*Article* **Operation of Submerged Anaerobic Membrane Bioreactors at 20** ◦**C: Effect of Solids Retention Time on Flux, Mixed Liquor Characteristics and Performance**

**Santiago Pacheco-Ruiz 1,2,\*, Sonia Heaven <sup>2</sup> and Charles J. Banks <sup>2</sup>**


**Abstract:** Four flat-sheet submerged anaerobic membrane bioreactors ran for 242 days on a simulated domestic wastewater with low Chemical Oxygen Demand (COD) and high suspended solids. Organic loading was maintained around 1.0 g COD L−<sup>1</sup> day−<sup>1</sup> , while solids retention time (SRT) was varied from 20–90 days. This was achieved at a constant membrane flux, maintained by adjusting transmembrane pressure (TMP) in the range 1.8–9.8 kPa. Membrane fouling was assessed based on the required TMP, with mixed liquors characterised using capillary suction time, frozen image centrifugation and quantification of extracellular polymeric substances (EPS). SRT had a significant effect on these parameters: fouling was least at an SRT of 30 days and highest at 60 days, with some reduction as this extended to 90 days. Operation at SRT < 30 days showed no further benefits. Although operation at a short SRT was optimal for membrane performance it led to lower specific methane productivity, higher biomass yields and higher effluent COD. Short SRT may also have accelerated the loss of essential trace elements, leading to reduced performance under these conditions. A COD-based mass balance was conducted, including both biomass and methane dissolved in the effluent.

**Keywords:** anaerobic membrane bioreactors; ambient temperature; membrane fouling; mean cell residence time; wastewater treatment

### **1. Introduction**

Anaerobic technologies for wastewater treatment may offer advantages over aerobic systems, as they produce methane-rich biogas and have much lower sludge yields [1,2]. Anaerobic processes are normally operated at around 35 ◦C, which is known to be an optimum for maintaining a high metabolic activity [3]. It is, however, rarely economical to work at this temperature when treating low-strength municipal wastewaters, as the energy yield may be lower than the parasitic energy demand for heating [4].

Although it is well-known that lower temperatures reduce the rate of biological reaction, there is increasing awareness that effective operation is possible using retained and acclimated biomass [5]. Lower temperature operation does, however, raise other issues such as the increased solubility of methane in the effluent stream; potentially lower removal efficiencies for Chemical Oxygen Demand (COD); and an increase in water viscosity that can reduce the membrane flux and change the settling characteristics of biological solids [3,6–9]. The development of the anaerobic membrane bioreactor (AnMBR) has made it possible to produce high-quality effluents while operating at ambient temperatures and at a reasonably short hydraulic retention time (HRT). A number of reviews [10–12] have demonstrated a growing interest in the application of AnMBRs to a variety of wastewater types. The successful treatment of real and simulated municipal wastewaters with biogas production in AnMBR at operating temperatures as low as 15 ◦C has been demonstrated [13,14].

**Citation:** Pacheco-Ruiz, S.; Heaven, S.; Banks, C.J. Operation of Submerged Anaerobic Membrane Bioreactors at 20 ◦C: Effect of Solids Retention Time on Flux, Mixed Liquor Characteristics and Performance. *Processes* **2021**, *9*, 1525. https://doi.org/10.3390/pr9091525

Academic Editor: Bipro R. Dhar

Received: 19 July 2021 Accepted: 24 August 2021 Published: 29 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Yet, there still remain considerable knowledge gaps and technical challenges before the technology can become more widely adopted at full scale [10,15].

The retention of biomass in an AnMBR allows mixed liquor suspended solids (MLSS) concentrations to be maintained without carry-over of solids into the effluent; it also gives the potential for selecting the MLSS concentration for optimum organic matter degradation and membrane performance [7,15]. The decoupling of solids retention time (SRT) from HRT provides a way to control the mean cell residence time (MCRT) in the reactor [8]. This approach to process control has been used extensively in aerobic wastewater treatment, as MCRT is the reciprocal of biomass growth rate and, therefore, directly influences both metabolic activity and sludge yield [16]. The simplest way to control MCRT, and, therefore, gain kinetic control over the treatment process, is through proportional biomass wastage [17]. This important kinetic control parameter has not, however, been widely used in anaerobic systems, even though MCRT has been shown to influence the production of extracellular polymeric substances (EPS) and soluble microbial products [3,18]. These are particularly important in submerged AnMBR (SAnMBR), where the membrane is directly immersed in the mixed liquor, and both EPS and soluble microbial products are known to affect membrane fouling [19,20]. In aerobic treatment systems, it is thought that a long MCRT reduces the concentration of EPS and soluble microbial products and because these are considered to be more critical in inducing membrane fouling than is the MLSS concentration, a long MCRT is usually chosen for operation [3,20,21].

Relatively few AnMBR studies to date have used MCRT or solids retention time (SRT) as a control parameter, while many have operated at very extended or near-infinite solids retention times, with the SRT in some cases determined only by the need to remove small volumes of MLSS for analysis. Studies in which SRT has been varied, however, have not always run for long enough to reach steady-state conditions: this is often defined as operation for at least 3 SRT. Without this, it is unlikely that MLSS properties will be representative in each case. Baek et al. [22] operated a 10-L AnMBR with a sidestream membrane filter on settled municipal wastewater over a range of SRT from 19–217 days but only achieved 3 SRT at the lowest value of 19 days. Yeo and Lee [23] tested SRT of 20 and 40 days in a 5-L SAnMBR operated at 23 ◦C on a feed of glucose at 5 g COD L−<sup>1</sup> . It is likely that only 1 SRT was achieved in each case, although the duration of this experiment was not explicitly stated. Dong et al. [24,25] used the bench and pilot-scale AnMBRs with external hollow fibre membrane (HFM) units fed on municipal wastewater at 23 ◦C. They tested SRT of 100, 70 and 40 days, but none of these appear to have run for 3 SRT. Thanh et al. [26] reduced the SRT from 100 to 75, 50 and then 25 days over a 60-day period in a flat-sheet SAnMBR at 35 ◦C fed on dilute synthetic wastewater. Yurtsever et al. [27] treated highsalinity synthetic textile wastewater in aerobic and anaerobic flat-sheet MBRs at 60-days, 30-days and near-infinite SRT, but in each case for much less than 3 SRT. Ji et al. [28] altered the SRT from 53 days to near-infinite in response to changes in MLSS caused by varying load: this is a familiar alternative method for process control in aerobic systems, but they did not attempt to use SRT itself as a control parameter.

A small number of studies using SRT as the main control parameter have run for long enough to approach steady-state operation. In some cases, these have used highstrength effluents under mesophilic conditions. Dereli et al. [29,30] operated two mesophilic cross-flow AnMBR on high-strength corn stillage for 3 months at SRT of 20 and 30 days, respectively, then ran for a further 3 months at a 50-day SRT. Szabo-Corbacho et al. [31] investigated the effect of SRT in a crossflow AnMBR treating high-strength synthetic dairy wastewater at 35 ◦C and completed ~3 SRT at 20 and 40-day SRT. Pacheco-Ruiz et al. [32] used a low-to-intermediate strength synthetic wastewater at 36 ◦C and showed that a short SRT gave enhanced membrane performance but resulted in lower specific methane production and higher waste sludge yield. Very few such studies have used low-strength wastewaters at lower temperatures, however, and the limited results to date have been conflicting. When treating synthetic low-strength wastewater in a flat-sheet membrane at 25–30 ◦C, Huang et al. [33] noted that a longer SRT led to greater fouling when using the

same AnMBR to treat municipal wastewater, however, they also found higher fouling at a short SRT [34]. While more studies are coming through using real municipal wastewaters, at ambient temperatures, and/or in larger-scale systems [1,6,15,24,35], data interpretation and performance prediction can be challenging due to the number of influencing variables. This is especially the case where multiple factors can alter simultaneously [3,4,7], and there is thus a clear need for studies focusing on the effects of individual parameters such as SRT.

The current research aimed to assess the effect of changes in SRT on the performance of a SAnMBR treating low-strength wastewater at 20 ◦C, and, in particular, the influence on membrane fouling, sludge yield, COD removal efficiency, and physico-chemical properties of the MLSS once steady-state operation has been achieved. The results may thus contribute towards resolving some conflicting reports in the literature and could help to establish MCRT or SRT as a principal control parameter for anaerobic wastewater treatment systems, in much the same way as it is considered the control parameter of choice in aerobic systems.

#### **2. Materials and Methods**

#### *2.1. SAnMBR Design and Operation*

Four SAnMBRs (S1, S2, S3 and S4) were used, in which the transmembrane pressure (TMP) was applied via a gravity head [28]. Each SAnMBR had a working volume of 9.6 L, with 2.1 L of headspace (Figure 1). Each was fitted with a chlorinated polyethylene flat-plate membrane cartridge (Type 203, Kubota Corporation, Osaka, Japan) with nominal pore size 0.4 µm and effective surface area 0.113 m<sup>2</sup> , at a membrane packing density of 0.012 m<sup>2</sup> L −1 . The SAnMBRs were fed continuously with the chilled substrate at a controlled flow achieved by maintaining a constant head differential with the outlet from the membrane cartridge lumen. The driving force for the passage of effluent through the membrane was thus gravity-induced, and the TMP could be altered within a range of 1.8–9.8 kPa by changing the differential head. The SAnMBRs could thus be operated at a constant membrane flux, allowing control of the organic loading rate to be maintained as the membranes became progressively fouled. Flow was measured on a weighing scale with 32 kg capacity and readable to 1.0 g (Model CBK 32, Adam Equipment Co Ltd., Milton Keynes, UK), with the weight of collected permeate logged automatically at 5 min intervals. The internal headspace pressure was maintained at around 0.3 kPa, with biogas released via a fermentation gas lock into a gas-impermeable collection bag.

In situ cleaning of the membrane was achieved by recirculating headspace biogas at approximately 0.57 L L−<sup>1</sup> of reactor min−<sup>1</sup> (corresponding to 48.5 L min−<sup>1</sup> m−<sup>2</sup> of membrane surface area) using a stainless-steel sparger and a diaphragm pump (AIRPO, Ningbo forever Electronic Appliances Co Ltd., Ningbo, Zheijiang, China). This gas flow also maintained the mixed liquor in suspension. SAnMBR temperature during the experimental period was maintained at 20.5 ± 0.5 ◦C by a stainless-steel internal heat exchange coil coupled with a thermocirculator (FC15 and FH15V, Grant Instruments Europe BV, Amsterdam, Netherlands). Feed and SAnMBR temperature were continuously recorded by solid-state IC temperature probes (LM35DZ, Texas Instruments, Dallas, TX, USA) and a datalogger (Model U3-LV, Labjack Corporation, Lakewood, CO, USA).

#### *2.2. Inoculum and Substrate*

The substrate used was a synthetic wastewater [36], which was prepared fresh each day as a concentrate and diluted to the required working strength. It was designed to give a C:N:P ratio of around 100:20:4 and solids contents similar to those of typical municipal wastewaters [37,38].

The four AnMBRs were seeded from a mesophilic anaerobic digester at a municipal wastewater treatment plant in Southampton, UK, and the headspace was purged with nitrogen. To allow temperature acclimatisation they were initially fed with the concentrated substrate at an organic loading rate (OLR) of 0.5 g COD L−<sup>1</sup> day−<sup>1</sup> for 48 days (not counted as part of the trial). A total of 50% of the MLSS was then removed and replaced by tap water to reduce the solids content. From this point onwards (taken as day 0 of the experimental

period) the SAnMBRs were fed with substrate prepared by diluting the concentrate to the required COD strength. The SAnMBRs were then run at different SRTs in 3 experimental phases (EP), giving a total operational period of 242 days. Details of the applied SRT, TMP and other operational parameters are given in Table A1. *Processes* **2021**, *9*, 1525 4 of 26

Two trace element stock solutions were used containing (g L−<sup>1</sup> ): Al 0.1, B 0.1, Co 1.0, Cu 0.1, Fe 10.0, Mn 1.0, Ni 1.0, Zn 1.0; and Se 0.1, Mo 0.1, W 0.1.

**Figure 1.** Schematic of experimental set-up for SAnMBRs.

#### **Figure 1.** Schematic of experimental set-up for SAnMBRs. *2.3. Performance and Stability*

*2.2. Inoculum and Substrate*  The substrate used was a synthetic wastewater [36], which was prepared fresh each day as a concentrate and diluted to the required working strength. It was designed to give a C:N:P ratio of around 100:20:4 and solids contents similar to those of typical municipal wastewaters [37,38]. The four AnMBRs were seeded from a mesophilic anaerobic digester at a municipal wastewater treatment plant in Southampton, UK, and the headspace was purged with nitrogen. To allow temperature acclimatisation they were initially fed with the concentrated substrate at an organic loading rate (OLR) of 0.5 g COD L−1 day−1 for 48 days (not counted as part of the trial). A total of 50% of the MLSS was then removed and replaced by tap water to reduce the solids content. From this point onwards (taken as day 0 of the experimental period) the SAnMBRs were fed with substrate prepared by diluting the concentrate to the required COD strength. The SAnMBRs were then run at different SRTs in 3 experimental phases (EP), giving a total operational period of 242 days. Details of the applied SRT, TMP and other operational parameters are given in Table A1. Two trace element stock solutions were used containing (g L−1): Al 0.1, B 0.1, Co 1.0, Membrane performance was evaluated on the basis of the flux achieved at constant TMP, calculated as described in Pacheco-Ruiz et al. [36]. SAnMBR performance and operational stability were assessed on the basis of COD percentage conversion, specific methane production (SMP) per g of COD removed, MLSS and mixed liquor volatile suspended solids (MLVSS) concentrations, mixed liquor pH, EPS content and composition, capillary suction time (CST) (Triton Electronics Ltd., Dunmow, Cambridge, UK), and frozen image centrifugation (FIC) (Triton Electronics, UK). FIC uses a technique in which a 'frozen image' of the sample is generated by matching the frequency of a stroboscopic light to the centrifuge rotor speed, allowing measurement of the height of the solid-liquid interface in real time without interrupting the test. The centrifugation speed was fixed at 660 ± 10 rpm with observations made every minute up to 8 min. COD of fresh feed, feed after chilled storage for 24 h, and AnMBR effluent was measured using a closed-tube reflux method with titrometric end-point determination [39]. Biogas composition was analysed by gas chromatography (GP-3400, Varian Inc., Palo Alto, CA, USA) using 36% CO<sup>2</sup> with 64% CH<sup>4</sup> (*v*/*v*) (BOC, Guildford, Surrey, UK) as a standard gas. Biogas volumes were determined by a weight-type gasometer [40] and are reported at standard temperature and pressure (STP, 0 ◦C and 101.3 kPa).

Cu 0.1, Fe 10.0, Mn 1.0, Ni 1.0, Zn 1.0; and Se 0.1, Mo 0.1, W 0.1. *2.3. Performance and Stability*  Organic loading rate (OLR) and SMP were calculated using the average of the COD values for the fresh and stored feed. The reported SMP value includes both methane in gas

thane production (SMP) per g of COD removed, MLSS and mixed liquor volatile suspended solids (MLVSS) concentrations, mixed liquor pH, EPS content and composition,

Membrane performance was evaluated on the basis of the flux achieved at constant TMP, calculated as described in Pacheco-Ruiz et al. [36]. SAnMBR performance and oper-

collected from the reactor headspace and dissolved methane in the effluent. This allows comparison of SAnMBR performance under different sconditions since the proportion of methane that leaves the system in the effluent will differ at different flux rates. Dissolved methane content was estimated based on Henry's Law using 20 ◦C saturation concentration. The resulting average value 29.0 mL CH<sup>4</sup> L <sup>−</sup><sup>1</sup> was confirmed by empirical measurement, using the method in Walsh and McLaughlan [41].

MLSS and MLVSS concentrations were quantified using Standard Method 2540-D [42]. pH measurements were made using a pH meter (3310, Jenway Ltd., London, UK) calibrated in pH 4, 7 and 9.2 buffers (Fisher Scientific UK Ltd., Loughborough, Leicestershire, UK). For the COD mass balance, the only input was influent COD while outputs were taken as effluent COD, COD in gaseous or dissolved methane and COD in biomass. The last of these was estimated from daily MLSS removal (g MLVSS day−<sup>1</sup> ) divided by the average ratio of COD/VSS in the mixed liquor (g COD g−<sup>1</sup> MLVSS). COD balances did not consider changes in stored biomass, as small errors in MLSS and MLVSS measurement could introduce large variations in the overall result. Thus the COD balances were only valid for steady-state conditions or other periods with stable solids contents. The COD value of methane was taken as 2.855 g COD L−<sup>1</sup> CH<sup>4</sup> based on stoichiometric considerations.

EPS was extracted using the formaldehyde plus NaOH procedure [43], modified in accordance with Domínguez et al. [44] and Liang et al. [45] to enable identification of bound and soluble components. Soluble EPS was extracted by centrifugation of mixed liquor, as suggested by Chabaliná et al. [46]. EPS composition was quantified by measuring the concentration of carbohydrate and protein using colorimetric methods. Carbohydrate was determined by the phenol-sulphuric acid method [47], using a glucose standard. Protein contents were analysed according to the modified Lowry Folin–Ciocalteu method suggested by Frølund et al. [48], with bovine serum albumin as a standard.

#### **3. Results and Discussion**

#### *3.1. Operational Performance*

Figure 2 presents graphical data on the operation and performance of the four AnMBRs during the 242-day experimental period, with results summarised in Tables A1 and A2. Discussion of key parameters in relation to operating conditions during each phase of the experimental period is provided below.

#### 3.1.1. Membrane Flux, TMP, MLSS, HRT and OLR

Start-up (days 0–59). After the 48-day temperature acclimatisation phase and following dilution, the MLSS in each SAnMBR was initially 16.2 g L−<sup>1</sup> . No sludge was wasted for the next 10 days, after which time proportional wasting was introduced with the aim of applying a SRT of 90 days (Figure 2a). In the following period, a number of changes were made to the TMP and feed concentration in order to find a combination of conditions that allowed acclimatisation to an increase in OLR. The TMP was initially set at 7 kPa, giving an initial flux of 16.1 L m−<sup>2</sup> h −1 , which after 5 days had decreased to 10.2–11.0 L m−<sup>2</sup> h −1 (Figure 2b). This high flux produced short HRTs with low COD removal rates of between 54–64%, and on day 9 TMP was reduced from 7.0 to 2.5 kPa with the aim of reducing the flux (Figure 2c). The flux remained high, however, at around 9.9–10.6 L m−<sup>2</sup> h −1 . Thus to avoid organic overloading, the COD of the feed was reduced to around 160 mg L−<sup>1</sup> . The pH immediately rose to 6.9–7.0, and the COD removal rate increased to 74–80%. The OLR was subsequently increased to 0.9, 1.0, 1.1 and 1.5 g COD L−<sup>1</sup> day−<sup>1</sup> stepwise at 3-day intervals by increasing the feed concentration. This led to changes in gas composition, with a higher methane content indicating greater biogas production, but the last increment in OLR resulted in a fall in pH to 6.6 and a decline in %COD removal. The OLR was, therefore, reduced on day 22 to 0.75 g COD L−<sup>1</sup> day−<sup>1</sup> by reducing the feed concentration, whilst the flux was maintained at 8.0–8.8 L m−<sup>2</sup> h −1 . To improve COD removal, on day 28, the flux was then further reduced to around 6 L m−<sup>2</sup> h <sup>−</sup><sup>1</sup> by reducing the TMP to 2.2 kPa (Figure 2c), and from this time onward, SAnMBR performance improved to a

point where between day 40 and 52 the OLR was successfully raised to 1.0 g COD L−<sup>1</sup> day−<sup>1</sup> . By day 60, stability had been achieved in all four SAnMBRs, with flux rates between 6.3–6.7 L m−<sup>2</sup> h −1 , and MLSS contents from 12.4–12.7 g L−<sup>1</sup> (Figure 2d), COD removal rates between 88–93% (Figure 2e), and stable biogas composition and production. At this point, the start-up phase was considered complete.

Experimental phase 1 (EP1, days 60–111). On day 60, the SRT was reduced to 30, 45 and 60 days in S1, S2 and S3, respectively, by increasing the volume of mixed liquor removed each day, while the SRT in S4 remained unchanged at 90 days. At the start of this phase, the TMP in all four SAnMBR was 2.5 kPa, while flux rates were 6.7, 6.4, 6.3 and 6.4 L m−<sup>2</sup> h −1 in S1, S2, S4 and S4, respectively (Figure 2b,c). As the flux gradually declined, TMP was adjusted in each SAnMBR individually, with the aim of stabilising at a value of around 5 L m−<sup>2</sup> h −1 . Feed concentrations were also modified slightly with respect to the achieved flux, with the aim of providing a consistent OLR of around 1 g COD L−<sup>1</sup> day−<sup>1</sup> . The amount of adjustment applied during this phase was, however, much less than that required in the start-up period.

By day 67, the effect of the applied changes in SRT was already evident in S3, where the onset of membrane fouling was indicated by the need for substantial increases in TMP to maintain a flux of 5 L m−<sup>2</sup> h −1 . Similar behaviour soon followed in S2 and S4, which from day 80 also needed higher TMPs to achieve the target flux value, while in S1 it became necessary to reduce the TMP as the flux was increasing. At this stage, it was concluded that a flux of 5 L m−<sup>2</sup> h <sup>−</sup><sup>1</sup> was too high for these experimental conditions. On day 100, the target was, therefore, reduced to 4 L m−<sup>2</sup> h −1 , achieved by applying reduced TMPs of 1.8, 2.9, 4.3 and 2.2 kPa in S1, S2, S3 and S4, respectively (Figure 2c). This gave HRTs of approximately 20 h and a OLR of around 1.0 g COD L−<sup>1</sup> day−<sup>1</sup> , at an average COD concentration in the feed of <sup>860</sup> <sup>±</sup> 23 mg COD L−<sup>1</sup> . Minor variations in feed concentration here were caused by slight day-to-day variations in batch preparation and in analytical results, with no further deliberate adjustments being made to influent strength. In S1, S2 and S4 a flux of 4 L m−<sup>2</sup> h −1 was maintained at constant TMP until the end of EP-1. In contrast, the flux in S3 continued falling irrespective of increases in the TMP, which at the end of this phase had reached 6.3 kPa (Figure 2c), giving a final OLR of 0.9 g COD L−<sup>1</sup> day−<sup>1</sup> at an HRT of 22.5 h in this reactor. The changes in SRT were also reflected in MLSS concentrations, with values between 12.4–12.7 g L−<sup>1</sup> in all SAnMBR at the start of the phase diverging to 6.2, 7.3, 8.7 and 11.7 g L−<sup>1</sup> in S1, S2, S3 and S4, respectively (Figure 2d).

Experimental phase 2 (EP-2, days 112–160). In response to the changes observed in EP-1, at the beginning of EP-2 the SRT in S3 was reduced in one step from 60 to 30 days, at an initial TMP of 6.3 kPa. S1, S2 and S4 remained at their previous SRT of 30, 45 and 90 days with corresponding initial TMPs of 1.7, 2.9 and 2.2. While flux in S1 and S2 remained steady at 4 L m−<sup>2</sup> h <sup>−</sup><sup>1</sup> without any adjustment of TMP, flux in S3 continued falling, despite continuing rises in TMP (Figure 2c). On day 143, the TMP in S3 reached the limiting value of 9.8 kPa, at a flux rate of 3.6 L m−<sup>2</sup> h −1 . The flux in S4 remained steady until day 115 when it began to decline, necessitating increases in the TMP. By day 160, S4 had also reached the maximum TMP at a flux of 3.2 L m−<sup>2</sup> h −1 (Figure 2b). In S1 and S2, the constant flux rates resulted in stable HRT of around 20.5 h and an OLR of 1.0 g COD L−<sup>1</sup> day−<sup>1</sup> for both SAnMBRs. In contrast the continuous decline in flux in S3 and S4 led to higher HRTs of 28.5 and 26.6 h with OLR of 0.7 and 0.8 g COD L−<sup>1</sup> day−<sup>1</sup> , respectively, at the end of EP-2.

MLSS concentrations in this phase continued to reflect the different SRT, with values in S1 and S2 stabilising at around 4.4 and 5.8 g L−<sup>1</sup> , respectively (Figure 2d). The fall in MLSS in S3 reflected the change in SRT to 30 days at the start of the phase, and the reduction in applied OLR due to lower flux, which resulted in a MLSS concentration of 4.3 g L−<sup>1</sup> in S3 at the end of EP-2. MLSS concentrations in S4 initially stabilised at 10.8 g L−<sup>1</sup> (Figure 2d) but by the end of the phase had fallen to 9.5 g L−<sup>1</sup> as a result of the decrease in flux rate and consequently in OLR.

**Figure 2.** Operational parameters in SAnMBRs during the whole experimental period: (**a**) SRT; (**b**) TMP; (**c**) flux (daily average); (**d**) MLSS; (**e**) COD removal; (**f**) SMP; (**g**) specific SMP (MLSS basis). Vertical dotted lines indicate start of exper-**Figure 2.** Operational parameters in SAnMBRs during the whole experimental period: (**a**) SRT; (**b**) TMP; (**c**) flux (daily average); (**d**) MLSS; (**e**) COD removal; (**f**) SMP; (**g**) specific SMP (MLSS basis). Vertical dotted lines indicate start of experimental phases.

imental phases.

At the end of EP-2, S1 had completed 101 days (equivalent to 3.4 solid retention times) at a 30-day SRT and was, therefore, regarded as having achieved steady-state operation in these conditions.

Experimental phase 3 (EP-3, days 161–242). At the start of this phase, SRT was reduced to 20 days in S1, maintained at 45 days in S2 and at 30 days in S3, and reduced to 30 days in S4. At this point, S4 had completed only 160 days at a 90-day SRT, equivalent to 1.78 solid retention times or approximately 83% washout of the MLSS originally present at the start of EP-1, but it was clear that operation was no longer sustainable within the limits of the target flux and TMP. As an alternative to the slow reduction in MLSS content occurring in S3, however, MLSS in S4 was reduced abruptly from 9.1 to 7.2 mg L−<sup>1</sup> , targeting a value close to that expected for a 30-day SRT.

In the case of S4, this reduction was achieved by removing a known volume of mixed liquor and replacing it with influent, to maintain a constant working volume in the SAnMBR. The process was repeated on four consecutive days, until the MLSS concentration approached the target value. The daily volume of mixed liquor replaced was 700 mL on days 160 and 161, and 400 mL on days 162 and 163.

In S1 at a 20-day SRT, the flux during this phase was successfully maintained at 4 L m−<sup>2</sup> h −1 , with only a slight TMP increase to 2.2 kPa. In contrast, flux in S2 at a 45-day SRT remained steady at 4 L m−<sup>2</sup> h <sup>−</sup><sup>1</sup> until day 180 when it began to fall, necessitating repeated increases in TMP (Figure 2c). By day 216, TMP in S2 had reached the limiting value of 9.8 kPa, and from then on flux continued gradually to decline, stabilising at around 3.2 L m−<sup>2</sup> h <sup>−</sup><sup>1</sup> by the end of EP-3. Flux in S3 at a 30-day SRT continued to fall (Figure 2b), reaching its lowest value of 2.7 L m−<sup>2</sup> h <sup>−</sup><sup>1</sup> on day 183. After this it slowly recovered, up to the 4 L m−<sup>2</sup> h −1 target by day 228 (Figure 2b). Over the next 5 days, TMP in S3 was reduced to 6.5 kPa, while flux remained steady at 4.1 L m−<sup>2</sup> h −1 to the end of the experiment.

The sharp reduction in MLSS in S4 produced an immediate increase in flux, reaching 4 L m−<sup>2</sup> h <sup>−</sup><sup>1</sup> by day 200 (Figure 2b). Over the next 10 days, the TMP was reduced to 3.8 kPa, while a steady flux of 4.1 L m−<sup>2</sup> h <sup>−</sup><sup>1</sup> was maintained until the end of EP-3.

In S1 at a 20-day SRT, the HRT and OLR remained steady at around 21.0 h and 1.0 g COD L−<sup>1</sup> day−<sup>1</sup> . Similar values were also achieved in S3 and S4 at a 30-day SRT once they approached the same sustainable rate of 4 L m−<sup>2</sup> h −1 . In S2 at a 45-day SRT a continuing slow decline in flux increased the HRT to 26.3 h with a corresponding fall in OLR to 0.8 g COD L−<sup>1</sup> day−<sup>1</sup> by the end of the phase.

At the end of EP-3, reactors S1, S2 and S3 had, respectively, completed 82, 183 and 131 days at 20-, 45- and 30-day SRT (equivalent to 4.1, 4.1 and 4.4 solid retention times in each case), making the results representative of steady-state operation in these conditions. S4 had completed 82 days at a 30-day SRT, corresponding to 2.73 solid retention times with removal of around 93.5% of the MLSS present at the start of EP-3. It was, therefore, regarded as approaching steady-state operation.

The above results indicate the complexity of interactions between SRT, required TMP, flux, MLSS, HRT and OLR, and factors associated with this are discussed in the following sections.

#### 3.1.2. COD Removal Rates and TE Requirements

During the start-up period COD removal initially showed quite a high variability, which stabilised towards the end of the period. COD removal rates were similar in all four SAnMBRs for the first 20 days of EP-1 at 89–92% (Figure 2e). Differences in SRT began to affect COD removal during EP-1, however, with S1, S2, S3 and S4 at 30-, 45-, 60- and 90-day SRTs reaching the end of the phase with removal rates of 88%, 92%, 96% and 98%, respectively. In EP-2, COD removal rates in S1, S2 and S4 stabilised at 85 ± 2%, 93 ± 1% and 96 ± 1%, respectively. In contrast, the removal rate for S3, in which the SRT had been reduced from 60 to 30 days, fell steadily from 95% to 91%.

In the first 40 days of EP-3, there was a slight decrease in COD removal in all four SAnMBRs (Figure 2e), followed by significant falls of 8% and 20% in S1 and S3, respectively,

between days 204 and 222. It was hypothesised that this was due to a lack of trace elements (TE), since the shorter SRT in these reactors meant they had the highest overall biomass turnover, and thus the greatest risk of depletion of essential micro-nutrients. Based on this, for a 3-day period from day 223, trace elements were added to the feed for all four SAnMBRS at a dosage of 0.1 mL of each TE solution per litre of dilute influent. This led to an immediate rise in COD removal in all SAnMBRs, which stabilised at 92%, 97%, 95% and 96% in S1, S2, S3 and S4, respectively, by the end of EP-3. It was, therefore, likely that some of the decline in COD removal during this period was due to TE deficiencies, but removal rates after the TE supplementation were still slightly lower at the shorter SRTs. Other authors [30,31] have also reported slightly higher COD removal rates at longer SRT, though these studies involved high-strength substrates. In contrast, Huang et al. [33] observed no significant differences in COD removal rates for low-strength synthetic wastewater at 30-day, 60-day and near-infinite SRT.

Determination of TE requirements in AnMBR is likely to be especially challenging due to the combined effects of uncoupled liquid and solid retention times with uncertainties on bioavailability and partitioning. A number of studies have reported the addition of trace elements to feed without providing detailed justifications of the concentration or dosages used [14,19,23,33,49]. Yu et al. [50] investigated the effects of individual and combined TE supplementation on samples from the methanogenic reactor in a 2-phase AnMBR treating industrial starch wastewater at 37 ◦C.

They reported little effect from low doses, but with an SRT of 200 days and inoculum from a municipal sewage treatment plant, it is possible that washout of some TE to critical levels had not yet occurred. Sierra et al. [51] looked at the partitioning of trace elements B, Co, Cu, Mg, Mn, Mo, Ni, Se and W and at the effect of additional supplementation with Co and W in an AnMBR treating a highly saline phenolic wastewater. Additional Co had little effect, but W was beneficial. Doubling the overall TE dose also improved performance, and they concluded that bioavailability and partitioning were affected by high salinity. Thanh et al. [26] examined the effects of HRT, SRT and pH on bioavailability and speciation of Co, Fe, Mn, Mo, Ni and Zn. When reducing the SRT from 100 to 75, 50 and then 25 days over a 60-day period, they found varying depletion rates for different metals depending on affinity and previous accumulation. While the total trace metal content fell with the reduction in MLSS, this was partly countered by a change in speciation towards more bioavailable forms for all metals except Mn and Ni. While these studies have shown the importance of TE in ANMBR, relationships between TE, SRT and other operating parameters and optimum dosing strategies for these systems are likely to be key areas for future work [51,52].

