GB2584704A - Gas treatment process and gas treatment apparatus - Google Patents

Abstract

Hollow fibre membrane contactors in series can be used to remove carbon dioxide from input gas (e.g. biogas from an aerobic digester), using an ammonia solution, to purify the methane gas recovered. A first carbon dioxide absorption solution, comprising ammonia, absorbs some of the carbon dioxide in a first gas-liquid contact stage. In a second stage, a second carbon dioxide absorption solution, comprising ammonia, absorbs carbon dioxide from the intermediate gas. The volume of the second carbon dioxide absorption solution may be at least twice the volume of the first carbon dioxide absorption solution. The hollow fibre membrane may be PTFE or polypropylene. Recirculated ammonium bicarbonate solution can be chilled to help prevent fouling of the membrane. Lumen-side crystallisation can be precluded, and shell-side crystallisation facilitated, by suppressing ammonia volatility through absorbent refrigeration and shifting the ammonia-ammonium equilibrium.

Description

GAS TREATMENT PROCESS AND GAS TREATMENT APPARATUS

Field of the Invention

The present invention relates to a gas treatment process and to a gas treatment apparatus. Of particular, although not necessarily exclusive interest, is the upgrading of biogas. Also of particular, although not necessarily exclusive interest, is carbon capture and storage.

Background

The application of gas-liquid contacting technology to the selective separation of CO2 from a multi component gas such as biogas is of interest for bulk gas separation. Biogas is the gaseous product from anaerobic digestion processes. Gas-liquid absorption columns are known for biogas upgrading. However, as columns rely on dispersive gas-liquid contact they suffer practical limitations such as flooding, channelling, entrainment and foaming which limit their operational range. Furthermore, their relatively low interfacial area coupled with the common use of water as the solvent necessitates pressurisation to increase absorption capacity of the fluid and reduce process scale through augmenting mass transfer.

WO 2006/022885 discloses a process for the treatment of combustion exhaust gas to capture CO2. CO2 absorption takes place in the aqueous NH3-0O2-H20 system, the absorption solution or slurry being chilled in order to reduce loss of NH3.

US 9,227,889 discloses a process for sweetening natural gas to reach liquefied natural gas specifications. A feed gas is introduced to the lumen side of a membrane contactor and an absorption solvent is introduced to the shell side of the membrane contactor. The absorption solvent is diethanolamine (DEA) or methyldiethanolamine (MDEA).

US 2016/0206993 discloses a closed cycle system for CO2 separation from CH4 for example for treatment of biogas using first and second membrane contactors. An ionic liquid absorber is used, comprising an organic salt.

US 2017/0274318 discloses methods for sequestering CO2. Aqueous ammonia is contacted with a gas containing CO2 under conditions sufficient to produce an aqueous ammonium carbonate. A microporous membrane contactor is used.

The use of a hollow fibre membrane contactor (HFMC) is of interest in gas-liquid separation technology, as an alternative to the use of columns. CO2 diffuses through the pores of a membrane into a liquid absorbent. The membrane therefore establishes non-dispersive contact between the two fluids, and can provide interfacial areas an order of magnitude higher that known absorption columns.

However, to date, attempts to implement CO2 separation from biogas, for example, using HFMC and using aqueous ammonia as a liquid absorbent have demonstrated uncontrolled precipitation of ammonia salts at the gas side of the HFMC, leading to membrane fouling and loss of function.

The present invention has been devised in light of the above considerations.

Summary of the invention

The present inventors have studied the precipitation of ammonia salts in the context of biogas treatment using aqueous ammonia, contacted via HFMC. They have realised that it is possible to provide first and second gas-liquid contact stages, fulfilling different functions, so that the ammonia salt precipitation can be controllably carried out to avoid uncontrolled nucleation and growth of crystallised reaction product at the HFMC, leading to a reliable gas treatment process with reduced risk of fouling of the HFMC. Furthermore, the present inventors have realised that their invention is not necessarily limited to biogas, but can be used for the treatment of an input gas comprising carbon dioxide, such as a combustion exhaust gas for example.

Accordingly, in a first aspect, the present invention provides a process for treatment of an input gas, the input gas comprising carbon dioxide, the process comprising: providing a first gas-liquid contact stage, the first gas-liquid contact stage comprising one or more gas-liquid hollow fibre membrane contactors permeable to carbon dioxide, the first gas-liquid contact stage having a gas side and a liquid side; flowing the input gas through the gas side of the first gas-liquid contact stage; providing a first carbon dioxide absorption solution, comprising ammonia; circulating the first carbon dioxide absorption solution through the liquid side of the first gas-liquid contact stage, to absorb some of the carbon dioxide from the input gas to leave an intermediate gas flowing out of the first gas-liquid contact stage; providing a second gas-liquid contact stage, the second gas-liquid contact stage comprising one or more gas-liquid hollow fibre membrane contactor permeable to carbon dioxide, the second gas-liquid contact stage having a gas side and a liquid side; flowing the intermediate gas through the gas side of the second gas-liquid contact stage; providing a second carbon dioxide absorption solution, comprising ammonia; circulating the second carbon dioxide absorption solution through the liquid side of the second gas-liquid contact stage, to absorb carbon dioxide from the intermediate gas to leave an output gas flowing out of the second gas-liquid contact stage, wherein the process is controlled so that more precipitation of ammonium bicarbonate occurs in the first carbon dioxide absorption solution, per unit volume of first carbon dioxide absorption solution, than in the second carbon dioxide absorption solution, per unit volume of second carbon dioxide absorption solution.

In a second aspect, the present invention provides a gas treatment system for treatment of an input gas, the input gas comprising carbon dioxide, the system comprising: a first gas-liquid contact stage, the first gas-liquid contact stage comprising one or more gas-liquid hollow fibre membrane contactors permeable to carbon dioxide, the first gas-liquid contact stage having a gas side and a liquid side, the first gas-liquid contact stage being adapted to receive a flow of input gas through the gas side of the first gas-liquid contact stage, the first gas-liquid contact stage being adapted to circulate a first carbon dioxide absorption solution, comprising ammonia, through the liquid side of the first gas-liquid contact stage, to absorb some of the carbon dioxide from the input gas to leave an intermediate gas to flow out of the first gas-liquid contact stage; a second gas-liquid contact stage, the second gas-liquid contact stage comprising one or more gas-liquid hollow fibre membrane contactor permeable to carbon dioxide, the second gas-liquid contact stage having a gas side and a liquid side, the second gas-liquid contact stage being adapted to receive a flow of the intermediate gas through the gas side of the second gas-liquid contact stage, the second gas-liquid contact stage being adapted to circulate a second carbon dioxide absorption solution, comprising ammonia, through the liquid side of the second gas-liquid contact stage, to absorb carbon dioxide from the intermediate gas to leave an output gas to flow out of the second gas-liquid contact stage, wherein the system is controllable so that more precipitation of ammonium bicarbonate occurs in the first carbon dioxide absorption solution, per unit volume of first carbon dioxide absorption solution, than in the second carbon dioxide absorption solution, per unit volume of second carbon dioxide absorption solution. 20 The present inventors consider that the first gas-liquid contact stage serves the purpose of allowing and promoting precipitation of ammonium bicarbonate, this in turn allowing sequestration of carbon dioxide from the system. The second gas-liquid contact stage, however, is necessary in order to further reduce the carbon dioxide concentration in the liquid, but with reduced or no precipitation of ammonium bicarbonate. In this manner, a continuous or quasi-continuous process can be operated, taking advantage of the features of hollow fibre membrane contactors.

Preferred or optional features of the invention will now be set out. These are applicable singly or in any combination with any aspect of the invention, unless the context demands otherwise.

In some embodiments, the concentration of ammonia in the first and second carbon dioxide absorption solutions is at least 2 mol L-1. The concentration of ammonia can be at least 2.2 mol L-1, at least 2.4 mol L-1, at least 2.6 mol L-1, at least 2.8 mol L-1, at least 3.0 mol L-1, at least 3.2 mol L-', at least 3.4 mol at least 3.6 mol L-1, at least 3.8 mol L-1, or at least 4.0 mol U1. It is considered that operating at a high concentration provides a higher product yield. The concentration of ammonia in the first and second carbon dioxide absorption solutions may be not more than 10 mol L-1.

The temperature of the first and second carbon dioxide absorption solutions is preferably controlled to be in the range of not more than 15°C. This may be achieved using one or more chillers. This temperature may be more than 0°C. Preferably, the temperature is not more than 14°C, not more than 13°C, not more than 12°C, not more than 11°C, not more than 10°C, not more than 9°C, not more than 8°C, not more than 7°C, not more than 6°C, or not more than 5°C. Controlling the temperature in this way is found to reduce the vapour pressure of ammonia, lowering its volatility and therefore its migration into the gas phase. In turn, this reduces ammonia slip.

The volume of the second carbon dioxide absorption solution in the process and system may be at least 2 times greater than the volume of the first carbon dioxide absorption solution. More preferably, the volume of the second carbon dioxide absorption solution in the process and system may be at least 3 times or 4 times greater than the volume of the first carbon dioxide absorption solution. This assists in controlling the conditions in the second carbon dioxide absorption solution to reduce or prevent precipitation of ammonium bicarbonate.

There may be provided a first carbon dioxide absorption solution chamber via which the first carbon dioxide absorption solution is circulated and recirculated through the liquid side of the first gas-liquid contact stage. There may also be provided a second carbon dioxide absorption solution chamber via which the second carbon dioxide absorption solution is circulated and recirculated through the liquid side of the second gas-liquid contact stage. In order to provide the effects identified above, the second chamber may have a volume of at least 2 times, at least 3 times or at least 4 times, the volume of the first carbon dioxide absorption solution chamber.

It is preferred that precipitated ammonium bicarbonate is removed from the first carbon dioxide absorption solution. This can be done continuously, semi-continuously or batch-wise. As discussed, the precipitated ammonium bicarbonate is of commercial interest as a product in its own right.

The first carbon dioxide absorption solution may be supplemented by some of the second carbon dioxide absorption solution. This can be achieved in the system by a line connecting the first carbon dioxide absorption solution chamber with the second carbon dioxide absorption solution chamber. This supplementation can be carried out continuously, semi-continuously or batch-wise.

The second carbon dioxide absorption solution may be supplemented by additional carbon dioxide absorption solution. For example, the additional carbon dioxide absorption solution may comprise ammonia thermally recovered from liquid digestate from an anaerobic digester.

The input gas may be a biogas from an anaerobic digester.

The present invention may therefore provide a convenient approach to the treatment of gas generated by an anaerobic digester, optionally using as the carbon dioxide absorption solution ammonia thermally recovered from liquid digestate from the anaerobic digester. In a further aspect, therefore, the present invention may provide an anaerobic digester plant comprising a system according to the second aspect, operatively connected to and anaerobic digester.

The input gas may comprise at least 40 mass % methane, or at least 45 mass % methane, at least 50 mass % methane, or at least 55 mass % methane. The input gas may comprise at least 20 mass cro carbon dioxide, or at least 25 mass % carbon dioxide, at least 30 mass % carbon dioxide, or at least 35 mass 96 carbon dioxide.

The output gas may comprise at least 90 mass % methane. The output gas may comprise at least 92 mass % methane, at least 94 mass % methane, at least 96 mass % methane, or at least 98 mass % methane. The output gas may comprise not more than 5 mass c/o carbon dioxide, more preferably not more than 4 mass % carbon dioxide, not more than 3 mass % carbon dioxide, or not more than 2 mass % carbon dioxide. In this way, the output gas can be provided in a manner to allow injection into utility gas supplies.

At steady state operation of the process and/or system, it is typical that the intermediate gas has a carbon dioxide concentration that is only slightly lower than that of the input gas. For example, the intermediate gas may comprises at least 0.8 times, or at least 0.9 times, or at least 0.95 times, the concentration of carbon dioxide in the input gas. We say "at steady stage" because at the outset of the process this limitation may not be reached. Another way to view this would be to consider the average carbon dioxide concentration of the intermediate gas during operation of the process and/or system.

The first gas-liquid contact stage may comprise two or more gas-liquid hollow fibre membrane contactors.

Similarly, the second gas-liquid contact stage may comprise two or more gas-liquid hollow fibre membrane contactors.

The (or each) membrane contactor may comprises a hydrophobic polymer.

The (or each) membrane contactor may have an average pore size of not more than 5 pm, more preferably not more than 4 pm, more preferably not more than 3 pm, more preferably not more than 2 pm, more preferably not more than 1 pm, more preferably not more than 0.5 pm.

The (or each) membrane contactor may comprise a thin film composite. In such a system there are no observable pores in the membrane surface. Thin film composite membranes may provide a degree of selectivity of CO2 over ammonia.

During operation of the process and/or system, gas-side crystallisation of ammonium bicarbonate in the gas-liquid hollow fibre membrane contactors may be substantially prevented. This can prevent fouling of the membrane contactors and can therefore allow the system and process to run continuously and efficiently.

The invention includes the combination of the aspects and preferred features described except where such a combination is clearly impermissible or expressly avoided.

