https://doi.org/10.1016/j.seppur.2022.120786
”
Microporous hollow fibre membrane contactors (HFMC) can support the same overall chemical reaction between CO2 and NH3, but the introduction of a hydrophobic membrane is used to initiate indirect contact between gas and liquid phases. The CO2 therefore freely diffuses from the gas phase on the lumen side, through the gas-filled pores of the membrane, and into an aqueous NH3 liquid on the shell-side of the membrane [18]. Mediating the inclusion of CO2 through the membrane, can improve governance over the rate of nucleation and crystal growth of ammonium bicarbonate [3]. Whilst the membrane introduces an additional resistance to mass transfer, the high specific surface area provides a process intensification 15 times greater than packed column technology [5]. McLeod et al. [27] also identified more than an order of magnitude reduction in NH3 slip with HFMC. The authors attributed the reduced slip to the developed laminar flow in the liquid phase, as this constrains NH3 transport via radial diffusion through the liquid film, stabilising NH3 through chemical transformation to the non-gaseous protonated ammonium form (NH4+). Simple recovery of an aqueous NH3 product from the gas-side of the membrane has also been proposed to further limit slip [40].
By decoupling gas and liquid phases, several authors have also successfully evidenced how a HFMC can permit formation of solid NH4HCO3 in the liquid phase following CO2 absorption into NH3 without causing membrane scaling [27], [2], [3]. The reduced mixing imposed by the characteristically laminar liquid phase, combined with a differential flow path across the interface, that is distinct from the gas phase, may also advantage solids transmission through the membrane. McLeod et al. [26] indicated how membrane hydrophobicity could benefit the induction of crystalline ammonium bicarbonate due to a reduction in activation energy for nucleation. This membrane also creates a unique concentration boundary layer at the three-phase line (gas-liquid-membrane) through the counter diffusion of CO2 and NH3, to create a supersaturated state which can control both nucleation and crystal growth. The crystal size distribution can be particularly important in this application, since this will govern the efficacy of transmission through the process, in addition to the kinetics of dissolution during regeneration of the solvent [2], [3].
However, several studies have also identified crystalline ammonium bicarbonate primarily forming in the gas phase flowing through the membrane lumen following CO2 absorption into NH3 (Table 1) [24], [26], [9], [40]. To illustrate, Makhloufi et al. [24] identified NH4HCO3 precipitation in the lumen-side (gas phase) which resulted in unstable CO2 separation, gradually leading to process failure, and has been similarly observed by other authors [26], [9]. Several mechanisms have been proposed which include: (i) a two-step wetting mechanism, in which the microporous membrane structure is wetted by the solvent, followed by breakthrough of aqueous NH3 into the gas phase which quickly approaches sufficient supersaturation in the CO2 rich gas phase to initiate nucleation [26]; (ii) a gas phase reaction between ‘slipped’ gaseous NH3 and CO2 to produce crystalline NH4HCO3; and (iii) a two-step condensation mechanism in which a binary water vapour/NH3 mixture is co-transported from the solvent to the gas phase, before condensing and becoming rapidly supersaturated with sufficient CO2 to initiate nucleation [26], [40]. The probability for gas phase crystallisation is therefore dependent upon solvent chemistry (NH3 concentration, reactant equilibria, fluid surface tension and temperature) which determines the rate and method of NH3 transport into the gas phase, and the membrane properties (pore geometry, contact angle), which can enhance the probability for wetting and heterogeneous nucleation [26]. However, as each study applies different conditions, it is difficult to establish the extent to which each parameter contributes to the probability for inducing ammonium bicarbonate crystallisation and does not explain why some studies favour gas-phase crystallisation over liquid-phase crystallisation.
Table 1. Literature review of ammonia-CO2 absorption in hollow fibre membrane contactors where crystallisation has been observed.