#### 3.1.3. COD Balances and Dissolved Methane

COD balances (shown in Figure 3) indicated that a substantial fraction of the COD was converted to methane, including a considerable proportion dissolved in the effluent. The response to the addition of TE on day 223 can also be clearly seen. COD balances for steady-state periods closed at between 92–96% in this work.

Small errors in COD balances are typical, and studies reporting better closure tend to operate with higher strength substrates and/or temperatures or at a larger scale [31]. Closures of 93–94% were obtained for a 20-L SAnMBR treating municipal wastewater at 25 ◦C at HRT from 4–12 h [28], while a slight excess of around 101% was reported for a 5 m<sup>3</sup> SAnMBR operating at the same temperature on a similar municipal effluent [1]. The missing fraction in the COD balance could be attributable in part to small quantities of biogas leaving the SAnMBRs through the permeate line in gaseous form as bubbles [36] and by any H2S fraction in the biogas. H2S was not taken into account in the COD balance as concentrations in the biogas were considered to be low, but the inclusion of this component may be essential in the treatment of protein- or sulphate-rich wastewaters [53]. Using Henry's Law to estimate the amount of CH<sup>4</sup> in the effluent may also lead to underestimation, as the calculation assumed a saturation concentration, while the elevated

TMP could cause additional dissolution and apparent methane supersaturation in the effluent [8]. In such cases, the actual volume of dissolved methane lost would be greater than the estimated value, especially at lower operating temperatures where saturation concentrations are higher. Differences between observed and simulated mass flows of methane in pilot and bench-scale sidestream AnMBRs treating screened municipal wastewater at ambient temperature were attributed to oversaturation [24], but were insufficient to account for errors of ~20% in COD balances, which may have been linked to the absence of steady-state conditions as well as to the reduction of sulphates and ferric iron. Methane oversaturation was also reported in a pilot-scale gas-sparged sidestream An-MBR operating on screened municipal wastewater at temperatures of 18.8–31.5 ◦C [12]. Yeo and Lee [23] noted variations in dissolved methane content with SRT in SAnMBR at 23 ◦C fed on glucose at 5.2 g COD L−<sup>1</sup> with an HRT of 10 days. They reported oversaturation of methane and a lower biogas methane percentage at a 20-day SRT and presented COD-based balances closing at 3% and 7% for the 20 and 40-day SRT, respectively. *Processes* **2021**, *9*, 1525 10 of 26 the importance of TE in ANMBR, relationships between TE, SRT and other operating parameters and optimum dosing strategies for these systems are likely to be key areas for future work [51,52]. 3.1.3. COD Balances and Dissolved Methane COD balances (shown in Figure 3) indicated that a substantial fraction of the COD was converted to methane, including a considerable proportion dissolved in the effluent. The response to the addition of TE on day 223 can also be clearly seen. COD balances for steady-state periods closed at between 92–96% in this work.

**Figure 3.** COD balances in SAnMBRs during the whole experimental period: (**a**) S1, (**b**) S2, (**c**) S3 and (**d**) S4. Vertical dotted lines indicate the start of experimental phases. Values for control parameters (SRT and TMP) in each phase are shown in Figure 2 and Table A2. **Figure 3.** COD balances in SAnMBRs during the whole experimental period: (**a**) S1, (**b**) S2, (**c**) S3 and (**d**) S4. Vertical dotted lines indicate the start of experimental phases. Values for control parameters (SRT and TMP) in each phase are shown in Figure 2 and Table A2.

Small errors in COD balances are typical, and studies reporting better closure tend to operate with higher strength substrates and/or temperatures or at a larger scale [31]. Clo-

°C at HRT from 4–12 h [28], while a slight excess of around 101% was reported for a 5 m3

#### 3.1.4. Specific Methane Productivity

Biogas composition for all SAnMBRs remained stable throughout the three experimental phases at 78% CH4, 8% CO<sup>2</sup> and 14% nitrogen, the latter from headspace equilibration with atmospheric gases dissolved in the influent. SMP in all SAnMBRs showed the same trend during start-up, beginning at zero, increasing sharply to between 0.29–0.42 L CH<sup>4</sup> g −1 COD as residual feed and intermediate products were consumed, then stabilising at 0.20–0.22 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> COD. As can be seen from Figure 2f and Table A1, during the stable operating periods of each phase, the SMP in L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> COD removed (CODrem) was around 0.23 ± 0.01 for operation at a 20-day SRT in S1; 0.22 ± 0.01, 0.23 ± 0.01 and 0.24 ± 0.01 at a 30-day SRT in S1, S3 and S4, respectively; 0.25 ± 0.01 at a 45-day SRT in S2; and 0.26 ± 0.02 at a 90-day SRT in S4. SMP at a 60-day SRT in S3 was around 0.23 ± 0.02 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> CODrem, although full steady-state operation under these conditions was not achieved. These results again demonstrate a decline in SMP at the shorter SRT, most likely due to the incorporation of a greater proportion of carbon into microbial biomass at higher growth rates.

This outcome was consistent with the results of earlier research at 36 ◦C [32], with the lower SMP values most probably due to reduced rates of reaction at a lower operating temperature. Huang et al. [33,34] also found that SMP rose with increases in SRT in the treatment of low-strength wastewaters. They reported values of 0.13, 0.20 and 0.22 L CH<sup>4</sup> g −1 CODrem at SRT of 30, 60 and infinite days respectively in treatment of synthetic wastewater at OLR from 1.1–1.67 g COD L−<sup>1</sup> day−<sup>1</sup> [33]; and around 0.03, 0.08 and 0.08 L CH<sup>4</sup> g −1 CODrem at SRT of 30, 60 and 90 days, respectively, when treating municipal wastewater at an OLR of around 1 g COD L−<sup>1</sup> day−<sup>1</sup> [34]. Their suggested explanation was that the longer SRT would provide better conditions for methanogenesis, allowing higher biogas productivity. No reason was given, however, for the very low values of SMP per g of COD removed when treating real municipal wastewater [34]. SMP values obtained at 20 ◦C in the current work are higher than those found by Huang et al. [33,34] at 25–30 ◦C. This may be partly due to the loss of dissolved methane in the effluent, which was not considered in their first study [33]. The pattern of lower SMP values at shorter SRTs is nevertheless evident in both of these studies and in this and earlier work at 36 ◦C [32]. Dong et al. [24] also reported a decline in SMP from 0.13 to 0.10 and then 0.08 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> CODrem as SRT was stepped down from 100 to 70 and then 40 days over a 400-day experimental period when treating municipal wastewater at 23 ◦C.

Consideration of the SMP normalised to MLSS (SMPMLSS) indicated that the biomass methane conversion efficiency was higher at shorter SRT (Figure 2g), as also seen in prior work at 36 ◦C [32]. SMP values obtained were similar to those at 36 ◦C, with a maximum of 0.095 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> CODrem L g−<sup>1</sup> MLSS for operation at a 20-day SRT in S1; and around 0.2 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> CODrem L g−<sup>1</sup> MLSS during a stable performance at a 90-day SRT between days 110–130 in S4. Data in Huang et al. [33] indicated a higher SMPMLSS value at longer SRTs, in contradiction to the outcomes both of this study at 20 ◦C and of the earlier research at 36 ◦C [32]. While the values reported by Huang et al. [33] did not include dissolved CH4, the same trend was repeated for all trials at a given HRT, whereas in theory, the effluent methane content in the effluent should be the same in each case. The absence of steady-state conditions under which stable MLSS concentrations can be achieved makes it difficult to identify reliable values for normalised SMP from other similar studies. For higher-strength wastewaters in mesophilic conditions, however, the same trends can be seen as in the current work. During AnMBR treatment of thin corn stillage, the SMPMLSS of 0.016 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> CODrem L g−<sup>1</sup> MLSS at day SRT of 20 and 30 days was higher than that of 0.010 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> CODrem L g−<sup>1</sup> MLSS for a 50-day SRT [29], while for dairy wastewater 20 and 40-day SRT gave SMPMLSS values of 0.046 and 0.026 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> CODrem L g−<sup>1</sup> MLSS, respectively [31].

#### *3.2. Membrane Performance and Fouling Phenomena*

This work at 20 ◦C showed that shorter SRT resulted in better performance with respect to flux but indicated that there was no clear advantage in operating at SRT of <30 days (Figure 2b). In EP-1, however, a more rapid increase in TMP was needed to maintain the flux at a 45-day SRT in S2 and a 60-day SRT in S3 (Figure 2c), compared to that required in S4 at a 90-day SRT, despite considerably higher MLSS concentration in the latter (Figure 2d). This suggests that the onset of membrane fouling was slower at SRT of 30 and 90 days than at 45 or 60 days. In EP-2, the required TMP at a 45-day SRT in S2 was above that for a 90-day SRT in S4 until day 120, while the TMP in S3 began rising sharply from day 106, despite the introduction of a 30-day SRT on day 100, indicating that performance was better at the longer 90-day SRT. TMP in S4 with a 90-day SRT started to increase slightly from day 80, although it was only after day 120 that it began to rise at a rate similar to that seen previously in S3 with a 60-day SRT (Figure 2c). When SRT in S3 and S4 was reduced to 30 days, this led to a complete recovery in flux to the target value of 4 L m−<sup>2</sup> h −1 (Figure 2b), together with reductions in required TMP. Faster recovery was seen in S4, however, most likely because of the abrupt drop in MLSS compared to the slower transition in S3. It should also be noted that the SAnMBR were operated with in situ gas cleaning only through the experimental period, with no external or chemical cleaning. These responses thus not only demonstrate the considerable effect of SRT on membrane fouling but also show that fouling of this type can be at least partially reversed if an optimal SRT is applied.

The above results confirmed that the effect of SRT on membrane fouling is not simply due to the corresponding changes in MLSS concentration. Other research shows that fouling is also affected by components such as EPS, the production of which is strongly related to microbial growth and hence to SRT [3,19]. Research into the effects of SRT on fouling and overall performance in AnMBR is still scarce, however, and the interactions between multiple different factors are often unclear. Work by Huang et al. [33] with synthetic wastewater as well as prior work at 36 ◦C [32], found that membrane fouling was more severe at longer SRTs. When treating real sewage, however, Huang et al. [34] obtained the highest flux rate at a 60-day SRT, with more severe fouling at both longer and shorter SRTs.

#### *3.3. Mixed Liquor Characteristics*

#### 3.3.1. Capillary Suction Time

Results of CST measurements are given in Figure 4a and Table A1. CST values ranged between 94–569 s, and SRT had a strong effect on the mixed liquor's ability to retain moisture: samples taken from SAnMBR operating at shorter SRT released liquid much more readily than those at longer SRT. During the start-up period, when all four SAnMBR were operating as replicates at a 90-day SRT, the increase in CST was similar in all cases. Differences in CST began to appear in EP-1, with values generally higher at longer SRT. By the end of EP-1, however, CST values in S3 and S4 were similar at 438 and 430 s, respectively, despite a 25% difference in MLSS concentrations. This result indicates that the 60-day SRT in S3 represented the least favourable conditions for moisture removal and is reflected in the fall in membrane flux during this period despite continuing increases in TMP. After SRT in S3 was reduced to 30 days at the start of EP-2, the CST began to fall. The value of CST per unit of MLSS (CSTMLSS, Figure 4b) continued to rise, however, showing that normalised mixed liquor filterability was still dropping. At the end of EP-2, the CSTMLSS value in S3 was twice that in S4, although S3 was operating at one third of the SRT in S4.

At the start of EP-3, SRT was reduced to 20 days in S1 and to 30 days in S4, in conjunction with a sharp reduction in MLSS content in the case of the latter. While CST values in S1 fell gradually, reaching around 100 s by the end of the phase, the sharp MLSS reduction in S4 produced a much steeper decline in CST, which fell from over 550 to 259 s by day 207 then stabilised at around 250 s. CST values continued to fall in S3, while in

in Figure 2 and Table A2.

*Processes* **2021**, *9*, 1525 13 of 26

*3.3. Mixed Liquor Characteristics*  3.3.1. Capillary Suction Time

SRT in S4.

contrast, in S2 the CST rose significantly, stabilising at around 480 s, in parallel with a major fall in flux rate. trast, in S2 the CST rose significantly, stabilising at around 480 s, in parallel with a major fall in flux rate.

At the start of EP-3, SRT was reduced to 20 days in S1 and to 30 days in S4, in conjunction with a sharp reduction in MLSS content in the case of the latter. While CST values in S1 fell gradually, reaching around 100 s by the end of the phase, the sharp MLSS reduction in S4 produced a much steeper decline in CST, which fell from over 550 to 259 s by day 207 then stabilised at around 250 s. CST values continued to fall in S3, while in con-

Results of CST measurements are given in Figure 4a and Table A1. CST values ranged between 94–569 s, and SRT had a strong effect on the mixed liquor's ability to retain moisture: samples taken from SAnMBR operating at shorter SRT released liquid much more readily than those at longer SRT. During the start-up period, when all four SAnMBR were operating as replicates at a 90-day SRT, the increase in CST was similar in all cases. Differences in CST began to appear in EP-1, with values generally higher at longer SRT. By the end of EP-1, however, CST values in S3 and S4 were similar at 438 and 430 s, respectively, despite a 25% difference in MLSS concentrations. This result indicates that the 60 day SRT in S3 represented the least favourable conditions for moisture removal and is reflected in the fall in membrane flux during this period despite continuing increases in TMP. After SRT in S3 was reduced to 30 days at the start of EP-2, the CST began to fall. The value of CST per unit of MLSS (CSTMLSS, Figure 4b) continued to rise, however, showing that normalised mixed liquor filterability was still dropping. At the end of EP-2, the CSTMLSS value in S3 was twice that in S4, although S3 was operating at one third of the

**Figure 4.** Mixed liquor filterability in SAnMBRs during the whole experimental period: (**a**) CST; (**b**) CSTMLSS. Vertical dotted lines indicate the start of experimental phases. Values for control parameters (SRT and TMP) in each phase are shown **Figure 4.** Mixed liquor filterability in SAnMBRs during the whole experimental period: (**a**) CST; (**b**) CSTMLSS. Vertical dotted lines indicate the start of experimental phases. Values for control parameters (SRT and TMP) in each phase are shown in Figure 2 and Table A2.

The sharp MLSS reduction in S4 at the start of EP-3 resulted in an almost-immediate rise in CSTMLSS values, indicating an increase in normalised filterability of the mixed liquor. This was reflected in the observed recovery in flux, which allowed significant reductions in the required TMP (Figure 2). In contrast, although CST values in S3 levelled off The sharp MLSS reduction in S4 at the start of EP-3 resulted in an almost-immediate rise in CSTMLSS values, indicating an increase in normalised filterability of the mixed liquor. This was reflected in the observed recovery in flux, which allowed significant reductions in the required TMP (Figure 2). In contrast, although CST values in S3 levelled off and then began to decrease after SRT was reduced to 30 days in EP-2, it was a further 100 days before the CSTMLSS started to fall. A reduction in SRT evidently produces only a gradual change in MLSS characteristics, as time is required for any 'fouling substances' present or microbial species particularly responsible for their production to be eliminated from the reactor. Conversely, if the MLSS concentration is decreased abruptly at the same time as SRT is reduced (e.g., by replacing a proportion of the mixed liquor as in S4), MLSS filterability can improve almost instantaneously as fouling materials are removed or disrupted, and the enhancements in CST and flux occur much more rapidly.

The steady-state CST values of around 100, 480 and 190 s in S1, S2 and S3 operating at 20, 45 and 30-day SRTs, respectively, at the end of EP-3 was higher than those generally reported for aerobic membrane bioreactors (MBR) treating municipal wastewater, which are typically on the order of 10 s [54,55]. CSTMLSS values in S1, S2 and S3 ranged from 39 to 113 s L g−<sup>1</sup> MLSS, again higher than the normalised values of around 1–2 s L g−<sup>1</sup> MLSS for aerobic MBR. CST and CSTMLSS values for combined or co-mingled primary and secondary sewage sludges show more variability but typically range from 50–200 s and 2–10 s L g−<sup>1</sup> , with digestion sometimes leading to an increase in one or both parameters [56–58]. CST values of around 70 s were reported for MLSS from a full-scale SAnMBR treating sourceseparated blackwater in Spain [59], with the corresponding CSTMLSS of around 15–17 L g−<sup>1</sup> MLSS, both similar in scale to those found in the current work. Dong et al. [25] operated a pilot-scale AnMBR on screened municipal wastewater at 23 ◦C at SRT of 40, 70 and 100 days, and reported a reduction in CST at the shorter SRT, although the system did not reach steady-state in each case. A similar pattern of reduction in CST at shorter SRT was

also noted in a cross-flow AnMBR operating at 37 ◦C on thin corn stillage [29]. The authors reported that CSTMLSS values were similar at a 30 or 50-day SRT, and lower at 20 days, but again did not complete a full three SRT in all conditions.

#### 3.3.2. Frozen Image Centrifugation

Frozen Image Centrifugation was developed by the UK's Water Research Centre to allow assessment of sludge dewaterability and thickening characteristics [60]. Although it has never been widely adopted in the water industry, it provides the basis for approaches to the design and operation of dewatering facilities [61,62].

Results from FIC testing are shown in Figure 5a, with the centrifuged sludge volume expressed as a percentage of the original MLSS sample at one-minute intervals during the FIC test. Values normalised against the original MLSS concentration are shown in Figure A2 in Appendix A. Similar to the CST values, these FIC results show clearly that SRT had a significant effect on solid-liquid separation, with MLSS samples taken at short SRT much more readily separable than those at longer SRT. As can be seen from Figure 5a, the rate of separation in the first few minutes of centrifugation was also considerably more rapid in SAnMBRs at shorter SRT, and the final sludge volumes were achieved in a shorter time. At the end of EP-3 the final sludge volumes for FIC tests were 13%, 18% and 29% at 20, 30 and 45-day SRT in S1, S3 and S2, respectively, while the corresponding final centrifuged solids concentrations were 20.1, 19.6 and 14.9 mg L−<sup>1</sup> and separation rates in the first minute were 5.25, 3.75 and 0.50 mm min−<sup>1</sup> . After one minute of centrifugation, S1 and S3 were close to their final volumes, while S2 was still compacting at the end of the test. Similar values at a 30-day SRT were also found in S1 at the end of EP-2 and S4 at end of EP-3, indicating some replicability of results.

One benefit of the FIC test is its ability to provide information on several aspects of dewaterability, all of which are potentially important with respect to equipment design and operating costs. The potential for a high ultimate solids concentration, for example, may be less significant if the time required to achieve it is inordinately long [62]. FIC can also reveal details and nuances not readily observable in other tests. In the current work, for example, from day 154 onwards, three separate phases were seen in many of the test samples. These consisted of a clear supernatant, then a distinct cloudy white layer overlying the darker layer of centrifuged sludge solids (Figure 5b). Times at which these layers were present are indicated by solid bar-colouration in Figures 5a and A1, and it can be seen that they remained visible for longer in samples taken from the SAnMBRs operating at longer SRT. The proportion of the test run during which three phases could be distinguished increased in S2 at 45-day SRT, but fell in S3 where the SRT had been reduced from 60 to 30 days, corresponding to improvements in TMP and CST. The appearance of the third phase of this type was also noted in FIC tests on digestate from sugar beet pulp [63]: it was referred to as the light fraction and constituted a high molecular weight material most probably composed of EPS and/or soluble microbial products. Its presence was more evident during periods of foaming, which, such as poor dewaterability, is associated with high sludge viscosity and EPS content. After the change from 60- to 30-day SRT in S3 at the end of EP-1, the CST remained high while final sludge volumes in the FIC test fell. Normalised values of both parameters continued to rise for almost 100 days, however, before dropping quite sharply around day 211. In S4 where the change to 30-day SRT at the end of EP-2 was achieved by MLSS dilution, a sharp fall in CSTMLSS after day 189 corresponding to a rapid increase in the separation rate at the start of the FIC test, matched by a reduction in time needed to reach the final volume. The reasons for these delayed but sudden shifts are unknown: they may be related to the passing of a threshold concentration for some component in the MLSS, or even to quorum behaviour of MLSS and biofilm organisms, a topic of growing interest for AnMBR performance and fouling [10,64].

*Processes* **2021**, *9*, 1525 16 of 26

**Figure 5.** Mixed liquor centrifugation during experimental period: FIC test—Separated sludge volumes as % of original sample volume in (**a**) S1, (**b**) S2, (**c**) S3 and (**d**) S4, (**e**) separation phases during FIC test in centrifuged mixed liquor from S2 on day 204. Vertical dotted lines indicate start of experimental phases. **Figure 5.** Mixed liquor centrifugation during experimental period: FIC test—Separated sludge volumes as % of original sample volume in (**a**) S1, (**b**) S2, (**c**) S3 and (**d**) S4, (**e**) separation phases during FIC test in centrifuged mixed liquor from S2 on day 204. Vertical dotted lines indicate start of experimental phases.

Several researchers have investigated relationships between parameters such as CST or specific resistance to filtration (SRF) and MLSS dewaterability, membrane flux and fouling in both aerobic and anaerobic MBR [25,29,59,65]. CST is generally considered to be one of the most useful parameters because it shows a reasonably good correlation with membrane performance and is relatively easy to evaluate on-site [27,35]. In the current work, regression analysis was carried out to investigate any relationship between applied TMP (as an indicator of filterability and membrane fouling) with CST and FIC. However, in both cases, interpretation was made difficult by periods at maximum TMP. Although neither showed a very strong correlation, the relationship between TMP and CST was generally stronger than for TMP and FIC. This is as might be expected as the

water removal mechanism in a CST test where liquid is drawn through the MLSS is more similar to that in membrane bioreactor than in FIC. Several authors have noted the importance of using a type of test, which reflects the dewatering technology [66,67], and FIC testing has greater similarity to both sludge thickening and centrifugation. The value of expanding the range of tests currently utilised for sludge characterisation was emphasised by Spinosa and Doshi [68], and further work on FIC and other tests is needed to provide an enhanced understanding of different parameters and the significance of the relationships between them.

#### 3.3.3. Extracellular Polymeric Substances

Under steady-state conditions at the end of EP-3, bound EPS concentrations at 20, 30 and 45-day SRT were 378, 483 and 511 mg L−<sup>1</sup> in S1, S3 and S2, respectively. Specific values normalised against MLSS were around 165 mg g−<sup>1</sup> VSS for the SAnMBR at 20 and 30-day SRT, and slightly lower at 134 mg g−<sup>1</sup> VSS at a 45-day SRT. Other authors have also reported higher EPS content at shorter SRT: examples include treatment of municipal wastewater in a sidestream AnMBR at SRT ranging from 19–217 days, although steady-state operation was achieved only at the 19-day SRT [22]; and of synthetic wastewater at 31 ◦C in a AnMBR with a ceramic membrane at 100, 50 and 25-day SRT [69], although with full steady-state operation for the 25-day SRT only.

As can be seen in Figure 6a,b, for much of the experimental period, the specific concentrations of bound protein and carbohydrate were highest in S1 operating at a 30-day and 20-day SRT. When SRT in S3 was reduced from 60 to 30 days at the start of EP-2, however, specific concentrations increased in this reactor, with the carbohydrate content rising first. Specific protein and carbohydrate concentrations also increased in S4 after the SRT was reduced to 30 days at the start of EP-3. At the end of the experimental period, both the specific bound protein and the bound carbohydrate concentrations were lowest in S4 at a 45-day SRT, and higher in the SAnMBR at shorter SRT. Huang et al. [33,34] found higher specific protein contents in EPS at a 30-day SRT than at longer SRT, and suggested that this might produce larger flocs with a lower membrane fouling propensity. They also noted that long SRTs were associated with smaller median particle size, as the lower EPS content reduces flocculation, and hence promotes more rapid fouling. These suggestions appear to be consistent with the outcomes of the current work.

There was also evidence of a relationship between soluble EPS content and membrane performance at longer SRTs. As can be seen in Figure 6c,d, the specific carbohydrate and protein concentrations of soluble EPS in S3, which had shown the most rapid fouling of all at a 60-day SRT, were generally higher than in the other SAnMBRs. After SRT in S3 was reduced to 30 days from the start of EP-2, the specific soluble protein and carbohydrate contents continued rising until around day 205 when they finally reduced. At that point, the CST value also fell, the rate of separation in the FIC test increased, and the flux began to recover. Conversely, towards the end of EP-3, the specific soluble protein and carbohydrate contents in S2, operating at a 45-day SRT, increased relative to the other SAnMBRs, which at that point were all running at shorter SRTs. This corresponded to the period in which CST values in S2 began to rise and flux to reduce, despite an increase in TMP, to 9.8 kPa. This reactor reached the end of the experiment with the highest soluble protein and carbohydrate contents (specific and absolute), highest TMP, highest CST and FIC values, lowest separation rate in the FIC test and lowest flux.

Huang et al. [33] found that the concentration of soluble microbial products was inversely related to SRT, while higher SMP carbohydrate and protein contents at longer SRT resulted in higher fouling rates. In contrast, when real municipal wastewater feed was used, the minimum specific and absolute values for SMP carbohydrate and proteins were found at a 60-day SRT [34]. This was attributed to incomplete substrate degradation at the 30-day SRT, and an increased concentration of residual cell materials at the 90-day SRT. Although EPS and MLSS concentrations were lowest at the shortest SRT, this was insufficient to counteract the fouling effect of the higher SMP content. Soluble microbial

products were not measured in the current study, but this explanation may be partly supported by the results of this experiment and of the previous work at 36 ◦C [32], which showed no further enhancement in membrane performance at SRT of <30 and <25 days, respectively. Laspidou and Rittmann [70] proposed that under some conditions soluble EPS and SMP may be similar, although other researchers were unable to confirm this [71]. Trends in bound and soluble EPS observed in this work are consistent with those in other studies carried out under steady-state conditions, where the consensus is that bound EPS tends to be high at shorter SRTs while microbially-induced SMP concentrations tend to decrease [33,34]. Given that soluble EPS also decreased at shorter SRTs in the current work, it can be concluded that even if it is not identical to SMP, its relationship to SRT and its effects on membrane fouling are closely similar. *Processes* **2021**, *9*, 1525 19 of 26

**Figure 6.** EPS composition during experimental period: (**a**) bound protein, (**b**) bound carbohydrate; (**c**) soluble protein, (**d**) soluble carbohydrate. Vertical dotted lines indicate start of experimental phases. Values for control parameters (SRT and TMP) in each phase are shown in Figure 2 and Table A2. **Figure 6.** EPS composition during experimental period: (**a**) bound protein, (**b**) bound carbohydrate; (**c**) soluble protein, (**d**) soluble carbohydrate. Vertical dotted lines indicate start of experimental phases. Values for control parameters (SRT and TMP) in each phase are shown in Figure 2 and Table A2.

The current study showed changes occurring over different timescales, which could be categorised as follows: (i) those which happen slowly, such as stabilisation of MLSS concentrations after a change in SRT; these changes cannot easily be accelerated as they

EPS content was not measured in the previous work with the same wastewater at 36 ◦C [32], and thus the effect of temperature in the current study is unknown. EPS and soluble microbial products normalised to MLSS content were reported to increase with decreasing temperature in a flat-sheet SAnMBR fed on a synthetic municipal wastewater at 25, 15 and 10 ◦C [14]. The ratio of protein to carbohydrate also rose with decreasing temperature, leading to higher rates of fouling at low temperatures. The applied SRT was not reported, however, thus it is likely that this was high and determined only by the withdrawal of samples for analysis. Similar trends in EPS and soluble microbial product concentrations were observed in two hollow fibre SAnMBR fed on synthetic wastewater at 25 and 35 ◦C with an SRT of 370 days [72], and in a hollow fibre SAnMBR fed on municipal wastewater at 25, 20 and 15 ◦C at SRT of 93.9, 40.3 and 20.7 days, respectively [73], although the operating periods under each set of conditions were less than 3 SRT. Kong et al. [1] operated a 5 m<sup>3</sup> pilot-scale SAnMBR with a HFM unit on municipal wastewater for over 200 days at 25 ◦C. No change in EPS concentration was observed when HRT was varied between 6–24 h, but SRT was not reported.

EPS and soluble microbial products were characterised in flat-sheet SAnMBR treating low-strength synthetic wastewater at 25 ◦C and operated at various OLR [19]. Properties were linked to membrane fouling mechanisms, but the SRT of the system was near-infinite as no MLSS was discharged other than for sampling.

While some differences in EPS production and composition were seen in the current study, these were not sufficiently marked enough to show a clear correlation between SRT and observed fouling behaviour. Nonetheless, the findings are consistent with those reported elsewhere. While there may be a specific range of SRT for each system in which membrane performance can be optimised, however, the effect of EPS concentrations and compositions on fouling in AnMBR is still not well understood, as a function of SRT or other operating parameters. For this reason, further work may be needed to assess the interaction between operational variables, MLSS characteristics and membrane fouling.

#### 3.3.4. Biomass Growth and Kinetics

As well as having a significant impact on mixed liquor characteristics, the MCRT or SRT controls microbial growth rates, and thus potential growth yield. As previously seen for the work carried out at 36 ◦C [32], in the current experiment, the biomass yield appeared to show a sharp increase at each reduction in SRT. This reflected the fact that, as noted in the Methods section, changes in stored biomass were not taken into account to eliminate major variations. The apparent increases in yield were followed by a gradual decrease as conditions stabilised. The stable value was taken as the representative yield for each corresponding SRT and was equal to 0.131 <sup>±</sup> 0.008 g VSS g−<sup>1</sup> CODrem at a 20-day SRT, 0.124 <sup>±</sup> 0.012 g VSS g−<sup>1</sup> CODrem at a 30-day SRT, and 0.114 <sup>±</sup> 0.004 g VSS g−<sup>1</sup> CODrem at a 45-day SRT. Biomass yields for the 60- and 90-day SRTs are not reported here as steady-state operation of the SAnMBRS was not achieved for these conditions, and thus representative values were not available.

The above results confirmed that shorter SRT gave higher biomass yields, in agreement with prior work at 36 ◦C [32]. These outcomes provide further support for the idea that at shorter SRT a higher proportion of the available carbon is directed into biomass growth rather than methane production, thus reducing the substrate SMP. Conversely, at longer SRT, lower growth rates and increased hydrolysis of MLSS leads to reduced biomass yields and higher SMP values [24]. In a full-scale system, a shorter SRT with a higher biomass yield would be linked to larger volumes of waste sludge for disposal and potentially to a greater requirement for TE supplementation, both leading to higher operating costs.

These low yields are typical of those reported elsewhere for similar systems, including 0.11 g VSS g−<sup>1</sup> COD for low-strength synthetic wastewater treatment at 25 ◦C and nearinfinite SRT [74]. Ji et al. [73] found sludge yields of 0.12, 0.19 and 0.388 g VSS g−<sup>1</sup> CODrem with real municipal wastewater at SRT of 93.9, 40.3 and 20.7 days; but the effect was confounded by accompanying changes in temperature from 25 to 20 and 15 ◦C and the

system did not operate for 3 SRT in each set of conditions. Sludge yields of 0.07–0.11 g VSS g <sup>−</sup><sup>1</sup> CODrem were reported using the same wastewater at 25 ◦C for SRT from 65 days to near-infinite with HRT from 6–12 h and OLR between 0.7–1.5 g COD L−<sup>1</sup> day−<sup>1</sup> [28].