Summary of the Figures

Embodiments and experiments illustrating the principles of the invention will now be discussed with reference to the accompanying figures in which: Fig. 1 shows the influence of ammonia concentration on lumen side crystallisation observed with PTFE membrane following absorbent recirculation. Supersaturation ratio C/C* calculated assuming CO2 absorbed fully converted in HCO3. Hydrodynamic conditions: G/L 11; VG 0.2 m VL 1.4 x 10-3 m s-1.

Gas and absorbent at 20°C. Error bars indicate standard deviation.

Fig. 2 shows the effect of absorbent temperature (3 moINH3 L-1, in recirculation) upon lumen side crystallisation observed with PTFE membrane. Supersaturation ratio C/C" calculated assuming CO2 absorbed fully converted in HCO3. (Left) Gas and absorbent at 20°C. (Right) Gas temperature 20°C; Liquid temperature °5C. Error bars indicate standard deviation.

Fig. 3 shows the impact of pore size reduction (from 3.4 pm in PTFE to 0.36 pm in PP membrane) on wetting prevention, following NH3 absorbent recirculation at 3M. Hydrodynamic conditions: G/L 5; VL 0.06 m S-1, VG 14.7 m S-1 for PP and 3.7 m S-1 for PTFE. Shell-side crystallisation occurred for both the materials. Membrane wetting observed only with PTFE following shell-side crystallisation. Error bars indicate standard deviation.

Fig. 4 shows the transition in crystal number determined at progressive levels of supersaturation (saturation demands conversion of 1.5M CO2 into HCO3-) for an NH3 concentration of 3.3M. Bicarbonate concentration shown for reference as this is important for induction. Error bars indicate standard deviation obtained from sacrificial experiments carried out in triplicate for each ammonia concentration.

Fig. 5 demonstrates suitable exemplary conditions for co-production of ammonium bicarbonate and high purity (98%) CH4. Two stage system comprising two 1.2 m2 HFMCs operated in series. Ammonia liquid is recirculated independently on the shell-side of each module whilst the gas flow passes successively between modules. The crystallising HFMC had an ammonia recycle volume of 0.6L whilst the secondary HFMC had an ammonia recycle volume of 4L. Inlet gas was 50/50 CO2/CH4 at a PG of 0.5 BarG (20°C) with 3.3 mol L-1 NH3 used as absorbent (5°C). Gas flow rate (Qc) was fixed at 2 L min-1. Liquid flow rate (QL) was fixed in recycle at 2 L min-1. Gas outlet is defined from the exit of the second HFMC.

Fig. 6 shows a schematic outline of a system and process according to an embodiment of the invention for the co-production of high purity methane gas product (stage two) and recovery of ammonium bicarbonate crystals (stage one).

Fig. 7 shows the influence of NH3(aq) concentration in the feeding solution on dual stage ammonia concentration recovered (black points) and volume recovered (red points), and on dual stage overall liquid phase mass transfer (blue points). Operational conditions: desorption time = 45 minutes; temperature = 80°C; vacuum pressure = 470mbar; volume initial feed = 1000mL; pH initial feed = 12 - 12.5. Error bars indicate standard deviation.

Fig. 8 shows the reduction in carbon dioxide flux reduction following recirculation of synthetic and real thermally recovered aqueous ammonia solutions at 3.3M and 5°C on the shell-side of the PP HFMC. Supersaturation ratio calculated assuming CO2 absorbed fully converted into HCO3-. Hydrodynamic conditions: G/L 5; VL 0.06 m s-1; VG 14.7 m s-1. Shell-side crystallisation occurred at C/C* of about 1.7 for both solutions. Error bars indicate standard deviation from sacrificial experiments undertaken in triplicate.

Detailed Description of the Invention

Aspects and embodiments of the present invention will now be discussed with reference to the accompanying figures. Further aspects and embodiments will be apparent to those skilled in the art. All documents mentioned in this text are incorporated herein by reference.

Further discussion of background and insight supporting the present disclosure Biogas produced through anaerobic digestion is typically a mixture of methane (CH4, at about 60% v/v) and carbon dioxide (CO2, at about 40% v/v). Through facilitating CO2 removal, biogas can be upgraded into biomethane for injection into the national grid (Persson et al., 2007). Following introduction of incentivisation for biomethane, there are now a large number of biogas upgrading plants operating across Europe. Most large scale facilities advocate absorption technology in which packed columns are used in conjunction with either water or chemical solvents as the scrubbing solution. Chemical solvents provide a higher absorption capacity and increased rate of mass transfer, which introduces considerable process intensification. VVhilst a relatively new technology for water utilities, chemical absorption is a mature technology in other sectors and has been extensively investigated for carbon capture and storage (CCS).

In CCS, ammonia (NH3) has been identified as an alternative solvent to organic amines since it exhibits greater absorption capacity, lower energy requirement for regeneration and is favourable from a cost perspective. However, due to the low molecular weight of ammonia, the saturation vapour pressure is high in comparison to alkanolamines, which introduces NH3 slip into the gas phase (Kozak et al., 2009).

In previous research from the same research group as the present inventors (McLeod et al., 2014), it was evidenced that NH3 slip could be limited by using a membrane contactor rather than a packed column.

Hollow fibre membrane contactors (HFMC) employ a hydrophobic microporous membrane to facilitate non-dispersive contact between gas and liquid phases, permitting only gases to diffuse through the pores. The minimisation of NH3 slip in HFMC was proposed to be due to the promotion of laminar flow in the liquid phase which promoted radial diffusion of NH3 from the absorbent bulk to the reactive boundary layer, facilitating overall stabilisation through the overall reaction (Budzianowski, 2011a): CO2 + NH3 +H20 -> NH4 HCO3(s) (1) Whilst the membrane introduces an additional resistance to mass transfer, the increased specific surface area in the HFMC provides substantive process intensification when compared to packed columns (Belaissaoui and Favre, 2018).

In biogas facilities operated by water utilities, there are ammonia rich wastewaters which have the potential to act as a sustainable source of absorption solvent for biogas upgrading. McLeod et al. (2014) illustrated an enhancement in CO2 separation of up to 15 times versus water when employing NH3 rich sludge liquor returns (around 2000 mgNH4*-N L1, 0.11mol L-1). He et al. (2017) introduced vacuum membrane distillation (VMD) to recover NH3 from return liquors prior to their introduction into a packed column for biogas upgrading. Pre-concentration using VMD achieved a recovered ammonia concentration of 18,300 mgN L-1 (1.3 mol L-1), markedly enhancing CO2 separation during biogas upgrading versus the untreated liquor. This remains below the ammonia concentration generally considered viable for CCS (4 to 6 mol L-1) (Shuangchen et al., 2013; Budzianowski, 2011b). The cost of ammonia removal from wastewater has been valued between £ 1.7 and £ 73.1 per kgNI-leremoved, dependent upon process scale (Oxera, 2006). Consequently, separation of the NH3-0O2 reaction product from solution presents several potential advantages, listed below: (i) avoiding the cost of wastewater treatment for ammonia; (ii) recovery of ammonium in a crystalline solid ammonium bicarbonate (NH4FIC03) product, which affords a potential resale value of up to 0.11 kg-1 due to its value as an industrial feedstock (e.g. fertiliser) (Budzianowski, 2011b); (Hi) no thermal requirement for solvent regeneration; (iv) minimisation of disposal volume, reducing transport costs; (v) locking CO2 into a solid phase which can improve environmental sustainability.

Crystallisation of the ammonium bicarbonate reaction product was first considered in the chilled ammonia process for CCS as a method to decrease NH3 slip in the absorber (WO 2006/022885). However, the homogeneous nucleation and growth of crystals within the mixed phase of a packed columns introduced clogging of equipment, making implementation difficult (Sutter et al., 2015).

McLeod et al. (2015) successfully demonstrated the feasibility to control the initiation of ammonium bicarbonate crystallisation in HFMC using a synthetic 2M ammonia solution. The high contact angle exhibited by the hydrophobic PTFE membrane employed promoted heterogeneous nucleation at the micropores where gas, liquid and membrane interact to induce surface growth of the reaction product.

This enhanced nucleation potential offered by the HFMC versus packed columns, coupled with the phase separation fostered by the membrane, enabled recovery of the reaction product in the liquid phase (Mcleod et al., 2015).

Therefore, HFMC technology offers the opportunity to simultaneously accomplish enhanced CO2 separation, NH3 removal from wastewater and the production of crystalline ammonium bicarbonate (comprising a potential resale value) within a single stage process, but the known processes outlined above cannot achieve this satisfactorily.

Previous studies that have sought to produce a solid crystalline product following reaction of ammonia with a counter oxyanion in real solutions have shown that the concentration of the reaction product is generally below the solubility limit, such that further processing is demanded before crystallisation can proceed. To illustrate, Ukwuani and Tao, (2016) coupled vacuum thermal stripping with acid absorption to recover ammonium sulphate ((NH4)2SO4) as the crystalline product. However, due to the high solubility limit of ammonium sulphate (706 g(NH4)2SO4 L-1), the introduction of an antisolvent or thermal gradient was needed to initiate the phase change to a crystalline product (Ukwuani and Tao, 2016), introducing a third processing stage, which increases capital cost and process complexity. Similarly, following facilitation of an NI-13-0O2 reaction in real wastewater using a HFMC, McLeod et al. (2014) identified that to achieve the phase change into a crystalline product, thermal treatment was required. However, due to the high solubility of ammonium bicarbonate (182 gNH4HCO3 L-1, 42 gNH4+ L-1 or 2.3M at 20°C) and the presence of other cations, more thermodynamically favourable products such as calcium carbonate and sodium bicarbonate were also formed, which limited ammonia transformation into the preferred solid reaction product.

Early research of NH3-0O2 absorption in packed columns focussed on forming ammonium bicarbonate (NH4FIC03) as the solid phase reaction product, to reduce the mass flow of stripper solution thereby reducing the heat requirement but also limiting NH3 slip (Gazzani et al., 2014). However, problems in managing solid precipitate within the column was demonstrated to be difficult due to clogging (Sutter et al., 2015). Consequently, NH3-0O2 column studies have typically sought to implement strategies to eliminate formation of a solid phase reaction product (Yu et al., 2012; Gazzani et al., 2014).

In studies utilising hollow fibre membrane contactors to facilitate NH3-0O2 absorption, the formation of a problematic solid phase reaction product has been observed to preferentially form in the gas-phase side of the membrane (typically the lumen-side), which blocked the lumen, subsequently limiting further gas-side flow (Makhloufi et al., 2014; Cui and deMonfigny, 2017; Villeneuve et al., 2018). To illustrate, Makhloufi et al. (2014) was unable to generate a stable CO2 absorption flux due to lumen-side (gas side) NH4HCO3 precipitation, which has been similarly evidenced elsewhere (Cui and deMontigny, 2017).

In the original investigation carried out in the inventors' research group (Mcleod et al., 2015), we were able to initiate shell-side nucleation through the introduction of sodium chloride into solution which modified the fluid surface tension and limited membrane wetting. However, the CO2 flux was insufficiently stable to enable complete crystallisation.

Overview of the advantages of the present disclosure Based on the insight of the inventors, control over the crystallisation within the liquid phase can be achieved as follows, thereby allowing control to ensure that crystallisation occurs away from the HFMC.

In the present work, the inventors have implemented four modifications to allow consistent and reproducible crystallisation in the liquid phase (typically in the shell side) in hollow fibre membrane contactors: * Reduced the liquid temperature to around 5°C to reduce the vapour pressure of ammonia, lowering its volatility and hence its migration into the gas phase; * Reduced the ammonia concentration in solution which constrained surface nucleation rate, the negative result of which is a modification in contact angle, resulting in disadvantageous fluid breakthrough into the gas phase; * Selected membrane properties (specifically pore size but also contact angle) which limit surface wetting and lower the energy barrier for nucleation; and * Operated the absorbent solution in recycle which forces a progressive reduction in pH with the continued absorption of CO2, subsequently converting free ammonia to non-volatile ammonium (which is the correct compound to induce crystallisation in solution).

By controlling the parameters identified above, it was possible to selectively control ammonium bicarbonate nucleation and crystal growth in the liquid phase, through a combination of membrane properties and specific operating conditions. However, the ammonia concentration was considerably lower than has been studied for CO2-NH3 crystallisation and so introduced an unsteady conversion. This means that the biogas being treated does not meet the required standards as biomethane (ordinarily greater than 90% methane).

