alonso-Vicario et al., 2010; Tippayawong and Thanompongchart, 2010; Grande, 2011; Montanari et al., 2010; Grande, 2012; Ryckebosch et al., 2011; Yuan et al., 2013; Andriani et al., 2014). The techniques used include PSA and TSA, both involving two packed-bed columns in a swing-type arrangement employing appropriate valve sequencing, as shown schematically in Figure 2. The operating principal is similar in the two cases, the main difference being the method used for adsorbent regeneration, i.e., either pressure or temperature variation. This two-column arrangement allows continuous-flow steady-state operation of the process, although the operation of individual columns is of course discontinuous and involves a multistep cycle. The simplest case uses a two-step cycle. One column is maintained at low temperature to continuously trap CO₂ from the biogas flow (allowing free passage of CH₄ through the packed bed) whilst the other is heated (TSA) or evacuated (PSA) to release previously adsorbed CO₂. Note that the temperature of the bed determines its adsorption capacity—the lower the temperature, the greater the adsorption capacity of an adsorbent for a given molecule. A CO₂ collection vessel is typically used, as well as a small portion of the purified methane stream (or an inert gas) in order to assist CO₂ flow into the collection vessel and to purge the column, preparing it for the next cycle. In the second step, the two 4-port valves are synchronous turned to a second position such that the role of each column is reversed: the cleaned column #2 now traps CO₂, whilst the previously CO₂-saturated column #1 commences regeneration. The duration of the two-step cycle is a crucial design parameter that depends on the CO₂ capacity of the columns. This must be appropriately determined in order to allow continuous utilization of the feedstream with continuous production of CH₄ in the system's exhaust. The basic concept can be varied, especially in regard to the number of steps required for a complete cycle, and with respect to the operating sequence of the valves during each step (Grande, 2012). In addition to CH₄-CO₂ separation, the concept has been successfully applied in a number of other separations, for example in the purification of H₂-containing streams for fuel cell applications (Nikolaidou et al., 2015); for C₂H₄ separation from CH₄ and O₂ in the oxidative coupling of methane, thus providing a one-step transformation of methane to ethylene with extremely high yields (Jiang et al., 1994); for air separation (Jee et al., 2001); for noble gas purification (Das et al., 2008); for n-paraffin/iso-paraffin separation (Yang, 2003), and others. The simplicity of PSA and its low capital and operational costs makes it very attractive in comparison to other fluid separation technologies.

Adsorbent characteristics are crucially important in adsorption-based biogas upgrading units. Their CO₂ capacity and the temperatures required for sufficient CO₂ adsorption and desorption are critical parameters that determine the economics and engineering aspects of the separation process. A wide variety of porous materials, including many kinds of zeolites, mesoporous materials, activated carbons and more recently high-surface area coordination polymers can be effectively applied in PSA and TSA. Yet other materials for this purpose are continuously under development, although only a few are used in current commercial PSA units. New research avenues are aimed at the simultaneous removal of CO₂ and H₂S from biogas (Belmabkhout et al., 2009; Tippayawong and Thanompongchart, 2010). A significant cost-effectiveness parameter for potential adsorbents is their ease of regeneration, rather than their ultimate CO₂-capacity. Accordingly, materials that exhibit quasi-linear CO₂ adsorption isotherms at low pressures, as opposed to very steep ones that rapidly flatten above a certain pressure, are far preferable in PSA applications, even if the latter offer much higher ultimate capacities (Grande, 2012). Materials which meet the requirements for effective, low-cost CH₄-CO₂ separation are zeolites, carbon molecular sieves and activated carbons (Pandey and Fabian, 1989; Cebula, 2009; Montanari et al., 2010; Ryckebosch et al., 2011; Alonso-Vicario et al., 2010).

Compact PSA plants for CH₄-CO₂ separation with yields of the order of 250 m³/h have been already commercialized. They involve low capital and installation costs and are well suited to small scale applications. A comprehensive overview of the fundamentals of the PSA process and its evolution with time, with emphasis on CH₄-CO₂ separation, has recently been published by Grande (2012).

The selective permeation of a molecule through a solid renders it as a potential membrane for the separation of that molecule from a gas mixture (Bernardo et al., 2009). Gas-gas and gas-liquid membrane separation processes have been developed for practical applications (for example, in the latter case CO₂ is extracted from a CO₂+CH₄ gas mixture into a liquid phase from which it is subsequently extracted by a second gas-liquid membrane thus achieving separation of the two components, Figure 3; Simons et al., 2009).

For biogas upgrading, membranes typically consist of materials that are permeable to CO₂, water and NH₃, partially permeable to H₂S and O₂ and essentially non-permeable to CH₄ and N₂ (Favre et al., 2009; Petersson and Wellinger, 2009; Deng and Hagg, 2010; Makaruk et al., 2010; Ryckebosch et al., 2011; Andriani et al., 2014). However, water and H₂S are usually removed from biogas before CO₂ separation as they can adversely affect membrane performance and efficiency. Due to the less than ideal efficiency of practical membranes, multi-stage separators are commonly used for efficient CH₄-CO₂ separation (Wellinger and Lindberg, 2005; Makaruk et al., 2010) and such biogas upgrading units with yield capacities >200 m³/h are in current use.

Recent developments in biogas upgrading technology include cryogenic separation, in situ biological methane enrichment and the so-called ecological lung (Petersson and Wellinger, 2009; Ryckebosch et al., 2011; Kao et al., 2012). These methods, although promising and providing better performance than traditional well-established technologies already operating in micro-, meso-, and macro-scale biomethane production plants, are still under development. Both traditional and more recent biogas upgrading methods are being continuously improved and developed, so that the subject is a very active area of applied technology research. It is worth noting that all biogas upgrading technologies are capable of producing RNG for vehicle applications containing >97% CH₄, which purity is superior to that of all globally produced natural gases. Ryckebosch et al. (2011) and Andriani et al. (2014) have reviewed biogas upgrading techniques currently in use or under development, providing comprehensive comparative technical and operational details, advantages/disadvantages, energy and technical requirements, maintenance and operational costs and other techno-economic information. More recently, Sun et al. (2015) published a comprehensive review covering biogas-upgrading technologies. As these authors point out, upgrading technology is site-specific, case-sensitive, and dependent on utilization requirements and local circumstances. They have critically evaluated state-of-the-art purification and upgrading processes, providing much useful information in regard to product purity, methane recovery and loss, process efficiency, as well as the investment and operating costs of the various alternatives.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.