The current study showed changes occurring over different timescales, which could be categorised as follows: (i) those which happen slowly, such as stabilisation of MLSS concentrations after a change in SRT; these changes cannot easily be accelerated as they are both growth-mediated and affected by washout rate, although interventions such as partial removal of MLSS may reduce the time needed to approach stable values. (ii) Those which can happen more rapidly, such as a response to TE addition or to other factors affecting the microbial population; as these are metabolically mediated, they can trigger an immediate response through stimulation or inhibition. (iii) Those which may happen rapidly but after a delay, such as the observed sharp changes in CST and FIC values following reductions in SRT, which could be metabolically or physico-chemically determined in relation to threshold concentrations. The factors that cause these changes may also be interdependent, and hence to allow full evaluation of their effects on system performance and mixed liquor characteristics long-term operation is advisable, wherever possible for at least 3 SRT under a given set of operating conditions.

#### **4. Conclusions**

Extended operation of four SAnMBRs at 20 ◦C on a low-to-intermediate strength substrate with a high suspended solids content was conducted at different SRTs. This enabled accurate determination of flux rates at specific SRT, accompanied by evaluation of COD removal efficiencies; estimation of biomass yields; a COD-based mass balance; physical characterisation of mixed liquors using CST and FIC tests; and analysis of EPS concentration, type and composition. The results showed that SRT had a considerable effect on flux rates, with shorter SRT allowing enhanced membrane performance and improved mixed liquor filterability, at a higher bound EPS content but with lower soluble EPS concentrations in the MLSS. Operation at shorter SRT led to a reduction in specific methane productivity and in COD removal rates, accompanied by higher biomass yields. Whilst no further enhancement of membrane performance was found at SRT of <30 days, operation at a 60-day SRT resulted in more rapid onset of membrane fouling and declining performance than at a 90-day SRT. Reduced COD removal rates at shorter SRTs were probably due to the increased washout of essential trace elements caused by the higher biomass turnover: this was supported by a rapid recovery in COD removal efficiency after TE supplementation was carried out. COD removal efficiencies remained slightly lower at shorter SRTs, however, suggesting that the lower available biomass concentration may also have affected this parameter. Overall COD removal efficiencies achieved after TE addition were very close to those seen in an earlier study using the same substrate at 36 ◦C, and the effect of the lower operating temperature on this parameter was, therefore, considered to be negligible. CST values gave some indication of changes in membrane performance, while frozen image centrifugation provided additional insights into MLSS properties and fouling behaviour, with three separate phases clearly visible at longer SRT. The ability of the FIC test to identify several parameters such as final solids concentration and rates of solid separation may also make it especially appropriate for assessing sludge dewaterability in gravity thickening or centrifugation, and further exploration of this approach is recommended. Operation over the full experimental period, without chemical or external cleaning, not only demonstrated the effects of SRT on performance parameters but also indicated that membrane fouling could be at least partially reversed if an optimal SRT is applied. These results indicated that responses to a change in SRT may be significantly delayed, probably as a result of the different timescales on which growth-related, metabolic and physicochemical and/or the various interactions between them. Thus confirming the importance of long-term operation to allow full evaluation of system performance and mixed liquor characteristics under steady-state conditions. Together with earlier findings from operation at 36 ◦C, this work confirmed that there are potential trade-offs to be made between

membrane performance, specific methane productivity and sludge yields when selecting a suitable SRT for AnMBR systems of this type.

*Processes* **2021**, *9*, 1525 21 of 26

**Author Contributions:** Conceptualization and methodology, S.P.-R., S.H., C.J.B.; experimental work, S.P.-R.; data analysis and interpretation, S.P.-R., S.H., C.J.B.; writing—original draft preparation, S.P.-R.; writing—review and editing, S.H., C.J.B.; supervision, C.J.B., S.H.; funding acquisition, C.J.B. All authors have read and agreed to the published version of the manuscript. **Author Contributions:** Conceptualization and methodology, S.P.-R., S.H., C.J.B.; experimental work, S.P.-R.; data analysis and interpretation, S.P.-R., S.H., C.J.B.; writing—original draft prepara-

**Funding:** This research was supported by the Mexican National Council on Science and Technology (CONACYT), the Faculty of Engineering and Physical Sciences at the University of Southampton, and the BBSRC ERA-Net AmbiGAS project BB/L000024/1. tion, S.P.-R.; writing—review and editing, S.H., C.J.B.; supervision, C.J.B., S.H.; funding acquisition, C.J.B. All authors have read and agreed to the published version of the manuscript. **Funding:** This research was supported by the Mexican National Council on Science and Technology

(CONACYT), the Faculty of Engineering and Physical Sciences at the University of Southampton,

**Institutional Review Board Statement:** Not applicable. and the BBSRC ERA-Net AmbiGAS project BB/L000024/1.

**Informed Consent Statement:** Not applicable. **Institutional Review Board Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are openly available on DOI: 10.5258/ SOTON/D1950. **Informed Consent Statement:** Not applicable. **Data Availability Statement:** The data presented in this study are openly available in [to be added

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. if paper is accepted for publication]. **Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **Appendix A**

**Appendix A** 

**Figure A1.** Flowchart for EPS extraction procedure. **Figure A1.** Flowchart for EPS extraction procedure.



6. Lim, K.; Evans, P.J.; Parameswaran, P. Long-term performance of a pilot-scale gas-sparged anaerobic membrane bioreactor under ambient temperatures for holistic wastewater treatment. *Environ. Sci. Technol.* **2019**, *53*, 7347–7354. (→) Variable trend: initial → middle → final; (±) Stable performance: One standard deviation to show spread of data from the average value under stable performance; (\*) Daily average.

7. Ozgun, H.; Dereli, R.K.; Ersahin, M.E.; Kinaci, C.; Spanjers, H.; van Lier, J.B. A review of anaerobic membrane bioreactors for municipal wastewater treatment: Integration options, limitations and expectations. *Sep. Purif. Technol.* **2013**, *118*, 89–104.


**Table A2.** Experimental results; SRT effect on: CH<sup>4</sup> content in biogas, SMP, MLSS, MLVSS, CST and pH.

(→) Variable trend: initial → middle → final; (±) Stable performance: One standard deviation to show spread of data from the average value under stable performance; (\*) Normalised to total biogas content in sample (i.e., neglecting air introduced dissolved through feed); (\*\*) Takes into account the methane dissolved in the effluent.

#### **References**


## *Article* **Exploring Farm Anaerobic Digester Economic Viability in a Time of Policy Change in the UK**

**Angela Bywater 1,\* and Sigrid Kusch-Brandt 1,2,\***

	- 89075 Ulm, Germany

**Abstract:** The combination of a post-Brexit agricultural policy, the Global Methane Pledge announced during the last United Nations Climate Change Conference in Glasgow (COP26), and urgency of meeting climate goals means the UK has a unique opportunity to create an exemplar through recognition of the benefits of small-scale farm anaerobic digesters that valorise on-site wastes for renewable electricity and heat, cushioning agri-businesses against energy perturbations. To explore economic viability of farm-based biogas production, combinations of support levels, energy prices, capital cost, internal rate of return (IRR), and digestate value were analysed, employing a 550-cow dairy farm with access to other agricultural wastes. A 145 kWe system utilising 100% of CHP electricity (grid value: £0.1361 per kWh) and 70% of the heat (heating oil value: £0.055 per kWh) could achieve an IRR above 15.5% with a median electricity tariff of £0.1104 per kWh at a heat tariff from £0.0309 to £0.0873 per kWh thermal. Under a subsidy-free regime, the same system could achieve a 10% IRR with electricity prices in the range £0.149 to £0.261 per kWh. High fertiliser prices could increase digestate value, further improving viability. With late-2021 high energy prices, the technology approaches subsidy-free viability, but uptake is unlikely unless wider environmental and societal benefits of on-farm systems can be explicitly valued.

**Keywords:** agricultural wastes; biogas production; anaerobic digestion costs; economic viability; UK policy; Brexit; feed-in tariff; renewable heat incentive

### **1. Introduction**

In the last decade, financial incentives based on energy production have created significant growth in anaerobic digestion (AD) installations in the United Kingdom (UK), with the number of agricultural plants increasing from 25 in 2010 to 344 in 2020 [1]. AD can capture uncontrolled greenhouse gas (GHG) emissions from the biodegradation of organic wastes and from farm management activities [2,3], making a useful contribution to overall GHG reductions (potentially by 6% for the UK, according to the Anaerobic Digestion and Bioresources Association [4]), thus helping the UK to meet its Paris Agreement and COP26 Global Methane Pledge commitments. Farm AD, particularly using on-site/local wastes for biogas generation and recovery of digestate, has myriad benefits beyond renewable energy generation alone: it can reduce GHG emissions [3,5–7], improve soil organic matter [8], facilitate improved nutrient management [9–11] thereby reducing the need for artificial fertiliser [12,13], kill pathogens and weed seeds if appropriately applied [14–16], provide opportunities for skilled rural employment [17,18], and create additional revenues in rural areas [19].

Clearly, a breakthrough in deployment of anaerobic digestion does not only depend on technical aspects and is strongly moderated by incentives provided [20]. Growth in the UK's on-farm AD industry started with the introduction of the Pollution Control Grants in the late 1980s, then stagnated when these were withdrawn in the mid-1990s [21,22]. The

**Citation:** Bywater, A.; Kusch-Brandt, S. Exploring Farm Anaerobic Digester Economic Viability in a Time of Policy Change in the UK. *Processes* **2022**, *10*, 212. https://doi.org/ 10.3390/pr10020212

Academic Editor: Elsayed Elbeshbishy

Received: 31 December 2021 Accepted: 17 January 2022 Published: 24 January 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

digesters were relatively small in size, usually under 350 m<sup>3</sup> , and the biogas was primarily used in boilers for heating [21,23]. During the eight years prior to the introduction of the 2010 Feed-In Tariff (FIT), only seven projects [24] running on farm feedstocks had been commissioned under the 2002 Renewables Obligation system, which was designed for production of renewable electricity.

The AD FIT incentivised electricity production from biogas via combined heat and power (CHP) units. It was followed by the introduction in 2011 of the Renewable Heat Incentive (RHI), which was designed to encourage heat use and biomethane production. Since CHP has relatively low electrical efficiencies of approximately 30% for smaller units and up to 45% for larger ones, the RHI introduced the possibility of improved AD economics if most of the CHP heat produced could be beneficially utilised at a relatively low cost. The policy, however, primarily encouraged the introduction of biomethane-to-grid plants, with 108 biomethane plants operational when it closed in March 2021, compared to seven heat-only plants [1].

Incentives that recognise only the energy contributions of AD have encouraged the construction of larger plants with a greater proportion of crop inputs. Agricultural CHP plants over 250 kWe comprise 68% of the UK total at an average of 1 MWeeq, with agricultural biomethane plants averaging 789 m<sup>3</sup> biomethane per hour or approximately 3.1 MWeeq [1]. Concern over using land to provide feedstock for crop-only digesters led to the requirement for sustainability criteria to encourage the use of 'waste' AD feedstocks and to limit the amount of crop material fed to digesters [25], which can result in indirect land use change when land for biofuel crops displaces that used for food or feed. The sustainability criteria have been built into the Green Gas Support Scheme (GGSS), which was introduced in 2021, funded by a gas consumer levy and specifically designed for AD biomethane grid injection. The policy aims to support larger biomethane plants, with the highest out of three tariffs being paid to installations injecting up to 60,000 MWh year−<sup>1</sup> of biomethane into the gas grid (approximately 750 m<sup>3</sup> biomethane per hour).

Most farmers are unwilling or unable to find sufficient feedstocks or to raise the significant investment required for such a large plant [26], due to factors such as uncertainty about the future, where best to target new investments and concerns over the underlying profitability of their farm businesses. Thus, there remains an AD policy gap for UK farmers who wish to valorise their on-site feedstocks by installing smaller, less capital-intensive AD systems. Additionally, because on-farm organic wastes such as slurries and manures contain less energy than an equivalent tonnage of crop biomass, the challenge is to introduce a mechanism that can make such systems a sufficiently attractive economic proposition for farm businesses to invest in.

In their life cycle assessments on AD systems, Mesa-Dominguez et al. [27] and Styles et al. [28] noted that the greenhouse gas balances of AD plants improved by maximising waste and minimising crop inputs, utilising any CHP heat produced and covering digestate stores. Mesa-Dominguez et al. [27] concluded that energy-based incentives do not create the most sustainable deployment of AD. To illustrate this context, it is worth highlighting that a cubic metre of digester space costs the same whether it is fed on 1 m<sup>3</sup> of cow slurry at 23 m<sup>3</sup> biogas tonne−<sup>1</sup> of fresh feedstock or 1 m<sup>3</sup> of maize at 220 m<sup>3</sup> biogas tonne−<sup>1</sup> of fresh feedstock, but the waste slurry produces only one-tenth of the energy of the crop. The slurry digester, however, has better overall environmental credentials, mitigating 1449 kg CO2eq tonnne−<sup>1</sup> of dry matter [27], because it utilises material that would otherwise create greenhouse gas emissions through its production, storage, and handling. Mesa-Dominguez et al. [27] concluded that public FIT/RHI funding should integrate consequential life cycle assessment (CLCA) and eco-systems services criteria into sustainability criteria for the most effective climate protection. In line with this, a CLCA conducted by Beausang et al. [29] for co-digestion of grass silage and cattle slurry found lower proportions of grass silage to be more sustainable, thus highlighting the environmental benefits of slurry-based farm AD plants.

Sustainability criteria were designed into the UK's non-domestic RHI and GGSS from the outset, and subsequently added to the accreditation process for the FIT from April 2017. With the 2020 Brexit withdrawal of the UK from the European Union (EU), spelling the end of EU agricultural support mechanisms in the UK, an opportunity arises to include anaerobic digestion in post-Brexit environmental schemes on UK livestock farms, particularly dairy farms, where it can make significant positive impacts [30]. The policy dilemma, therefore, is to address how on-farm AD, which utilises local or on-site wastes can be encouraged in a cost-effective way, particularly to mitigate the GHG emissions associated with livestock farming, thus valuing the technology for its wider ecosystem services deliverables (such as nutrient recycling, pollution mitigation, GHG reduction), rather than simply the production of energy [31]. While this is particularly acute in the UK due to current changes in regulation and support, similar challenges exist in many countries trying to encourage the implementation of AD by using financial support mechanisms that directly influence the digester size [32,33].

A unique opportunity to advance on-farm AD in the UK is currently given for policymakers. This opportunity is characterised by the following: a prevailing policy gap for small farm-scale AD to valorise on-site materials such as slurries and manures; the need to introduce a completely new agricultural policy that values the environment; increased impetus to meet climate targets; and high current fossil energy prices, which also means elevated prices for artificial fertilisers. Such prices favour the on-site production of renewable electricity and heat, including through AD.

In the light of COP26 aspirations, the introduction of new post-Brexit agricultural support mechanisms and a current UK policy vacuum for small farm AD that utilises on-site/local waste feedstocks, the aim of this study is to determine what level of policy support might be required against the high energy prices such as those that characterize the end of 2021. The economics of implementing an on-farm AD installation under current and potential future UK policy regimes are explored, using the Leckford Estate as an example. An economic model has been developed for this study, underpinned by AD process data from the University of Southampton's mass and energy balance model ADAT (Anaerobic Digestion Assessment Tool), available at http://borrg.soton.ac.uk/resources/ adat (accessed on 18 January 2022). Combinations of heat and electricity tariff support levels, capital cost, and digestate value that could make farm AD projects viable are explored.

#### **2. Materials and Methods**

#### *2.1. The Leckford Estate as Site under Study*

To ensure that the situation in practice is adequately captured and the farm perspective appropriately considered, this study is based on analysing the Leckford Estate by way of an example. The estate, known as the 'Waitrose Farm', covers approximately 1600 hectares and is part of the John Lewis Partnership (JLP). In line with the ethos of Waitrose and the JLP, the estate is 'passionate about sustainable farming' [34] and the production of quality food. With an engaged, environmentally aware, and supportive management, availability of on-site wastes primarily from livestock operations and further feedstocks from diversification activities, AD is a technology that should be accessible to this and similar farm businesses.

At the Leckford Estate, an AD case study (not published) was originally carried out in 2017 by the first author of this paper. For the purpose of this current research, the situation at the Leckford Estate has been revisited because of the unique opportunity that the combined conditions under the current post-Brexit policy changes present.

#### *2.2. Feedstock Pre-Assessment and Selection*

To enable an assessment of the feasibility of AD at the Waitrose Farm, the various agricultural operations were examined to identify a range of potential digester feedstocks, their volumes, and seasonal availability.

The primary feedstock for the proposed AD installation is slurry and farmyard manure from the 550-cow Holstein/Friesian dairy herd. This herd size is larger than the UK average of 155 animals [35], the EU average of 45 [36], and the US average of 297 animals [37], although it should be noted that such average figures hide large variations in herd sizes, particularly in the US and UK.

The cows are housed for part of the year (1 November–1 April), with access to grazing for at least 120 days a year. Approximately 50 of these cattle will not be in milking, so will be housed on straw yards, producing 75% farmyard manure (FYM) and 25% slurry. The remaining 450 will be producing slurry. A further 50 cows will be in the 'dry cow yard' and will also be producing FYM during the winter housing season. In the summer, half of these will be out to grass. Of the 500 cows normally in the main accommodation shed, on average, 300 cows will spend 10 h per day at grass through this period, so 10/24 of the manure from these animals will not be collected.

Other farm operations generating waste feedstocks include free-range chickens; apple and pear orchards; grape vines; beef cattle and sheep, which are permanently out to pasture; a mushroom growing operation; and a cold-pressed rapeseed oil production facility. Further enterprises that produce small amounts of organic material include a golf course, a nursery, a farm shop, a lodge/campsite, a guest house, and a café.

In order to provide a conservative biogas production figure, control capital expenditure and minimise potential process problems, a number of potential feedstocks were excluded from the final calculation. These included feedstocks that were:


Although there is an arable part of the business and energy crops could be a potential choice to underpin biogas production at agricultural AD installations [38], the John Lewis Partnership did not wish to introduce purpose-grown crops as an AD feedstock, thus maintaining their key focus on food production. Maize silage was, however, considered for inclusion as an option to improve the economic case for the digestion plant and/or level out seasonality.

#### *2.3. AD Plant Site Selection and Request for Supplier Quotes*

Before contacting potential AD plant suppliers active in the UK market, a suitable site for the plant installation was identified, based on the following main criteria: a large open area with suitable road access and proximity to the main feedstocks, existing feedstock storage, a water supply, and an electricity grid connection.

Three types of AD technology supplier were contacted to provide 'budget' quotes based on the digester location and feedstock types and volumes. These were: a supplier who designs simple cost-effective farm-based digesters (CAPL), one whose mid-range farm digester offering could provide automated de-gritting (CAPM), and a third who provides mainly industrial digesters (CAPH). As all suppliers provided quotes in British pound sterling (GBP) (£), monetary figures are shown in this currency.

Best efforts were made to ensure that the three quotes encompassed a similar scope of supply. It is important to note that, at this stage, suppliers use their own feedstock biogas production values for CHP and digester sizing. Whilst main components such as the tank and pipework, CHP container, feed system, control system, and any associated pumps will be included in a quote, many site-specific items may be approximate estimates only or may be considered out of scope at the budget quotation stage. These items could include planning, professional fees, road access, security, hardstanding, water supply, drainage, feedstock storage (e.g., silage clamps), rainwater/effluent catchment, data cabling, electricity supply, cost of electricity grid access, operator amenities, modifications to buildings or slurry/materials handling systems, permitting, compliance (e.g., bunding), commissioning

(including cost of initial heating), digestate storage, and separation equipment. Therefore, quotes received were evaluated against literature data to check for reliability and whether they were realistic.

#### *2.4. Digester System Modelling Using the ADAT Tool*

The University of Southampton's ADAT mass and energy balance modelling tool [39] was used for the digester modelling. While several biogas production models are available online and offline, the ADAT tool, developed under the leadership of the University of Southampton in the context of various academic projects and support from the IEA (International Energy Agency) Task 37 UK, has a number of advantages, such as transparency about data assumptions and calculations, inclusion of fertiliser values in the underlying database, the option to easily add feedstocks as required by the specific user case, and availability of a comprehensive manual. The ADAT tool has been used in several studies [40–43].

Feedstock volumes were entered into the ADAT model. Where specific feedstocks were not available within the model, values for total solids, volatile solids, and methane production were taken from a range of other sources and added to the model as a 'userdefined' feedstock.

Two scenarios for energy use were considered:


Table 1 shows the parameters used in the ADAT model for a steel construction digester with integral gas storage operating at a mesophilic temperature of 37 ◦C located in the Southampton climatic region. Default ADAT values [39] were used for all parameters unless otherwise indicated.


**Table 1.** Main parameters for the farm anaerobic digestion system from the ADAT modelling tool.

The ADAT modelling tool was used to calculate the total feedstock volume, the digester size based on loading rate, the retention time, biomethane production, digestate output, CHP size, CHP electricity production, digester parasitic electrical energy, and net heat output after allocating required digester heating. The ADAT model can also calculate embodied energy [39], but as the aim of the study was to establish the potential economic viability of an on-farm digestion system rather than its carbon footprint, this was not considered.

#### *2.5. Economic Modelling*

Based on the ADAT model output and the three supplier quotes, a purpose-built economic model was created in Excel to explore the economic outcomes of the two energy options (CHP-E scenario, CHP-EH scenario). The economic model integrates the following three key parameters: (1) capital expenditure; (2) energy production expectations; (income and savings); and (3) operation and maintenance costs as described below.

#### 2.5.1. Capital Expenditure

Budget digester cost from the suppliers is the main cost element considered here. It is also possible to include other costs such as planning, grid connection, consultants, digestate storage, groundworks, water supply, commissioning boiler fuel, permitting, and where applicable, heat meters, heat pipe, and trenching. In order to maintain a similar scope of supply and to minimise any skewing of budget costs, none of these other costs were specifically included in the economic viability calculations in the current study.

For the CHP-EH scenario, the following costs were added: heat pipe at £100 m−<sup>1</sup> , pipe trenching at £8 m−<sup>1</sup> (based on Estate costs), heat meters at £1000, plus a contingency margin of 3.7% of these total costs.

#### 2.5.2. Energy Production

Figures derived from the ADAT model were included here: gross energy production, CHP size, CHP electrical efficiency, load factor, and electricity production based on the load factor.

For the CHP-EH scenario, the following additional parameters were considered: CHP heat efficiency and heat production based on the load factor.

#### 2.5.3. Income and Savings

In order to ascertain what savings might be achieved through electricity and heat generation, net heat and electrical energy values are calculated after deduction of ADATderived digester parasitic heat and electricity values, respectively. The model allows a percentage of the net electricity to be used on site to displace bought-in electricity, with the remainder exported at £0.0557 kWh−<sup>1</sup> [44]. Where excess CHP heat can be beneficially utilised on site, the model allows a percentage of heat to displace fossil fuels such as heating oil, liquefied petroleum gas (LPG), or natural gas. In this case study, heating oil was displaced.

Although the UK no longer has incentives designed specifically for heat or electricity production from biogas, historical tariff values exist, and provide an indication of the support levels that government deems practicable in terms of both budget and levels of deployment. Therefore, the model utilised selected tariff levels from the Feed-in Tariff (FIT) for electricity production [44] and the Renewable Heat Incentive (RHI) for heat production [45]. FIT levels were banded in order to provide smaller systems with greater tariff support. Between the FIT introduction in April 2010 and September 2011, these systems would have fallen into a sub-500 kWe CHP band; thereafter, they would be included in the sub-250 kWe CHP band.

The FIT tariff levels currently available [44] are higher than they were at the inception of the scheme, since they are index-linked and therefore adjusted annually in line with the UK Retail Prices Index (RPI). AD FITs for the relevant CHP capacity band were extracted and analysed for the highest (FITH) and lowest (FITL) values. To facilitate fiscal management of this fixed budget, the FIT scheme also included a degression mechanism that reduced tariffs as deployment increased, meaning that there was a significant difference between these two figures, so a median figure was also calculated (FITM).

A similar exercise was carried out to ascertain the minimum and maximum RHI tariff levels, RHIL and RHIH, respectively. Although the RHI scheme also included a degression mechanism, the differential between these two figures was not large, so a median figure was not calculated. Both tariff levels are given in Table 2.

Nutrients in digestate have a potential value because they can reduce or eliminate the need for fossil fertilisers [46]. The model calculates potential savings in fertiliser costs associated with digestate utilisation, but inclusion of these is optional since this value has historically been hard to realise.

Fertiliser values in terms of nutrients contained can be based either on ADAT figures [39] or on the RB209 farm nutrient management guide standard values for farm sourced digestate [47]. For the purposes of the case study, RB209 figures of 3.6 kg tonne−<sup>1</sup> (total N), 1.7 kg tonne−<sup>1</sup> (P2O5), and 4.4 kg tonne−<sup>1</sup> (K2O) [47] were used since reliable nutrient data for the user-defined feedstocks in the ADAT model could not be ascertained.

Current market prices for equivalent fossil fertilisers were obtained [48] in order to provide a value for each kilogram of each nutrient. These fertilisers were ammonium nitrogen at 34.5% N content, triple super phosphate (TSP) at 46% P2O<sup>5</sup> content, and muriate of potash at 60% K2O content. The sum of these individual nutrient prices represents the displaced cost of synthetic fertiliser.

**Table 2.** Economic model parameter values.


#### 2.5.4. Operation and Maintenance (O&M) Costs

As outlined in Table 2, these include AD operator wages, as well as annual permitting, insurance, and rates. Digester and CHP maintenance costs are also included here.

#### 2.5.5. Economic Viability Calculations

An internal rate of return (IRR) is calculated based on the capital cost, total income/savings, total operational/maintenance costs, and a Retail Price Index (RPI) of 102%. It is assumed that short-lived equipment will be replaced every 5 years at 4% of the original capital expenditure and that, additionally, 20% of the original capital expenditure will be spent every 10 years [23]. The IRR is calculated over 20 years.

The internal IRR target for JLP in 2017 was 16%, although, the company recognised the wider environmental benefits of projects such as an AD installation, and so was prepared to consider lower returns.

Based on the ADAT model output and the three supplier quotes, the economic model was then used to explore the economic outcomes of the two energy options (CHP-E scenario, CHP-EH scenario) with the aim of answering the following questions:


The capital cost of digesters has not historically decreased, particularly for large systems that are complex and expensive civil engineering projects. Nevertheless, precedent exists for small digesters, which are modular and off-the-shelf, with minimal expensive site work, thus reducing capital costs [21].

#### **3. Results and Discussion**

#### *3.1. Methane Generation and Digester Design*

Table 3 shows the feedstocks as entered into the ADAT model and the resultant methane production estimates; where values were not available within the model, methane yields were taken from the sources shown in the notes. Feedstocks themselves can be variable within a single farm or even within a silage clamp, for example, so some suppliers recommend lab tests be carried out to ascertain site-specific values. Feedstock characteristics for mushroom waste were estimated based on typical energetic nutritional values for raw mushrooms with a 50% reduction for spoilage to give a conservative estimate of methane yields. Thus, this waste accounted for approximately 1.4% of the total methane production. For regulatory purposes, all feedstocks are likely to be considered as wastes, thereby meeting incentive scheme sustainability criteria, although interpretation across different schemes varies in practice.



<sup>1</sup> Default ADAT values used; <sup>2</sup> Figure refers to whole milk (as reported by LfL [54] for "Vollmilch, Kuh, Frisch", English translation: whole milk, cow, fresh); <sup>3</sup> Without maize silage.

Cows are bedded on 1.5 tonnes day−<sup>1</sup> of a paper pulp/sawdust product, which has a dry matter content of 95%. Whilst both paper [57] and paper sludge [58] can produce biogas, production from sawdust in negligible [59], so a conservative assumption was made that no additional biogas was produced from the bedding.

Dairy farm substrates (slurry, FYM, waste milk and waste feed) from the 550-cow herd contribute to 82% of the yearly methane production. In the UK, approximately 1.5% of the UK's holdings house 12.3% of the nation's dairy cows in herds greater than 499 cows [35]. Initial targeting of larger dairy farms (such as Leckford Estate) to improve slurry management by the introduction of AD makes logistical and financial sense, particularly where they are also able to utilise local substrates such as chicken manure in case that the farm may not have a land base sufficient to spread those nutrients.

Volumes of the following feedstocks vary throughout the year at the Leckford Estate:


The seasonality of these feedstocks does not affect the biomethane production calculations, as these are on a yearly basis. However, when planning such a scheme, the estate may need to consider the costs and benefits of strategies such as storage or ensiling in order to include them in a way that maintains the relatively consistent gas production that a CHP requires.

For completeness, it is noted that some potentially available biomass types were not included in the calculation due to small quantities/unreliable availability, poor quality of material (i.e., too old), or low suitability for AD (e.g., strongly lignocellulosic biomass), thus relying on conservative figures to avoid any overestimation of the methane generation to be expected. The following potential feedstocks were not included:


Although these materials were not included in the following calculations, it is worth noting that with an operational digester, the first two feedstocks could also be trialled in order to level seasonal feedstock fluctuations.

To ensure proximity to the bulk of the feedstocks, the AD site selected was on the dairy unit, adjacent to silage clamps, two digestate storage lagoons, a holding tank, a separator, and with good road access. The site was approximately 2 km from the mushroom farm where CHP heat could potentially be utilised year-round.

Based on the above quantities of feedstocks included in the assessment, Table 4 summarises the results from the ADAT tool that were utilised in the economic modelling.

The on-site baseload electricity usage, as reported by Leckford Estate (figures not provided as commercially confidential), comfortably exceeds that of the potential CHP electrical production so all electricity could be used on site continuously. It was also confirmed that the mushroom farm could utilise 100% of the heat produced year-round: this is unusual, but many farms have a heat requirement for agricultural or diversification activities [65,66]. This is discussed further in the CHP-EH scenario below.


**Table 4.** Resultant ADAT model data (to be further utilised for the economic modelling).

#### *3.2. Capital Cost*

Supplier quotes received to install the AD plant at the site under study were £900,000 (CAPL), £1,400,000 (CAPM), and £1,600,000 (CAPH). It was assumed that the farm would provide 100% of the funding without applying for bank loans. If finance through bank loans were required, as is likely to be the case for many typical UK farm businesses, the business case would be further detrimentally affected, and thus the assumption of no bank loan is in line with this study's approach to analyse whether economic viability of AD is at all possible.

Despite the fact that suppliers were contacted with a detailed set of data describing the specific case, it is important to note that they will use their own procedures to frame the AD plant sizing and to estimate the site-specific requirements, and these assumptions might be rather complete or not, depending on the level of detail and the degree of complexity a supplier is ready to include. In general, the cost of such site-specific items means that the actual capital cost of an AD plant installation varies greatly because the assumed complexity of the scope of the project varies: for example, a project with a complex planning application where a road, extensive bunding, and several silage clamps are included as part of the AD plant finance figure will be considerably more expensive than one of the same size where these are not required. Additionally, data on costs and their breakdown are often not readily available, as they are considered to be commercially sensitive information.

Because of this uncertainty around digester costs, quotes were cross-checked against two sources that provide UK digester historical cost data, albeit based on relatively small sample sizes. The first was a tool developed by the World Biogas Association, based on actual historical costs for 64 UK plants [67]. Quotes were also checked against the relevant CHP capacity band for low, central, and high digester cost cases analysed by Parsons Brinckerhoff in a 2015 report to the UK government [68].

The costs indicated by the obtained supplier quotes for a system based on a 145 kWe appeared broadly in line with the central and high Parsons Brinckerhoff historical costs shown Table 5. In this context, it is interesting to note that the digester cost model developed by the World Biogas Association [67] suggests that a plant of 145 kWe is likely to have a cost of £449,500; but the relatively small number of digesters analysed at this scale may account for the lack of alignment with the supplier quotes.

Thus, the three supplier quotes (capital costs of £900,000, £1,400,000, £1,600,000) obtained for the site under study are considered realistic, indicating a lower-cost solution for a simple cost-effective farm-type digester (CAPL), a medium-cost solution (CAPM), and a solution at industrial standard with elevated costs (CAPH). These figures are therefore used in the following as a basis for the economic viability calculations.

In addition, for the scenario with heat valorisation, with the mushroom farm located 2 km from the proposed digester site, the capital cost of providing heat to the farm was calculated to be £225,000 (assumptions are documented in Section 2).