The solution chemistry was further studied to provide quantitative evidence of the onset of membrane induced nucleation and crystallisation that could be used to scale the process within this dynamic state, where the bicarbonate solution concentration varied due to a progressive CO2 dissolution and a dynamic pH which caused a shift in the carbonate-bicarbonate-carbonic acid equilibrium over time. This evidence was then used to develop and test a two-stage configuration which could enable the crystallisation of ammonium bicarbonate and the complete treatment of biogas to the appropriate gas phase concentration: * In the first stage, CO2 rich biogas enters into the hollow fibre membrane contactor and contacts a small volume of recirculating solution comprising ammonia. Due to the progressive increase of CO2, the solution acidifies (pH reduces) which subsequently shifts the ammonia-ammonium equilibrium toward ammonium which is less chemically reactive with CO2 but is favourable for the induction of ammonium bicarbonate crystallisation; * Once sufficient bicarbonate is developed in the first stage, heterogeneous nucleation is promoted at the membrane surface, and ammonium bicarbonate crystallisation proceeds; * Whilst the CO2 develops within the solution of the first stage, the chemical reactivity declines and the gas phase exits stage one still containing a considerable fraction of CO2 gas; * The CO2 gas enters into a second stage hollow fibre membrane contactor, which also has a cooled recirculating solution of ammonium bicarbonate but a greater volume of this solution is provided to limit the overdevelopment of CO2 concentration in solution. The sustained solution reactivity ensures that the second stage provides sufficient CO2 separation to achieve the specified gas treatment objective; * Once crystallisation in complete in the first stage, the solution is drained down, and topped up with solution from the second stage. This ensures no CO2 losses. Fresh ammonia absorbent is then added to stage two. In a continuous process, ammonium bicarbonate crystals can be harvested continuously or semi-continuously from the first stage and the first stage reservoir can be topped up continuously or semi-continuously with solution from the second stage.

Whilst not explicitly demonstrated, Teichgraber and Stein (1994) proposed that using thermal packed column stripping for direct return liquor treatment, separation and concentration of the ammonia fraction could be achieved to deliver a final concentration of up to 130 gNH3*-N L-1 (about 9 moINH3 L-1), which is considerably higher than has been previously identified in the literature to date. Our research has demonstrated the thermal stripping of ammonia to achieve equivalent concentrations to those proposed (Figure 7). In preferred embodiments of the present invention, such a liquid can act as an environmentally sustainable source of ammonia for enhanced CO2 separation (Figure 8). Our research evidenced three key results: * In the thermal pre-concentration step, lower temperatures enable improved selectivity toward ammonia which can advantage reactivity of the final produced solution; * The reactivity of ammonia concentrate recovered from return liquors behaves similarly to synthetic solutions. Whilst we and others previously observed similar results, the use of thermal pre-concentration to achieve such high initial (and hence reactive) ammonia concentrations has not previously been achieved; and * Due to thermal pre-concentration, the impact of other cations present in the mother solution is negligible as they are not carried over which ensures a pure ammonium bicarbonate crystal can be produced and recovered.

In more detail, Figure 7 demonstrates that during thermal desorption, mass transfer is controlled by initial concentration. Whilst mass transfer rate changed, the recovered ammonia concentration did not which indicates that the solution was at saturation for ammonia which for the present conditions was around 6 mo1/1. This is the highest concentration recorded to date in the literature to our knowledge and based on the disclosure provided here can be increased significantly.

For Figure 8, we compared synthetic ammonia solution to real ammonia solution recovered from return liquors, and identified a common trend between the synthetic and real solutions; a slight reduction in CO2 mass transfer was observed with the real solution due to the carryover of surface active agents Identification of boundary conditions and membrane properties to promote crystallisation of ammonia-0O2 reaction products in membrane contactors for bionas upgrading Materials and methods Fabrication, equipment setup and operation: The micro-porous PTFE hollow fibre (Zeus Industrial Products, Letterkenny, Ireland) comprised pores with a nominal pore size of 0.5 pm. The maximum stretched pore length was also characterised (dmax, 3.4 pm) as the pores were oval rather than circular. The PP hollow-fibre (Membrana GmbH, Wuppertal, Germany), comprised a nominal pore size of 0.2 pm (dmax, 0.36pm) (Table 1).

Table 1 -Dimensions and surface characteristics of the single membrane fibre (2) where B (-) is the pore geometry coefficient (B=1 for perfectly cylindrical pores, O<B<1 for non-cylindrical pores), yL(N m-1) is the liquid surface tension, 0 (°) is the contact angle between the liquid absorbent and the membrane surface and dirait (m) is the maximum pore diameter of the membrane. For each experiment, single hollow-fibres with an active length of 165mm were fixed into a Perspex cell, with a 12mm diameter channel. Hollow-fibres were potted in epoxy resin (Bostick Ltd., Stafford, UK) and sealed into the crossflow cell.

Carbon dioxide (99.8%, BOC gases, Ipswich, UK) was introduced into the lumen of the hollow-fibre at a flow rate of 1000 ml min-1 using a laminar mass flow controller (0.01-1 L min-1, Roxspur Measurement and Control Ltd., Sheffield, UK). Absorbent was pumped counter-current on the shell-side of the membrane at 200 ml min-1 with a peristaltic pump (520Du, Watson-Marlow Ltd., Falmouth, UK). The absorbent temperature was fixed with a refrigerated bath and circulator (R1 series, Grant Instruments Ltd.,

PIPE

ane mate min ctuter iiem.eter:NM thicitness Aouve angth Surface r eat lm.un1 pore stze;tOiciiii) Maximum pore. size, (dmkrj Geumefrical facts, B Cuctaa argle (a't Breattthrougtt pressure Lumen cross seettunsu are PntMráfluwoethylene (FIFE) 22o 3..44 x 10'3 60X101 9.6 tim Shell side characteristics Height 1,-Vidth treciti)nai area dome OpratioSl Flow eime Brfa side Lumen -side tiguld temperature Gas temottratare Data c>f 'Witted b'V ma Based on fore outer dat < Dafa efelistfcaNdetertnin Bowe and Mtita (2013) " For comparison ith PP and ol The liquid entry pressure (Pa) for each membrane was estimated according to the Laplace-Young equation (Franken et al., 1987): n PIFE was quit} recor.isalat unter-currer 3 -23 06 0 Es 09 8% CD: 20i1' Cambridge, UK). Thermocouples (K-type, Thermosense Ltd., Bucks, UK) were sited in the gas and liquid phase both upstream and downstream of the cell.

Chemical preparation, sampling and analysis: Absorbent ammonia concentration was fixed by addition of aqueous NH3 concentrate (35% Fisher Chemicals, Loughborough, UK) to de-ionised water (15.0 MD cm-1) (Table 1). The absorbent pH was fixed at pH 10 with addition of hydrochloric acid (HCI, 37%, Fisher Scientific, Loughborough, UK), which is typical of that employed in aqueous ammonia packed column processes (Yeh et al., 2005). Ammonia concentration was confirmed using ammonium cell test (VWR International Ltd., Poole, UK) coupled with spectrophotometric determination (Spectroquant Nova 60, Merck-Millipore, Darnstadt, Germany). Manual volumetric flow meters of 50 mL (Restek, Bellefonte, US) and 1000 mL volume (SKC, Blandford Forum, UK) were used to measure the gas flow rate and to calculate the carbon dioxide flux (J002, MOI rn-2 S-1).

where QG in and QG out are gas flow rate (m3 s-1) before and after HFMC, Air is the membrane surface area for absorption (m2) and TG is the gas temperature (K) (Atchariyawut et al., 2007). The error for gas flow measurement was around 2% of the reported value.

Solution pH was monitored using a Jenway epoxy bodied pH electrode (bulb end type) connected to a pH meter (Jenway 4330, Cole-Parmer, Stone, UK). The solution UV absorbance was determined at 215 nm (Jenway 6715 UVNis. Spectrophotometer, Cole-Parmer, Stone, UK) which corresponds to the absorption of bicarbonate. To characterise the development of the crystalline solids phase, experiments were terminated at different levels of supersaturation (C/C"), which was defined as the ratio between the CO2 absorbed into solution and the CO2 required to form ammonium bicarbonate at the solubility limit. For each level of supersaturation, sacrificial experiments were undertaken in triplicate and the crystal size distribution (CSD) determined. The absorbent was initially filtered through a 0.45 pm VVhatman filter and weighed (Camlab Ltd., Cambridge, UK). Crystals were then immersed in anhydrous alcohol to minimise agglomeration during counting and transferred to a microscope slide consisting of a 1 mm grid for counting under an optical microscope (Optech Microscope Services Ltd., Thame, UK), equipped with PL 5/0.12 lens and digital camera (Infinity 3, Lumenera, Ottawa, Canada). Images were analysed with an image processing software (Image Pro Plus, Media Cybernetics, Cambridge, UK) to determine crystal numbers and sizes. Each image analysed corresponded to around 50 crystals. For each sacrificial test, at least 600 crystals were classified to achieve sufficient accuracy for the CSD, which was determined assuming a log-normal distribution.

Chemistry of the NH3-0O2-H20 system: In the liquid phase, Wang et al. (2011) explains the kinetics of the reactions of NH3(aq) with CO2(aq) and carbonate species, comprising the reversible reactions of ammonium carbarnate/ carbamic acid formation.

The relative proportion of dissolved CO2 available as carbonic acid (H2CO3) and the relative proportion of ammonia present in protonated form (NH4) are dependent on both solution pH and temperature.

Whilst five solid reaction products may potentially form, ammonium bicarbonate is thermodynamically favoured, due to the lower solubility of the salt, which exhibits some dependency upon temperature.

Ammonium bicarbonate is formed from the ionic bond between NH4 1-and HCO3-.

In the gas phase, carbon dioxide and ammonia can produce ammonium carbamate (NH2COONH4). Through hydrolysis, NH2COONH4 successively converts into ammonium bicarbonate (NH4HCO3).

Nevertheless, due to the high solubility of ammonium carbamate in water, ammonium bicarbonate formation in the gas phase is more likely to proceed.

An increase in aqueous ammonia concentration raises ammonia volatility, thus facilitating NH3 slip in the gas phase and increasing the likelihood for gas side crystallisation of NI-141-1CO3.

Results Diagnosis of liquid absorbent operation to reduce/prevent lumen side crystallisation: The PTFE HFMC was initially tested in single pass operation with absorbent and gas temperatures at 20±1°C. Ammonia concentration was varied between 0.6 and 3 moINH3 L-1 whilst liquid and gas velocity (VL and VG) were fixed at 0.2 m s-1 and 1.4 x 10-3 m s-1, respectively.

Lumen-side (i.e. gas-side) crystallisation was observed at the two highest ammonia concentrations examined (2.3 and 3 moINH3 L-1), which corresponded to 10 and 15 gCO2 absorbed L-1. EDX analysis of the crystals grown in the lumen using absorbent at 3 moINH3 L-1, demonstrated that they comprised 0, N and C in a ratio indicative of ammonium bicarbonate.

During tests under identical hydrodynamic conditions, ammonia concentrations and gas/liquid temperatures, but following absorbent recirculation (rather than single pass), lumen side crystallisation was not observed at 0.6 and 2.3 moINH3 L-1 corresponding to 26 and 120 gCO2 absorbed L-1, respectively (where NH4HCO3 solubility is equivalent to 101 gCO2 absorbed L-1 at 20°C and 1 atm), while it was detected when using absorbent at 3 moINH3L-1, below NH4HCO3 solubility limit (Fig. 1).

Chilled absorbent recirculation at 5±1°C and 3 moINH3 L-1 within the same PTFE HFMC induced multiple advantages, including reduction of NH4HCO3 solubility to 75 gCO2 absorbed L-1, prevention of lumen-side crystallisation and promotion of shell-side crystallisation, which commenced at 104 gCO2 absorbed L-1 (Fig. 2). Specifically, the ammonia saturated vapour pressure dropped from 6.11 to 2.9 kPa when refrigerating the aqueous ammonia solution at 3 moINH3 L-1 from 20 to 5°C, which significantly reduced ammonia slip in the gas phase. Nevertheless, at 113 gCO2 absorbed L-1, wetting occurred, facilitating CO2 mass transfer (Figs. 2 and 3).

Investigation of vapour-liquid-solid equilibria throughout CO2 absorption within chilled aqueous ammonia in recirculation: Reduction of pH from 10 to 7.3 (0<C/C*<1.7) was observed when recirculating chilled aqueous ammonia at 3 moINH3 L-1, corresponding to ammonia-ammonium transition up to 100% ionised ammonia (hence, 3 moINHC-N L-1).

In the same pH range, bicarbonate concentration initially raised up to 3 moIHCO3-L-1 at pH 8 (where HCO3-100% within the carbonic acid speciation), corresponding to a supersaturation ratio C/C*=1 (HCO3-saturation), followed by a decline, where shell-side crystal nucleation and growth of ammonium bicarbonate occurred, coupled with 13% reduction of bicarbonate converted into carbonic acid. In particular, in C/C* range between 1 and 1.7, overall bicarbonate drop was approximately 85%, furtherly exacerbated in C/C* range between 1.7 and 2.5, with an overall HCO3-decrement of 3-fold.