| Ref. | Operating conditions | Empty Cell | Absorbent | Empty Cell | Feed gas | Ammonium Bicarbonate crystallisation observed | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Membrane/ ID (mm)b | Pore size (µm) |
Gas CSAc (x106 m2) |
Liquid flow/ Gas side |
Empty Cell | [NH3]d (mol L−1) |
C*a (mol L−1) |
pH – |
Te (°C) |
Empty Cell | [CO2]/ RHinlet (%) |
Te (°C) |
Gas velocity | Liquid velocity (x103 m s−1) |
||
| Empty Cell | Empty Cell | (m s−1) | |||||||||||||
| [24] | PP/0.3 | 0.2 | 0.07 | S.P. f / lumen | 0.6–2.9 | 2.4 ± 0.2 | >11 | 21 | 15/0 | 21 | 0.4–1.1 | 0.086–0.24 | Gas side | ||
| [26] | PTFE/1.5 | – | 1.8 | Recycle / lumen | 2 | 2.4 ± 0.2 | >11 | 20 | 50/0 | 20 | 0.93 | 20 | Liquid side | ||
| PTFE/1.5 | – | 1.8 | Recycle / lumen | 3–5 | 2.4 ± 0.2 | >11 | 20 | 50/0 | 20 | 0.93 | 20 | Gas side | |||
| [9] | PTFE/1.5 | 2.6 | 450 | S.P. f / shell | 1–3.7 | 2.4 ± 0.2 | >11 | 21 | 10–15-20/0 | 21 | 0.11 | 5–40 | Gas side | ||
| [40] | PP + PMP/0.2 | – | 0.03 | S.P. f / lumen | 2.9 | 1.9 ± 0.1 | >11 | 10 | 15/100 | 10 | 2 | 2 | Gas side | ||
| PP + PMP/0.2 | – | 0.03 | S.P. f / lumen | 2.9 | 1.9 ± 0.1 | >11 | 10 | 15/0 | 10 | 2 | 2 | Gas side | |||
| PP + PMP/0.2 | – | 0.03 | S.P. f / lumen | 2.9 | 2.4 ± 0.2 | >11 | 21 | 15/100 | 21 | 2 | 2 | Gas side | |||
| PP + PMP/0.2 | – | 0.03 | S.P. f / lumen | 2.9 | 2.4 ± 0.2 | >11 | 21 | 15/0 | 21 | 2 | 2 | Gas side | |||
| PP + PMP/0.2 | – | 0.03 | S.P. f / lumen | 2.9 | 2.4 ± 0.2 | >11 | 21 | 15/0 | 21 | 1.3 | 2 | Gas side | |||
| PP + PMP/0.2 | – | 0.03 | S.P. f / lumen | 2.9 | 2.4 ± 0.2 | >11 | 21 | 15/100 | 21 | 1.3 | 2 | Gas side | |||
| PP + PMP/0.2 | – | 0.03 | S.P. f / lumen | 2.9 | 3.8 | >11 | 40 | 15/0 | 40 | 1.3 | 2 | Not observed | |||
| This study | PP/1.2 | 0.36–0.2 | 1.1 | Recycle / lumen | 3 | 1.6 ± 0.1 | 10 | 5 | 100/0 | 20 | 14.7 | 60 | Liquid side | ||
| PTFE/2.4 | 3.4–0.5 | 4.5 | Recycle / lumen | 3 | 1.6 ± 0.1 | 10 | 5 | 100/0 | 20 | 3.7 | 60 | Liquid side | |||
| PTFE/2.4 | 3.4–0.5 | 4.5 | S.P. f / lumen | 2.3–3 | 2.4 ± 0.2 | >11 | 20 | 100/0 | 20 | 0.2 | 1.4 | Gas side | |||
Delineating the mechanisms that can promote the crystallisation of ammonium bicarbonate preferentially within the liquid phase of a HFMC is therefore critically important in delivering sustainable NH3-CO2 absorption comprising consistent CO2 separation, with the energetic benefit provided by formation of a solid phase NH4HCO3 reaction product. In this study, we therefore systematically explore critical process parameters that enable switching from lumen-side crystallisation to shell-side crystallisation, to sustain CO2 separation by avoiding problematic gas phase reactions and improve recoverability of crystalline NH4HCO3 in the absorption solvent which could then be exploited to lower the energy barrier for solvent regeneration. The specific objectives are to: (i) determine the solvent conditions that primarily govern ammonia transport into the gas phase to prevent lumen side crystallisation; (ii) compare membranes of different properties to evidence how pore size and surface chemistry may promote lumen-side or shell-side crystallisation; and (iii) characterise reactant equilibria chemistry, nucleation kinetics and crystal growth phenomena to describe the mechanism underpinning preferential shell-side nucleation that can enable sustained recovery of crystalline NH4HCO3 in the solvent to promote low energy solvent regeneration.
“