**Table 5.** Literature-based data indicating capital costs for a range of digester sizes (derived from figures available in [68]) and, thus, calculated capital cost for a 145 kWe plant (PBL: Parsons Brinckerhoff Low Case figure; PBM: Parsons Brinckerhoff Central Case (i.e., median) figure; PBH: Parsons Brinckerhoff High Case figure).


*3.3. Economic Viability of the AD Plant with CHP Electrical Production Only (CHP-E Scenario)*

Table 6 shows the IRR based on the three capital expenditure levels given by suppliers. The £900,000 digester (CAPL) would easily reach the 16% IRR under levels FITM (median feed-in-tariff applicable for UK situation) and FITH (highest feed-in-tariff applicable for a UK situation).

**Table 6.** Internal rate of return (IRR) based on three quoted capital expenditure levels (supplier quotes obtained for the site under study), electricity production only (CHP-E scenario).


<sup>1</sup> FITL: £0.0465 kWh−<sup>1</sup> (lowest electricity feed-in-tariff applicable for UK situation); <sup>2</sup> FITM: £0.1104 kWh−<sup>1</sup> (median electricity feed-in-tariff applicable for UK situation); <sup>3</sup> FITH: £0.1814 kWh−<sup>1</sup> (highest electricity feed-intariff applicable for UK situation).

Because of their broad alignment and inclusion in UK Government AD viability calculations, the calculations were repeated using the Parsons-Brinckerhoff cases [68], shown in Table 7. Due to the slightly lower capital costs, the model showed an increased IRR in all use cases.

**Table 7.** Internal rate of return based on historical Parsons Brinckerhoff figures (as available in [68]), electricity production only (CHP-E scenario).


<sup>1</sup> FITL: £0.0465 kWh−<sup>1</sup> (lowest electricity feed-in-tariff applicable for UK situation); <sup>2</sup> FITM: £0.1104 kWh−<sup>1</sup> (median electricity feed-in-tariff applicable for UK situation); <sup>3</sup> FITH: £0.1814 kWh−<sup>1</sup> (highest electricity feed-intariff applicable for UK situation).

These figures provide an indication of where a positive business case may lie for such systems. The capital cost of £1.2 million (PBH) to £1.4 million (CAPM) combined with the mid-range FIT of £0.1104 kWh−<sup>1</sup> appears to provide IRRs large enough to warrant investment, but not so large as to create a 'boom' in construction activities.

Aligning historical deployment data with FIT levels at small-scale could shed further light on where an economic FIT level might lie, but this analysis is complicated by several factors:

• There is typically a lag time between the announcement of a FIT level and a digester being commissioned. Due to the long lead-in times (typically a year or more) required for financing, planning, and constructing a digester, the UK FIT scheme had a preliminary accreditation mechanism that allowed applicants to 'lock in' at a given tariff level in order to provide certainty of income, and therefore make a project financially

attractive to lenders. Thus, a digester could have locked into a tariff of £0.16 kWh−<sup>1</sup> on a certain date, but by the time the digester was actually built and included in deployment figures a year later, the tariff on that date could have been much lower.


The average FIT level has undergone considerable changes during its lifetime [69]. While the average FIT for an AD plant of the scale 146 kWe was above £0.10 kWh−<sup>1</sup> in the years 2010 to 2015, it fell to lower levels in later years, and was below £0.06 kWh−<sup>1</sup> in the years 2018 and 2019. In any case, when the FIT had dropped to less than £0.05 kWh−<sup>1</sup> , the UK's Anaerobic Digestion and Bioresources Association (ADBA) noted in its April 2017 policy report [70] that the smaller-scale end of the market 'has been decimated', adding in the November 2017 report [71] that only 13 plants under 250 kWe had been commissioned in 2016. This coincided with the timing of the original case study carried out for JLP in 2017 (report not published), which concluded that the desired IRR could not be achieved at the more realistic CAPM and CAPH levels using these parameters: a FIT rate of £0.0499 kWh−<sup>1</sup> , an RHI rate of £0.0226 kWhth −1 , and energy prices of £0.10 kWh−<sup>1</sup> and £0.42 kWh−<sup>1</sup> for electricity and heating oil, respectively.

The 1 April 2015 FIT was set at £0.1013 kWh−<sup>1</sup> , which when modelled using current energy prices, provides an IRR in double figures for all cases except CAPH; this therefore provides a further clear indication of the level of support below, for which few plants of this size are likely to be built.

#### *3.4. Economic Viability of the AD Plant with CHP and Heat Use (CHP-EH Scenario)*

In the case of the site under study in this research, the mushroom farm is able to use all of the CHP heat produced year-round, and thus full valorisation of the available heat is assumed. This represents a best-case setting with view to heat usage. The improved IRRs under this setting, shown in Table 8, reflect the financial benefits of utilising both heat and electricity.


**Table 8.** Internal rate of return based on three quoted capital expenditure levels, electricity production and 100% heat use (CHP-EH scenario).

<sup>1</sup> FITL: £0.0465 kWh−<sup>1</sup> (lowest electricity feed-in-tariff applicable for UK situation); <sup>2</sup> FITM: £0.1104 kWh−<sup>1</sup> (median electricity feed-in-tariff applicable for UK situation); <sup>3</sup> FITH: £0.1814 kWh−<sup>1</sup> (highest electricity feed-intariff applicable for UK situation).

Overall CHP efficiency is increased if a beneficial heat use can be found. It is easier to use the bulk of the heat produced by a smaller CHP, thus maximising the overall system efficiency. Leckford Estate is relatively unusual in its ability to utilise heat year-round. Nevertheless, a dairy farm could use CHP heat for hot water in buildings [65,66], for dairy washing (which is a highly relevant cost factor for dairy farms, typically amounting to nearly one third of electricity costs of a dairy unit [72]), space heating, crop drying, and to improve milk yields by warming cattle drinking water [73,74]. A number of increasingly common farm business diversification activities also use heat, e.g., greenhouses, campsites (space and water heating), and local food production operations.

A FIT tariff of £0.1013 kWh−<sup>1</sup> has historically resulted in modest digester deployment. In a scenario where 70% of the heat could be used, it was decided to calculate what level of

RHI support would be required in order to achieve an IRR of 16% at CAPM and CAPH. This resulted in an RHI rate of £0.0459 kWh−<sup>1</sup> and £0.0859 kWh−<sup>1</sup> at CAPM and CAPH, respectively. This result is broadly in line with the RHIL of £0.0309 kWhth <sup>−</sup><sup>1</sup> and RHIH of £0.0873 kWhth −1 , a level of support which has historically been considered reasonable.

With the end of the FIT in April 2019, electricity generation from biogas has fallen from regulatory favour, not least because the bulk of the energy produced from even the most electrically efficient CHP is in the form of heat. In the UK and countries with similar climates, it is relatively rare to find a year-round heat use, nevertheless, system economics can be improved by utilising a greater proportion of the CHP heat production. Particularly in colder climates where there is a much wider variation in the heat demand, not least for digester heating, the challenge is to ensure that there is sufficient heat produced to meet the demand. This means that in seasons of low demand, there is a greater excess of heat. Strategies to address this might include finding a further seasonal heat use, for example, a diversification activity such as hot water for campers. Conversely, any seasonal shortfall could be met through other renewables: biomass, solar PV, solar thermal, battery storage [75].

#### *3.5. Analysis of the Impact of Energy Prices*

In view of the recent sharp increases in energy prices across the UK and worldwide, it was decided to explore the potential for subsidy free support for the production of electricity while utilising 70% of the heat. Table 9 illustrates what electricity prices (required non-domestic prices) would have to be in order to achieve economic viability of an AD plant at the studied site, and includes a less ambitious IRR of 10% in a subsidy-free scenario where CHP heat is replacing heating oil (at a cost of £0.55 per litre [50]).


**Table 9.** Minimum economic electricity prices at varying capex and IRR under a tariff-free regime with 70% heat use.

<sup>1</sup> Replacing heating oil with an average price of £0.055 kWh−<sup>1</sup> [50].

With an average non-domestic electricity price for very small non-domestic users of £0.1734 kWh−<sup>1</sup> over the period from the fourth quarter (Q4) of 2019 to the second quarter (Q2) of 2021 [49], future electricity prices in the region of approximately £0.20 kWh−<sup>1</sup> seem increasingly likely, which would make some scenarios an economic proposition.

These figures [49] show a gradual increase in non-domestic electricity prices from an average of £0.0416 kWh−<sup>1</sup> in 2004 to £0.1361 kWh−<sup>1</sup> in Q2 of 2021, an increase of 327%, with natural gas rising from £0.01254 kWh−<sup>1</sup> to £0.0259 kWh−<sup>1</sup> , an increase of 207% over the same period. These figures do not reflect the steep rise in wholesale gas prices experienced in the last quarter of 2021 [76]. Due to worldwide increases and volatility in the wholesale gas price market, UK prices reached as high as £0.042 kWh−<sup>1</sup> [77], affecting both gas and electricity prices (as 35.7% of electricity is generated using gas [78]).

Support at the lowest FIT rate of £0.1013 kWh−<sup>1</sup> , coupled with the lowest RHI rate at £0.0309 kWh−<sup>1</sup> , an electricity price of £0.1492 kWh−<sup>1</sup> and £0.1840 kWh−<sup>1</sup> at CAPM and CAPH, respectively, would provide an IRR of 16%. These values are greater than the ten-year (2010 to 2019) non-domestic electricity price average of £0.1030 kWh−<sup>1</sup> . However, the value of £0.1492 kWh−<sup>1</sup> , although higher than the current average electricity price modelled of £0.1361 kWh−<sup>1</sup> , is certainly within the range of 3rd quarter 2021 prices for small/medium users (£0.1502 kWh−<sup>1</sup> for 500–1999 MWh annual usage) and less than

that for small users (£0.1534 kWh−<sup>1</sup> for 20–499 MWh annual usage) and very small users (£0.1818 kW−<sup>1</sup> for 0–20 MWh annual usage) [49]. This provides a further indication of the support required if the general trend for average non-domestic electricity prices continue their gradually increasing trajectory, which has characterised the past seventeen years (2004–2020) [49].

UK domestic electricity prices are typically higher than the European average, whereas gas prices are lower [79,80]. This price differential when coupled with the poor building fabric tends to discourage electrification, and therefore the decarbonisation of heat, particularly in rural areas where electricity grids may be weak and there is no access to the gas grid for alternative fuels such as hydrogen. In the face of climate change perturbations and the complexity associated with fossil energy market forces, it is unclear whether such volatility will continue [81], since multiple factors such as supply chain issues, energy systems decarbonisation, and emissions trading scheme carbon prices affect costs [76].

Where energy costs are both high and volatile over a long period, on-site de-centralised energy generation becomes particularly important and AD at small scale using mainly wastes becomes increasingly viable, cushioning these important farm businesses against such volatility and helping the UK and other countries meet their COP26 Methane Pledge and wider climate goals.

However, unless there is a guarantee that prices will be sustained at such levels, AD projects are still unlikely to be considered worthy of investment, so some form of support for technology implementation would still be required, e.g., a minimum price guarantee or floor price. Energy based incentives which also value the non-energy contributions of AD to such farms could be an option [7,82]. These could, for example, combine a smallscale electricity/heat tariff to improve on-site/local energy utilisation. There is an added opportunity to consider the benefits of AD in the context of the post-Brexit agricultural policy support schemes.

#### *3.6. Digestate Savings/Income*

The value of digestate as a source of nutrients that can displace fossil fuel based synthetic fertilisers (which can be costed) and as a soil conditioner, which replenishes soil carbon (which is difficult to cost) was not taken into account in any of the above calculations. Digestate is often regarded as an expense to the business, particularly for large digesters where large volumes of digestate require longer transport distances in order not to cause soil nutrient overloading [83]. Whilst the ADAT tool can include the energetic cost of transport, this was not included within this economic case study, as digestate was not being transported any further than the slurry and wastes otherwise would have been.

Digestate can also be separated into liquid and solid fractions to facilitate differential nutrient application [16]. Prior to the introduction of end of waste criteria, digestate fibre mixes used to be sold to gardeners [21] as a peat-free option, but the costs now associated with meeting the requirements [84] are likely to preclude small digesters from this market, unless costs could be defrayed through, for example, an aggregation mechanism whereby several smaller operations could be considered as one for the purposes of regulation.

Against a backdrop of high fossil fuel energy prices, the value of digestate can offset the commensurately high price of fossil fertilisers [48]:


Using the above fertiliser costs [48] and RB209 nutrient guide values [47], a tonne of digestate was calculated to be worth £9.39 or a total of £119,160 in this case study, potentially further improving the economics of the project under study. For a CAPM RHI digester using 70% heat with no subsidies and digestate at this price, the IRR is a respectable 12.76%.

The results thus confirm previous reports that recycling such organic materials back to land in the form of digestate can potentially add value [33,53] while improving the GHG credentials of the business [75,85].

The post-Brexit agricultural policy aims to reward public goods, particularly those related to the environment [86,87], including improved soil health, water quality improvement, and reduced GHG emissions. A well-run AD plant that recycles carbon back to soils as digestate and captures or avoids otherwise uncontrolled emissions occurring from biodegradation of organic matter or storage of materials (particularly methane from slurry stores) can contribute to these aims [1] and as such should be recognised within that policy [4].

#### **4. Conclusions**

As illustrated in this research, small farm-scale AD installations that utilise diffuse on-site or locally sourced wastes to produce both electricity and heat from a CHP can improve the overall efficiency and economics over those that do not valorise the heat. Additionally, they can provide heat at a scale that can often be beneficially utilised. AD can help cushion such food production businesses against high and/or volatile energy prices, as well as helping to decarbonise a sector that often uses carbon intensive fuels for heating. Where rural electricity grids are weak and/or electricity is expensive, on-site electricity production also provides a route to farm business electrification, increasing economic options for utilisation of electrical vehicles, agritech, robotics, and more.

The Leckford Estate is a large, diversified farming business owned by a major retailer with a commitment to sustainability and access to a wide range of local waste feedstocks, but under the current UK policy regime, which is characterised by a lack of support at this relatively small scale (145 kWe, 1500 m<sup>3</sup> digester), it is uneconomic for the estate to utilise these resources through AD in order to reduce its carbon footprint and improve their sustainability in the face of climate change.

As energy prices increase, the required level of support for such projects decreases, but projects also struggle to put an economic business case together where there is no long-term pricing clarity or alternative support mechanisms that value the many benefits of AD. The results of the current study can be used to provide an indication of recommended levels of support, and a basis for varying them in response to shifting energy prices.

Designing a policy that values these benefits is a challenge for all countries and regions [33], but with the introduction of a new agricultural policy, the UK has an opportunity to valorise these diffuse organic wastes through small AD, and support farm businesses in the face of the urgency of the climate crisis and high energy prices.

One potential support mechanism could be to include the 'public goods' benefits of on-farm AD (including greenhouse gas reduction, improved nutrient management, positive soil organic carbon impact, and strengthened rural development) in the UK post-Brexit agricultural policy support scheme.

**Author Contributions:** Conceptualization, A.B. and S.K.-B.; Methodology, A.B.; Software, A.B.; Validation, S.K.-B.; Formal analysis, A.B. and S.K.-B.; Investigation, A.B.; Data curation, A.B.; Writing original draft preparation, A.B. and S.K.-B.; Writing—review and editing, A.B. and S.K.-B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by a Business Interaction Voucher (BIV2016020) from the Anaerobic Digestion Network (grant number BB/L013835/1), a Network in Industrial Biotechnology and Bioenergy (NIBB) funded by the Biotechnology and Biological Sciences Research Council.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data supporting this study are openly available from the University of Southampton repository at https://eprints.soton.ac.uk/ access on 10 December 2021.

**Acknowledgments:** The authors wish to thank the Leckford Estate and the John Lewis Partnership for giving us access to their estate for the purpose of this study, for making available data about potential AD feedstocks at their estate and for the assistance provided to collect the required data; Yue Zhang (University of Southampton) for scientific advice; James Murcott (Methanogen UK Ltd.), Chris Morris (Fre-Energy Ltd.), Michael Chesshire (Lutra), Chris Cooper (CPN Projects Ltd.) for providing data about commercial costs in the AD sector; and the Environmental Biotechnology Network's (a BBSRC/EPSRC NIBB) Anaerobic Digestion Working Group for expert knowledge that contributed to ensure robustness of the assessments made. Special thanks to Sonia Heaven (University of Southampton) for scientific advice and discussions, which have enriched this work in a very valuable way.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Review* **Recent Advances in Membrane-Based Biogas and Biohydrogen Upgrading**

**Cenit Soto 1,2 , Laura Palacio 1,2 , Raúl Muñoz 2,3 , Pedro Prádanos 1,2,\* and Antonio Hernandez 1,2,\***


**Abstract:** Biogas and biohydrogen, due to their renewable nature and zero carbon footprint, are considered two of the gaseous biofuels that will replace conventional fossil fuels. Biogas from anaerobic digestion must be purified and converted into high-quality biomethane prior to use as a vehicle fuel or injection into natural gas networks. Likewise, the enrichment of biohydrogen from dark fermentation requires the removal of CO<sup>2</sup> , which is the main pollutant of this new gaseous biofuel. Currently, the removal of CO<sup>2</sup> from both biogas and biohydrogen is carried out by means of physical/chemical technologies, which exhibit high operating costs and corrosion problems. Biological technologies for CO<sup>2</sup> removal from biogas, such as photosynthetic enrichment and hydrogenotrophic enrichment, are still in an experimental development phase. In this context, membrane separation has emerged as the only physical/chemical technology with the potential to improve the performance of CO<sup>2</sup> separation from both biogas and biohydrogen, and to reduce investment and operating costs, as a result of the recent advances in the field of nanotechnology and materials science. This review will focus on the fundamentals, potential and limitations of CO<sup>2</sup> and H<sup>2</sup> membrane separation technologies. The latest advances on membrane materials for biogas and biohydrogen purification will be systematically reviewed.

**Keywords:** biogas; biomethane; biohydrogen; membrane separation; mixed matrix membranes; upgrading technologies; thermal rearrangement

#### **1. Biogas and Biohydrogen as Green Energy Vectors**

Biogas is produced via Anaerobic Digestion (AD) of residual biomass from diverse origins such as urban solid waste, livestock waste, agricultural waste, and wastewater. AD is a biological process (based on the action of micro-organisms) able to convert this residual biomass, by means of oxidations and reductions of organic carbon, to carbon dioxide and methane (CO<sup>2</sup> and CH4, respectively) in the absence of oxygen [1,2]. This biological conversion is carried out through a sequence of hydrolysis, acidogenesis, acetogenesis and methanogenesis steps in an anaerobic digester [3]. Biogas is typically composed of CH<sup>4</sup> and CO<sup>2</sup> in a concentration range of 45–85% and 25–50%, respectively, and minor concentrations of other components such as H2O (5–10%), N<sup>2</sup> (~0–1%), O<sup>2</sup> (~0–0.5%), H2S (0–10,000 ppm), NH<sup>3</sup> (0–100 ppm) and hydrocarbons (0–200 mg Nm−<sup>3</sup> ) [4,5]. The biogas produced by AD represents an excellent alternative to fossil-based energy vectors [2], since biogas can be employed for the production of electricity, steam and heat, as a feedstock in fuel cells, as a green substitute of natural gas for domestic and industrial use or as a vehicle fuel [1]. The contribution of biogas in the European Union could account for 10% of the natural gas demand by 2030 and up to 30–40% by 2050.

Based on the latest report of the World Biogas Association [6], 50 million micro-scale digesters generating biogas for cooking or heating were in operation, mainly in China

**Citation:** Soto, C.; Palacio, L.; Muñoz, R.; Prádanos, P.; Hernandez, A. Recent Advances in Membrane-Based Biogas and Biohydrogen Upgrading. *Processes* **2022**, *10*, 1918. https://doi.org/10.3390/pr10101918

Academic Editors: Sonia Heaven, Sigrid Kusch-Brandt and Charles Banks

Received: 29 August 2022 Accepted: 18 September 2022 Published: 22 September 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

(42 million) and India (4.9 million). On the other hand, 18,774 large-scale plants devoted to generating 11 GW (a biomethane plant produces an average of 36 GWh per year) of electricity were in operation in 2021 in Europe, Germany being the leader in the European market with 11,279 in 2020 plants (140 plants/1 Mio capita), followed by Italy (1666 in 2020) and France (833 new plants in 2020) [4,7]. China with 6972 large scale digesters and the USA with 2200 AD plants in 2015 represented the second and third largest biogas producer in the world, respectively. The global electricity generation from biogas increased by 90% in six years (from 46,108 GWh in 2010 to 87,500 GWh in 2016) and by 11.5 % from 2016 to 2020 (from 87,500 GWh in 2016 to 96,565 GWh in 2020) [6,8].

Biogas can be purified and converted into a high-quality biomethane via three sequential processes: desulfurization (elimination of the H2S), CO<sup>2</sup> removal and biomethane polishing (removal of the minor biogas contaminants) [9]. The European EN-16723 Standard for biomethane introduction into natural gas networks (UNE-EN 16723-1-2016) and automotive/vehicle fuel (UNE-EN 16723-2-2017) requires an effective cleaning of biogas. This UNE-EN 16723-1-2016 standard has resulted in a specific Spanish standard for biomethane injection into the natural gas grid, requiring a minimum methane content of 90% and a maximum CO<sup>2</sup> content of 2% (*v*/*v*) [10]. In 2017, the number of biogas upgrading plants in the world accounted for 700 plants, Europe being the leading region with 540 upgrading plants in operation.

At the end of 2020 (the most recent data available), 880 biogas upgrading plants with a production capacity of 2.43 billion m<sup>3</sup> were in operation in Europe (161 additional plants relative to 2019) [4,7]. By 2021 the increase in the number of biomethane plants is expected to be even faster since 115 plants have started operation by August 2021 [7].

On the other hand, biologically produced hydrogen (commonly referred to as biohydrogen) generated via Dark Fermentation (DF) represents another alternative bioenergy source [11]. Biohydrogen (bioH2) has the potential to become a relevant H<sup>2</sup> generation platform for the creation of a green economy [12]. In this context, hydrogen has multiple advantages as a clean energy vector such as: (i) the combustion of H<sup>2</sup> gas can be pollutionfree in fuel cells, (ii) its energy efficiency in H<sup>2</sup> fuel cells is approx. 50% higher than that of gasoline, (iii) it has a specific energy content of 122 kJ/g (~2.75-fold larger than conventional fossil fuels), (iv) its conversion efficiency to electricity could be doubled using fuel cells instead of gas turbines, and finally (v) it can be stored as a metal hydride.

Dark fermentation is based on hydrogen and carbon dioxide (CO2) production via anaerobic bacteria [13] and/or algae growing in the absence of light and with high carbohydrate content as substrate [14,15]. The biohydrogen produced is mainly composed of hydrogen (40–60%) and carbon dioxide (47–60%) with traces of methane and H2S [16,17]. Currently, only 1% of hydrogen is produced from biomass [15]. This fact is probably due to the relatively late research on bioH<sup>2</sup> production by dark fermentation, where research is still conducted at a laboratory scale with a limited number of experiments at pilot scale [18]. Despite the fact that the H<sup>2</sup> yield from dark fermentation is higher than that of other processes, the main disadvantage of the gas generated during dark fermentation is its low hydrogen concentration (40–60%; *v*/*v*) [19], which hinders its direct use in fuel cells for electricity generation (where the purity of hydrogen is crucial to achieve high energy yields) [16]. Therefore, it is crucial to separate H<sup>2</sup> from the multiple gas by-products from DF, mainly CO2, in order to obtain purified hydrogen. For instance, a hydrogen content of 73% can be obtained in a two-step gas membrane separation module [19].

The sustained use of non-polluting renewable energy vector such as biogas and bioH<sup>2</sup> is required to reduce the demand and dependence from fossil fuels [20]. Based on the International Energy Agency, the share of renewable and low-carbon transport fuels should increase up to 6.8% in 2030 in Europe, with advanced biofuels representing at least 3.6% of the total fuel consumption. The development of low footprint and cost technologies for the conversion of biogas to a purified biomethane and bioH<sup>2</sup> to pure H<sup>2</sup> is essential to guarantee the competitiveness of these green gas vectors as an energy source.

#### **2. Biogas and Biohydrogen Purification with Membrane Technology**

Nowadays, there are two main types of technologies for biogas purification, physicochemical and biological methods, while bioH<sup>2</sup> purification is only performed by physicochemical methods. Physicochemical technologies exhibit high energy and chemical demand, and therefore they present large operating costs and environmental impacts. As an example, this section will only focus on CO<sup>2</sup> removal technologies.

Pressure swing adsorption (PSA), cryogenic CO<sup>2</sup> separation, scrubbing with H2O, chemical solutions or organic solvents, and membrane separation, dominate the biogas upgrading market nowadays [21], while cryogenic distillation, PSA and membrane separation are the most popular processes for H<sup>2</sup> purification at commercial scale [22–24].

Separation of gas mixtures through membranes has become a relevant unitary operation for the recovery of valuable gases and mitigation of atmospheric pollution, which offers several advantages over conventional gas separation methods [25]. Indeed, Membrane Separation (MS) is considered nowadays the most promising gas purification technology. Membrane separation relies on the interaction (physical or chemical) of certain gases with the membrane material [26]. The membranes used are selective physical barriers with certain components that permeate across them [27]. Gas separation by membrane technology is characterized by selectivity properties and flux, which supports a functional transport of the target gases across the barrier (permeability). This technology presents a low energy consumption, a simple operation, cost effectiveness, smaller footprint, a negligible chemical consumption and low environmental impacts [28,29]. The potential of MS to achieve high efficiencies of gas separation foster their use in different industrial applications including refineries and chemical industries, and recent advances in material science render MS a competitive technology [30]. The lifetime of commercial membranes account for 5–10 years [31]. Today, the use of membranes in industry includes the separation of N<sup>2</sup> or O<sup>2</sup> from air, separation of H<sup>2</sup> from gases such as CH4, separation of CH<sup>4</sup> from biogas, separation of H2S and CO<sup>2</sup> from natural gas, etc. The use of membranes in separation processes is rapidly growing, especially in Europe (Figure 1). Among the available technologies for the purification of biogas to biomethane, membrane separation is currently the most widely used technology (39%), followed by water scrubbing (22%) and chemical scrubbing (18%). Pressure swing adsorption (12%), cryogenic separation (1%) and physical washing (1%) complete the market share (with the exception of 7% of European biomethane plants, with no data available in the EBA database) [7]. For instance, Baker (2002), calculated that the market share of membrane gas separation processes in 2020 would be five times higher than that of year 2000 [32]. Indeed, the market share of MS for biogas upgrading application has increased from 10% in 2012 to 25% in 2017 [33]. Likewise, MS has grown exponentially since the initial commercial application of Prism membranes by Permea (Monsanto) for H<sup>2</sup> separation from the off-gas stream of NH<sup>3</sup> production plants [26].

A detailed economic study of the total costs of biogas purification is a difficult task nowadays due to the large number of parameters to be considered. However, Miltner and co-workers (2017) have published some general estimates and a comparison of the most common physicochemical technologies such as pressurized water scrubbing, amine scrubbing, pressure swing adsorption and gas permeation. This study included investment costs (15 years' depreciation), plant reliability of 98%, operational consumptions in terms of electricity and consumables (electricity price 15 €ct/kWh), as well as maintenance and overhaul (without engineering costs, taxes and revenues). Thus, the costs for an installation with a capacity of 250 m<sup>3</sup> STP/h are in the range of 25 €ct/m<sup>3</sup> STP, while these costs drop below 15 €ct/m<sup>3</sup> STP for capacities above 2000 m<sup>3</sup> STP/h. This work concluded that gas permeation is slightly more advantageous for sizes below 1000 m<sup>3</sup> STP/h. Overall, smallscale biogas upgrading entails higher capital and operational costs [34].

permeation is slightly more advantageous for sizes below 1000 m3 STP/h. Overall, small-

scale biogas upgrading entails higher capital and operational costs [34].

**Figure 1.** Market share of different upgrading technologies in Europe in 2020. Figure adapted from EBA, 2021. **Figure 1.** Market share of different upgrading technologies in Europe in 2020. Figure adapted from EBA, 2021.

Ideally, membrane materials for gas separation should exhibit a high selectivity and big fluxes, excellent chemical, mechanical and thermal stability, a defect-free production and be cost effective. Membranes are classified according to the type of material, configuration, structure, composition, support material and industrial reactions, among others (Figure 2) [35–37]. Four kinds of membranes are typically proposed for development and commercialization in hydrogen purification: (i) polymeric membranes (organic), (ii) porous membranes (ceramic, carbon, metal) (iii) dense metal membranes and (iv) ion conductive membranes, the last three also referred to as inorganic membranes [27]. In this context, dense-metal membranes and polymeric have experienced the largest advances in terms of scale-up [38]. The most commonly used polymeric membranes for gas separation are nonporous membranes, which are classified as glassy or rubbery. Of them, glassy polymers are most typically used for gas separation applications. These polymers include polysulfones (PSF), polycarbonates (PC) and polyimides (PI), which are often employed for the separation of H2/CH4, H2/N2 and O2/N2 [39]. On the other hand, membranes can be configured as hollow fibers, capillaries, flat sheets and tubular and can be installed in a suitable membrane module. The most commonly used modules are pleated cartridges, Ideally, membrane materials for gas separation should exhibit a high selectivity and big fluxes, excellent chemical, mechanical and thermal stability, a defect-free production and be cost effective. Membranes are classified according to the type of material, configuration, structure, composition, support material and industrial reactions, among others (Figure 2) [35–37]. Four kinds of membranes are typically proposed for development and commercialization in hydrogen purification: (i) polymeric membranes (organic), (ii) porous membranes (ceramic, carbon, metal) (iii) dense metal membranes and (iv) ion conductive membranes, the last three also referred to as inorganic membranes [27]. In this context, dense-metal membranes and polymeric have experienced the largest advances in terms of scale-up [38]. The most commonly used polymeric membranes for gas separation are nonporous membranes, which are classified as glassy or rubbery. Of them, glassy polymers are most typically used for gas separation applications. These polymers include polysulfones (PSF), polycarbonates (PC) and polyimides (PI), which are often employed for the separation of H2/CH4, H2/N<sup>2</sup> and O2/N<sup>2</sup> [39]. On the other hand, membranes can be configured as hollow fibers, capillaries, flat sheets and tubular and can be installed in a suitable membrane module. The most commonly used modules are pleated cartridges, tubular and capillary, hollow-fiber and plate-and-frame and spiral-wound systems [40].

tubular and capillary, hollow-fiber and plate-and-frame and spiral-wound systems [40]. H2 separation was one of the pioneered applications in gas separation membranes, DuPont (E. I. du Pont de Nemours and Co., Wilmington, DE, USA) being the pioneer in manufacturing small-diameter hollow-fiber membranes. Due to the limited productivity (or permeance) of these membranes and their high cost, Monsanto Co. (Monsanto Company, St. Louis, MO, USA) developed polysulfone hollow-fiber membranes, which considerably increased the transport through the fibers, and consequently were successfully implemented at industrial-scale for hydrogen recovery from ammonia purge gases [41]. Then, Separex Corp (Champigneulles, France) developed Separex® spiral-wound cellulose acetate membranes (including separations for natural gas and dehydration [41] providing better performance than hollow fiber membranes due to their high resistance of hydrogen impurities [42]. Polymeric membranes, especially polyimides, have been employed to separate hydrogen from gaseous mixtures (N2, CO and hydrocarbons) based on H<sup>2</sup> separation was one of the pioneered applications in gas separation membranes, DuPont (E. I. du Pont de Nemours and Co., Wilmington, DE, USA) being the pioneer in manufacturing small-diameter hollow-fiber membranes. Due to the limited productivity (or permeance) of these membranes and their high cost, Monsanto Co. (Monsanto Company, St. Louis, MO, USA) developed polysulfone hollow-fiber membranes, which considerably increased the transport through the fibers, and consequently were successfully implemented at industrial-scale for hydrogen recovery from ammonia purge gases [41]. Then, Separex Corp (Champigneulles, France) developed Separex® spiral-wound cellulose acetate membranes (including separations for natural gas and dehydration [41] providing better performance than hollow fiber membranes due to their high resistance of hydrogen impurities [42]. Polymeric membranes, especially polyimides, have been employed to separate hydrogen from gaseous mixtures (N2, CO and hydrocarbons) based on their economic viability, easy processibility and satisfactory thermal stability (350–450 ◦C) [43]. Polyimide membranes with excellent heat resistances were introduced by Ube in Japan (Ube Industries, Ltd., Tokyo, Japan), and the refinery at Seibu Oils (Seibu Oil Company Limited, Onoba, Japan) was the first facility to apply them commercially [41]. Commercial

2017.