Influence of membrane properties to prevent wetting and promote shell-side nucleation and growth: Membrane pore size reduction from 3.4 to 0.36 pm, following recirculation of aqueous ammonia at 5°C (G/L = 5), enabled stable CO2 flux and prevented wetting until complete exhaustion of the absorbent (C/C*= 2.5) (Fig. 3), facilitating consistent and reproducible shell-side crystallisation. In particular, the decrement in pore size accomplished (from PTFE to PP), induced an increase in the geometrical factor B from 0.15 to 0.56, boosting the liquid breakthrough pressure by 20-fold (Table 1). Once exceeded bicarbonate saturation limit (C/C"=1), heterogeneous nucleation and growth commenced on both the fibres (AGhet/AGI., of about 90% for PTFE and about 60% for PP; Table 1), followed by collection in the bulk. When using the membrane with wide pore structure, wetting occurred subsequent to growth (Fig. 3), resulting in crystal precipitation of 100 crystals cm-3 of 440 pm mean size at C/C"=1.7, compared with only 1 crystal cm-3 of 230 pm mean size at the same supersaturation ratio within the tight pore structure.

In the latter case, in particular, ammonium bicarbonate crystal number increased from 1 crystal cm-3 at C/C*=1.7 to 10 crystal cm-3 at C/C*=2.5, while crystal size raised from 233 to 709 pm within the same C/C* range. Analogously, a gain in mean crystal mass recovered was detected within the 0.36 pm pore size fibre from 12 mg at C/C"=1.7 to 491 mg at C/C"=2.5, whilst 815 mg have been recovered from the test with 3.4 pm pore size fibre at C/C"=1.7.

Discussion This study has demonstrated that by limiting the transport of free ammonia into the gas phase through suppressing ammonia vapour pressure using temperature, coupled with recirculation of the absorbent to shift the equilibrium toward the non-volatile ionised ammonium form, consistent and reproducible induction and growth of ammonium bicarbonate crystals can be achieved on the shell-side (i.e. liquid-side) of the membrane.

When absorbent flow was operated in single pass, NH4HCO3 crystallisation was observed in the gas phase, on the lumen-side of the membrane, and corresponded to conditions in which the 2.3 M NH3 absorbent solution was below the solubility limit (C/C*, 1) required to facilitate crystallisation within the liquid phase. This is similar to earlier investigation in which CO2 mass transfer was studied using single pass NH3 absorbent flow (Makhloufi et al., 2014; Cui and deMontigny, 2017; Villeneuve et al., 2018). A previous investigation has demonstrated crystalline ammonium bicarbonate can result from gas phase reaction via formation of ammonium carbamate (NH2COONH4) as an intermediate solid. Consequently, gas-phase crystallisation could proceed through one of two mechanisms: direct reaction between CO2 and slipped NH3 in the gas phase; or wetting and breakthrough of solvent into the gas phase followed by CO2 dissolution. At an equivalent absorbent concentration, lumen side crystallisation was prevented through recycling the absorbent (Fig. 1). As CO2 absorbs, it quickly reacts with free NH3(aq) to form carbamic acid (NH2000H), which then transforms to carbamate liberating a hydrogen ion. The resultant reduction in pH shifts the ammonia-ammonium equilibrium toward the non-volatile ionised ammonium form. It is this conversion of NH3 into its ionised form which dissipates NH3 slip, reducing the probability for ammonium bicarbonate formation in the gas phase. Since 2.3 moINH3 L-1 was not sufficient to induce crystallisation of the reaction product on the shell-side at ambient temperature (20±1 CC) an increase in aqueous ammonia to 3 moINH3 L-1 was trialled which again promoted lumen side crystallisation (Fig. 1). At 3 moINH3 L-1, the saturated vapour pressure for ammonia is 65% higher than for 2.3 moINH3 L-1 which increased the probability for transport to the gas phase. This is supported by the observations of McLeod et al. (2015) in which lumen side (gas-phase) crystallisation occurred during absorbent recirculation at ambient temperature for 3 to 5 moINH3 L-1 but not at 2 moINH3 L-1. For the Chemically Reactive membrane crystallisation reactor (CR-MCr), the reduction in absorbent temperature affords two advantages: a reduction in NH3 vapour pressure will limit the probability for gas-phase crystallisation; and at lower temperature NH4HCO3 becomes less soluble, which should advantage induction and crystal growth on the shell-side (absorbent side) of the membrane. This was demonstrated through recirculation of 3moINH3 L-1 absorbent at 5±1°C, where lumen-side crystallisation was prevented in favour of shell-side nucleation (Fig. 2). The point of induction was observed close to the solubility limit for CO2 (C/C"), which was coincidental with around 100% of the nitrogen present as ammonium since solution pH had reduced to pH 8 at this point both ammonium and bicarbonate concentrations were twice the solubility limit (1.5mol L-1), enabling shell side nucleation to occur on the PTFE fibre.

Shortly after saturation was achieved for the PTFE membrane, shell-side crystal growth was determined. Polytetrafluoroethylene exhibits a contact angle of around 135° (Bougie and lliuta, 2013) which is thought to promote heterogeneous nucleation at the membrane wall (AGhei/AGhom, of about 90%): Once nucleation and growth was observed, a sudden peak in CO2 flux occurred due to breakthrough of ammonia solution into the gas phase, in response to surface welling subsequently enabling direct chemical reaction with CO2 (Fig. 3). Through microscopic surface characterisation, McLeod et al. (2015) were able to confirm nucleation to initiate at the edge of the pore, where the membrane (solid phase), liquid and gas phases intersect. It is proposed that the concentration gradient developed within the boundary layer coupled with the material promotion of heterogeneous nucleation, initiated substantive nucleation within the membrane pores. The extent of nucleation was confirmed in the bulk, where the crystal number was two orders of magnitude higher than for the polypropylene membrane at an equivalent supersaturation level. Crystallisation around the membrane pores modified the contact angle, reducing the breakthrough pressure which was already limited by the large pore size, subsequently inducing wetting and the transfer of solvent into the gas phase. Application of a smaller pore size membrane (polypropylene) in conjunction with chilled recirculating absorbent to limit slip, enabled consistent and reproducible crystallisation on the shell-side (liquid side) of the membrane (Fig. 3).

Specifically, the stretched pore length of the membrane reduced from 3.4 to 0.36 pm, which increased the breakthrough pressure by 20-fold. The lower crystal number and size observed for the polypropylene membrane can be ascribed to the smaller contact angle of the material (117°) which reduced the thermodynamic potential for induction. Analogous to the trend identified for crystal growth with the 0.36 pm stretched pore size membrane, Di Profio et al. (2003) also recorded an increase in crystal number and crystal size over time when using a similar polypropylene membrane in osmotic crystallisation of organic macromolecules, and inferred the trend to be that of heterogeneous nucleation. Nitrogen mass balance demonstrated that 5 96 of the nitrogen lost from the NH3 absorption solvent was recovered within the crystallised product in the bulk solution, which suggests negligible NH3 slip to have occurred and insignificant crystalline surface deposition on the surface of the membrane.

Accordingly, recirculation of a chilled absorbent within HFMCs can reduce or prevent lumen-side blockage and enable consistent and reproducible shell-side nucleation and growth of ammonium bicarbonate through chemical CO2 absorption from the gas phase.

Lumen-side crystallisation can be precluded and shell-side crystallisation facilitated by (1) suppressing ammonia volatility through absorbent refrigeration and (2) shifting ammonia-ammonium equilibrium toward the non-volatile NH4* through liquid recirculation. Additionally, selection of a membrane with tight and regulated pore structure quenches solvent ingress into the pores, thus promoting sustained crystallisation on the membrane-liquid interface. Preferential heterogeneous nucleation on the tight pore structure fibre, driven by a concentration gradient, favours nucleation over growth.

Chemically reactive membrane crystallisation reactor for CO2-M-I3 absorption and ammonium bicarbonate crystallisation: Kinetics of heterogeneous crystal growth Materials and methods Fabrication, equipment setup and operation: The module comprised a single 165mm long polypropylene (PP) micro-porous hollow-fibre membrane (Membrana GmbH, Wuppertal, Germany) with a nominal pore size of 0.2 pm and membrane area of 9.33x10-4 m2 (based on the outer diameter) (Table 2).

Table 2-Dimensions and characteristics of the hollow-fibre membrane Fibre characterstics Membrane rnateual mm mm um mm rn2 P II/Prot:Ilene * Inner diameter Outer diameter Wall thickness Active length Surface are& um 1.2 Nominal pore srce mm mm 1.8 Lumen cross sectionaL area Shell side characteristics * Height VVidth Shell cross sectional area * Priming volume Operational character Flow regime Shell-side Lumen-side Liquid temperature Gas temperature Data provided by manufacturer b Based on fibre cuter diameter 0.2 1.13 x Box 10' *11.0 Counter-curren 1 1933 4.6 M NH (ag) 99.8% CO: 5..-7 8 -20 The hollow-fibre membrane was potted in epoxy resin (Bostick Ltd., Stafford, UK) and sited within a 12mm diameter channel within the Perspex cell. To allow direct observation of shell-side crystallisation, a viewing window was engineered into a recess within the upper section of the cell (Autin et al., 2016). A Nikon SMZ-2T stereomicroscope with a 0.5x objective lens (Nikon UK Ltd., Surrey, UK) was fixed above the viewing window and images captured with a high-resolution camera (Leica EC3 Microsystems, Milton Keynes, UK). Membrane experiments were conducted sacrificially for each supersaturation level reached (C/C"). The supersaturation level (C/C*) is defined as the ratio between the CO2 absorbed into solution and the CO2 required to form ammonium bicarbonate at the solubility limit. Thus for each supersaturation level, a new single hollow-fibre was mounted into the Perspex cell. Sacrificial experiments were conducted in triplicate for each supersaturation level, and the mean and standard deviation reported. Carbon dioxide (99.8%, BOC gases, Ipswich, UK) was introduced into the lumen of the hollow-fibre at a flow rate of 1000mImin-1 using a laminar mass flow controller (0.01-1 L min-1, Roxspur Measurement and Control Ltd., Sheffield, UK). Absorbent was pumped counter-current on the shell-side of the membrane at 200m1min-1 with a peristaltic pump (520Du, Watson-Marlow Ltd., Falmouth, UK). The absorbent temperature was fixed at 5°C with a refrigerated bath and circulator (R1 series, Grant Instruments Ltd., Cambridge, UK). Gas and liquid phase temperatures were measured across the membrane system (K-type thermocouples, Thermosense Ltd., Bucks, UK). Inlet temperatures for the gas and liquid phase were 20°C and 5°C respectively, with losses across the membrane noted to be around 0.4±0.2°C for both liquid and gas.

Chemical preparation, sampling and analysis: Absorbent ammonia concentrations ranging between 1 and 4.6 moINI-13 L-1, were prepared through addition of aqueous NI-13 concentrate (35% Fisher Chemicals, Loughborough, UK) to de-ionised water (15.0 Ma cm-1). The absorbent pH was fixed at pH 10 with addition of hydrochloric acid (HCI, 37%, Fisher Scientific, Loughborough, UK) as this is typical of the pH employed in aqueous ammonia packed column processes. Ammonia concentration was confirmed using an ammonium cell test which pre-acidifies the sample (VWR International Ltd., Poole, UK) before determination by spectrophotometry (Spectroquant Nova 60, Merck-Millipore, Darnstadt, Germany). A 1000 ml bubble flow meter (SKC, Blandford Forum, UK) was used to measure gas flow rate in order to calculate CO2 flux (J002, mol m-25-1): where QG in and QGput are the inlet and outlet gas flow rates (m3 S-1) respectively, Am is the membrane surface area (m2) and To is the gas temperature (K). The error for gas flow measurement was less than 2% of the reported value. Absorption solvent pH was monitored using a Jenway epoxy bodied pH electrode (Jenway 4330, Cole-Farmer, Stone, UK). The development of bicarbonate in solution was determined by UV absorbance at 215 nm (Jenway 6715, Cole-Farmer, Stone, UK) which corresponds to the absorption region for bicarbonate. A crystal size distribution was developed for each sacrificial experiment which corresponded to a specific level of supersaturation (C/C*). Prior to counting, the absorbent was filtered through a 0.45 pm filter (VVhatman, Camlab Ltd., Cambridge, UK) and then immersed in anhydrous alcohol to minimise agglomeration. The crystals were transferred onto a microscope slide consisting of a 1 mm counting grid, and placed under an optical microscope (Optech Microscope Services Ltd., Thame, UK) equipped with PL 5/0.12 lens and digital camera (Infinity 3, Lumenera, Ottawa, Canada). Images were analysed with image processing software (Image Pro Plus, Media Cybernetics, Cambridge, UK) to determine crystal number and size. Around 50 crystals were quantified with each image and a minimum of 600 crystals counted to develop one size distribution to minimise standard error. A Siemens D5005 X-ray diffractometer with Cu Ka1 radiation (Bruker UK Ltd., Coventry, UK) was used for crystal analysis, using a step size of 0.04° with diffraction patterns recorded in the 20 range 15-40°.

Chemistry of the NH3-0O2-H20 system: Upon absorption of CO2 into solution, the relative proportion of dissolved CO2 available as carbonate (C032-), bicarbonate (HCO3) or carbonic acid (H2CO3) is dependent upon solution pH, assuming a fixed liquid temperature of 5°C.

As explained above, ammonium bicarbonate is thermodynamically favoured, due to the lower solubility of the salt, which exhibits some dependency upon temperature. Ammonium bicarbonate is formed from the ionic bond between NH4 4-and HCO3.