*Processes* **2022**, *10*, x FOR PEER REVIEW 5 of 41

membrane systems provide a H<sup>2</sup> purity of 90–95% during hydrogen purification with a moderate recovery of 85–90% [44]. pany Limited, Onoba, Japan) was the first facility to apply them commercially [41]. Commercial membrane systems provide a H2 purity of 90–95% during hydrogen purification with a moderate recovery of 85–90% [44].

their economic viability, easy processibility and satisfactory thermal stability (350–450 °C) [43]. Polyimide membranes with excellent heat resistances were introduced by Ube in Japan (Ube Industries, Ltd., Tokyo, Japan), and the refinery at Seibu Oils (Seibu Oil Com-

**Figure 2.** Classification of membranes for gas separation. Adapted from Vinoba and co-workers, **Figure 2.** Classification of membranes for gas separation. Adapted from Vinoba and co-workers, 2017.

At the beginning of the 1990s, gas mixture separation membranes with a poor recovery of methane and low selectivity were installed for the upgrading of landfill biogas [45]. In 2007, Air Liquide MedalTM further developed and tested new selective membranes com-At the beginning of the 1990s, gas mixture separation membranes with a poor recovery of methane and low selectivity were installed for the upgrading of landfill biogas [45]. In 2007, Air Liquide MedalTM further developed and tested new selective membranes combining high CH<sup>4</sup> recoveries with high CH<sup>4</sup> concentrations.

bining high CH4 recoveries with high CH4 concentrations. Today, membrane-based biogas upgrading can provide methane concentrations of 97–98% in the biomethane with a concomitant methane recovery above 98%, based on the high permeabilities of CO2 in commercial membrane materials. The permeation rate mainly depends on the molecular size of the gas components and on the membrane construction material [46]. Membrane-based biogas upgrading at commercial scale is carried out at 6–20 bar, which entails energy consumption of 0.18–0.20 kWh/Nm3 of raw biogas Today, membrane-based biogas upgrading can provide methane concentrations of 97–98% in the biomethane with a concomitant methane recovery above 98%, based on the high permeabilities of CO<sup>2</sup> in commercial membrane materials. The permeation rate mainly depends on the molecular size of the gas components and on the membrane construction material [46]. Membrane-based biogas upgrading at commercial scale is carried out at 6–20 bar, which entails energy consumption of 0.18–0.20 kWh/Nm<sup>3</sup> of raw biogas or 0.14–0.26 kWh/Nm<sup>3</sup> of biomethane [9].

or 0.14–0.26 kWh/Nm3 of biomethane [9]. In this regard, despite polymeric membranes having consistently demonstrated promising results and being commercially available at large-scale for hydrogen and biogas purification, their use is limited to 8–9 polymeric materials (e.g., cellulose acetate, polyimides, perfluoropolymer etc.) [47,48]. Therefore, further research in the field of material science needs to be conducted to achieve new membranes with superior gas separation In this regard, despite polymeric membranes having consistently demonstrated promising results and being commercially available at large-scale for hydrogen and biogas purification, their use is limited to 8–9 polymeric materials (e.g., cellulose acetate, polyimides, perfluoropolymer etc.) [47,48]. Therefore, further research in the field of material science needs to be conducted to achieve new membranes with superior gas separation properties: higher permeability, selectivity and stability (mainly restricted plasticization) [47].

properties: higher permeability, selectivity and stability (mainly restricted plasticization)

#### [47]. **3. Fundamentals of Membrane-Based Gas Separation**

**3. Fundamentals of Membrane-Based Gas Separation**  The membrane gas separation process is based on the separation of gases by selective permeation of one or more gaseous components through a thin membrane (porous or dense membrane) [49]. The separation potential of the membrane is governed by its The membrane gas separation process is based on the separation of gases by selective permeation of one or more gaseous components through a thin membrane (porous or dense membrane) [49]. The separation potential of the membrane is governed by its transport properties of the components of a mixture. This transport rate is in turn dictated by the permeability and selectivity of the membrane material and its driving force [38].

transport properties of the components of a mixture. This transport rate is in turn dictated by the permeability and selectivity of the membrane material and its driving force [38]. Gas separation takes place according to the morphology of the membrane materials and can be based on three transport mechanisms depending on the porous size: solutiondiffusion, molecular sieve and Knudsen diffusion (Figure 3). In this context, the transport Gas separation takes place according to the morphology of the membrane materials and can be based on three transport mechanisms depending on the porous size: solutiondiffusion, molecular sieve and Knudsen diffusion (Figure 3). In this context, the transport of gases by Knudsen diffusion takes place in porous membranes (pore diameter in the range of 50–100 Å), with smaller pore size than the gas molecules. In this mechanism, gas molecules interact more frequently with the pore walls, colliding with each other, allowing diffusion of lighter molecules to occur through the pores. The molecular sieving mechanism, with pore size between 3.0–5.2 Å, is based on the size exclusion of gas molecules, leading to

*Processes* **2022**, *10*, x FOR PEER REVIEW 6 of 41

the separation of gas molecules of different kinetic sizes. Indeed, the pores only allow the passage of molecules smaller than that size, preventing the passage of larger ones [26,29]. allow the passage of molecules smaller than that size, preventing the passage of larger ones [26,29].

of gases by Knudsen diffusion takes place in porous membranes (pore diameter in the range of 50–100 Å), with smaller pore size than the gas molecules. In this mechanism, gas molecules interact more frequently with the pore walls, colliding with each other, allowing diffusion of lighter molecules to occur through the pores. The molecular sieving mechanism, with pore size between 3.0–5.2 Å, is based on the size exclusion of gas molecules, leading to the separation of gas molecules of different kinetic sizes. Indeed, the pores only

**Figure 3.** Schematic representation of the three mechanisms for gas mixture separation in membranes: diffusion—Knudsen diffusion, molecular sieving, and solution-diffusion. Orange circle (molecule A), Red circle (molecule B). The figure was adapted from Sridhar and co-workers, 2014, and Ismail and co-workers, 2015. **Figure 3.** Schematic representation of the three mechanisms for gas mixture separation in membranes: diffusion—Knudsen diffusion, molecular sieving, and solution-diffusion. Orange circle (molecule A), Red circle (molecule B). The figure was adapted from Sridhar and co-workers, 2014, and Ismail and co-workers, 2015.

Gas transport in non-porous dense polymeric membranes is most commonly described by solution-diffusion mechanisms (used exclusively in current commercial devices), which allows gases to pass through the membrane free volume units and consists of three steps [50]: (i) sorption in upstream side; (ii) diffusion through the membrane and (iii) desorption at the downstream side. Figure 4 shows a schematic overview of mass transfer by solution-diffusion, where gas molecules sorb into the high-pressure face of the membrane, then diffuse through along the membrane and later desorb from the low-pressure face of the membrane [51]. This mechanism of solution-diffusion is determined by the occurrence of differences in the thermodynamic activities at the upstream and downstream faces of the membrane, and the interacting force working among the gas mole-Gas transport in non-porous dense polymeric membranes is most commonly described by solution-diffusion mechanisms (used exclusively in current commercial devices), which allows gases to pass through the membrane free volume units and consists of three steps [50]: (i) sorption in upstream side; (ii) diffusion through the membrane and (iii) desorption at the downstream side. Figure 4 shows a schematic overview of mass transfer by solution-diffusion, where gas molecules sorb into the high-pressure face of the membrane, then diffuse through along the membrane and later desorb from the low-pressure face of the membrane [51]. This mechanism of solution-diffusion is determined by the occurrence of differences in the thermodynamic activities at the upstream and downstream faces of the membrane, and the interacting force working among the gas molecules, which depends on the membrane components and the permeate molecules [52].

cules, which depends on the membrane components and the permeate molecules [52]. A key parameter to evaluate membrane transport properties is the Permeability coefficient (*P*), which refers to the gas flux across a membrane considering the membrane thickness and pressure gradient (*pi,0-pi,l*) through the membrane (Equation (1)).

$$P = \frac{N\_i \, l}{\Delta p} \tag{1}$$

where *N<sup>i</sup>* is molar flux of a gas component *i* through the membrane, *l* is the membrane thickness and ∆*p* is the pressure gradient, calculates as the difference between *pi,0* (the upstream pressure) and *pi*,*<sup>l</sup>* (the downstream pressure) [53].

The Permeability coefficient ranges from 10−<sup>4</sup> to 10<sup>4</sup> Barrer as a function of the gas component and the polymer structure [52]. Permeability coefficients are expressed in mol (m<sup>2</sup> ·s·Pa) in the international system of units. However, *P* is typically given in Barrer, where <sup>1</sup> *Barrer* = <sup>10</sup>−<sup>10</sup> *cm*<sup>3</sup> *STP*· *cm cm*2·*s*·*cmHg* .

**Figure 4.** Detailed overview of mass transfer by solution-diffusion model. **Figure 4.** Detailed overview of mass transfer by solution-diffusion model.

where

*cm cm Barrer STP*

<sup>⋅</sup> <sup>=</sup> <sup>−</sup> 2 3

⋅ ⋅

A key parameter to evaluate membrane transport properties is the Permeability coefficient (*P*), which refers to the gas flux across a membrane considering the membrane thickness and pressure gradient (*pi,0-pi,l*) through the membrane (Equation (1)). *<sup>N</sup> <sup>l</sup> <sup>P</sup> <sup>i</sup>* <sup>Δ</sup> <sup>=</sup> (1) The solution-diffusion model considers that the conditions of equilibrium between sorption and desorption are maintained. In this context, a solubility coefficient, *S<sup>i</sup>* , is introduced, which is the ratio between the concentration of gas component dissolved in the membrane material, *C<sup>i</sup>* , and the pressure of the gas, *p<sup>i</sup>* , in contact with the polymer (Equation (2)). The solubility of a gas component *i* in the polymeric material depends mainly on the gas molecule condensability.

$$\mathbf{S}\_{i} = \frac{\mathbf{C}\_{i,0}}{p\_{i,0}} = \frac{\mathbf{C}\_{i,l}}{p\_{i,l}} \tag{2}$$

stream pressure) and *pi***,***l* (the downstream pressure) [53]. The Permeability coefficient ranges from 10−4 to 104 Barrer as a function of the gas where *Ci*,0 and *Ci*,*<sup>l</sup>* stand for concentration of the gas component *i* at the feed and permeate side, respectively.

component and the polymer structure [52]. Permeability coefficients are expressed in mol (m2·s·Pa) in the international system of units. However, *P* is typically given in Barrer, On the other hand, the molar flux, *N<sup>i</sup>* , can be expressed as a function of the diffusivity coefficient (*D<sup>i</sup>* ) described by the Fick's Law (Equation (3)):

$$N\_{l} = \,^{D}D\_{l}S\_{l}\frac{p\_{i,0} - \,^{p}p\_{i,l}}{l} \tag{3}$$

The solution-diffusion model considers that the conditions of equilibrium between sorption and desorption are maintained. In this context, a solubility coefficient, *Si*, is introduced, which is the ratio between the concentration of gas component dissolved in the membrane material, *Ci*, and the pressure of the gas, *pi*, in contact with the polymer (Equation (2)). The solubility of a gas component *i* in the polymeric material depends mainly on the gas molecule condensability. *C C* According to the solution-diffusion model, the ability of a gas molecule to pass through the membrane depends on a kinetic factor, the diffusivity, (*D<sup>i</sup>* ), which characterizes the movement of the gas molecules diffusing through the polymer, and a thermodynamic factor, the solubility, (*S<sup>i</sup>* ), which characterizes the number of gas molecules passing through the membrane. Thus, *P* can be represented as the product of the diffusion coefficient, *D<sup>i</sup>* , and gas solubility coefficient, *S<sup>i</sup>* (Equation (4)) [53,54].

$$P = D\_i \mathcal{S}\_i \tag{4}$$

where *Ci*,0 and *Ci*,*l* stand for concentration of the gas component *i* at the feed and permeate side, respectively. On the other hand, the molar flux, *Ni*, can be expressed as a function of the diffusivity coefficient (*Di*) described by the Fick's Law (Equation (3)): (4)On the other hand, a parameter characteristic of gas separation is the ability of a membrane to separate two gas components (*A* and *B*). Typically, selectivity is also treated as a material property of the polymer and is represented by Equation (5). The parameter α is defined as the permeability ratio of the faster permeable gas (*PA*) between the slower permeable gas (*PB*), so that αAB > 1 [52].

*i*

,0

*i*

*S*

*i l*

,

$$
\mathfrak{a}\_{AB} = \frac{P\_A}{P\_B} \tag{5}
$$

Usually, pure gas permeabilities are used in Equation (5) giving the so-called ideal selectivity (*α*). According with the solution-diffusion model, Equation (5) can be reworded using Equation (4), and the selectivity of diffusivity and solubility can be expressed using Equation (6):

$$
\mathfrak{a}\_{AB} = \frac{D\_A}{D\_B} \frac{S\_A}{S\_B} = \mathfrak{a}\_{AB}^D \mathfrak{a}\_{AB}^S \tag{6}
$$

where *DA*/*D<sup>B</sup>* stands for the diffusion coefficients ratio of gas components *A* and *B*, while *SA*/*S<sup>B</sup>* is the ratio of their solubility coefficients. Membrane selectivity determines the energy needed to support gas separation and directly impacts on the operating cost of a membrane system [55].

#### **4. Challenges in Polymer Membranes for Gas Separation**

The membrane transport properties are governed by factors such as the change in feed composition and the degree of swelling at the gas–membrane interface. In addition, other phenomena such as plasticization and ageing influence the transport properties of membranes. In this context, large contents of condensable gases such as CO<sup>2</sup> may plasticize the membrane material. Nowadays, research in membrane-based gas separation targets the development of new membranes with increasing permeabilities and selectivities, with increasing permeabilities without compromising the selectivity or improving the selectivity maintaining the permeability values. Indeed, the increase in permeability without compromising selectivity is typically considered one of the main target routes to expand the market share of membrane materials for gas separation [56].

#### *4.1. Trade-Off Relationship*

Membrane gas separation has been used for the purification of hydrogen (in H2/CO2, H2/CH<sup>4</sup> and H2/N<sup>2</sup> gas mixtures) in refineries and the petrochemical industry, for the separation of CO2/CH<sup>4</sup> mixtures (in natural gas sweetening and biogas upgrading) and for the treatment of flue gas (CO2/N2) [24,30,57–59].

As stated above, permeability and selectivity represent key parameters for optimal gas separation. However, these parameters typically experience a trade-off relationship since highly permeable polymers tend to have less selectivity and vice versa. In this context, an experimental upper-bound relationship between selectivity and permeability was proposed by Robeson in order to benchmark membranes for gas separation [60,61]. This upper bound has been employed to relate gas permeability values in a different format. Later on, Robeson 1991 and Freeman provided a fundamental theory for this observation [62]. As more data on the gas separation characteristics of the polymers employed in the analysis published in 1991 were available, an updated compilation was published in 2008 [60], where the most significant changes were triggered by the information of perfluorinated polymers not reported in 1991. These data confirmed that when the permeability of a gas increases, the permeability of other gases also increases, since the diffusion coefficient of gases is related to the polymer free volume [53]. Figure 5, displays an example of a Robeson-type trade-off graph for CO2/CH4, where the CO2/CH<sup>4</sup> selectivity is shown against the CO<sup>2</sup> permeable support material [60,61].

Swaidan reported in 2015 new permeability/selectivity "upper bounds" for commercial membrane modules for air and hydrogen separation (H2/N2, H2/CH<sup>4</sup> and O2/N2) [63]. The Robeson upper bound behavior was redefined by Comesaña-Gandara in 2019 for CO2/N<sup>2</sup> and CO2/CH<sup>4</sup> separations using ultra-permeable Polymers of Intrinsic Microporosity (PIM) [64].

By transferring this trade-off relationship to the Robeson upper bound, the optimum balance involving a high selectivity in combination with a high permeability is determined. Nowadays, the research in this field is focused on developing new polymer materials capable of exceeding the upper bounds for the most relevant gas pairs. The key variables of the upper bound plots from the upper bound correlations *P<sup>i</sup>* = *k α n i j* are tabulated in Table 1, for the present upper bound data against the previous upper bound data.

[62]. As more data on the gas separation characteristics of the polymers employed in the analysis published in 1991 were available, an updated compilation was published in 2008 [60], where the most significant changes were triggered by the information of perfluorinated polymers not reported in 1991. These data confirmed that when the permeability of a gas increases, the permeability of other gases also increases, since the diffusion coefficient of gases is related to the polymer free volume [53]. Figure 5, displays an example of a Robeson-type trade-off graph for CO2/CH4, where the CO2/CH4 selectivity is shown

Swaidan reported in 2015 new permeability/selectivity "upper bounds" for commercial membrane modules for air and hydrogen separation (H2/N2, H2/CH4 and O2/N2) [63]. The Robeson upper bound behavior was redefined by Comesaña-Gandara in 2019 for CO2/N2 and CO2/CH4 separations using ultra-permeable Polymers of Intrinsic Mi-

against the CO2 permeable support material [60,61].

croporosity (PIM) [64].

**Figure 5.** CO2/CH4 Robeson plot for conventional glassy polymers. CA: cellulose acetate; PPO: poly(phenylene oxide); PTMST: Poly(trimethylsilylpropyne); TB-Bis A-PC: tetrabromobisphenol A poly(carbonate), Matrimid®: commercial poliimide. Permeabilities for single gases were measured in the range 25–35 °C and pressures from 1 to 20 bar. The continuous line stands for the 2008 upper bound, while dashed line represents the 1991 upper bound (adapted from Galizia and co-workers, 2017). **Figure 5.** CO2/CH<sup>4</sup> Robeson plot for conventional glassy polymers. CA: cellulose acetate; PPO: poly(phenylene oxide); PTMST: Poly(trimethylsilylpropyne); TB-Bis A-PC: tetrabromobisphenol A poly(carbonate), Matrimid®: commercial poliimide. Permeabilities for single gases were measured in the range 25–35 ◦C and pressures from 1 to 20 bar. The continuous line stands for the 2008 upper bound, while dashed line represents the 1991 upper bound (adapted from Galizia and coworkers, 2017).





#### *4.2. Physical Aging and Plasticization*

Plasticization is a frequently observed problem affecting the performance of membranes for gas separation (mostly from glassy polymers) [66,67]. Plasticization occurs when the gas concentration inside a polymer increases, causing swelling. As a result, the free volume and chain movement in the polymer material increase and in turn, the coefficients of gas diffusion increase and diffusion selectivity decreases [53,68]. A typical phenomenon observed during plasticization of glassy polymers is the increase in the permeability of a pure (or mixed) gas as the partial pressure (upstream) of the gas increases [67] caused by the loss of the polymer selectivity. The permeability increase is driven by the increase in diffusion coefficient, which in turn is governed by the penetrant (upstream) pressure [69]. CO<sup>2</sup> is the gas most commonly investigated in plasticization studies [70–73]. Gas sorption is known to increase after exposing a glassy polymer to CO<sup>2</sup> at a given pressure for a certain timeframe, which can even affect the mechanical properties of the polymer [74]. For glassy polymers, plasticization typically occurs at pressures of 10–40 bars and CO<sup>2</sup> concentration of 38 <sup>±</sup> 7 cm<sup>3</sup> (STP)/cm<sup>3</sup> polymer. Since pressure is related to CO<sup>2</sup> concentration in the polymer, it has been hypothesized that each polymer needs the same CO<sup>2</sup> concentration to induce plasticization but a different pressure to achieve it. As a rule of thumb, polymers that absorb more CO<sup>2</sup> are more likely to plasticize than those that absorb less CO<sup>2</sup> at a given pressure [53]. The thickness of a glassy polymer film (membrane) represents a key factor in the plasticization process because thinner films tend to be more sensitive to CO<sup>2</sup> pressure changes. Thus, a thin film tends to plasticize more quickly [75].

There is a wide variety of glassy polymers with outstanding performance in gas separations. These materials, by their nature, are not in equilibrium and have a high free volume due to their inefficient packing (caused by the movement of their chains), which avoids fully equilibrium properties to be reached [76]. This gradual approach to equilibrium influences various properties that change over time and consequently the material undergoes "physical aging". This frequent drawback affecting the membrane performance is a steady continuation of the glass transition that sets in around Tg. Thus, physical ageing influences all temperature-dependent properties that change significantly and sharply at Tg. Ageing can be explained by the free-volume theory (Figure 6). The free-volume concept assumes that the transport mobility of the particles depends mainly on the degree of packing of the system. If packing is efficient, the number and size of free volume elements are reduced, and thus the gas diffuses slower through the membrane over time [76]. The rate of physical ageing should then decrease over time because, when the free volume gradually decreases, the driving force governing physical ageing decreases, and also the pace of segmental movements that help reorganize the polymer chains decreases [53].

Physical ageing, apart from reducing gas permeability, also impacts other physical properties with an increase in internal energy concomitant with an increase in entropy [77]. Therefore, as the polymer ages, the free volume decreases along with permeability (although at slower rates as time goes on), which is accompanied by an increase in selectivity as a consequence of the reduction of membrane flux over time [53].

chains decreases [53].

**Figure 6.** The qualitative free volume concept. adapted from Struik, 1978. An increasing degree of packing entails a decrease in the mobility. At a critical degree of packing, the mobility steeply falls to zero. **Figure 6.** The qualitative free volume concept. adapted from Struik, 1978. An increasing degree of packing entails a decrease in the mobility. At a critical degree of packing, the mobility steeply falls to zero.

less CO2 at a given pressure [53]. The thickness of a glassy polymer film (membrane) represents a key factor in the plasticization process because thinner films tend to be more sensitive to CO2 pressure changes. Thus, a thin film tends to plasticize more quickly [75]. There is a wide variety of glassy polymers with outstanding performance in gas separations. These materials, by their nature, are not in equilibrium and have a high free volume due to their inefficient packing (caused by the movement of their chains), which avoids fully equilibrium properties to be reached [76]. This gradual approach to equilibrium influences various properties that change over time and consequently the material undergoes "physical aging". This frequent drawback affecting the membrane performance is a steady continuation of the glass transition that sets in around Tg. Thus, physical ageing influences all temperature-dependent properties that change significantly and sharply at Tg. Ageing can be explained by the free-volume theory (Figure 6). The freevolume concept assumes that the transport mobility of the particles depends mainly on the degree of packing of the system. If packing is efficient, the number and size of free volume elements are reduced, and thus the gas diffuses slower through the membrane over time [76]. The rate of physical ageing should then decrease over time because, when the free volume gradually decreases, the driving force governing physical ageing decreases, and also the pace of segmental movements that help reorganize the polymer

Physical ageing, apart from reducing gas permeability, also impacts other physical properties with an increase in internal energy concomitant with an increase in entropy [77]. Therefore, as the polymer ages, the free volume decreases along with permeability (although at slower rates as time goes on), which is accompanied by an increase in selec-

tivity as a consequence of the reduction of membrane flux over time [53].

Membrane thickness represents another factor influencing physical ageing. According to Baker and Low (2014), the free volume elements migrate to the surface as bubbles, leaving a viscous liquid, with the migration distance being proportional to the square of the thickness of the membrane. Therefore, rearrangement and loss of permeability occurs in a short time in thin membranes [46]. In this context, Tiwari and co-workers investigated Membrane thickness represents another factor influencing physical ageing. According to Baker and Low (2014), the free volume elements migrate to the surface as bubbles, leaving a viscous liquid, with the migration distance being proportional to the square of the thickness of the membrane. Therefore, rearrangement and loss of permeability occurs in a short time in thin membranes [46]. In this context, Tiwari and co-workers investigated the impact of physical ageing on gas permeability in thin and thick membranes manufactured with "high free-volume" glassy polymers (e.g., PIM-1). The results of this study showed a dominant ageing effect in thin films, where even physical ageing overcame the CO<sup>2</sup> plasticization effects [71]. Figure 7 displays an example of the time course of the decrease in membrane permeability. This effect, using Matrimid® coated with polydimethylsiloxane (PDMS) membranes, was investigated by Rowe and co-workers (2009), who observed that ageing rapidly increases in thinner membranes [76]. Likewise, Xia and co-workers (2014) investigated both the effect of the membrane thickness on ageing and the influence of the ageing time on the plasticization using a commercial polyimide membrane, Matrimid®, for gas separation [78]. This study concluded that membranes become more vulnerable to CO<sup>2</sup> plasticization as their thickness decreases and the ageing time increases [78]. Finally, it is worth mentioning that the ageing process can be reversed by heating the membrane above Tg [79].

#### *4.3. Novel Polymeric Membrane Materials for Gas Separation*

Good mechanical strength, sorption capacity and chemical resistance rank among the most relevant criteria for selecting polymeric material for gas separation. However, the membrane permeability, the capacity of the polymer to withstand swelling mediated plasticization and the processability of the polymer into a useful asymmetric or thin film composite morphology have been identified as key properties of membrane materials. Moreover, the polymer material should exhibit a good interaction with at least one of the components of the mixture in order to induce an effective separation [29]. Today, research in the field of gas separation is devoted to the development of novel membranes materials with superior permeability and selectivity performance exceeding the latest published Robeson upper bound limit, and consequently overcome the trade-off effect of conventional membranes [60,61,63–65].

heating the membrane above Tg [79].

the impact of physical ageing on gas permeability in thin and thick membranes manufactured with "high free-volume" glassy polymers (e.g., PIM-1). The results of this study showed a dominant ageing effect in thin films, where even physical ageing overcame the CO2 plasticization effects [71]. Figure 7 displays an example of the time course of the decrease in membrane permeability. This effect, using Matrimid® coated with polydimethylsiloxane (PDMS) membranes, was investigated by Rowe and co-workers (2009), who observed that ageing rapidly increases in thinner membranes [76]. Likewise, Xia and coworkers (2014) investigated both the effect of the membrane thickness on ageing and the influence of the ageing time on the plasticization using a commercial polyimide membrane, Matrimid®, for gas separation [78]. This study concluded that membranes become more vulnerable to CO2 plasticization as their thickness decreases and the ageing time increases [78]. Finally, it is worth mentioning that the ageing process can be reversed by

**Figure 7.** Effect of physical aging on CH4 permeability in Matrimid® films as a function on time and thickness at 35 °C and 2 atm (Adapted from Rowe and co-workers, 2009). **Figure 7.** Effect of physical aging on CH<sup>4</sup> permeability in Matrimid® films as a function on time and thickness at 35 ◦C and 2 atm (Adapted from Rowe and co-workers, 2009).

*4.3. Novel Polymeric Membrane Materials for Gas Separation*  Good mechanical strength, sorption capacity and chemical resistance rank among the most relevant criteria for selecting polymeric material for gas separation. However, the membrane permeability, the capacity of the polymer to withstand swelling mediated plasticization and the processability of the polymer into a useful asymmetric or thin film composite morphology have been identified as key properties of membrane materials. Moreover, the polymer material should exhibit a good interaction with at least one of the components of the mixture in order to induce an effective separation [29]. Today, research in the field of gas separation is devoted to the development of novel membranes materials with superior permeability and selectivity performance exceeding the latest published Robeson upper bound limit, and consequently overcome the trade-off effect of conventional membranes [60,61,63–65]. According to Galizia and co-workers [55], most of the polymers developed for gas According to Galizia and co-workers [55], most of the polymers developed for gas separation membranes in the last 30 years were evaluated without systematically proving their superior performance compared to the existing materials. Due to their high flexibility, one of the most synthesized families of materials for creating and understanding structureproperty relationships are polyimides [55]. However, it has not been possible to significantly improve the structure-property balance of polyimides-based membranes. Therefore, despite polymeric membranes can be utilized in the separation of almost any gas mixtures such as O2/N<sup>2</sup> separation, hydrogen purification (H2/N2, H2/CH4, and H2/CO), CO2/CH<sup>4</sup> biogas mixtures and vapor/gas separation, it is necessary to move beyond conventional polymers. In this context, new membrane materials for gas separation must provide higher permeabilities and permselectivities than conventional membranes. In addition, the production of new membranes for gas separation must consider good film-forming, good mechanical properties, absence of microdefects in the thin film, outstanding thermal and chemical stability, and absence of ageing [52].

separation membranes in the last 30 years were evaluated without systematically proving Poly(benzimidazoles) (PBIs) often exhibit glass transition temperatures (Tg) greater than 400 ◦C, and a good thermal, mechanical and chemical stability, which is not typical among glassy polymers. Celazole® (PBI Performance Products, Inc., Charlotte, NC, USA) (sometimes named as m-PBI) is an example of membranes derived from PBIs that exhibit promising gas transport properties. However, Celazole® exhibits a low solubility in common solvents due to its structural features and intermolecular hydrogen bonding forces [80,81]. Borjigin and co-workers synthesized a novel PBI with sulfonyl moieties by performing a structural modification using 3,30 ,4,40 -tetraamino-diphenylsulfone (TADPS) as monomer, which entailed a good solubility in common solvents such as N-methyl-2- Pyrrolidinone, NMP, N,N-dimethylacetamide, DMAc and dimethylsulphoxide, DMSO. Unfortunately, despite the good thermal stability and high permeabilities of PBIs, these materials are still susceptible to physical ageing [82].

Aromatic Polyamides (PA) were one of the first aromatic linear polymers considered thermally stable. PA typically exhibit a high cohesive energy density, a strong tendency for highly efficient polymer chain packing and a semicrystalline morphology [83]. Additionally, PA reported also a fair balance of properties: good chemical stability, high thermal resistance, good mechanical properties and an easier processability than aromatic polyimides [84,85]. However, PA support a low gas permeability of small molecules compared to polyimides. In recent years, there have been many attempts to improve PA gas separation performance by introducing bulky moieties, contoured structures or by introducing hexa-fluoropropane parts into the macromolecular chain, but with a limited success [86,87]. Likewise, Lozano and co-workers carried out in situ sialylation of diamines by adding trimethylchlorosilane (TMSCl) to the diamine solutions that, after the addition of a diacid chloride, resulted in high molecular weight aromatic polyamides, which guarantees high performance [88].

On the other hand, the so-called nanoporous polymers, as a result of their extremely fine nanoporous structure, have shown an outstanding performance in terms of gas separation. Examples of these materials are:


Recently, significant advances were achieved in the optimization of Mixed Matrix Membranes (MMMs) [35]. MMMs allow tuning the transport properties of conventional polymers for target applications by combining the high permeability of the polymer and the good selectivity of the filler materials.

#### **5. Mixed Matrix Membranes for Gas Separation**

Polymeric membranes have been successful in some gas separation processes such as natural gas sweetening but are still subject to the trade-off between permeability and selectivity, and the impact of physical ageing and plasticization, which makes them unstable for industrial applications. Recently, Barker (2014) reviewed the barriers limiting the development of membranes with high selectivity and permeance from the last 35 years and identified the need to develop new materials for new and future membrane applications [46]. Therefore, most research efforts are devoted nowadays to the development of new polymeric materials and membranes material such as zeolites, metal organic framework (MOF), carbon molecular sieves, carbon nanotubes and graphenes to improve the gas separation performance of membranes [93].