Crystallisation kinetics: In a supersaturated solution, the nucleation rate and crystal growth rate can be determined by consideration of rate constants for nucleation and growth respectively, along with the CO2(aq) concentrations adjacent to the membrane surface and the equilibrium-saturated concentration at the membrane wall.

Results Impact of ammonia concentration on shell-side ammonium bicarbonate crystallisation: Carbon dioxide absorption was determined at four liquid phase ammonia concentrations ranging 1 to 4.6 moINH3 L-1 with the NH3 solution in recirculation. For each NH3 concentration, CO2 flux followed a similar two-stage profile, characterised by a rapid initial decline in CO2 flux followed by a slow progressive decline in CO2 flux. Whilst the initial CO2 flux was lower for the 1M NH3 solution, the flux profile was similar for NH3 solutions ranging 1.9 to 4.6M. A spike in CO2 flux was observed for the 4.6M NH3 solution, corresponding to a supersaturation level of 1.2 and was coincident with the onset of substantive shell-side crystallisation which covered the membrane surface. Within the 3.3M NH3 solution, shell-side crystallisation was observed at C/C" 1.7, with no observable impact on CO2 flux. Shell-side crystallisation did not occur in NH3 solutions ranging 1-1.9M.

Chemistry governs induction of NH4HCO3 in CO2-NH3-H20 system: As CO2 absorption progressed, the solution pH declined. The reduction in pH was more evident for the lower ammonia concentrations (1 and 1.9M) despite the absorption of an equivalent amount of CO2. To illustrate, a minimum pH value of 7.4 was reached at C/C" 0.55 with 1M NH3 absorbent, whilst that pH was reached at C/C" 2.6 for the 4.6M NH3 solution. However, normalisation of the cumulative CO2 absorption data to the initial free ammonia concentration of solution (CO2 solution loading, mol mol-1) evidenced that the reduction in pH exhibits an analogous trend independent of initial NH3 solution concentration. Shell-side crystallisation occurred at CO2 loadings of 0.77 and 0.44 for 3.3 and 4.6 moINH3 L-1 respectively.

The arising pH data was developed to estimate ammonia and bicarbonate equilibria during CO2 absorption. The ammonia-ammonium equilibrium had shifted primarily toward NH4* for each absorbent at supersaturation (C/C", 1), independent of initial NH3 concentration. During the initial CO2 absorption phase, the carbonate equilibrium shifts toward bicarbonate, due to the decrease in solution pH. However, as solution pH continued to decline with further CO2 absorption, the water-0O2 equilibrium shifts away from bicarbonate toward carbonic acid. For the lower NH3 solution concentrations, this results in a peak in HCO3-concentration in advance of supersaturation (C/C*, 1). This was corroborated by UV215nm in which the bicarbonate concentration was noted to peak for both the 1 and 1.9M NH3 solutions prior to supersaturation, and was evidently a contribution of the initial ammonia concentration. The UV215nm absorbance provided a surrogate for HCO3-concentration and demonstrated the 4.6M NH3 solution to comprise the highest concentration at the point of supersaturation.

Kinetics of nucleation and growth for chemically assisted membrane crystallisation: For the 3.3M NH3 solution, an increase in crystal number and crystal size was observed with an increase in supersaturation level and was coincident with a decrease in UV215nm (Fig. 4).

Nitrogen mass balance indicated that >99% of the nitrogen reduction in the absorbent was contained within the recovered crystalline product. X-ray diffraction was used to compare pure ammonium bicarbonate to crystals produced following CO2 absorption into 3.3M NH3 solution, which evidenced crystals obtained through the CR-MCr to be ammonium bicarbonate.

Discussion In this study, the underpinning chemistry required to simultaneously absorb CO2 and initiate nucleation of crystalline ammonium bicarbonate in a hollow-fibre membrane contactor has been ascertained, and the kinetics of heterogeneous nucleation and crystal growth determined, which evidence the mechanism for chemically reactive membrane crystallisation. Both crystal population density and crystal size increased with increasing levels of supersaturation. This is in contrast to Veiga et al. (1999) who crystallised ammonium bicarbonate in batch and observed a decline in crystal population density with particle size which is characteristic of conventional crystalliser configurations where homogeneous primary and secondary nucleation is favoured. The increase in population density observed in the present study, in parallel with crystal growth, is indicative of sustained primary heterogeneous nucleation, promoted by the membrane substrate through a reduction in the free energy barrier.

In the present work, it is suggested that the increasing nucleation rate and growth rate observed at increasing levels of supersaturation is due to: (i) the low driving force of the chemical reaction, which limited consumption of NH3 in the initial phase of crystallisation; and (ii) the counter-current diffusion of CO2 and NH3 into the concentration boundary layer, which provided consistent replenishment of reactant to the site of preferential nucleation. This latter contribution is a unique facet of this chemically reactive membrane crystallisation reactor configuration.

For the 4.6M NH3 solution, encrustation (or scaling) of the fibre occurred too quickly for the nucleation rate to be determined. Such crystallisation induced scaling may arise from Ostwald ripening in which an agglomeration of fine particles occurs immediately after nuclei breeding, followed by the supported growth of coarser crystals through supply of small crystal particles that become more stable on the membrane surface. As the CO2 flux profiles for both 3.3 and 4.6 M NH3 solutions were similar, it is the free ammonia concentration and not CO2 transport which determines the nucleation rate. It is suggested for this study that the higher NH3 concentration increased nucleation rate which induced an analogous effect to Ostwald ripening. The resultant effect was a change in surface contact angle which subsequently induced welling and a breakthrough of solution NH3 into the gas phase which temporarily increased flux. In contrast, whilst several discrete crystals were observed to have formed on the membrane fibre at 3.3M, most crystals were collected downstream of the membrane, which was confirmed by nitrogen mass balance. When coupled with the consistent flux profiles sustained over the duration of experiments, this would suggest that provided nucleation rate can be specific below a critical threshold, sustained membrane operation is achievable with minimum reactive maintenance intervention.

The highest CO2 flux recorded was similar to that of McLeod et al. (2015) who used a wider pore size PTFE membrane for CO2-NH3 absorption. For initial NH3 concentrations equal to or above 1.9M, CO2 flux was apparently independent of NH3 concentration which implies the liquid phase imparted negligible resistance to mass transfer, as is commonly reported for chemical absorption where the reactant is in excess. However, a two-stage decline in CO2 flux was observed following progressive CO2 absorption.

In the initial phase of CO2 absorption, carbamic acid forms alongside ammonium carbonate. The carbamic acid deprotonated to form carbamate, whilst a fraction of the CO2 reacted with hydroxide to form bicarbonate (HCO3), the cumulative effect being a reduction in solution pH. The lower pH shifted the equilibrium toward NH4* which is favourable for crystallisation but reduced the rate of reaction with CO2. The onset of the second slower phase of CO2 flux decline occurred at a pH of between 8.2 and 8.5 for each NH3 concentration studied which corresponded to a shift in equilibrium from NH3 to NHat prior to supersaturation. It is this reduction in reactivity which constrained nucleation rate. The rate of pH decline was dependent upon CO2 loading (CO2/NH3) where a relative excess of CO2 drives a faster decline in pH as was observed in this study with a lower NH3 absorbent concentration. The 4.6M NH3 solution therefore sustained a higher pH at supersaturation which subsequently favoured HCO3-formation at induction. We suggest that it is the increased availability of HCO3-provided by the buffering capacity of the 4.6M NH3 solution which increased nucleation rate and induced surface scaling.

Evaluation of the thermodynamics of the crystalline reaction product show that the intermediate solid products, ammonium carbonate monohydrate ([NH4]2CO3H20), sesquicarbonate ([NH4]2C032NH41-1CO3) and ammonium carbamate (NH2COONH4) could form before the final reaction product (ammonium bicarbonate) dependent upon solution temperature and CO2 loading (the molar ratio of CO2/NH4. At an absorbent temperature of 5°C, a CO2/NH3 loading exceeding 0.53 has been shown to be required to induce crystallisation of ammonium bicarbonate. In this study, this CO2 loading was exceeded at each ammonia concentration, however, crystallisation did not occur for NH3 concentrations below 3.3M. For 1M absorbent, the solid phase transition is thermodynamically limited since the free ammonia concentration is below the solubility limit for NH4HCO3 of 1.5 M. At 1.9M, NH3 is not thermodynamically limiting. Instead the solid phase transition is kinetically limited by the pH transient which forces the equilibrium toward carbonic acid (H2CO3) such that upon reaching supersaturation (C/C" -1), there is an insufficiency of bicarbonate to initiate heterogeneous nucleation. For NH3 concentrations of 3.3 and 4.6M, the excess NH3 reduced the CO2/NH3 ratio and sustained pH through to supersaturation such that sufficient HCO3-was available to induce crystal growth at CO2 loadings of 0.77 and 0.44 respectively. Confirmation of crystalline ammonium bicarbonate was provided by XRD analysis.

As demonstrated in this work, an ammonia concentration of 3.3M permits maximum attainable flux to be achieved which will limit membrane area, whilst assuring continuity in the crystallisation of the ammonium bicarbonate product of reaction.

Accordingly, it has been demonstrated that a minimum free ammonia concentration is demanded to sustain pH to ensure there is sufficient bicarbonate in the reaction zone at supersaturation to facilitate heterogeneous nucleation. Free ammonia concentration greater than 1.9M did not improve CO2 flux, which is similar to observations in CO2-NH3 packed column investigations. Selection of the upper threshold for ammonia should therefore be based on limiting nucleation rate to avoid crystallisation induced scaling of the membrane. The observed increase in crystal population density and crystal size is indicative of the membrane acting as a physical substrate for heterogeneous nucleation during chemically reactive crystallisation. Both nucleation rate and crystal growth rate increased with progressing levels of supersaturation, which can be ascribed to the relatively low chemical reactivity and the unique counter diffusion mechanism fostered by this type of membrane crystallisation technology which provides continued reactant replenishment within the boundary layer where heterogeneous nucleation is promoted by the membrane through reduction in the free energy barrier. Provided an appropriate nucleation rate was specified, only limited crystals were observed to form at the membrane and were instead collected downstream of the membrane.

Further study of practical multi-fibre HMFC and operation of a two stage HMFC system for gas treatment Following on from the work reported above, we now disclose a study assessing the merit of employing a multi-fibre HFMC as reactor for implementing ammonium bicarbonate recovery within a chilled ammonia process whilst maintaining an industrially relevant gas purity (>98% CH4), relative to existing technology (absorption columns) applied to the same process. Specific objectives include (i) comparing the performance of a commercially available HFMC and bespoke absorption column with identical interfacial area during physical and chemically reactive (3.3 mol L_, NH3) absorption (ii) investigate crystal growth within a CO2-NH3-H20 chemically reactive system for both absorber technologies (iii) demonstrate an applied biogas upgrading-CR-MCr process for the co-production of NH4HCO3 and L98% CH4 from a synthetic biogas applying an ammonia absorbent within a multi-fibre HFMC.

Materials and methods Experimental set-up: The performance of dispersive and non-dispersive gas-liquid contacting technologies was undertaken utilising two commercial HFMC designs and a bespoke low-pressure absorption column. HFMC from a commercial module range, a small-scale parallel flow module and larger-scale transverse flow module (3M Industrial Group, Charlotte, USA) were employed, both fibre bundles comprised polypropylene X50 fibres with an outer diameter and inner diameter of 300 pm and 240 pm respectively. Each parallel flow module comprised a tube-in-shell design with poor shell-side liquid phase distribution, each fibre bundle of 0.13m length compromised 7400 fibres cased in a polycarbonate shell yielding an interfacial area of 0.5 m2, whereas each transverse flow module employed shell-side transverse flow around a central baffle to overcome maldistribution and comprised a 0.16 m fibre bundle of 10200 fibres cased in a polypropylene shell yielding an interfacial area of 1.2 m2. A 0.3 m acrylic absorption column of 0.05 m diameter was filled with 0.005 m diameter borosilicate glass raschig rings with a specific surface area of 984 m2m3 (Hilgenberg GmbH, Malsfeld, Germany) to yield an interfacial area of 0.5m2. Absorption solvent was stored in an 85 L PVC tank and applied in counter current mode using a peristaltic pump (up 2000mL min-1, 5308, Watson-Marlow Ltd, Falmouth, UK), 15.0 Mf2 3m-1 de-ionised water was applied as physical absorbent and a 3.3 mol L-1 NH3 solution applied as chemically reactive absorbent, ammonia was prepared through the addition of NH3 concentrate (35% Fisher Chemicals, Loughborough, UK) to de-ionised water (15.0 MO cm-1). The absorbent was selectively directed to the lumen-side and shell-side of the parallel flow module but only to the shell-side of the transverse flow modules. Absorbent was applied from the top of the absorption column, and flowed counter-current to the gas phase. All experimentation was undertaken in a temperature-controlled environment (20°C). Synthetic biogas, comprising a 0.5 BarG, 50/50 CH4/CO2 mixture, was prepared in-line by mixing carbon dioxide (002, 99.7%) and methane (CH4, 99.995%) (BOO gases, Ipswich, UK) using mass flow controllers (0.01 -5.0 L min-1, Roxspur Measurement and Control Ltd., Sheffield, UK) to provide combined flow rates between 0.05 and 4 L min* The effect of absorption technology type; non-dispersive HFMC and dispersive absorption column on NH4HCO3 crystallisation under conditions of supersaturation (C/C*), defined as the ratio between the CO2 absorbed in solution and the CO2 required to form ammonium bicarbonate at the solubility limit, was determined in a 0.6L, 3.3 mol L-1 NH3 reaction solution (5°C) with an initial pH 13. QG,in and absorbent flow rate (Qin) were fixed at 2 L min-1 respectively during HFMC operation, with QLm operated as recycling batch flow, whilst QG,In and QL,,, were fixed at 2 L min-1 and 0.7 L min-1 respectively during column operation due to column hold up at higher Chm. A transverse flow contactor was utilised due to the resistance of the polypropylene casing to 3.3 mol L-1 NH3, with the absorbent applied shell-side counter current to the gas flow. Absorbent pH was monitored using a Jenway epoxy bodied pH electrode (Jenway 4330, Cole-Farmer, Stone, UK) whilst bicarbonate concentration in solution was determined by UV absorbance at 215 nm (Jenway 6715, Cole-Farmer, Stone, UK), the absorption region for bicarbonate. Absorbent temperature was maintained at 5°C with a circulating chiller (R1 series, Grant Instruments Ltd., Cambridge, UK) and gas phase inlet temperature maintained at 20°C, gas and liquid phase temperatures were measured during operation (K-type thermocouples, Thermosense Ltd., Bucks, UK). Mettler-Toledo Focused Beam Reflectance Measurement (FBRM®) G 400 and ParticleView (V19®) with PVM (Particle Vision and Measurement) probes provide in-situ nucleation detection. FBRM detection of nucleation represents an approximation as crystal nuclei must reach a detectable size and number. The FBRM employs a laser beam, rotated with high speed optics set at 2 m s-1, to scan a circular path in the flow, with particles measured via diffuse backscattered light at a rate of five measurements per second, providing a robust particle chord length distribution (OLD) measurements at a measurement frequency of 15 s. Three particle detection settings, detection, which gives maximum sensitivity to particle count, fine (Primary), which yields an enhanced sensitivity to fine particle, and coarse (Macro), which gives a lower sensitivity to fine particles were employed, with stuck particle correction employed to mitigate background noise due to sensitivity to impeller agitation, yielding a background count of from 0-0.05 counts 5-1. The OLD was determined with iC FBRM) software using a window size of 10 sample smoothing correspond to a 50 bins number for a range of 1-100 microns with logarithmic regulation spacing applied. The ParticleView V19 immersion probe-based video microscope utilises reflected light illumination to collect high-resolution (2 pm) images of liquid suspensions under harsh conditions with respect to temperature and pressure with a field of view 1300 pm x 890 pm (± 50 pm).

Analysis: Gas flow rate was measured by means of a manual bubble flow meter (up to 300 mL min-1, 50 mL, error ±2%, Restek, Bellefonte, USA; from 300-3000 mL min-1, 1000 mL, Model 311, error ±2%, Blandford Forum, UK) and gas composition determined using an in-line infrared CO2 analyser (BCP-CO2, accuracy <0.5% full-scale, Bluesens gas sensor GmbH, Herten, Germany). A gas phase mass-balance was completed according to: [(QG XCG in) WG,outXCG,out)11 (Equation Cl) where q is the dimensionless CO2 capture ratio, QG,in and Qaout are inlet and outlet gas flow rates respectively (m3 s-1), and Cam and Came are inlet and outlet gas phase concentrations respectively (mol m-3). Carbon dioxide flux (J002, mol m-2 s-') was calculated using: [(QG,inxcG.in)-(Qacutxcc,,,uol.273.1sx moo 1CO2 (Equation C2) (22.4xAmxTG) Where Am refers to available membrane surface area for absorption (m2) and TG is gas temperature (K).

Chemical enhancement of JCO2 was characterised using the 'enhancement factor' (E): 1c02(chem (cal) (Equation C3) lc02(pnysical) The CO2 absorption rate and the enhancement factor are dependent on reaction and diffusion rates in the liquid film, often expressed as the Hatta number, the ratio of maximum reactive conversion rate to the maximum diffusion mass transfer rate is defined as: Ha =,okNH3DC0LCNH3 (Equation C4) kL,ett where k2.IVH3 (M3 krnal S-1) is the second-order reaction rate constant as taken from from Liu et al., and CNH3 is the NI-13 concentration (kmol m-3). kL,e,d(m 5-1) is the liquid phase mass transfer coefficient, which during liquid phase control, a non-wetted membrane and the presence of 100% CO2 is equivalent to the overall mass transfer coefficient, Key (m s-1) [35], and experimentally measured as: i = 1-exp Koval:.) \ (Equation C5) where VG, L and a are the superficial gas velocity (m s-1), fibre length (m) and interfacial area (m2 m-3) respectively.

Crystal nucleation kinetics (Equation C7) and growth (Equation C6) in a supersaturated solution for both dispersive and non-dispersive contacting was undertaken according to the methodology proposed by Chen et al., 2014 [38]: G = K G (Lim -CT) (Equation C6) B = KBGb (Equation C7) Where KG is he kinetic rate constant for crystal growth (kg-g m3r1s-1), KB the nucleation rate constant (No m-3-b s-l+b), and g and b exponent constants. The parameters cm, and c* (kg m-3) represent the CO2 (aq) concentrations adjacent to the membrane surface and the equilibrium-saturated concentration at the membrane wall temperature Tim. Defining the crystal growth rate as the size variation from L (m) at time t to L+AL at t+At, allows the average crystal growth rate G, over Lit (s), to be calculated through the linear equation: G(L, = AL/At (Equation C8) When experimentally determining G at different experimental time intervals and converting Equation 8 into its logarithmic form: log G = log KG + g log(cf," -c*) (Equation C9) The crystal growth rate constant KG and the exponent g can be derived from the intercept and slope of the linear regression respectively.

Results Chemical reactivity is comparable for CO2 separation with HFMC and column technology: Absorber performance was evaluated applying water as a physical absorbent in counter current operation within a column and HFMC, of 0.5m2 interfacial area respectively. During HFMC operation, gas flow was passed to the module shell-side. The influence of contacting mechanism on the CO2 capture 0(3002) attained under variable liquid-to-gas ratios (US) was assessed at differing gas flow rates by initially fixing QG and subsequently varying liquid flow rate for both absorber types. A comparable impact of the L/G, as indicator of absorbed CO2 load, is observed as a linearised logarithmic relationship with qco2,with both absorbers reaching a target 0.98(3002.,002 as demanded as an industrial biomethane standard at a LJG of 8.2.

HFMC absorber performance, /7002, was evaluated applying 3.3 mol L-1 NH3 as absorbent with gas flow applied in counter current operation to the shell-side. A linearised logarithmic relationship between the L/G and /7002 is only partially observed for HFMC operation from an L/G of 0.1 upwards, in contrast to physical absorption which displays a linear relationship across the entire operational envelope. The likely cause of deviation from a linear relationship is the increased relative contribution of membrane resistance to the overall resistance to mass transfer, as consequence of increased membrane wetting over time. As comparison, reactive absorption during column contacting maintains the linear relationship down to an L/G of 0.03, the lower boundary condition as dictated by column entrainment at higher gas flow, due to obviation of membrane resistance. However, the impact of membrane welling is limited at the UG ratio (0.2 -0.3) required for a 0.98 n002, with both absorbers displaying comparable performance. Chemically reactive absorption is responsible for reducing the operational L/G required to reach a 0.98 /7002 from an LIG of 8.2 for physical absorption to an LJG of 0.3 under enhancement, increasing the biogas treatment capacity within the example system from a Qan of 0.08L min-1 for physical absorption to a Qan of 1L min-1 for reactive absorption, a 12.5 fold increase in treatment capacity.

Enhancement of mass transfer, Equations C2 and C3, was measured and the influence of contacting mechanism on the CO2 capture ((3002) under variable L/G was assessed at differing absorbent flow rates by initially fixing QL and subsequently varying QG for both absorber types. The influence of absorbent flow rate on the Hatta number as calculated from the overall mass transfer coefficients (Kay) for physical absorption, indicates parity between the two absorbers. Introduction of an increasing gas-phase flow rate and as consequence an increasing mean average gas-phase CO2 partial pressure, results in increasing parity between the Hatta number and experimentally derived enhancement to mass transfer. The relationship is borne out across both absorber types with parity between Ha and E achieved at a QGin of 4 L min-1, corresponding to a ri002 of 0.79 to 0.9 and likely a result of the Hatta number approximation to mass transfer including an assumption of a CO2 partial pressure of 1. However, wetting during HFMC operation limits gas flow rate specific parity between the two technologies to flow rates at which membrane welling has yet to occur, Qc,,,= 0.5, 1, and 1.5 L min-1 above which disparity is noted, with increasing Ckir, not corresponding to an increased E, indicating rate limiting phase resistance to no longer lie in the gas-phase.

HFMC at scale apply liquid shell-side to take advantage of lower pressure drop. However, the small-scale commercial 0.5m2 HFMC modules employed in this study were constructed with a polycarbonate (PC) shell, susceptible to material degradation in the presence of aqueous ammonia, limiting operation to an hour before catastrophic casing degradation. Scaled module designs, 1.2m2 and above utilise an ammonia resistant polypropylene casing, as such were employed for long term operation. HFMC absorber performance, 0.5m2 lumen-side absorbent, 0.5m2 shell-side absorbent and 1.2m2 shell-side absorbent, were compared during wetted operation, with performance of 1.2m2 shell-side NH3> 0.5m2 lumen-side NF-I3> 0.5m2 shell-side, congruent with literature assertions of enhanced performance at scale.

Lower nucleation rates in HFMC sustain CO2 separation during crystal growth: Crystal growth kinetics were measured during batch operation of both absorber types applying a recycling chilled (5°C) 3.3 mol L-1 NH3 solution to limit ammonia volatility and breakthrough into the gas phase. FBRM derived chord length distributions at increasing supersaturation levels for column contacting and HFMC contacting were evaluated utilising the primary setting to detect small particulates below 10nm, corresponding to crystal nucleation and macro for larger particle growth and agglomeration. A primary particle count is detected at a C/C* of 1.05 for column operation, raising to a stable maximum after a C/C* of 1.43, indicating crystal nucleation within the bulk solution. Further crystal growth is seen with a macro CLD centred at about 65pm appearing at a C/C* of 1.43 and increasing in particle count to a maximum at a C/C" of 1.44, indicating further crystal growth in solution. In contrast, during HFMC operation no discernible increase in the primary CLD particle count was evident, indicative of crystal nucleation outside of the batch reactor volume housing the FBRM, this corroborates literature assertions that primary nucleation is initiated within membrane module, at the micropore gas-liquid interface. Crystal growth is then intimated at higher C/C*by the increased particle count within the macro CLD, reaching a maximum count at an C/C" of 1.2, indicating secondary crystal growth within the batch reactor volume, corroborating the explanations set out above regarding the decoupling of primary nucleation at the membrane surface and crystal growth in solution. However, the macro CLD during HFMC contacting shifts downwards, centring at about 53pm, lower than the about 65pm observed during column contacting. Additionally, the maximum observed particle size reduces from a maximum of 299 pm during column contacting to a maximum of 130 pm during HFMC contacting. This decrease in maximum particle size is attributed to the agglomeration of crystals in solution during dispersive contacting and was confirmed using PVM and a preference for single crystal growth was confirmed during non-dispersive contacting. The average crystal growth rate (G) was determined at each supersaturation level for dispersive and non-dispersive contacting, which enabled derivation of the crystal growth rate constant KG, 2.8x1 026 and 5.7x1021 Kg-g ITI3g+1 5-1 and exponent g, of -45.3 and -44.1 respectively.

In-series HFMC can achieve simultaneous gas separation and crystallisation: The CO2 loading ratio and supersaturation ratio for both absorber types was determined through an integrated gas phase CO2 mass balance over reaction time, and bicarbonate concentration (UV215) measured at each supersaturation point, during batch operation applying a recycling chilled (5°C) 3.3 mol L-1 NH3 solution. As a function of the supersaturation ratio (C/C1 and CO2 loading ratio (CO2/NH3), the point of crystallisation measured through macro detection is C/C" = 1.18, CO2/NH3 = 0.61 and 0/0" = 1.2, CO2/NH3= 0.64 for HFMC and column contacting respectively, coincident with a maximum absorbance reading of 2.5 to 2.75. However, the non-dispersive mechanism of contacting within the gas-liquid column resulted in the agglomeration of crystalline reaction product and detrimental precipitation with the absorber packing material leading to column flooding. The detection of primary nucleation within the reaction volume at a C/C" of 1.05 corresponds with the initiation of column flooding at a C/C" of 1.06, complete column blocking occurs at a 0/0" of 1.27 whereas particle growth is not detected until a C/C" of 1.43.