In this context, hybrid materials known as MMMs have been manufactured by adding inorganic materials as the disperse phase into polymers in order to take advantage of the processability of polymers and simultaneously overcome the trade-off between permeability and selectivity. The mixed matrix membranes concept has been described in multiple scientific publications. According to the most recent definitions, MMMs results from the combination of an inorganic or inorganic-organic hybrid material (micro or nanoparticles)—in the form of dispersed particles called additive or filler—and a polymeric matrix-continuous phase (Figure 8) [30,93]. PIMs and HPI are the most commonly used polymeric matrices, and zeolites the most common fillers. Moreover, MMMs have been recently thermally treated to obtain MMM-TR with outstanding gas transport properties for gas pairs such as CO2/CH4, O2/N2, H2/CO2, etc. [94–98].

for gas pairs such as CO2/CH4, O2/N2, H2/CO2, etc. [94–98].

**5. Mixed Matrix Membranes for Gas Separation** 

gas separation performance of membranes [93].

Polymeric membranes have been successful in some gas separation processes such as natural gas sweetening but are still subject to the trade-off between permeability and selectivity, and the impact of physical ageing and plasticization, which makes them unstable for industrial applications. Recently, Barker (2014) reviewed the barriers limiting the development of membranes with high selectivity and permeance from the last 35 years and identified the need to develop new materials for new and future membrane applications [46]. Therefore, most research efforts are devoted nowadays to the development of new polymeric materials and membranes material such as zeolites, metal organic framework (MOF), carbon molecular sieves, carbon nanotubes and graphenes to improve the

In this context, hybrid materials known as MMMs have been manufactured by adding inorganic materials as the disperse phase into polymers in order to take advantage of the processability of polymers and simultaneously overcome the trade-off between permeability and selectivity. The mixed matrix membranes concept has been described in multiple scientific publications. According to the most recent definitions, MMMs results from the combination of an inorganic or inorganic-organic hybrid material (micro or nanoparticles)—in the form of dispersed particles called additive or filler—and a polymeric matrix-continuous phase (Figure 8) [30,93]. PIMs and HPI are the most commonly used polymeric matrices, and zeolites the most common fillers. Moreover, MMMs have been recently thermally treated to obtain MMM-TR with outstanding gas transport properties

**Figure 8.** Illustration of a mixed matrix membrane (adapted from Lin and co-workers, 2018) [99]. **Figure 8.** Illustration of a mixed matrix membrane (adapted from Lin and co-workers, 2018) [99].

MMMs have emerged as a promising material for gas separation in membrane technology. The main objective of the manufacture of MMMs is to provide solutions to the existing permeability versus selectivity trade-off relationship of gas separation polymeric membranes by taking advantage of the superior properties of inorganic particles [100,101]. In addition, MMMs compensate the unavoidable fragility limitation of inorganic membranes using a flexible polymer as the continuous matrix. These features provide MMMs with the potential to achieve a greater selectivity, permeability (caused by increasing the diffusion coefficients) or both, compared to existing polymeric membranes and to exceed the upper limit proposed by Robeson. These organic and inorganic materials employed as fillers could have a unique structure, surface chemistry and mechanical strength. Inorganic fillers contribute to enhanced diffusivity selectivity by acting as molecular sieves due to their precise pore size and shape and geometry, thus overcoming the MMMs have emerged as a promising material for gas separation in membrane technology. The main objective of the manufacture of MMMs is to provide solutions to the existing permeability versus selectivity trade-off relationship of gas separation polymeric membranes by taking advantage of the superior properties of inorganic particles [100,101]. In addition, MMMs compensate the unavoidable fragility limitation of inorganic membranes using a flexible polymer as the continuous matrix. These features provide MMMs with the potential to achieve a greater selectivity, permeability (caused by increasing the diffusion coefficients) or both, compared to existing polymeric membranes and to exceed the upper limit proposed by Robeson. These organic and inorganic materials employed as fillers could have a unique structure, surface chemistry and mechanical strength. Inorganic fillers contribute to enhanced diffusivity selectivity by acting as molecular sieves due to their precise pore size and shape and geometry, thus overcoming the properties of common polymeric membranes [55,102]. Overall, MMMs support unprecedented increases in permeability while maintaining selectivity by introducing fillers into the polymeric matrix, due to the increase in diffusion coefficients.

The first reports of the manufacture of MMMs were published in the 1970s. For instance, Paul and Kemp (1973) added a commercial zeolite (Molecular Sieve Type 5A) as a filler to a PDMS rubber used as polymer matrix [103]. A good interaction between the polymer and the zeolite was observed due to the flexibility of the rubber polymer and a large increase of a delayed diffusion time lag effect. However, high fluxes of gas in the polymer matrix can result in a low improvement in the selectivity [30]. In the last decade, manufacture of MMMs, researcher on of their mechanical and transport properties, as well as the investigation of their nanostructure have increased a significant attention in the membrane research field [52].

#### *5.1. Factors Influencing Mixed-Matrix Membrane Manufacture*

Multiple factors during the preparation of MMMs can cause: interfacial defects caused by particle sedimentation (due to the differences in physical properties and density with the polymer), migration of filler particles or agglomeration in the surface, especially when the fillers load is high due to the fact that this scenario increases the diffusion distance within the solid phase agglomerate [100].

According to Noble (2011), the compatibility between the disperse and continuous phases in terms of permeability is an important factor to consider due to the fact that the resistance to mass transfer is typically much higher in phases with much lower gas permeability [104]. In addition, there is a relationship between the filler particle size and membrane thickness, as smaller particles provide a higher surface area/volume ratio, which supports a greater mass transfer between phases. Finally, an effective contact between the two phases is necessary to prevent any gaps between them that could block the access to the pores [104].

Today, the achievement of the desired morphology, mechanical/chemical stability, and gas separation properties in MMMs requires overcoming multiple manufacturing challenges such as: obtaining a flawless interface to guarantee a good separation performance of the membranes, obtaining a homogeneous dispersion between the two phases, avoiding agglomerations responsible for low selectivity and finally selecting materials with excellent separation properties and good compatibility between the phases [102,105].

#### 5.1.1. Morphologies of the Mixed-Matrix Membrane

The desired morphology of MMMs would include the absence of defects in the polymer–particle interface and must ensure gas transport across the dispersed phase instead through the continuous phase (polymeric matrix) (Figure 9) [101]. The advantages of morphology can be understood in terms of the ideal Maxwell model that represents the simplest case for mixed matrix transport properties [106]. This model, described by Robeson as a dilute suspension of particles in a polymeric matrix, was mainly developed for estimating dielectric properties of composites and describes the effective permeability of MMMs, *Pe f f* as follows [107,108]:

$$P\_{eff} = P\_c \left[ \frac{P\_d + 2P\_c - 2\,\Phi\_d \left(P\_c - P\_d\right)}{P\_d + 2\,P\_c + \Phi\_d \left(P\_c - P\_d\right)} \right] \tag{7}$$

where *P<sup>c</sup>* is the continuous phase permeability (i.e., polymer matrix), represents the dispersed phase permeability (i.e., filler) and *Φ<sup>d</sup>* is the dispersed phase volume fraction. Note that Equation (7) goes to the appropriate value of *P* in the limits as *Φ<sup>d</sup>* = 0 or 1. Maxwell's model can be complicated by assuming that the dispersed phase, being uniformly distributed, is encapsulated by an "interface" (region between polymer matrix and inorganic fillers) with characteristics different from both the dispersed and continuous phases [104,106]. The formation of the interface is attributed to the inhibition of the mobility of the polymer chains in compressive stress near the polymer–particle interface. Figure 10 shows a representation of the polymer matrix, the dispersed phase and the rigidified interface (three-phase MMM system) [109].

One of the disadvantages of this model for MMMs is the need to determine the transport properties (e.g., through kinetic sorption in monodispersed crystals) in order to obtain a good characterization of the dispersed phase [106]. Moreover, it is also only applicable to low filler loadings with free volume fractions lower than 0.2. In this context, high values of *Φ<sup>d</sup>* render the ideal Maxwell model useless. In addition, the Maxwell model does not consider the morphological properties of the filler such as particle shape, particle size distribution or the aggregation of filler particles [100].

Thus, the preparation of ideal MMMs entails a difficult procedure as a result of the formation of defects at the polymer–particle interface, which are typically caused by a weak particle-polymer adhesion, induced by the difference in properties between both phases [102]. These interface defects between the continuous and dispersed phases can impact membrane properties such as the membrane separation performance.

The most common factors responsible for interfacial defects can be divided into three main categories: (i) Interfacial voids or sieves-in-a-cage, (ii) Rigidified polymer layer around the inorganic fillers, and (iii) Particle pore blockage [100,105,106].

A low linkage between the continuous phase and the dispersed phase could lead to the formation of non-selective voids in the interfacial region (Figure 11, case i). Other factors responsible for interfacial voids formation are the modification of the polymer packing in the vicinity of the dispersed phase, the repulsive force between the two phases, the different thermal expansion coefficients and the elongation stress during fiber spinning [100]. In addition, interfacial voids or sieves-in-a-cage are attributed to the de-wetting of the polymeric chains on the external surface of the particles [101]. Moore and Koros (2005) observed that solvent evaporation, thermal effects and the resulting stresses at the

ers, 2010.

polymer-disperse phase interface cause defects such as interface void formation, due to the partial or apparent clogging of the dispersed phase [106]. The formation of these defects allows the gases to pass and, hence, deteriorates the apparent selectivity and increases the permeability of MMMs. [104,106]. The formation of the interface is attributed to the inhibition of the mobility of the polymer chains in compressive stress near the polymer–particle interface. Figure 10 shows a representation of the polymer matrix, the dispersed phase and the rigidified interface (three-phase MMM system) [109].

fillers) with characteristics different from both the dispersed and continuous phases

*Processes* **2022**, *10*, x FOR PEER REVIEW 16 of 41

**Figure 9.** Schematic diagram of an ideal MMM. This figure was adapted from Aroon and co-work-**Figure 9.** Schematic diagram of an ideal MMM. This figure was adapted from Aroon and coworkers, 2010.

**Figure 10.** Schematic representation of polymer matrix, dispersed phase, and their interphase.

around the inorganic fillers, and (iii) Particle pore blockage [100,105,106].

**Figure 10.** Schematic representation of polymer matrix, dispersed phase, and their interphase.

The most common factors responsible for interfacial defects can be divided into three main categories: (i) Interfacial voids or sieves-in-a-cage, (ii) Rigidified polymer layer

A low linkage between the continuous phase and the dispersed phase could lead to

the formation of non-selective voids in the interfacial region (Figure 11, case i). Other fac-

ing in the vicinity of the dispersed phase, the repulsive force between the two phases, the different thermal expansion coefficients and the elongation stress during fiber spinning [100]. In addition, interfacial voids or sieves-in-a-cage are attributed to the de-wetting of the polymeric chains on the external surface of the particles [101]. Moore and Koros (2005) observed that solvent evaporation, thermal effects and the resulting stresses at the polymer-disperse phase interface cause defects such as interface void formation, due to the partial or apparent clogging of the dispersed phase [106]. The formation of these defects allows the gases to pass and, hence, deteriorates the apparent selectivity and increases the

permeability of MMMs.

**Figure 11.** Schematic diagram of an interface void (**i**), rigidified polymer (**ii**) and partial blockage (**iii**) in MMMs. (Adapted from Aroon and co-workers, 2010). **Figure 11.** Schematic diagram of an interface void (**i**), rigidified polymer (**ii**) and partial blockage (**iii**) in MMMs. (Adapted from Aroon and co-workers, 2010).

Rigidified polymer layer around the inorganic fillers occurs when the polymer matrix chains, in direct contact with the filler surface, are rigidified as compared with the bulk polymer chains, which reduces the free volume and is related to a uniform tension around the particles [102,105]. Moore and Koros (2005) hypothesized that polymer rigidification (Figure 11, case ii) enhanced the diffusive selectivity and decreased membrane permeability [106]. Rigidified polymer layer around the inorganic fillers occurs when the polymer matrix chains, in direct contact with the filler surface, are rigidified as compared with the bulk polymer chains, which reduces the free volume and is related to a uniform tension around the particles [102,105]. Moore and Koros (2005) hypothesized that polymer rigidification (Figure 11, case ii) enhanced the diffusive selectivity and decreased membrane permeability [106].

Particle pore blockage occurs when the surface pores of the filler are partially blocked by the rigidified polymer chains (Figure 11, case iii). This clogging is usually generated by the presence of sorbent, solvent traces, a contaminant or a minor component in the feed gas, before, during and after the manufacture of the MMMs [105,106,108,110]. However, there is no accurate methodology to differentiate the influence of these factors. Based on previous research, if the pores are completely blocked, the gas cannot pass through the particle fillers, and no enhancement in selectivity over the neat polymer is reached as in the case of MMMs filled with nonporous particles. Particle pore blockage occurs when the surface pores of the filler are partially blocked by the rigidified polymer chains (Figure 11, case iii). This clogging is usually generated by the presence of sorbent, solvent traces, a contaminant or a minor component in the feed gas, before, during and after the manufacture of the MMMs [105,106,108,110]. However, there is no accurate methodology to differentiate the influence of these factors. Based on previous research, if the pores are completely blocked, the gas cannot pass through the particle fillers, and no enhancement in selectivity over the neat polymer is reached as in the case of MMMs filled with nonporous particles.

The formation of a rigidified polymer layer around the inorganic fillers and particle pore blockage are caused by sorption of a strongly retained molecule. In the first case, the strongly retained molecule completely prevents the penetrants of interest from entering the dispersed phase, while in the second case, the penetrants of interest enter or pass through the dispersed phase more slowly than usual [105,106]. The formation of a rigidified polymer layer around the inorganic fillers and particle pore blockage are caused by sorption of a strongly retained molecule. In the first case, the strongly retained molecule completely prevents the penetrants of interest from entering the dispersed phase, while in the second case, the penetrants of interest enter or pass through the dispersed phase more slowly than usual [105,106].

In summary, poor adhesion, mobility of polymer matrix chains and pore clogging by the matrix are just some of the phenomena observed when incorporating a dispersed phase into a continuous phase during the fabrication of MMMs. In summary, poor adhesion, mobility of polymer matrix chains and pore clogging by the matrix are just some of the phenomena observed when incorporating a dispersed phase into a continuous phase during the fabrication of MMMs.

#### Methods for Manufacturing Defect-Free Membranes Methods for Manufacturing Defect-Free Membranes

Poor adhesion and repulsive forces between the continuous and disperse phases, incompatibility between polymer and filler, solvent evaporation during membrane formation, polymer packing disruption in the vicinity of the inorganic phase, and different thermal expansion coefficients for polymer and filler can induce multiple interfacial defects and non-ideal morphologies in MMMs [102]. In order to avoid these interfacial defects and manufacture defect-free MMMs, the following methodological strategies have Poor adhesion and repulsive forces between the continuous and disperse phases, incompatibility between polymer and filler, solvent evaporation during membrane formation, polymer packing disruption in the vicinity of the inorganic phase, and different thermal expansion coefficients for polymer and filler can induce multiple interfacial defects and non-ideal morphologies in MMMs [102]. In order to avoid these interfacial defects and manufacture defect-free MMMs, the following methodological strategies have been applied:

been applied: An important factor during the manufacture of an ideal MMM with optimal performance is the homogeneous distribution (or dispersion) of the filler within the continuous phase in order to guarantee an effective filler/polymer contact [101]. In fact, a poor filler distribution can affect membrane performance by agglomeration, which leads to the formation

of non-selective interfacial voids [99]. Unfortunately, high filler loadings can sometimes result in particle aggregation, which can form voids within the particle aggregates that cannot be reached by polymer chain segments and act as channels facilitating gas molecules transport, thus reducing the selectivity of the MMMs. Similarly, high filler loadings can cause sedimentation, which also contributes to the poor dispersion of the filler into the continuous phase [101]. This filler agglomeration entails the creation of pinholes that cannot be reached by polymer segments, resulting in non-selective defects in MMMs [105].

In this context, the so called "priming" method created by Mahajan and Koros (2002) is the most common strategy to avoid filler agglomeration [111]. This technique can reduce the stress at the filler/polymer interface, thus resulting in an improved interaction between the polymer primed filler and the bulk polymer, concomitantly with a reduced agglomeration of the filler [101,102]. This prime method consists in dispersing the particles in a suitable solvent, subjecting them to sonication followed by coating the surface of the filler in suspension. This coating is carried out by adding a small percentage of homogeneous polymer solution prior to the dispersion in the bulk polymer solution [110]. On the other hand, the preparation of polymer diluted solutions to increase the viscosity and decrease membrane thickness have been proposed to avoid agglomeration since this methodology can reduce particle sedimentation. Alternatively, the membrane can be cast "quickly" so that the fillers do not have time to precipitate.

Finally, another approach to achieve flexibility during membrane formation is to mimic the use of a low Tg polymer by forming the membrane close to the Tg of the polymer matrix used as a precursor of the MMMs. An obvious limitation of this strategy is the common tendency to use suitable casting solvents that boil at temperatures below the Tg of a typical rigid polymer such as Matrimid® [112].

#### 5.1.2. Polymer Materials

The optimum selection of materials for both the continuous phase and the dispersed phase is a key factor during the development of MMMs since the properties of the precursor materials can affect the morphology and separation performance of membranes [105]. Despite the selection of optimum fillers being the major concern in the early manufacture of MMMs, the selection of the polymer used as the matrix greatly impacts the gas separation performance of MMMs [105].

In the field of gas separation using membranes, rubbery and glassy polymers have been traditionally used. Rubbery polymers contain flexible polymer chain structures and have the ability to stretch the chains apart, the chains returning to their original position when tension is released. Rubbery polymers also exhibit a high permeability and a low selectivity for the separation of common gas pairs, as a result of the different condensability of the gas components [30]. On the contrary, glassy polymers possess rigid chain structures with restricted segmental motion. This rigid chain structure offers desirable separation properties such as high selectivity combined with medium/low permeability [26]. The high selectivity of glassy polymers can be attributed to their lower free volume, the narrower distribution of the free volume and the lower flexibility of the polymer chains compared to their rubbery counterparts.

Due to the high degree of mobility, rubbery polymers ensure good adhesion between the polymeric matrix and the fillers, which can avoid interfacial voids and facilitate the manufacture of defect-free MMMs. However, a high mobility also entails a high permeability, which suggests that gas transport is dominated mainly by the polymer matrix and only a small portion is attributed to the filler. On the other hand, although glassy polymers exhibit superior properties to rubbery polymers, their rigid chain structure typically results in a poor adhesion of the pair polymer-filler, thus generating voids at the interface [101]. Therefore, the gas transport properties of the materials and adhesion between the phases should be carefully considered when selecting the polymer matrix [102]. In this context, novel polymers capable of separating gas mixtures by solubility selectivity are needed.

Material selection to manufacture MMMs is a difficult task, especially for glassy polymers. However, a considerable number of glassy polymers are being employed as continuous phase in MMMs such as cellulose acetate (CA), polyimide (PI), polysulfone (PSU), polyamide (PA), polypropylene (PP), polyethersulfone (PES), poly-vinylidene fluoride (PVDF) and perfluorinated materials, etc. [95,97,113–116]. Polymers such as PMP (4-methyl-2-pentyne), PTBA (*tert*-butylacetylene) and PTMSP (1-trimethylsilyl-1-propyne), namely "reverse-selective polymers", have also been used as continuous phase due to their high fractional free volume. In the latter membranes, the gas transport mechanism shifts from being controlled by diffusivity to being controlled by solubility (contrarily to the observations in traditional low-free volume glassy polymers), and therefore, transport properties are favored for more condensable species (e.g., CO2) than for smaller molecules [117].

In recent years, the most common materials developed for the manufacture of MMMs are divided into three groups: (i) Advanced high permeability polymers (PIM, Polyimides and TR polymers), (ii) Polymers with moderate permeability and high selectivity and (iii) Ionic liquid/poly ionic liquids with high permeability and high selectivity [118]. For instance, a limited number of researchers have studied the transport properties using PIM-1 as a continuous phase for the separation of CO2/CH<sup>4</sup> with a notable increase in permeabilities compared to the matrix. These studies also demonstrated that the introduction of a filler (ZIF-8) to this polymeric matrix mediates an increase in free volume, as a result of the combination of the contributing cavities and looser polymeric chains at the boundary between the filler and the PIM-1 matrix [98,119,120]. These membranes represent good candidates for CO<sup>2</sup> removal from biogas, although they suffer from severe physical ageing. On the other hand, the introduction of TR materials (e.g., hydroxypolyimide, HPI, and hydroxypolyamide, HPA) as a continuous phase has been proposed as a promising alternative since TR polymers show superior gas separation properties and can help to reduce non-selective voids during the manufacture of MMMs [97,121–123].

The permeation properties of MMMs are mainly determined by the shape and size of the filler, its pore size, pore size distribution, sedimentation and agglomeration properties and the gas separation operational conditions (gas composition, pressure, and temperature). In addition, the permeability of both the continuous and disperse phase should be comparable since a continuous phase with a high permeability reduces the contribution of the filler to gas separation [100].

#### 5.1.3. Advanced Functional Fillers

The major challenges encountered during the manufacture of MMMs are the selection of adequate fillers that provide a good interaction with the polymer for the enhancement of gas separation properties. Indeed, the addition of suitable fillers in the polymer matrix results in a significant increase in the overall separation efficiency and therefore in a superior gas selectivity performance by MMMs [35].

There is a great variety of fillers that have been used in the development of MMMs as disperse phase. In recent years, the synthesis of novel organic/inorganic membrane materials has yielded in emerging materials used as high-performance fillers in MMMs for gas separation. Here, the most studied fillers (with a good compatibility with polymers) to date are briefly reviewed, particularly Zeolites, Metal Organic frameworks (MOFs) [124], Covalent Organic Frameworks (COFs) [125], Porous Aromatic Framework (PAFs) [126] and Porous Polymer Networks (PPNs) [127], recently named Porous Organic Polymers (POPs) [128] (Figure 12). Due to the fact that only few of the fillers used in the field of mixed matrix membranes have been mentioned, this article will only focus on representative work for the separation of CO<sup>2</sup> and H<sup>2</sup> from biogas and biohydrogen.

of mixed matrix membranes have been mentioned, this article will only focus on repre-

sentative work for the separation of CO2 and H2 from biogas and biohydrogen.

**Figure 12.** Some types of fillers used in MMM preparation (Adapted from Chakrabarty and coworkers, 2022) [129]. **Figure 12.** Some types of fillers used in MMM preparation (Adapted from Chakrabarty and coworkers, 2022) [129].

• Zeolites • Zeolites

Based on structural features, zeolites are an inorganic material frequently used as disperse phase for the manufacture of MMMs. Zeolites are hydrated aluminosilicate materials with opened three-dimensional framework structures that possess regular intracrystalline cavities and channels of molecular dimension (3–12 Å). Its structure is formed by SiO4 and AlO4 tetrahedral, by sharing an oxygen ion. Due to the presence of the tetrahedron AlO4, the chemical characteristics of the frameworks are determined, which tend to have negative charge compensated by alkali or earth alkali cations, located in the micropores [130]. Moreover, zeolites are materials with shape/size-selective nanopores [131,132]. The pore sizes of zeolites range between 0.3 and 1 nm, with pore volumes of about 0.10–0.35 cc/g [133]. There are 176 types of zeolite structures approved by the IZA Structures Commission (IZA-SC) in February 2007 and assigned with a 3-letter code [134], of which, according to Bastani and co-workers, the most common are: 4A (3.8 Å), 5A (4.3 Å), 13X (7.4 Å), NaY (7.4 Å), ZSM-5 (5.1 × 5.5 Å and 5.3 × 5.6 Å), SSZ-13 (3.8Å), Based on structural features, zeolites are an inorganic material frequently used as disperse phase for the manufacture of MMMs. Zeolites are hydrated aluminosilicate materials with opened three-dimensional framework structures that possess regular intracrystalline cavities and channels of molecular dimension (3–12 Å). Its structure is formed by SiO<sup>4</sup> and AlO<sup>4</sup> tetrahedral, by sharing an oxygen ion. Due to the presence of the tetrahedron AlO4, the chemical characteristics of the frameworks are determined, which tend to have negative charge compensated by alkali or earth alkali cations, located in the micropores [130]. Moreover, zeolites are materials with shape/size-selective nanopores [131,132]. The pore sizes of zeolites range between 0.3 and 1 nm, with pore volumes of about 0.10–0.35 cc/g [133]. There are 176 types of zeolite structures approved by the IZA Structures Commission (IZA-SC) in February 2007 and assigned with a 3-letter code [134], of which, according to Bastani and co-workers, the most common are: 4A (3.8 Å), 5A (4.3 Å), 13X (7.4 Å), NaY (7.4 Å), ZSM-5 (5.1 × 5.5 Å and 5.3 × 5.6 Å), SSZ-13 (3.8Å), etc. [135].

etc. [135]. Zeolites have received increasing attention due to the fact that they have a wide range of structures with different chemical compositions and physicochemical properties. Zeolites are widely used as catalysts, adsorbents and ion exchange media. The transport of gas molecules starts by molecular adsorption into the pores, diffusion onto the zeolite surface and desorption into the permeate. The gas molecules that have the strongest attractive force with the zeolite pores are those with the highest dipole moment, which is why CO2 is adsorbed most strongly on zeolites, followed by H2, CH4 and N2 [135]. The success of zeolite-based MMMs is attributed to the type of zeolites used and their adsorption capacity [133]. For example, zeolite 5A has a CO2 adsorption capacity of 222 mg CO2/g adsorbent at 0.1 MPa. Moreover, due to their good selectivity and adsorbent selection parameters, zeolite 5A turns out to be a better adsorbent for removing CO2 and N2O from Zeolites have received increasing attention due to the fact that they have a wide range of structures with different chemical compositions and physicochemical properties. Zeolites are widely used as catalysts, adsorbents and ion exchange media. The transport of gas molecules starts by molecular adsorption into the pores, diffusion onto the zeolite surface and desorption into the permeate. The gas molecules that have the strongest attractive force with the zeolite pores are those with the highest dipole moment, which is why CO<sup>2</sup> is adsorbed most strongly on zeolites, followed by H2, CH<sup>4</sup> and N<sup>2</sup> [135]. The success of zeolite-based MMMs is attributed to the type of zeolites used and their adsorption capacity [133]. For example, zeolite 5A has a CO<sup>2</sup> adsorption capacity of 222 mg CO2/g adsorbent at 0.1 MPa. Moreover, due to their good selectivity and adsorbent selection parameters, zeolite 5A turns out to be a better adsorbent for removing CO<sup>2</sup> and N2O from air and for separating CO2/CH<sup>4</sup> gas mixture compared with MOFs [136]. Likewise, NaX zeolites have an adsorption capacity of 263 mg CO2/g adsorbent, which renders it an excellent candidate for separating CO<sup>2</sup> from CH<sup>4</sup> [137]. Similarly, zeolite 13X is another kind of zeolite with a great CO<sup>2</sup> adsorption capacity of 324 mg CO2/g adsorbent, making it an excellent candidate for the purification of methane from natural gas [138].

One of the most relevant properties of this material is sorption and diffusion due to the different sizes of its channels and cavities, which determines the free space or void volume of the MMMs [132]. Zeolites possess interesting thermal and chemical stability, a well-defined microstructure and high mechanical strength, which makes them suitable candidates to be used as a dispersed phase in the manufacture of MMMs [135,139]. Interestingly, the low packing density of zeolites makes them an unsuitable material for gas separation, however their use as a dispersed phase in the fabrication of MMMs provides an opportunity to overcome this problem [101]. In addition, zeolites exhibit a permeability and selectivity superior to polymeric materials due to their unique molecular sieving characteristics.

Zeolites have traditionally received attention as potential fillers due to their thermal stability and promising separation and transport properties. Thus, the specific sorption properties and shape selectivity of zeolites, when applied to polymers with easy processability, provide superior gas separation properties to MMMs [132]. Several investigations have shown that the transport properties of MMMs are affected by the type of zeolite used. For instance, MMMs prepared with zeolite 4A support an effective O2/N<sup>2</sup> separation due to their adequate pore size (3.8–4.0 Å), with selectivities up to 37. Membranes with zeolite 5A as filler exhibit much higher H2/N<sup>2</sup> and O2/N<sup>2</sup> selectivity than membranes with zeolite 4A as filler. In this context, zeolites are still of interest for membrane investigations despite providing low permeabilities for O<sup>2</sup> (0.8 Barrer) [111,140]. Ahmad and co-workers (2021) investigated the CO2/CH<sup>4</sup> separation properties behavior of MMMs fabricated with SSZ-16 zeolite at different loading ratios as filler and 6FDA-DAM:DABA as polymeric matrix. As a result, MMMs loaded at 5 wt.% SSZ-16 supported up to two-fold higher CO<sup>2</sup> permeability with respect to pristine membranes, while maintaining the CO2/CH<sup>4</sup> selectivity. In addition, these authors found that a 5 wt.% loading provides an excellent filler dispersion [141]. Zhang and co-workers (2008) prepared MMMs based on Matrimid and ZSM-5 zeolite, increasing the H2/N<sup>2</sup> separation from 79 for Matrimid and 143 at 10% load, while the ideal H2/CH<sup>4</sup> separation factor increased from 83 to 169 at 20% load, further confirming the excellent interactions between the particles and the polymer [142]. Ebadi Amooghin and co-workers (2016) modified zeolite-Y by introducing silver cations (via ion-exchange method) to form Ag+ zeolite and use it as filler on Matrimid® to form novel Matrimid®/AgY MMMs. In this particular study, CO<sup>2</sup> permeability increased from 8.34 Barrer for the pure membrane up to 18.62 Barrer for Matrimid/AgY (15 wt.%) without affecting CO2/CH<sup>4</sup> selectivity, which increased from 36 to 60 for pure membrane and MMMs, respectively [143]. Finally, Montes Luna and co-workers modified the natural zeolite Clinoptilolite (CLINO) with CaCl<sup>2</sup> in an aqueous solution to replace Na+ ions with Ca2+ ions, thus enhancing gas separation properties for CH4/N2/CO<sup>2</sup> gas mixtures [144].

Despite the promising results obtained in the laboratory, MMMs with zeolites as the dispersed phase have not been commercially exploited due to the poor adhesion at the zeolite–polymer interface (especially glassy polymers), resulting in a "sieve-in-a-cage" morphology. This defect is responsible for the non-selective penetration of gas molecules, the reduction in selectivity and poor mechanical properties, especially in the formation of thin films. In addition, high zeolite loadings often result in non-uniform dispersions in MMMs [55].

• Metal Organic Frameworks

Metal Organic Frameworks, MOFs, are hybrid materials prepared by combining organic ligands with metal ions or metal-oxide clusters. Ligands play a key role in defining the final framework of MOFs, while metal ions also influence the structure of MOFs due to their tunable geometries [145]. MOFs are highly porous chemically mutable materials, with unique properties, different pore sizes and shapes, and multiple functional sites and high specific surface areas that allow creating a wide variety of crystals [93,118]. Compared to zeolites, the tunable structure of MOFs results in well-dispersed fillers, which allows

high affinity organic linkers in MOFs and polymer chains, thus reducing non-selective defects at the polymer–filler interface. The partially organic nature of MOFs supports a better polymer-filler interaction, which represents a structural advantage compared to other porous materials [55].

In order to optimize gas diffusion and selectivity, new strategies for the formation of high-performance MMMs using MOFs as dispersed phase have been assessed. A wide variety of MOFs subfamilies with ultrasmall aperture sizes have been chosen as potential fillers. The most typically studied MOFs are Zeolitic Imidazolate Frameworks (ZIFs), copper-based MOFs (Cu-MOFs), Materials Institute Lavoisier (MIL) series, MOF-74 series, and University of Oslo-66 (UiO-66) series [146]. ZIFs possess a similar topology to zeolites with tunable pore structures and with high thermal and chemical stabilities [35]. In this context, ZIF-8, HKUST-1, MIL-53, MIL-101, MOF-74, and UiO-66 have been specifically tested. For instance, ZIFs-8 are a new class of porous crystals (3.4 Å pore aperture and 11.6 Å cages) [147] composed of tetrahedral metal ions (typically zinc or cobalt) forming extended three-dimensional structures bridged by imidazolate (Im) [148].