To demonstrate the co-production of high purity CH4 and NH4HCO3within an HFMC cascade, two chilled ammonia batches (5°C) were applied to the shell-side of two separate 1.2m2 HFMC, creating two independent ammonia reservoirs. The inlet gas flow was passed sequentially from the first stage (crystallising) HFMC to a secondary stage (gas purity) HFMC with an integrated gas-phase CO2 mass balance completed over reaction time, and bicarbonate concentration (UV215) measured at each supersaturation point for the first stage and for the overall system (cascade). The results are shown in Fig. 5. As reported above, the UV215 absorbance was used as a surrogate for assessing crystal nucleation at supersaturation. The concurrence of 98% CH4 and a maximum absorbance reading at a C/C* of 1 indicative of the co-production of high purity CH4 and precipitation of NH4HCO3 is indicated in Figure 5. However, at the observed point of crystallisation, C/C" = 1.2 the outlet gas purity was measured at 95% CI-14.

Discussion A key finding of this study is that absorption columns as current generation gas-liquid contacting technology suffer process blocking at the point of NH4HCO3 crystallisation whereas multi-fibre HFMC are capable of maintaining continued crystallisation without suffering process blocking. Employing a two- module cascade design, a first stage crystallising module and second stage gas purity module can co-produce NH4HCO3 and a high purity gas product from biogas. The impact of a 3.3 mol L-1 NH3 reactive absorbent on mass transfer enhancement is comparable between absorption columns and HFMC, however the development of a concentration gradient and explicitly a clearly defined counter-current concentration gradient at a well-defined membrane surface area encourages crystal nucleation over crystal growth. In comparison the concurrence of crystal nucleation and crystal growth within the column packing media leads to rapid process blocking once supersaturation is reached.

Absorber performance parity is demonstrated across the operational spectrum of both absorber types applying the 002-H20 system, evidenced from the lower to upper boundary conditions, as limited by gas-phase bubbling into the liquid-phase (HFMC) and entrainment (column) at the lower boundary and maximum mass transfer (HFMC) and flooding at the upper boundary. Parity indicates the impact of an additional membrane resistance as negligible across investigated L/G set points. The minimal influence of membrane resistance is attributed to limited wetting, demonstrated in the literature to increase mass transfer relative to wetted operation. Wetting is evident within experimental data as the plateau of qco2 at a fixed VG, with an increased LJG representing an increased QL, however subsequent membrane drying between runs and VG increase eliminates wetting and restores parity between absorbers. As consequence of performance parity between the two absorber types, comparable Kay values (Equation C5) are expected at identical flow conditions. Previous studies have measured membrane resistance up to 85% of overall resistance to mass transfer and so attributed intensification factors of between 0.8 and 15 times between columns and HFMC to the detrimental impact of membrane resistance relative to advantageous increased interfacial area. However, this study shows that during non-wetted operation any intensification can be attributed to increased interfacial area alone.

Transition to a CO2-NH3-H20 chemically reactive system yields comparable performance at L/G above 0.1 for both absorber types. The Haifa number (Ha) as the ratio of maximum reactive conversion rate to the maximum diffusion mass transfer rate, and enhancement factor (E) are expected to display parity at Ha >3. However, in applying a binary gas mixture a gas-phase limitation to mass transfer is expected from gas diffusion limits at low CO2 partial pressure present during high CO2 capture. Disparity between Ha and experimental Eat reducing QG is therefore attributed to the observed high (>0.98) t/c02, and reduction in the mean average CO2 partial pressure along module length. Comparable E values at Ha in both absorber types indicates the system to be influenced by absorbent choice and not the dispersive or non-dispersive contacting mechanism. This conclusion is evidenced in the literature by the wide spread use of Hatta number estimations of mass transfer enhancement in both absorption columns and HFMC modelling studies. As consequence the 8 fold intensification factors reported for a chemically enhanced absorption column can be expected of HFMC under the same operational flow rates. However, some incongruences are visible at higher QG. Relative disparity between the two contacting mechanisms at higher QG is attributed to membrane wetting as evidenced by the disparity at lower L/G ratios, with HFMC performance decreasing relative to absorption column. To prevent bubbling of the gas-phase into the absorbent, the absorbent must remain at a higher pressure than the gas-phase however low enough to prevent pore wetting. As consequence of pressure drop the pressure differential varies with module length and due to counter-current operation, gas and liquid pressure loss is greatest at opposing ends.

Lumen-side pressure drop as experienced by the liquid-phase is greater than shell-side pressure drop as experienced by the gas-phase. The pressure differential at the point of gas entry therefore favours bubbling, subsequent increase in liquid-phase pressure to compensate results in overpressure at absorbent entry and leads pore wetting.

The fibre bundle of the parallel module was housed within a polycarbonate casing, severely affected by ammonia solutions, as such catastrophic module failure i.e. infiltration of the liquid-phase to the shell side through case degradation may not be prevented. The impact was limited through frequent end cap replacement and flushing the module with deionised water to remove residual free ammonia between runs. However, for longer term operation, transition to a transverse flow module with an ammonia resistant polypropylene case was required. Prior to the integration of the CO2-NH3-H20 system into a recycling batch crystallisation reaction, the transition from a tube-in-shell parallel module to a transverse flow HFMC was made, in particular liquid lumen-side to liquid shell-side. Relative performance of 1.2m2 shell-side NH3> 0.5m2 lumen-side NH3> 0.5m2 shell-side is in-line with previous literature assertions of enhanced mass transfer at scale, particularly in the presence of enhanced liquid phase distribution during transverse flow.

Integration of the CO2-NH3-H20 system within HFMC as CR-MCr for the precipitation of NH4HCO3 from ammonia solution and pure CO2 was previously demonstrated by Mcleod et al. (2015), which focused on the nucleation kinetics at the local gas-liquid interfacial contact zone of the shell-side micropore entrance and within a stirred dispersive batch reaction volume. They identified a decoupled nucleation-growth system for HFMC operation where nucleation is initiated at the micropore surface, as consequence crystal nucleation is favoured over growth which instead takes place downstream, in contrast to the dispersive volume where nucleation and growth occurs within the same system. As a consequence, HFMC operation can be expected to display a particle size distribution of smaller size than the dispersive absorption column. However this study does not evidence an elevated sub 10 pm CLD particle count expected of newly nucleated crystals around the C/C*= 1 expected to initiate crystallisation. A subsequent increase in particle count within the 10-100 pm range at an C/C" of 1.2 and above is however evidenced, in contrast to dispersive column contacting where a clear sub 10 pm CLD is evident from an C/C" of 1.05. This intimates that the increased interfacial area offered by a fibre bundle relative to a single fibre and development of a concentration boundary layer may slow reaction development.

However, the lack of quantifiable sub 10 pm particulates within the bulk solution may be attributed to the suspected mechanism of crystal nucleation and growth. When the membrane material itself acts as a substrate for nucleation, the local conditions of supersaturation at the micropore gas-liquid interface induce nucleation, limited growth occurs until enough shear is induced to initiate crystal detachment, freeing the pore as a future nucleation point and encouraging further nucleation. Further growth then occurs downstream in the bulk solution, evidenced in this study by the subsequent quantifiable CLD at elevated C/C* and particle size within the bulk reaction volume. The FBRM detection method requires a minimal crystal size and population density to be met before detection can occur, as sub 10 nm crystals are likely to be initially retained at the micropore entrance, reaction initiation may appear delayed. The work reported above suggests a 120-300 pm minimum crystal size for HFMC but a 60 pm minimum for a mixed reactor, interestingly the particle size corresponds with the expected boundary layer film thickness of about 139 pm whereupon the crystal would be subjected to higher liquid phase velocity and hence shear stress present in the bulk solution. As a consequence of decoupled growth, the reaction product is formed as single crystals in solution and further intimated by the reduced upper crystal size within the post 10 pm CLD relative to absorption columns, enabling crystal harvesting to occur before agglomeration as observed in absorption column operation and limit process blocking. In contrast, the dispersive absorption column evidences both sub 10 pm and post 10 pm CLD particle count increase after the C/C" 1.05 reaction initiation point, indicating concurrent nucleation and growth within the system. As a consequence of the dispersive contacting mechanism within a closely packed contact media, precipitate is readily captured within the packing media and rapidly contributes to column blocking and flooding.

Interestingly, column blocking begins at C/C" = 1.05 and correlates to a sub 10 pm CLD, indicating that column blockage precedes crystal growth. The subsequent two-fold increased crystal growth rate evidenced in the absorption column relative to HFMC accounts for the rapid process blockage by an C/C* of 1.27, close to the C/C" of 1.2 at which crystallisation is detected within the HFMC system.

At 5°C, a 3.3 mol L-1 NH3 absorbent at pH 13 has a ammonium bicarbonate solubility limit of 1.7 mol kg-1, enabling a maximum reaction yield of 48.5%. The obtained reaction yield of 41.8% during non-dispersive contacting represents an 87% yield relative to the solubility limit of ammonium bicarbonate. The increase in yield is attributed to both increased pH and subsequent elimination of bicarbonate transition into carbonic acid and the increased interfacial area of a multi-fibre bundle and hence nucleation rate relative to a single fibre approach.

Ammonia slip, reaction yield reduction and process blocking therefore limit the potential for current generation absorption columns at wastewater treatment works (WVVTVV) in operating as crystallising reactors.

To take advantage of greater reaction yield and lack of process blocking available through HFMC, a design for a biogas upgrading-CR-MCr process based on a crystallising-purity membrane cascade for in-series gas-phase contacting is demonstrated as illustrated in Fig. 6. The first stage (crystallising membrane) exploits the elevated CO2 partial pressure present in untreated biogas to maintain higher rates of mass transfer whilst an independent recycling absorbent reservoir (0.6L) ensures the 0.6 CO2/NH3 loading ratio required for precipitation can be quickly met. The second stage (gas purity membrane) employs a second larger recycling absorbent reservoir (2L) to ensure a low CO2/NH3 loading ratio to maintain a concentration gradient against the reduced CO2 partial pressure in the partially treat biogas (Fig. 5) without losing absorbed CO2. Once crystallisation is initiated, careful control of the crystallising reaction volume at a 0.6 CO2/NH3 loading ratio through in-line UV2isabsorbance and continued feed of partially saturated absorbent from the second module facilitates process control.

As demonstrated in this study, the initial crystal formation expected within the sub 10 pm range is not observed in HFMC, shifting the observed point of nucleation rightward from an C/C" of 1.05 to 1.2, however as nucleation is observed at the expected C/C* during dispersive contacting and is demonstrated in this study to correspond with a maximum UV2Thabsotance, membrane crystallisation is taken to occur at an identical C/C* but the reaction product to be retained at the point of nucleation until sufficient limited growth occurs to induce shear. Successful crystallisation within the CR-MCr membrane indicated by the concurrence of C/C* of 1 and successful biogas upgrading (a 0.98/7002) by the outlet CH4% of 98% represents an integrated biogas upgrading-CR-MCr process for the co-production of NH4HCO3 and n8% CH4. This process has demonstrated chemical reduction of the nitrogen load of the solution through an 87% reaction yield, upgrading of biogas and process intensification through chemical enhancement, representing a viable process for cost reduction and increase in resource recovery when applied at a VWVTW. **X

The features disclosed in the foregoing description, or in the following claims, or in the accompanying drawings, expressed in their specific forms or in terms of a means for performing the disclosed function, or a method or process for obtaining the disclosed results, as appropriate, may, separately, or in any combination of such features, be utilised for realising the invention in diverse forms thereof While the invention has been described in conjunction with the exemplary embodiments described above, many equivalent modifications and variations will be apparent to those skilled in the art when given this disclosure. Accordingly, the exemplary embodiments of the invention set forth above are considered to be illustrative and not limiting. Various changes to the described embodiments may be made without departing from the spirit and scope of the invention.

For the avoidance of any doubt, any theoretical explanations provided herein are provided for the purposes of improving the understanding of a reader. The inventors do not wish to be bound by any of these theoretical explanations Any section headings used herein are for organizational purposes only and are not to be construed as limiting the subject matter described.

Throughout this specification, including the claims which follow, unless the context requires otherwise, the word "comprise" and "include", and variations such as "comprises", "comprising", and "including" will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps.

It must be noted that, as used in the specification and the appended claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from "about" one particular value, and/or to "about" another particular value. When such a range is expressed, another embodiment includes from the one particular value and/or to the other particular value. Similarly, when values are expressed as approximations, by the use of the antecedent "about," it will be understood that the particular value forms another embodiment. The term "about" in relation to a numerical value is optional and means for example +/-10%.

References A number of publications are cited above in order to more fully describe and disclose the invention and the state of the art to which the invention pertains. Full citations for these references are provided below. The entirety of each of these references is incorporated herein.

Persson, M., Jonsson, 0., Wellinger, A., 2007. Task 37-Biogas Upgrading to Vehicle Fuel Standards and Grid Injection. EA Bioenergy. 1,8-9.