Khdhayyer and co-workers studied the gas transport properties of MMMs based on PIM-1 as polymeric matrix and three isoreticular MOFs (UiO-66, UiO-66-NH<sup>2</sup> and UiO-66-(COOH)2) as fillers, confirming the good prospects of these MMMs for CO<sup>2</sup> removal from biogas [98]. Ahmad and co-workers investigated the gas separation properties of MMMs using three types of Zr-based MOFs (UiO-66 and its functionalized derivatives, UiO-66-NH<sup>2</sup> and UiO-66-NH-COCH3) as fillers on 6FDA-DAM as a polymeric matrix. The addition of these particles improved both CO<sup>2</sup> permeability and CO2/CH<sup>4</sup> selectivity. For instance, permeabilities of the polymer 6FDA-DAM and its 14–16 wt.% Zr MOFs MMMs, tested with binary (30:70 vol.%; CO2:CH4) feed mixture, were 231, 541, 359 y 291 Barrer, and for tertiary (30:5:65 vol.%; CO2:H2S:CH4) feed mixture, permeabilities of 167, 385, 243 and 193 Barrer were recorded for neat membranes, UiO-66-based MMMs, UiO-66-NH2-based MMMs and UiO-66- NH-COCH3-based MMMs, respectively [149].

Recently, Kertik and co-workers (2017) created in-situ molecular sieves with controlled heat treatment up to 350 ◦C for 24 h for Matrimid®/ZIF-8, obtaining excellent selectivity for CO2/CH<sup>4</sup> gas mixtures due to the excellent interfacial filler-polymer adhesion [150]. Matrimid®/ZIF-8 (40 wt.%) thermally treated MMMs exhibited a CO<sup>2</sup> permeability of ~1.9 Barrer and a CO2/CH<sup>4</sup> selectivity of ~134 at 40 bar, 35 ◦C with gas mixtures containing 50 vol.% CO2/50 vol.% CH<sup>4</sup> [55,150]. ZIF-8 as inorganic filler was added into 6FDA-durene diamine, obtaining a notable increase in CO<sup>2</sup> permeability from 1468 Barrer to 2185 Barrer for pure membrane and 30 wt.% loaded ZIF-8 MMM, respectively, and 17.1 of selectivity for CO2/CH<sup>4</sup> gas pair [151].

Finally, it should be stressed that the preparation of membranes with well-dispersed fillers, good filler-polymer interfacial adhesion and a defect-free membrane surface represent nowadays the major challenges of MOF-based MMMs [152].

### • Covalent Organic Frameworks

Covalent Organic Frameworks, COFs, developed by Côté and co-workers in 2005 [125], have been recently proposed as a type of porous organic material used as a filler for the fabrication of MMMs. COFs are crystalline porous materials synthesized by the covalent combination of rigid and stable organic monomers (phenyl diboronic acid and hexahydroxytriphenylene), which offer superior chemical and thermal stability compared with MOFs [153,154]. COF materials have well-defined and predictable 2D or 3D crystalline structures as a result of the formation of strong covalent bonds [155]. COFs are classified into three groups, based on their uptake capacities, pore size and structural dimensions: (i) 2D structures featuring small 1D pores (9 Å for COF-1 and -6); (ii) 2D structures with large 1D pores (27, 16 and 32 Å for COF-5, -8 and -10, respectively) and (iii) 3D structures containing medium-sized 3D pores (12 Å for COF-102 and -103) [154]. Three-dimensional COFs, COF-1 and COF-5 presented a high hydrothermal stability, and regular and stable porosity, with surface areas ranging from 700 and 1600 m<sup>2</sup> g −1 [125], while two-dimensional COFs, COF-6, -8, and -10 showed structures with pore sizes ranging from 6.4 to 34.1 Å

and exhibited high thermal stability, low densities and high porosity with specific surface areas of 980, 1400, and 2100 m<sup>2</sup> g −1 for COF-6, -8, and -10, respectively [156]. The highest reported surface area for a COF is 4210 m<sup>2</sup> g (BET) in COF-103 [157]. Due to their properties such as low crystal density, ultrahigh accessibility and rich electronic lattice, COFs can be efficiently used for gas storage and selective adsorption. According to theoretical studies performed through grand canonical Monte-Carlo simulated calculations, the H<sup>2</sup> storage capacity with COF has been predicted, showing about 10% excess H<sup>2</sup> storage with COF-105 and 108 at 77 K, being the best-known organic materials for H<sup>2</sup> storage [154]. Han and coworkers (2008) conducted a study focused on the H<sup>2</sup> uptake capacities with experimental H<sup>2</sup> loading data for COF-5, achieving a total evacuation of the pores at 3.4 wt.% at 50 bar and 77 K. In the same study, a H<sup>2</sup> storage capacity of 8.9 wt.% at 77 K for COF-108 was observed, while the highest volumetric yield was shown for COF-102 (40.4 g L−<sup>1</sup> of H<sup>2</sup> at 77 K). [158].

Due to their variable structures, easily modifiable scaffold and high affinity to the polymeric matrix, good thermal stability, appropriate solvent compatibility, COFs have demonstrated to be excellent candidates in the field of gas separation [118,159–161]. Despite the advantages offered by COFs, a limited research has been conducted with COF-based MMMs. For instance, Wu and co-workers (2017) incorporated COFs as particles into PIM-1 as a polymeric matrix, obtaining a remarkable improvement in CO<sup>2</sup> permeability and CO2/CH<sup>4</sup> and CO2/N<sup>2</sup> selectivity compared to pure PIM-1 [153]. Likewise, Biswal and co-workers (2016) manufactured MMMs incorporating TpBD into polybenzimidazole (PBI), resulting in permeabilities above 18 Barrer for CO<sup>2</sup> and selectivities of ~48 and 23 for CO2/CH<sup>4</sup> and CO2/N2, respectively [162]. COF (imine-based COF with a twodimensional network) was also incorporated into poly(vinylamine) (PVAm) to enhance membrane performance for CO2/H<sup>2</sup> separation. As a result, a MMM at 10 wt.% COF load showed a CO2/H<sup>2</sup> selectivity of 15 and a CO<sup>2</sup> permeance of 396 GPU at 0.15 MPa, which further suggested that COFs possess good compatibility with polymers, thus enabling the fabrication of MMMs with a superior performance [163].

### • Porous Aromatic Framework

Porous Aromatic Framework, PAFs, are a subfamily of Covalent Organic Frameworks (COFs) that, unlike traditional COFs and MOFs, are stronger and more stable and exhibit a good physical-chemical stability [118]. PAFs are synthesized via irreversible cross-coupling reactions by aromatic rigid linkers [154] and constructed from carbon−carbon-bond-linked aromatic-based building units [164]. Moreover, compared to conventional porous materials (such as zeolites and MOFs), PAFs exhibit specificity in their chemistry and functionalities due to their strong carbon–carbon bonding, which makes them stable under severe chemical treatment [164]. Due to their covalent backbone, PAFs are chemically robust materials, although with a high irregular internal structure that reduces their porosity and associated crystallinity [93,154]. Indeed, these fillers exhibit a high porosity, narrow pore-size distributions for amorphous solids and Brunauer−Emmett−Teller (BET) surface areas as high as 5200 m<sup>2</sup> g −1 , which typically results in high affinities for adsorption of CO<sup>2</sup> and other gases [93,165]. PAF surface area and CO<sup>2</sup> capture may vary depending on the batch, tetrahedral core, phenyl chain length, functionalization and also the arrangement of the nanoparticles in the fillers [166]. PAFs, which are porous materials, have voids that serve to accommodate gas molecules, making them excellent absorbents. These PAFs are prepared with ultrahigh surface areas to enhance their H<sup>2</sup> storage capacity. For instance, the first reported PAF, PAF-1 with ultrahigh specific surface area (BET: 5600 m<sup>2</sup> g −1 , Langmuir: 7100 m<sup>2</sup> g −1 ) [167], exhibited a hydrogen adsorption capacity of 7.0 wt.% at 48 bar and 77 K [164]. On the other hand, due to their high surface area and stability, the capacity of PAFs as CH<sup>4</sup> sorbents has also been investigated. For instance, the CH<sup>4</sup> uptake capacity of PAF-1 is 18 cm<sup>3</sup> g <sup>−</sup><sup>1</sup> at 14 KJ mol−<sup>1</sup> heat adsorption, while PAF-3 (BET surface area of 2932 m<sup>2</sup> g −1 ) showed the highest uptake capacity at 27 cm<sup>3</sup> g <sup>−</sup><sup>1</sup> and 15 KJ mol−<sup>1</sup> heat adsorption and PAF-4 (BET surface area of 2246 m<sup>2</sup> g −1 ) presented a similar capacity to PAF-1 but at 23.2 KJ mol−<sup>1</sup> heat adsorption. With their well-defined networks, PAFs also

offer a high potential for CO<sup>2</sup> capture at low pressure. For example, CO<sup>2</sup> sorption capacities of 46 cm<sup>3</sup> g −1 (9.1 wt.%) for PAF-1, 78 cm<sup>3</sup> g −1 (15.3 wt.%) for PAF-3 and 54 cm<sup>3</sup> g −1 (10.7 wt.%) for PAF-4 were recorded at 273 K and 1 atm. [168].

However, despite their exceptional surface areas and good thermal and hydrothermal stability, PAFs exhibit weak interactions with gases, which limits their gas storage capacity and operating temperature [169]. However, Hou and co-workers (2022) added PAF-1 into PIM-1, which improved gas separation performance following the conventional method to manufacture MMMs and combining the filler with a post UV irradiation treatment. As a result, MMMs permeability showed a high permeability (i.e., P(H2) = 4800 Barrer) as well as a remarkable improvement in the separation factor (i.e., improvement for H2/CH<sup>4</sup> selectivity, from 5.4 to 90), surpassing the 2008 upper bounds for H2/CO<sup>2</sup> and CO2/CH<sup>4</sup> and 2015 upper bounds for H2/N<sup>2</sup> and H2/CH<sup>4</sup> [170]. Ben and co-workers (2009) synthesized a porous aromatic framework PAF-1 via phenyl-phenyl coupling with a Langmuir surface area of 7100 m<sup>2</sup> g −1 [171]. Likewise, Lau and co-workers (2014) demonstrated that the addition of PAF-1 as disperse phase into PTSMP and poly(methylpentyne) (PMP) can mitigate the permeability loss associated with physical ageing of these super glazed polymers [165].

• Porous Polymer Networks

Recent investigations have attributed new merits for gas separation to this family of adsorbents as a result of their high thermal and chemical stability, easy processing and low cost [172,173]. PPNs are synthesized by the homocoupling of tetrahedral monomers via the oxidative Eglinton coupling or Yamamoto-type Ullmann coupling reaction, exhibit high thermal and chemical stability and are insoluble in conventional solvents. PPNs possess Langmuir surface areas as high as 5323 m<sup>2</sup> g −1 . Between the first reported PPNs (ie. PPN-1, 2 and 3), PPN-1 showed the highest gas affinity and exhibited more micropores of less than 1 nm diameter than PPN-2 and PPN-3. Despite the lowest surface area (827 m<sup>2</sup> g −1 ), PPN-1 showed the best CO2/CH<sup>4</sup> selectivity. On the other hand, PPN-3 exhibited the highest H<sup>2</sup> uptake capacity (4.28 wt.%, 77 K) among these three (3.30 and 3.76 wt. % H<sup>2</sup> uptake for PPN-1 and PPN2, respectively) [172].

A new generation of PPNs, namely Porous Organic Polymers, POPs, was recently developed by reacting rigid trifunctional aromatic monomers with ketones exhibiting electron-withdrawing groups, in superacidic media via acid-catalyzed condensation (Lewis or Brönsted) at low temperatures. PPNs and POPs are microporous materials with Brunauer−Emmett−Teller (BET) surface areas ranging from 580 to 790 m<sup>2</sup> <sup>g</sup> <sup>−</sup><sup>1</sup> and from 760 to 935 m<sup>2</sup> g −1 , respectively, and have attractive properties such as: excellent CO<sup>2</sup> uptake capacity as solid adsorbents (up to 207 mg g−<sup>1</sup> (105 cm<sup>3</sup> (STP) g−<sup>1</sup> ) at 0 ◦C and 1 bar), ability to regenerate by vacuum without heating and an exceptional chemical and thermal stability [114,127]. Their ease of synthesis and high conversion render PPNs as materials easy to scale-up. In addition, these materials present a selective adsorption of CO<sup>2</sup> (32.7) superior to N<sup>2</sup> (22.5) under postcombustion conditions, which are higher when compared to other high-performance microporous materials [114]. In this context, Aguilar-Lugo and co-workers (2019) added PPNs (at different loads) as filler into Matrimid®, resulting in an improvement in the permeability of up to 700% for the gases tested without significantly affecting selectivity (CO2/N<sup>2</sup> and CO2/CH<sup>4</sup> selectivities decreased by 4% and 12%, respectively). These authors also observed a good filler-polymer adhesion, which was supported by the increase in the Tg of the MMMs compared to the pure polymer matrix [114]. Likewise, Rico-Martínez and coworkers incorporated bipyridine moieties-based on POPs into aromatic polyimides at different loads, which supported four- and seven-fold increases in CO<sup>2</sup> and CH<sup>4</sup> permeability, respectively [115].

#### **6. Thermally Rearranged Polymers**

As previously mentioned, new materials with superior gas separation performance, increased chemical/thermal resistance to aggressive feed conditions and high selectivity are needed. Significant advances have been generated in the chemistry of polymeric membranes for gas separation, mainly aimed at increasing the molecular stiffness and improving the free volume fraction (FVF) of membranes, leading to a high permeability without a significant decrease in selectivity [174,175]. In this context, glassy polymers such as polybenzoxazoles (PBO), polybenzothiazoles (PBT), polypyrrolones (PPL) or benzimidazoles (PBI), represent a class of rigid-rod ordered polymers with outstanding mechanical and thermal properties, and extreme rigidity [176]. However, these materials are unattractive in gas separation because their efficient packing entails few free volume elements accessible to gas penetration, which hinders their manufacture in the form of flexible and tough films. Moreover, the above-mentioned glassy polymers are soluble only in strong acids, and consequently not suitable candidates for membrane fabrication [177]. Therefore, the new strategies for the synthesis of rigid-rod polymers are mainly focused on enhancing solubility and processability.

In this regard, Park and co-workers (2007), based on the thermal conversion of imides containing hydroxides to benzoxazoles performed by Tullos and co-workers (1999) [92,177], demonstrated the occurrence of free-volume structures in dense glassy polymers that can be systematically modified by thermal rearrangement. This process enables an extraordinary gas separation performance and constitutes a novel method to prepare high-performance polymers for molecular-scale separations [92]. This successful research based on poly(1,3 benzoazole)s membranes was carried out by subjecting membranes to a thermal treatment in solid state of poly(o-hydroxyimide)s, containing *ortho* positioned functional groups (with respect to the amino group) [55,178]. This thermal rearrangement process involves a thermal cyclization step subjected to temperatures of 350–450 ◦C for a certain duration of time and under inert atmosphere or vacuum. The need for thermal processing to manufacture these materials is responsible of their name as 'thermally rearranged', or TR polymers. Depending on the functional group in the *ortho* position (-OH, -SH, or -NH2) of the precursor, the structures resulting from the cyclization process are PBO, PBT, PPL or PBI [92,179,180]. Since polybenzoxazoles may be a source of possible cross-linking as a consequence of the high temperature used during their conversion, which would also explain their insolubility, this material cannot be processed. In this sense, TR-precursors during the manufacture of these membranes can be *ortho*-hydroxyl polyimides (HPI) and *ortho*-hydroxyl polyamides (HPA) (also called α-TR and β-TR polymers, respectively, HPIs being the most studied [178]. Figure 13 shows the solid-state mechanism of a poly(hydroxyimide) (PI) and a poly(hydroxyamide) (PA) to form a TR-polymer with the proposed PBOs structure.

In both cases a final polyheterocycle of the polybenzoxazole type is reached by a cyclization process, where the heat treatment is carried out at different temperatures, depending on the TR-precursor. Additionally, the solid-state conversion process involves decarboxylation when the precursor is a polyimide, while the thermal reorganization phenomenon takes place through the loss of water molecules, or cyclodehydration, when using a polyimide as a precursor. The final PBO materials possess a chemically stable structure, resistance to CO<sup>2</sup> plasticization (likely due to their cross-linked structure) and excellent permeability and selectivity values due to the formation of a desirable free volume element distribution during thermal conversion [175,181–183].

Although in both cases the final structure of the PBO is similar, membranes exhibit different characteristics, especially in terms of gas transport properties. The TR precursor HPI can efficiently separate condensable gases, while TR precursor HPA has an outstanding ability to separate light gases. These aromatic polyamides exhibit an appropriate balance of properties such as good mechanical and chemical stability, high thermal resistance and easier processability when their precursor monomers are adequately selected [85].

Park and co-workers (2007) demonstrated that polymers with a medium cavity size, with a narrow cavity size distribution and a shape reminiscent of bottlenecks connecting adjacent chambers, possess high selectivity and high permeability [92]. Thus, TR polymers provide an increase in FFV as a consequence of the generation of microcavities with controlled size bimodal distribution in the range of 0.3–0.4 nm (which is beneficial for selective transport of gas molecules such as CO2) and 0.7–0.9 nm (which entails an enhanced

gas diffusion) [178]. The above bimodal cavity size distribution is governed by the structure of the precursor and the protocol of thermal treatment used to produce the TR-PBO [184]. *Processes* **2022**, *10*, x FOR PEER REVIEW 27 of 41

**Figure 13.** Mechanism of thermal conversion of a cyclodehydration of a hydroxypoliamide to polybenzoxazole (β‒TR‒PBO)**<sup>a</sup>** and thermal conversion of a hydroxypolyimide to polybenzoxazole (α‒TR‒PBO)**<sup>b</sup>**. **Figure 13.** Mechanism of thermal conversion of a cyclodehydration of a hydroxypoliamide to polybenzoxazole (β–TR–PBO)**<sup>a</sup>** and thermal conversion of a hydroxypolyimide to polybenzoxazole (α–TR–PBO)**<sup>b</sup>** .

In both cases a final polyheterocycle of the polybenzoxazole type is reached by a cyclization process, where the heat treatment is carried out at different temperatures, depending on the TR-precursor. Additionally, the solid-state conversion process involves decarboxylation when the precursor is a polyimide, while the thermal reorganization phenomenon takes place through the loss of water molecules, or cyclodehydration, when using a polyimide as a precursor. The final PBO materials possess a chemically stable struc-Gas transport in TR-polymer membranes depends on the degree of thermal conversion, the nature of the free volume elements and their size distribution [185]. It is assumed that the newly created micropores mediating the transport of gases in TR polymers are responsible for the usual molecular screening in the separation of gases by glassy polymeric membranes. The narrowest part of these micropores plays a role of a molecular size caliber. Today, there are consistent empirical proofs confirming the exceptional selective molecular

ture, resistance to CO2 plasticization (likely due to their cross-linked structure) and excellent permeability and selectivity values due to the formation of a desirable free volume

element distribution during thermal conversion [175,181–183].

transport performance and high permselectivity in small molecules because of the free volume structure of these polymers.

#### **7. Thermally Rearranged Mixed Matrix Membranes**

Recent investigations in membrane gas separations are focused on taking advantage of the MMMs and the Thermally Rearranged polymers' properties, in order to improve the performance of gas transport properties, mainly for CO2/CH4, H2/CO2, CO2/N<sup>2</sup> separation. MMMs manufactured with TR-able polymers are known as Thermally Rearranged Mixed Matrix Membranes (TR-MMMs). Membranes manufactured with thermally rearranged (TR) polymers result in unusually high selectivities and permeabilities, attributed to their unique hourglass configuration, while the addition of particles can add selective pathways for gas transport [97,185,186]. Despite this field of research being very recent, promising results in gas separations mixtures have been obtained. For instance, in 2017, Brunetti and co-workers manufactured the first TR-MMMs loaded with 0.5 wt.% of oxidized multi-wall carbon nanotubes (MWCNT) for CO<sup>2</sup> separation with an enhanced permselectivity and conducted an aging study. The addition of the nanotube entailed the increase of H<sup>2</sup> permeability followed by CO2, N<sup>2</sup> and CH<sup>4</sup> compared to the neat TR (increasing from 171 a 201 Barrer for H2, 105 to 126 for CO2, 9.2 to 9.3 for N<sup>2</sup> and 4.4 to 4.9 for CH4). Additionally, the influence of addition of nanotubes on aging resulted in a decrease in CO<sup>2</sup> permeability after 150 days of 13% [187]. Kim and co-workers (2019) fabricated TR-MMM for hydrogen separation using a TR-able Polyimide (HPI: HBA-DAM-6FDA) as polymeric matrix and a zeolitic imidazolate framework-8 (ZIF-8) as filler. As a result, MMMs loaded with 20% of ZIF-8 and thermally rearranged for 90 min Dwell time, exhibited excellent H<sup>2</sup> separation properties, with an increase from 365 to 1206 Barrers for H<sup>2</sup> permeability, before and after thermal treatment, respectively, and selectivity of 22.3 and 25.7 for H2/N<sup>2</sup> and H2/CH<sup>4</sup> gas pairs, respectively [121]. Smith and co-workers also carried out a pioneer study on TR-MMMs prepared by adding PAF-1 into 6FDA-HAB5DAM<sup>5</sup> (DAM) TR-able polyimide in order to improve permeation properties. As a result, TR-MMMs showed an increase of 37-fold H<sup>2</sup> permeability and 55-fold for CO<sup>2</sup> gas permeability with similar gas selectivities [97]. Soto and co-workers (2020) developed a new family of TR-MMMs to enhance CO2/CH<sup>4</sup> permselectivity using recent porous polymer networks (PPNs) as fillers on a polyamide, 6FCl-APAF, capable of producing benzoaxazoles, as a polymeric matrix. In this study, TR-MMMs showed a notable increase in gas permeability. For example, CO<sup>2</sup> permeability increased 34-fold for TR-MMM at 30% of filler compared to MMM at 30%, with a slight decrease in CO2/CH<sup>4</sup> selectivity (from 27.75 pristine membrane to 24.02 for TR-MMM). Similarly, TR-MMMs with PPNs as a filler and polyimides (*ortho*-hydroxypolyimide, PIOH, or an *ortho*-acetylpolyimide, PIOAc) as polymer matrix have been recently carried out by Aguilar-Lugo and co-workers (2021), where membranes loaded at 30% of filler showed 1036 Barrer for CO<sup>2</sup> permeability with a CO2/CH<sup>4</sup> selectivity of 28 for PIOAc-based TR-MMMs [122].

In general, TR-MMMs offer improvements in gas transport properties favored by the use of microporous materials with a high thermal stability. In addition, thermal treatment at high temperature contributes to eliminate the interfacial voids between the filler and the polymer matrix, leading to an increase in the gas selectivity of the membranes [121]. However, the excellent results in gas permeability can be accompanied by the loss of other desirable properties such as anti-aging permeability and pressure resistance, which requires further research [97].

#### **8. Membrane Modules and System Design**

#### *8.1. Membrane Modules*

The separation units in which the membrane surface is fitted are called membrane modules, which refer to the central part of a membrane device. The module should allow a separate conduction of the feed and permeate gas streams on both sides of the membrane. Figure 14 shows the schematic of the simplest design in which a single module is used.

A feed stream with a given composition and flow rate is introduced into the separation module (Feed), divided into two streams, one of which enters through the membrane (permeate) and the other (retentate) leaves the module in a smaller proportion [26,188]. The feed composition and flow rate within the module will change as a function of distance, as the membrane has the ability to transport one component more readily than another [188]. in two geometries: (1) flat sheet membranes, which include plate-and-frame and spirally wound modules and (2) tubular membranes, including tubular hollow fiber and modules based on fine capillaries or tubes housed such as a shell and tube heat exchanger [29]. The main difference between these types of membranes is based on their dimensions: tubular membranes exhibit diameters larger than 10 mm, diameters below 0.5 mm for hollow fibers and diameters between 10 mm and 0.5 mm for capillary membranes [189].

The modules engineered to date are based on the membrane configuration, classified

*Processes* **2022**, *10*, x FOR PEER REVIEW 29 of 41

of 28 for PIOAc-based TR-MMMs [122].

**8. Membrane Modules and System Design** 

quires further research [97].

*8.1. Membrane Modules* 

another [188].

For example, CO2 permeability increased 34-fold for TR-MMM at 30% of filler compared to MMM at 30%, with a slight decrease in CO2/CH4 selectivity (from 27.75 pristine membrane to 24.02 for TR-MMM). Similarly, TR-MMMs with PPNs as a filler and polyimides (*ortho*-hydroxypolyimide, PIOH, or an *ortho*-acetylpolyimide, PIOAc) as polymer matrix have been recently carried out by Aguilar-Lugo and co-workers (2021), where membranes loaded at 30% of filler showed 1036 Barrer for CO2 permeability with a CO2/CH4 selectivity

In general, TR-MMMs offer improvements in gas transport properties favored by the use of microporous materials with a high thermal stability. In addition, thermal treatment at high temperature contributes to eliminate the interfacial voids between the filler and the polymer matrix, leading to an increase in the gas selectivity of the membranes [121]. However, the excellent results in gas permeability can be accompanied by the loss of other desirable properties such as anti-aging permeability and pressure resistance, which re-

The separation units in which the membrane surface is fitted are called membrane modules, which refer to the central part of a membrane device. The module should allow a separate conduction of the feed and permeate gas streams on both sides of the membrane. Figure 14 shows the schematic of the simplest design in which a single module is used. A feed stream with a given composition and flow rate is introduced into the separation module (Feed), divided into two streams, one of which enters through the membrane (permeate) and the other (retentate) leaves the module in a smaller proportion [26,188]. The feed composition and flow rate within the module will change as a function of distance, as the membrane has the ability to transport one component more readily than

**Figure 14.** Module illustration of a membrane process. **Figure 14.** Module illustration of a membrane process.

*8.2. Plate-and-Frame Module*  Plate-and-frame modules represent the pioneer types of membrane unit, whose design is based on the conventional filter-press [190]. Plate and frame modules are the most common setups, which are similar to the flat membranes used in the laboratory. They can The modules engineered to date are based on the membrane configuration, classified in two geometries: (1) flat sheet membranes, which include plate-and-frame and spirally wound modules and (2) tubular membranes, including tubular hollow fiber and modules based on fine capillaries or tubes housed such as a shell and tube heat exchanger [29]. The main difference between these types of membranes is based on their dimensions: tubular membranes exhibit diameters larger than 10 mm, diameters below 0.5 mm for hollow fibers and diameters between 10 mm and 0.5 mm for capillary membranes [189].

#### *8.2. Plate-and-Frame Module*

Plate-and-frame modules represent the pioneer types of membrane unit, whose design is based on the conventional filter-press [190]. Plate and frame modules are the most common setups, which are similar to the flat membranes used in the laboratory. They can be mounted in plate, bag or spiral-wound form [26]. These module membranes are separated by a feed spacer, with the separate layers stacked towards each other, like a sandwich. These spacers serve to seal the module and allow the flow of material through the drilled holes and alternate channels [29]. The membrane surfaces per module volume (packing density) range from 100–400 m2/m<sup>3</sup> . A stop disc is used in order to favor the flow over the surface membrane and reduce the formation of preferential channels [189]. Plateand-frame modules present advantages such as: exchange ability of single membranes, low sensitivity to particulate blocking of the feed channels and usage of flat membranes without the usage of glue. In addition, they exhibit disadvantages such as: need of several sealings, high pressure drop and low packing density [191]. Currently, this kind of module is still used in ultrafiltration and pervaporation processes and represents the only plate-andframe configuration used in solution-diffusion membranes [192]. Since plate-and-frame modules present smaller membrane surface area per unit volume, they are effective in pervaporation applications [192]. However, compared to hollow fiber and spiral-wound modules, plate-and-frame modules are less applied in gas separation [190]. For instance, oxygen enrichment from air, organic vapor recovery and even medical applications are among the commercial applications of these modules in gas separation [193].

#### *8.3. Spirally Wound Modules*

The spiral-wound format was the first to be commercialized [26] and was initially developed for reverse osmosis applications. Spiral-wound modules are typically applied in CO<sup>2</sup> removal from natural gas and vapor/gas separations [32]. Currently, this configuration is also used in ultrafiltration and gas permeation applications, which render it an important module in membrane applications [194]. This kind of module is used when countercurrent flow is not required to increase separation efficiency and when pressure drop must be taken into account [50].

This configuration consists of a plate-and-frame system that wraps around a central collection tube, similar to a sandwich roll. A spacer material is placed between the membranes to prevent contact of the feed and permeate, as well as to allow free space for the interaction of gas molecules with the membrane [29]. The interleaved sheets are spirally wound around a central permeate collection channel [26,188]. The feed stream flows along the center tube in axial direction, while the permeate flows in radial direction towards the center tube and is collected on the inside of the envelope. The packing density of this spirally wound module (300–1000 m2/m<sup>3</sup> ) is greater than that of the plate-and-frame module. However, this parameter depends on the channel height, which in turn is determined by the permeate and feed-side spacer material. According to Caro and co-workers (2007), spiral-wound modules exhibit a good mass transfer due to feed spacers, are simple, and present a cost-effective fabrication and relatively high packing density/membrane area-tovolume ratio (up to 1000 m2/m<sup>3</sup> ). However, spirally wound modules exhibit disadvantages such as difficultly to be cleaned and long permeate pathway [195]. Less than 20% of gas separation membranes nowadays are manufactured as spiral-wound modules. Currently, spiral-wound modules are industrially used in natural gas processing.

#### *8.4. Tubular Modules*

Tubular membrane modules are based on cylindrical membranes, which consist of thin layers of selective membrane deposited on the two membrane faces of a porous stainless steel, ceramic, or plastic tubular support with a diameter superior to 10 mm. Tubular membranes can be manufactured with inner diameters ranging from 5–25 mm, with 12.5 mm being the most common diameter. Although the number of tubes placed in the module is not limited, it can vary from 4 to 18 tubes [26,188]. The feed flows through the center of the membrane tubes and the permeate moves across the membrane from the inside to the outside, subsequently flowing into the larger tube [189]. Ceramic membranes are mainly assembled in such tubular module configurations. The packing density is rather low, typically <300 m2/m<sup>3</sup> [29,188]. The main advantages of this module are: membrane fouling can be easily controlled, which reduces operating costs, as well as concentration polarization effects [26]. Thus, given their resistance to fouling (due to the effect of good fluid hydrodynamics), the use of tubular modules is often restricted to ultrafiltration applications [196].

#### *8.5. Capillary Module*

The capillary module consists of a large number of capillaries assembled together in a module with an inner diameter of 0.2–3 mm arranged in parallel as a bundle in a shell tube [26]. They are self-supporting and the free ends of the capillaries are encapsulated with agents such as epoxy resins, silicone rubber or polyurethanes. There are two types of module arrangements: (1) membranes where the feed passes through the bore of the capillary and the permeate exits through the side of the membrane and (2) membranes where the feed enters the module on the shell side and the permeate exits through the bores of the membrane [188,189]. The selection of the module arrangement will depend on the application, and parameters such as operating pressure, pressure drop, type of membrane material available, etc. Packing densities range from 600 to 1200 m2/m<sup>3</sup> [188].

#### *8.6. Hollow Fiber*

The hollow fiber module is similarly to the capillary module. Spiral-wound and hollow fiber modules are commercially available for gas separation. Hollow fibers are based on a porous, non-selective support layer (~200 µm) and an active layer (actual membrane) (<40 nm). As a result of the small thickness of the active layer, this must be supported by a

thicker layer in order to obtain mechanical strength, to withstand the pressure difference between the feed and permeate side [197].