Kozak, F., Petig, A., Morris, E., Rhudy, R., Thimsen, D., 2009. Chilled ammonia process for CO2 capture. Int. J. Greenh. Gas Con. 1, 1419-1426.

McLeod, A., Jefferson, B., McAdam, E., 2014. Biogas upgrading by chemical absorption using ammonia rich absorbents derived from wastewater. Water Res. 67, 175-186.

McLeod, A., Buzatu, P., Autin, 0., Jefferson, B., McAdam, E., 2015. Controlling shell-side crystal nucleation in a gas-liquid membrane contactor for simultaneous ammonium bicarbonate recovery and biogas upgrading. J. Membrane Sci. 473, 146-156.

Budzianowski, W.M., 2011a. Benefits of biogas upgrading to biomethane by high-pressure reactive solvent scrubbing. Biofuels Bioprod. Bioref. 6, 12-20.

Belaissaoui, B., Favre, E., 2018. Novel dense skin hollow fiber membrane contactor based process for CO2 removal from raw biogas using water as absorbent. Sep. Purif. Technol. 193, 112-126.

He, Q., Yu, G., Tu, T., Yan, S., Zhang, Y., Zhao, S., 2017. Closing CO2 loop in biogas production: recycling ammonia as fertilizer. Environ. Sci. Technol. 51, 8841-8850.

Shuangchen, M., Huihui, S., Bin, Z., Gongda, C., 2013. Experimental study on additives inhibiting ammonia escape in carbon capture process using ammonia method. Chem. Eng. Res. Des. 91, 27752781.

Budzianowski, W.M., 2011b. Mitigating NH3 vaporization from an aqueous ammonia process for CO2 capture. Int. J. Chem. React. Eng. 9, 58-60.

Oxera Consulting Ltd, 2006. What is the cost of reducing ammonia, nitrates and BOD in sewage treatment works effluent? Report prepared for Ofwat. (Accessed 13th March, 2018) Sutter, D., Gazzani, M., Mazzotti, M., 2015. Formation of solids in ammonia-based CO2 capture processes -Identification of criticalities through thermodynamic analysis of the CO2-NH3-H20 system. Chem. Eng. Sci. 133, 170-180.

Ukwuani, AT., Tao, W., 2016. Developing a vacuum thermal stripping-acid absorption process for ammonia recovery from anaerobic digester effluent. Water Res. 106, 108-115.

Gazzani, M., Sutter, D., Mazzotfi, M., 2014. Improving the efficiency of a chilled ammonia CO2 capture plant through solid formation: a thermodynamic analysis. Energy Proced. 63, 1084-1090.

Yu, H., Qi, G., Wang, S., Morgan, S., Allport A., Cottrell, A., Do, T., McGregor, J., Wardhaugh, L., Feron, P., 2012. Results from trialling aqueous ammonia-based post-combustion capture in a pilot plant at Munmorah Power Station: Gas purity and solid precipitation in the stripper. Int. J. Greenh. Gas Con. 10, 15-25.

Makhloufi, C., Lasseuguette, E., Remigy, JO., Belaissaoui, B., Roizard, D., Favre, E., 2014. Ammonia based CO2 capture process using hollow fiber membrane contactors. J. Membrane Sci. 455, 236-246.

Cui, Z., deMontigny, D., 2017. Experimental study of carbon dioxide absorption into aqueous ammonia with a hollow fiber membrane contactor. J. Membrane Sci. 540, 297-306.

Villeneuve, K., Roizard, D., Remigy, J.C., Iacono, M., Rode, S., 2018. CO2 capture by aqueous ammonia with hollow fiber membrane contactors: Gas phase reactions and performance stability. Sep. and Purif. Technol. 199, 189-197.

Teichgraber, B., Stein, A., 1994. Nitrogen elimination from sludge treatment reject water-comparison of the steam-stripping and denitrificafion processes. Wat. Sci. Technol. 30, 41-51.

Bougie, F., lliuta, M.C., 2013. Analysis of Laplace-Young equation parameters and their influence on efficient CO2 capture in membrane contactors. Sep. Pudf. Technol. 118, 806-815.

Yeh, J.T., Resnik, K.P., Rygle, K., Pennline, H.W., 2005. Semi-batch absorption and regeneration studies for CO2 capture by aqueous ammonia. Fuel Process. Technol. 86, 1533-1546.

Atchariyawut, S., Jiraratananon, R., Wang, R., 2007. Separation of CO2 from CH4 by using gas-liquid membrane contacting process. J. Membrane Sci. 304, 163-172.

Wang, X., Conway, W., Fernandes, D., Lawrance, G., Burns, R., Puxty, G., Maeder, M., 2011. Kinetics of the reversible reaction of CO2(aq) with ammonia in aqueous solution. J. Phys. Chem. 115, 6405-6412.

Di Profio, G., Curcio, E., Drioli, E., 2010. Supersaturation control and heterogeneous nucleation in membrane crystallizers: facts and perspectives. Ind. Eng. Chem. Res. 49, 11878-11889.

Veiga, A.R., Calmanovici, CE., Giulietti, M., 1999. Operational conditions evaluation in ammonium bicarbonate crystallisation. Proceedings of the 14th International Symposium on Industrial Crystallization. 65,1-12.

J. Liu, S. Wang, G. Qi, B. Zhao, C. Chen, Kinetics and mass transfer of carbon dioxide absorption into aqueous ammonia, Energy Procedia. 4 (2011) 525-532. doi:10.1016/j.egypro.2011.01.084.

G. Chen, Y. Lu, X. Yang, R. Wang, A.G. Fane, Quantitative Study on Crystallization-Induced Scaling in High-Concentration Direct-Contact Membrane Distillation, (2014). doi:10.1021/ie501610q.

Claims (28)

1. Claims: 1. A process for treatment of an input gas, the input gas comprising carbon dioxide, the process comprising: providing a first gas-liquid contact stage, the first gas-liquid contact stage comprising one or more gas-liquid hollow fibre membrane contactors permeable to carbon dioxide, the first gas-liquid contact stage having a gas side and a liquid side; flowing the input gas through the gas side of the first gas-liquid contact stage; providing a first carbon dioxide absorption solution, comprising ammonia; circulating the first carbon dioxide absorption solution through the liquid side of the first gas-liquid contact stage, to absorb some of the carbon dioxide from the input gas to leave an intermediate gas flowing out of the first gas-liquid contact stage; providing a second gas-liquid contact stage, the second gas-liquid contact stage comprising one or more gas-liquid hollow fibre membrane contactor permeable to carbon dioxide, the second gas-liquid contact stage having a gas side and a liquid side; flowing the intermediate gas through the gas side of the second gas-liquid contact stage; providing a second carbon dioxide absorption solution, comprising ammonia; circulating the second carbon dioxide absorption solution through the liquid side of the second gas-liquid contact stage, to absorb carbon dioxide from the intermediate gas to leave an output gas flowing out of the second gas-liquid contact stage, wherein the process is controlled so that more precipitation of ammonium bicarbonate occurs in the first carbon dioxide absorption solution, per unit volume of first carbon dioxide absorption solution, than in the second carbon dioxide absorption solution, per unit volume of second carbon dioxide absorption solution.
2. 2. A process according to claim 1 wherein the process is controlled so that substantially no precipitation of ammonium bicarbonate occurs in the second carbon dioxide absorption solution.
3. 3. A process according to claim 1 or claim 2 wherein the concentration of ammonia in the first and second carbon dioxide absorption solutions is at least 2 mol L-1.
4. 4. A process according to any one of claims 1 to 3 wherein the temperature of the first and second carbon dioxide absorption solutions is controlled to be in the range of not more than 10°C.
5. 5. A process according to any one of claims 1 to 4 wherein the volume of the second carbon dioxide absorption solution in the process is at least two times greater than the volume of the first carbon dioxide absorption solution.
6. 6. A process according to any one of claims 1 to 5 wherein there is provided a first carbon dioxide absorption solution chamber via which the first carbon dioxide absorption solution is circulated and recirculated through the liquid side of the first gas-liquid contact stage.
7. 7. A process according to any one of claims 1 to 6 wherein there is provided a second carbon dioxide absorption solution chamber via which the second carbon dioxide absorption solution is circulated and recirculated through the liquid side of the second gas-liquid contact stage.
8. 8. A process according to any one of claims 1 to 7 wherein precipitated ammonium bicarbonate is removed from the first carbon dioxide absorption solution.
9. 9. A process according to any one of claims 1 to 8 wherein the first carbon dioxide absorption solution is supplemented by some of the second carbon dioxide absorption solution.
10. 10. A process according to any one of claims 1 to 9 wherein the second carbon dioxide absorption solution is supplemented by additional carbon dioxide absorption solution.
11. 11. A process according to claim 10 wherein the additional carbon dioxide absorption solution comprises ammonia thermally recovered from liquid digestate from an anaerobic digester.
12. 12. A process according to any one of claims 1 to 11 wherein the gas is a biogas from an anaerobic digester.
13. 13. A process according to any one of claims 1 to 12 wherein the input gas comprises at least 40 mass % methane and at least 20 mass % carbon dioxide.
14. 14. A process according to any one of claims 1 to 13 wherein the output gas comprises at least 90 mass 'Yo methane and not more than 5 mass % carbon dioxide.
15. 15. A process according to any one of claims 1 to 14 wherein the output gas comprises not more than 2 mass % carbon dioxide.
16. 16. A process according to any one of claims 1 to 15 wherein the intermediate gas comprises at least 0.8 times the concentration of carbon dioxide in the input gas.
17. 17. A process according to any one of claims 1 to 16 wherein the first gas-liquid contact stage comprises two or more gas-liquid hollow fibre membrane contactors and the second gas-liquid contact stage comprises two or more gas-liquid hollow fibre membrane contactors.
18. 18. A process according to any one of claims 1 to 17 wherein the membrane contactor comprises a hydrophobic polymer.
19. 19. A process according to any one of claims 1 to 18 wherein the membrane contactor has an average pore size of not more than 5 pm.
20. 20. A process according to any one of claims 1 to 19 in which gas-side crystallisation of ammonium bicarbonate in the gas-liquid hollow fibre membrane contactors is substantially prevented.
21. 21. A gas treatment system for treatment of an input gas, the input gas comprising carbon dioxide, the system comprising: a first gas-liquid contact stage, the first gas-liquid contact stage comprising one or more gas-liquid hollow fibre membrane contactors permeable to carbon dioxide, the first gas-liquid contact stage having a gas side and a liquid side, the first gas-liquid contact stage being adapted to receive a flow of input gas through the gas side of the first gas-liquid contact stage, the first gas-liquid contact stage being adapted to circulate a first carbon dioxide absorption solution, comprising ammonia, through the liquid side of the first gas-liquid contact stage, to absorb some of the carbon dioxide from the input gas to leave an intermediate gas to flow out of the first gas-liquid contact stage; a second gas-liquid contact stage, the second gas-liquid contact stage comprising one or more gas-liquid hollow fibre membrane contactor permeable to carbon dioxide, the second gas-liquid contact stage having a gas side and a liquid side, the second gas-liquid contact stage being adapted to receive a flow of the intermediate gas through the gas side of the second gas-liquid contact stage, the second gas-liquid contact stage being adapted to circulate a second carbon dioxide absorption solution, comprising ammonia, through the liquid side of the second gas-liquid contact stage, to absorb carbon dioxide from the intermediate gas to leave an output gas to flow out of the second gas-liquid contact stage, wherein the system is controllable so that more precipitation of ammonium bicarbonate occurs in the first carbon dioxide absorption solution, per unit volume of first carbon dioxide absorption solution, than in the second carbon dioxide absorption solution, per unit volume of second carbon dioxide absorption solution.
22. 22. A system according to claim 21 wherein the system is controllable so that substantially no precipitation of ammonium bicarbonate occurs in the second carbon dioxide absorption solution.
23. 23. A system according to claim 21 or claim 22, the system further comprising at least one chiller, adapted to control the temperature of the first and second carbon dioxide absorption solutions to be in the range of not more than 10°C.
24. 24. A system according to any one of claims 21 to 23 wherein there is provided: a first carbon dioxide absorption solution chamber via which the first carbon dioxide absorption solution is circulated and recirculated through the liquid side of the first gas-liquid contact stage; and a second carbon dioxide absorption solution chamber via which the second carbon dioxide absorption solution is circulated and recirculated through the liquid side of the second gas-liquid contact stage, wherein the volume of the second carbon dioxide absorption solution chamber is at least two times greater than the volume of the first carbon dioxide absorption solution chamber.
25. 25. A system according to any one of claims 21 to 24, the system being adapted to permit precipitated ammonium bicarbonate to be removed from the first carbon dioxide absorption solution.
26. 26. A system according to any one of claims 21 to 25 wherein the first gas-liquid contact stage comprises two or more gas-liquid hollow fibre membrane contactors and the second gas-liquid contact stage comprises two or more gas-liquid hollow fibre membrane contactors.
27. 27. A system according to any one of claims 21 to 26 wherein the membrane contactor comprises a hydrophobic polymer.
28. 28. A system according to any one of claims 21 to 27 wherein the membrane contactor has an average pore size of not more than 5 pm.