The hollow fiber membrane module consists of a large number of hollow fibers assembled together into a bundle, which is encapsulated at the ends to prevent leakage between the feed and permeate chambers [29]. The fibers, arranged in parallel to pass through the tubular sheets or one or both ends of the device, range between 1.0 and 1.5 mm outside diameter and the bore of the fibers has a typical diameter of 0.5–1 mm. Two types of module arrangement can be distinguished: (1) membranes where the feed enters through the bore of the fiber ("inside-out") and the permeate is collected outside the membrane in the housing or (2) membranes where the feed enters on the outside ("outside-in") and the permeate passes into the membrane bore. Hollow fiber modules exhibit the highest packing density among all module configurations, reaching values of up to 30,000 m2/m<sup>3</sup> [188]. The high membrane area-to-volume ratio, together with their high packing density and cheap fabrication cost, are the main advantage of the hollow fiber module. The low-pressure resistance and mostly laminar flows, which increases mass transfer limitations, rank among the main disadvantages of this membrane module [190].

#### *8.7. Module Selection Criteria*

Gas separation systems are commercially available as hollow fiber or spirally wound modules and, in some applications, also in plate-and-frame modules. The selection of the appropriate membrane module is typically determined by cost considerations. Hollow fiber modules are more economical per square meter, however the fabrication of very thin selective layers in the form of hollow fibers is a difficult process. As a result, the permeance in this type of membrane tends to be lower than in flat sheet membranes based on the same polymer. Hollow fiber modules require more membrane surface area to achieve the same separation factor. They also require more feed pretreatment than spirally wound modules for the removal of particles, oil residues and other fouling components [32,59]. According to Ismail and co-workers (2015), the manufacturing cost (\$/m<sup>2</sup> ) for hollow fiber ranges from 2 to 10 \$ per m<sup>2</sup> , from 5 to 50 \$ per m<sup>2</sup> for capillary fibers, from 5 to 50 \$ per m<sup>2</sup> for spirally wound, and from 50 to 200 \$ per m<sup>2</sup> for plate-and-frame and tubular membranes [26,32]. However, capital costs are not the only factor to consider when selecting membranes modules. Therefore, it is necessary to consider that the choice of membrane module will also depend on the application (Table 2) [32,197].

In gas separation plants, especially refinery and petrochemical operations, the cost of the modules corresponds to only 10–25% of the total cost of gas separation. Indeed, even if the cost of the membrane modules was reduced, the total cost of the plant would decrease significantly [32].

The economics of the process of membrane-based separation is determined by process design. Single-stage configurations entail low capital costs and are only suitable when the required purity and product recovery are moderate. More demanding applications require multiple stages of separation and recycling. The design of a membrane system involves the configuration of the permeator network and the operating conditions of the individual permeator systems [198]. A key part of the membrane gas separation design is the selection of the separation configuration. Single-stage configurations without gas recycling are the most common and simplest design. However, the demand for higher product purity (for instance methane contents of 98–99.5% in biomethane) and the need for recovery target products makes the use of recycle streams as well as multi-stage configurations a must [199]. These multi-stage systems are typically designed using two, three or four stages [200]. Figure 15 displays the main process configurations.


**Table 2.** Characteristics and typical applications of the different modules for gas separation.

**Figure 15.** Different design configurations for biogas upgrading (**a**) single-stage configuration, (**b**) two-stage configuration with a recirculation loop, (**c**) two-stage configuration with sweep and (**d**) three-stage configuration with sweep. Adapted from Angelidaki and co-workers, 2018, and Bauer and co-workers, 2013. **Figure 15.** Different design configurations for biogas upgrading (**a**) single-stage configuration, (**b**) two-stage configuration with a recirculation loop, (**c**) two-stage configuration with sweep and (**d**) three-stage configuration with sweep. Adapted from Angelidaki and co-workers, 2018, and Bauer and co-workers, 2013.

**9. Conclusions**  The development of compact and low-cost biogas-to-biomethane and biohydrogento-high purity H2 conversion technologies is crucial to ensure the competitiveness of these green energy vectors, and to promote the implementation of anaerobic digestion and dark fermentation for organic waste treatment. Nowadays, the removal of CO2 from biogas at the industrial scale is carried out by physical/chemical technologies, which exhibit high operating costs and corrosion problems. In fact, CO2 removal at the commercial scale is performed using very energy-intensive technologies that require a prior removal of H2S, such as pressurized water scrubbers, chemical and organic solvent scrubbers, PSA adsorption systems or cryogenic CO2 separators. On the other hand, biological technologies The one-step system uses a single membrane and compressor (Figure 15a), which entails a low energy consumption, with no internal recirculation of the rejected gas [9,31,201]. This configuration needs less maintenance and reduces the operational cost compared to multistage membrane units [9]. The second-step system (Figure 15b) involves a gas recirculation loop for the gas retained to a second membrane installed to increase the purity of biomethane and the recovery of methane [9]. The third-step (Figure 15c) system is also based on two membranes, where the rejected gas from the first membrane is purified in a second membrane and recirculated to the first membrane [9]. The most complex configuration (Figure 15d) involves the purification of the permeate from the first membrane in a sequential membrane in order to increase the efficiency of the process, and the recovery of

for CO2 removal from biogas are still in an experimental development phase and require large areas of land or the availability of renewable hydrogen. In this context, the energy

decades, a wide variety of polymeric materials have been developed to increase the gas transport performance of membranes. However, several challenges remain in the field, such as the trade-off between permeability and selectivity (which often prevents overcoming the Robeson limits), the physical aging of membranes and material plasticization (which visibly affects membrane performance). In this context, novel inorganic materials, with outstanding chemical and thermal properties (superior to polymeric materials) and excellent performance in gas separation, have been recently synthesized. However, despite these materials being difficult to process, their combination with polymeric materials in order to develop MMMs has resulted in unprecedented gas separation performance. In addition, polymeric materials capable of producing benzoaxazoles have been recently used to develop thermally rearranged MMMs, leading to excellent gas separation properties, exceeding the Robeson limit, as well as delaying physical aging. Thus, the development of new materials with enhanced physical and chemical properties compared with the CH<sup>4</sup> from the gas rejected by the first membrane (using a third membrane) and from the rejected gas of the polishing membrane via recirculation [31].

#### **9. Conclusions**

The development of compact and low-cost biogas-to-biomethane and biohydrogen-tohigh purity H<sup>2</sup> conversion technologies is crucial to ensure the competitiveness of these green energy vectors, and to promote the implementation of anaerobic digestion and dark fermentation for organic waste treatment. Nowadays, the removal of CO<sup>2</sup> from biogas at the industrial scale is carried out by physical/chemical technologies, which exhibit high operating costs and corrosion problems. In fact, CO<sup>2</sup> removal at the commercial scale is performed using very energy-intensive technologies that require a prior removal of H2S, such as pressurized water scrubbers, chemical and organic solvent scrubbers, PSA adsorption systems or cryogenic CO<sup>2</sup> separators. On the other hand, biological technologies for CO<sup>2</sup> removal from biogas are still in an experimental development phase and require large areas of land or the availability of renewable hydrogen. In this context, the energy demand and effectiveness of membrane-based CO<sup>2</sup> separation from biogas and biohydrogen is gradually decreasing as a result of the rapid advance in material science. In the last decades, a wide variety of polymeric materials have been developed to increase the gas transport performance of membranes. However, several challenges remain in the field, such as the trade-off between permeability and selectivity (which often prevents overcoming the Robeson limits), the physical aging of membranes and material plasticization (which visibly affects membrane performance). In this context, novel inorganic materials, with outstanding chemical and thermal properties (superior to polymeric materials) and excellent performance in gas separation, have been recently synthesized. However, despite these materials being difficult to process, their combination with polymeric materials in order to develop MMMs has resulted in unprecedented gas separation performance. In addition, polymeric materials capable of producing benzoaxazoles have been recently used to develop thermally rearranged MMMs, leading to excellent gas separation properties, exceeding the Robeson limit, as well as delaying physical aging. Thus, the development of new materials with enhanced physical and chemical properties compared with conventional organic and inorganic membranes, providing a superior performance in terms of permeability and selectivity, represents the cornerstone in biogas and biohydrogen upgrading.

**Author Contributions:** Conceptualization, methodology, investigation, writing-original draft preparation, C.S.; validation and visualization, C.S., L.P., R.M., P.P. and A.H.; writing—review and editing and supervision, L.P., R.M., P.P. and A.H.; supervision, R.M., P.P. and A.H.; funding acquisition, L.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Spanish Government (AEI) through projects PID2019- 109403RB-C21/AEI/10.13039/501100011033; and by the Regional Government of Castilla y León and the EU-FEDER programme (CLU2017-09, UIC082, CL-EI-2021-07 and UIC 315) and CDTI (ECLOSION PROJECT).

**Acknowledgments:** C.S. acknowledges the Regional Government of Castilla y León for her Ph. D. contract.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Potential for Biomethanisation of CO<sup>2</sup> from Anaerobic Digestion of Organic Wastes in the United Kingdom**

**Angela Bywater \*, Sonia Heaven \* , Yue Zhang and Charles J. Banks**

Water and Environmental Engineering Group, University of Southampton, Southampton SO16 7QF, UK; y.zhang@soton.ac.uk (Y.Z.); c.j.banks@soton.ac.uk (C.J.B.)

**\*** Correspondence: a.m.bywater@soton.ac.uk (A.B.); s.heaven@soton.ac.uk (S.H.)

**Abstract:** The United Kingdom (UK) has a decarbonisation strategy that includes energy from both hydrogen and biomethane. The latter comes from the growing anaerobic digestion (AD) market, which in 2020 produced 23.3 TWh of energy in the form of biogas. According to the strategy, this must be upgraded to biomethane by removal of carbon dioxide (CO<sup>2</sup> ): a goal that could also be fulfilled through CO<sup>2</sup> biomethanisation, alleviating the need for carbon capture and storage. Results are presented from a survey of publicly available datasets coupled with modelling to identify potential scale and knowledge gaps. Literature data were used to estimate maximum biomethane concentrations by feedstock type: these ranged from 79% for food wastes to 93% for livestock manures. Data from various government sources were used to estimate the overall potential for CO<sup>2</sup> biomethanisation with current AD infrastructure. Values for the uplift in biomethane production ranged from 57% to 61%, but the need for more consistent data collection methodologies was highlighted. On average, however, if CO<sup>2</sup> biomethanisation was applied in all currently operating UK AD plants an energy production uplift of 12,954 GWh could be achieved based on 2020 figures. This is sufficient to justify the inclusion of CO<sup>2</sup> biomethanisation in decarbonisation strategies, in the UK and worldwide.

**Keywords:** in-situ biomethanisation; power-to-gas; anaerobic digestion; biomethane production; United Kingdom policy; energy security

#### **1. Introduction**

CO<sup>2</sup> biomethanisation is the microbially mediated transformation of carbon dioxide (CO2) to methane (CH4) via the addition of exogenous hydrogen (H2), according to the overall reaction shown in Equation (1):

$$4\text{ H}\_2 + \text{CO}\_2 \rightarrow \text{CH}\_4 + 2\text{ H}\_2\text{O} \tag{1}$$

The process has clear potential applications in the anaerobic digestion (AD) industry, which utilises microbial communities to transform a wide variety of organic materials into biogas, a mixture of biomethane and CO<sup>2</sup> [1]. These mixed communities already contain the hydrogenotrophic methanogens which catalyse the direct CO<sup>2</sup> biomethanisation route shown in Equation (1), as well as syntrophic organisms able to mediate indirect routes, e.g., via homoacetogenesis and acetoclastic methanogenesis [2,3]. The combination of AD with H<sup>2</sup> addition to promote CO<sup>2</sup> biomethanisation can thus improve the methane productivity of digesters fed on organic feedstocks [4], increasing the energy output (and therefore, viability) [5,6] and enhancing the carbon utilisation efficiency [7].

The United Kingdom (UK) has a well-established AD market, with initial incentives for distributed small-scale (<5 MWe) electricity production introduced in the 2002 Renewables Obligation (RO) and the 2010 Feed-In Tariff (FIT). It is, however, becoming increasingly common practice in many countries to upgrade biogas to biomethane by removal of CO<sup>2</sup> and other impurities [4,8]. This is because biomethane is valuable as a low carbon fuel that

**Citation:** Bywater, A.; Heaven, S.; Zhang, Y.; Banks, C.J. Potential for Biomethanisation of CO<sup>2</sup> from Anaerobic Digestion of Organic Wastes in the United Kingdom. *Processes* **2022**, *10*, 1202. https:// doi.org/10.3390/pr10061202

Academic Editor: Pietro Bartocci

Received: 31 May 2022 Accepted: 14 June 2022 Published: 16 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

can be utilised locally or upgraded to share existing natural gas storage and distribution networks for use in heating, transport and the centralised generation of electricity [4].

Biomethane production in the UK has been promoted by the Renewable Heat Incentive (RHI), which accepted applicants between 2011 and 2021. This was followed by the Green Gas Support Scheme (GGSS), introduced in late 2021 and designed to encourage the production of biomethane for gas grid injection, with a minimum of 50% of the gas coming from wastes. The GGSS reflects the UK Government's ongoing commitment to increasing the growth of biomethane production through AD, based on the advice and projections of the Climate Change Committee (CCC) in its Sixth Carbon Budget [9]. The CCC calculates that biomethane/biogas could displace up to 10% of UK demand for natural gas and predicts that by 2030, biomethane production could more than treble from 2020 levels. Furthermore, by 2035, it could abate the equivalent of 1.5 million tonnes year−<sup>1</sup> of CO<sup>2</sup> through fossil gas displacement [9].

Hydrogen also plays an important part in the UK's decarbonisation scenario, with plans to kick-start a mass market for hydrogen by reformation of fossil gas accompanied by carbon capture and storage (CCS). In addition, electrolytic hydrogen production from excess renewables will also be developed in a 'Balanced Pathway' scenario, in which reformed hydrogen will provide 60% of the requirement by 2035, after which, this proportion will decrease in favour of electrolytic hydrogen production, which will make up almost 50% of the rising target for production by 2050 [9].

The current UK strategy proposes that biomethane targets are met through refining biogas to remove the biogenic CO2, which will subsequently require permanent storage, i.e., through carbon capture and storage (CCS) solutions. As biomethane and electrolytic hydrogen production are already part of the UK's energy strategy, however, it also makes sense to consider the potential role of CO<sup>2</sup> biomethanisation of organic wastes. This could address the upgrading requirement, utilising existing AD assets at a relatively low capital cost [5,10] with the added bonus of increasing overall methane yields, and the prospect that the process may be more energetically efficient and cheaper than CCS with its as-yetunknown costs. CO<sup>2</sup> biomethanisation thus offers both a means to support the transition from carbon-based fuels to hydrogen, and a rational long-term solution to maximise the energetic value of biomass carbon-based renewable fuels in their own right.

Nevertheless, future policy which considers this option cannot be formulated and then prescribed without the necessary data for modelling and assessment. CO<sup>2</sup> biomethanisation is still in its early stages as a commercial process, with only a few examples of plants operating at scale [7], almost all of which are based on the use of separate dedicated bioreactors fed on gaseous inputs in ex situ processes [3,5]. There is, however, a growing body of research on in situ and hybrid processes in which H<sup>2</sup> alone or with additional CO<sup>2</sup> is injected directly into a digester, with a variety of organic feedstocks, equipment configurations and operating conditions. There is thus a need to assess the available data with regard to anticipated improvements in process performance, and to the scale of existing AD resources in the UK. The outcomes can then be considered in the context of dynamic economic, technology and infrastructure developments in renewable power and hydrogen production, and use.

As a first step in this complex journey, the current work has carried out a high-level review of the potential scale of application in the UK, supported by an assessment of reported data on in-situ CO<sup>2</sup> biomethanisation performance according to feedstock type. This provides an essential starting point for identifying research and policy needs for the integration of CO<sup>2</sup> biomethanisation as a contributor to the UK future energy mix and for assessment of its potential role in the transition from a fossil-based energy system.

Alternative routes to CO<sup>2</sup> methanisation via thermal catalytic conversion are also currently under development [4,7] but are not considered in this paper as the principle focus is on conversion of biogas from AD of organic wastes.

#### **2. Materials and Methods**

### *2.1. Performance of CO<sup>2</sup> Biomethanisation of Organic Feedstocks*

The objective of this part of the work was to establish values for the increases in methane production and biogas methane content that could be achieved through CO<sup>2</sup> biomethanisation of a range of organic feedstocks. To achieve this, experimental results were collated from relevant studies, focusing primarily on addition of exogenous H2: a small number of studies examining approaches such as syngas addition were also considered, but bioelectrochemical systems were not included as, although promising, the technology is some way from large-scale application [4,11].

Parameters taken from the literature included substrate type, operating temperature ( ◦C), digester configuration and characteristics, organic loading rate (OLR) expressed as g volatile solids (VS) L−<sup>1</sup> day−<sup>1</sup> or g Chemical Oxygen Demand (COD) L−<sup>1</sup> day−<sup>1</sup> , hydraulic retention time (HRT, days) and operating pH. Reported and/or calculated values for specific methane potential (SMP: L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS or L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> COD), volumetric methane production (VMP: L CH<sup>4</sup> L <sup>−</sup><sup>1</sup> digester day−<sup>1</sup> ), H<sup>2</sup> input (L H<sup>2</sup> L <sup>−</sup><sup>1</sup> day−<sup>1</sup> ) and output gas composition (% volume) were used to calculate the change in SMP and VMP due to CO<sup>2</sup> biomethanisation. Note that the term SMP is used here to refer to the total volume of methane produced, from both anaerobic degradation of organic material and biomethanisation of CO<sup>2</sup> where applicable, per unit of organic feed added. Similarly, the VMP is based on the total volume of methane produced from both sources, per unit working volume of reactor per unit time. The difference between VMP with and without addition of H<sup>2</sup> (or H<sup>2</sup> and exogenous CO2) is assumed to be methane produced from CO<sup>2</sup> biomethanisation and is referred to as the Methane Evolution Rate (MER: L CH<sup>4</sup> L <sup>−</sup><sup>1</sup> day−<sup>1</sup> ). The CO<sup>2</sup> removal rate (CRR) is similarly defined as the difference between volumetric CO<sup>2</sup> output (L CO<sup>2</sup> L <sup>−</sup><sup>1</sup> day−<sup>1</sup> ) with and without H<sup>2</sup> addition, adjusted for any additional exogenous CO<sup>2</sup> input. The terms volumetric biogas production and volumetric gas production (VBP and VGP, L L−<sup>1</sup> day−<sup>1</sup> ) were used to distinguish between the sum of methane and carbon dioxide outputs, and the total gas output including any residual H2, respectively. More detailed nomenclature and definitions are provided in Supplementary Materials S1.

A number of other parameters based on the above data were also evaluated. The H<sup>2</sup> transfer efficiency *E* (%) refers to the proportion of H<sup>2</sup> successfully transferred into the system, i.e., not leaving in gaseous form in the output gas; and is calculated from (H<sup>2</sup> input-H<sup>2</sup> output)/H<sup>2</sup> input, with inputs and outputs in L H<sup>2</sup> L <sup>−</sup><sup>1</sup> day−<sup>1</sup> . The expected MER is the amount of H<sup>2</sup> transferred divided by 4, based on reaction stoichiometry, and is equal to the expected CRR; these values can then be compared with the actual observed MER and CRR. The ratios between actual MER and actual CRR; actual MER and original volumetric CO<sup>2</sup> output (without CO<sup>2</sup> biomethanisation); VBP with and without CO<sup>2</sup> biomethanisation; H<sup>2</sup> transferred to actual MER and actual CRR; and H<sup>2</sup> input to original volumetric CO<sup>2</sup> output (without CO<sup>2</sup> biomethanisation) were also determined.

Where control reactors without H<sup>2</sup> addition were operated, the change in SMP or VMP due to CO<sup>2</sup> biomethanisation was taken as the difference between control and experimental values during the same period; in trials without controls, experimental values were compared with those of a baseline period without H<sup>2</sup> addition. Where multiple sets of conditions were tested, the best performance in terms of SMP and/or biogas methane content under apparently stable conditions is shown. As far as possible, calculations were carried out in a standard manner. In some cases, this means values differ slightly from those reported in the original paper. Wherever possible, the consistency of results was checked by comparing reported and calculated values: e.g., under steady state conditions SMP × OLR = VMP. If some aspect appeared unclear or inconsistent, the authors were contacted to request additional information. More details of calculation methods are provided in the Supplementary Materials.

Digestion systems operating in the temperature range 35–40 ◦C are referred to as mesophilic, and between 50–60 ◦C as thermophilic. Unless specified, standard temperature and pressure (STP) values of 0 ◦C and 101.325 kPa were assumed and applied throughout this work.

To provide indicative values for modelling, outputs from individual studies were grouped into the following feedstock types: livestock manures, crop materials and agrowastes, food wastes, the organic fraction of municipal solid waste (OFMSW) and sewage sludges. In some cases, there was no single study that directly demonstrated the maximum achievable SMP and biogas methane content for a given feedstock type. Where possible, this was estimated from baseline pH and biogas CO<sup>2</sup> concentrations (i.e., without CO<sup>2</sup> biomethanisation) and observed or assumed maximum pH values for stable operation with CO<sup>2</sup> biomethanisation, using an equation derived and tested for this purpose [12].

#### *2.2. AD Feedstock/Energy Production Data and Calculations*

Several potential sources of data on AD feedstock quantities and energy production are available and were assessed for use in the current work. Those selected for use were derived from freely available public sources. The most detailed is the dataset which accompanies the Renewable Obligation's Annual Report (2019–20) [13], produced by the Office of Gas and Electricity Markets (Ofgem), the regulator for the electricity and natural gas markets in Great Britain. All RO biomass electricity generators over 50 kWe are required to report on feedstock sustainability criteria in two areas: greenhouse gas (GHG) criteria for national emissions data collection; and land criteria, to assess land use and potential change [14]. The resulting data are expressed as volume of biogas produced for each of 10 feedstock categories aggregated from the RO feedstock consignment sustainability data (RO-SUS) [15]. In order to derive fresh feedstock tonnages, biomethane content and the energetic value of the biomethane from the biogas data, these feedstock categories were characterised using data from the University of Southampton's Anaerobic Digestion Assessment Tool (ADAT) [16], and from the literature review.

A similar feedstock characterisation and energy calculation exercise was carried out on AD feedstock volume data provided by the UK Government's Department for Environment, Food and Rural Affairs (Defra) [17]. Results were compared with the RO data where possible.

Values used in the RO/Defra feedstock calculations were applied to the energy production data in the 2021 UK Digest of UK Energy Statistics (DUKES) [18] for anaerobic digestion and sewage gas. The DUKES data, which are produced by the UK Government's Department for Business, Energy and Industrial Strategy (BEIS) from a wide array of other sources, were also used to provide a wider overview.

Maximum CH<sup>4</sup> values for CO<sup>2</sup> biomethanisation of organic feedstocks derived from the literature review were then utilised to calculate a potential uplift in energy for the RO, Defra and DUKES data [13,17,18].

#### **3. Results and Discussion**

#### *3.1. Performance of CO<sup>2</sup> Biomethanisation by Feedstock Type*

Several excellent reviews on CO<sup>2</sup> biomethanisation have recently been published, from the wider perspective of biogas upgrading [4,19] to those more specifically focused on the biological process [3,11] and ranging from technical assessment to fundamental biochemical aspects [2,5,6]. This part of the current work considered the same data from the viewpoint of assessing the achievable performance according to feedstock type. An ideal study for this purpose accurately simulates large-scale operation, but in practice, this can be challenging: common issues include the fact that laboratory-scale reactors may be fed only once per day and/or 5 days per week, while feeding of commercial digesters is usually more frequent. Arrangements for the supply and recirculation of external and headspace gases are also different at large and small scale, with laboratory experiments relying on a variety of approaches from continuous or intermittent input with or without recycling to daily injection under pressure. These affect the availability of exogenous gases over time, and further interact with different modes organic waste feeding which in turn

influence the in-situ gas composition. Perhaps most importantly, reactor mixing and mass transfer parameters can vary considerably with scale. All of these factors may affect the biochemical environment, microbial community and metabolic pathways which determine the achievable methane yields and concentrations. In addition, experiments are designed for a range of different purposes and do not necessarily provide data on the maximum stable methane yield or concentration; nonetheless, these studies provide an indication of the potential performance and what is known to date.

In some cases, issues were encountered when attempting to put data from different sources into a comparable format. Some studies include residual H<sup>2</sup> in the reported biogas production, while others do not: not all specify this. Most studies indicate that reported gas volumes have been normalised to STP, but not all state which STP conditions are used. When gas compositions are reported, the main components (CH4, CO2, H2) do not always sum to 100%: authors do not always state whether they are reporting dry and/or normalised gas volumes, or whether any discrepancies are due, e.g., to variability in experimental measurement or to the inclusion of water vapour and other trace gases. A range of different terms are used to describe the proportion of input H<sup>2</sup> that is consumed, and the resulting increase or decrease in CH<sup>4</sup> and CO<sup>2</sup> produced. There is a clear need for more consistent reporting, and a comprehensive set of standards for this purpose has been proposed [20].

#### 3.1.1. Livestock Manures

Starting with the earliest work on in situ CO<sup>2</sup> biomethanisation of waste feedstocks [1], several studies have been conducted using cattle manure (CM) as the sole or main substrate. Luo et al. [1] ran duplicate CSTR digesters for 3 HRT at an OLR of around 6 g VS L−<sup>1</sup> day−<sup>1</sup> using sieved CM diluted by a factor of 2 for ease of small-scale operation. H<sup>2</sup> was added continuously to one digester via two ceramic diffusers at around four times CO<sup>2</sup> production (H2/CO<sup>2</sup> ratio 4:1 *v*/*v*), while the other digester continued operating as a control. The experiment aimed to demonstrate the principle rather than to optimise the system: specific methane production per unit of organic feed (SMP) rose from 60 to 73 L CH<sup>4</sup> kg−<sup>1</sup> VS but the biogas methane concentration only increased from 62% to 65% due to unconverted H<sup>2</sup> in the output gas.

Two-stage systems in which the first digester was fed twice daily on CM were tested by Bassani et al. [21]. The second digester received biogas and digestate from the first, with exogenous H<sup>2</sup> added at four times the initial CO<sup>2</sup> volume. One system was operated under mesophilic and one under thermophilic conditions. As expected, the thermophilic system gave higher specific and volumetric methane yields than the mesophilic. Both systems had higher specific methane yields than those obtained by Luo et al. [1], probably at least in part due to the longer overall HRT; but VMP in the mesophilic system was lower than in Luo et al. [1] because of the requirement for a second reactor. The methane concentration was 88.9% in the mesophilic system and 85.1% for the thermophilic, reflecting a higher residual H<sup>2</sup> content in the thermophilic system. The same system configuration and operating conditions were used to investigate changes in performance and microbial community over an extended period [22]. Results for the early stages of stable operation were similar to those of Bassani et al. [21]; but after two years of operation, biogas methane content was increased to around 99%, with an associated rise in both pH and volatile fatty acid (VFA) concentration. The VFA was primarily acetate, and was attributed to enhanced homoacetogenesis accompanied by changes in the microbial population.

Wahid and Horn [23] used a 2-stage system with two 6-L CSTRs operating in series at 55 ◦C. The first reactor was fed on CM at an HRT of 15 days, while the second received biogas and digestate from the first at an HRT of 20 days. After 120 days of stable operation, H<sup>2</sup> was added initially at a 2/1 H2/CO<sup>2</sup> ratio, and subsequently at 4/1. Mixing speeds and gas recirculation rates were varied. There are some inconsistencies between reported values for OLR, SMP and VMP but the highest observed increase in SMP was 90%, similar

to the value found by Treu et al. [22], though H<sup>2</sup> transfer efficiency was lower, resulting in a higher residual H<sup>2</sup> content.

Lebranchu et al. [24] operated a 142-L CSTR with a working volume of 100 L at 40 ◦C. The digester was fed continuously on CM with a VS content of 11% at an HRT of 28 days; the source of inoculum is not stated. H<sup>2</sup> was added continuously via a silicone tube diffuser without gas recirculation, and digester contents were mixed by helical and Archimedes screw mixers. The trial ran for 2.5 HRT with H<sup>2</sup> addition rates increasing from zero to 0.17, 0.29 and 0.45 L H<sup>2</sup> L <sup>−</sup><sup>1</sup> day−<sup>1</sup> . This was followed by trials at different mixing speeds, then additional exogenous CO<sup>2</sup> was injected at 0.05 L CO<sup>2</sup> L <sup>−</sup><sup>1</sup> day−<sup>1</sup> with a matching stoichiometric increase in H2. SMP rose from 0.186 to 0.221 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS with H<sup>2</sup> addition, and to 0.236 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS with exogenous CO<sup>2</sup> and H2. The relatively large scale, simple configuration and operating mode of this study make these results particularly useful in estimating the possible performance of conventional commercial systems.

Diluted pig manure was digested at mesophilic and thermophilic temperatures under different mixing conditions in CSTRs operating at an OLR of 2 g VS L−<sup>1</sup> day−<sup>1</sup> and an HRT of 25 days [25–27]. Under the same operating conditions, thermophilic digestion gave higher specific and volumetric methane yields than mesophilic. The highest SMP was observed in thermophilic conditions with continuous mixing and addition of H<sup>2</sup> and sodium formate [25], although the biogas methane concentration was lower than without the H<sup>2</sup> addition due to the presence of residual H<sup>2</sup> in the output gas. The greatest increase in SMP was under mesophilic conditions with H<sup>2</sup> addition and intermittent mixing [26]; in this case, methane production decreased with continuous stirring. The different response of the two systems to a change from intermittent to continuous stirring was attributed to different effects on alkalinity, TAN, VFA concentration and pH associated with CO<sup>2</sup> dissolution in the liquid phase.

No studies appear to have used other livestock manures as a mono-substrate, although digestate from a 2-stage system fed on chicken manure was used to inoculate reactors receiving exogenous H<sup>2</sup> and CO<sup>2</sup> and to replenish buffering capacity during the enrichment stage [28].

Systems with CM as a major feedstock component have been studied by several researchers. Whey has been used as an acidogenic co-substrate to counteract the pH increase associated with in situ CO<sup>2</sup> conversion [29]. An increase of 80% in SMP was achieved in a single-stage thermophilic CSTR with ceramic diffuser and magnetic stirring [30]; the highest stirring speed led to a small reduction in methane production. The best results were achieved using a hollow fibre membrane (HFM) for H<sup>2</sup> transfer, with an increase in SMP from 0.288 to 0.516 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS, a biogas methane content of 96.1%, and no residual H<sup>2</sup> detected. A two-stage thermophilic system previously operated with CM as a mono-substrate [23] was also used to test CM + whey at different feedstock ratios, feeding frequencies and stirring speeds [31]: SMP increased by up to 23% with H<sup>2</sup> addition but residual CO<sup>2</sup> and H<sup>2</sup> contents were high, indicating poor mass transfer and leading to a rather low biogas methane concentration.

CM and potato starch were used as co-substrates in a 2-stage system consisting of a CSTR followed by an upflow anaerobic sludge blanket (UASB) reactor, with H<sup>2</sup> added to the first stage and gas recirculated between the stages [32]. H<sup>2</sup> transfer efficiencies were around 98.5% but VFA accumulation occurred, and stable operation with H<sup>2</sup> addition was not achieved: the results are therefore not shown in Table S1.

Mesophilic digestion of CM and vegetable wastes was carried out in a 2-stage reactor at a reported OLR of 3.5 g VS L−<sup>1</sup> day−<sup>1</sup> and HRT of 10 days, under a range of gas recirculation and H<sup>2</sup> addition rates [33]. H<sup>2</sup> addition gave increases in SMP of around 67% and 157% with and without optimised gas recirculation, respectively. The 2-stage system appeared capable of eliminating residual H<sup>2</sup> and was able to achieve biogas methane concentrations of 92% without and 99% with recirculation, respectively.

For studies using other co-substrates with CM as the minor component, see the sections on crops and agro-wastes and on food wastes below.

Several other studies not shown in Table S1 have used manure or manure-based digestates and feedstocks and/or inoculum. Batch experiments with CM as substrate were carried out to investigate the effect of H<sup>2</sup> addition at different ammonia concentrations [34,35]. Garcia-Robledo et al. [36] used micro-sensors to study the dynamics of H<sup>2</sup> and CO<sup>2</sup> conversion in fresh CM and CM digestate, but the experiment was not designed to establish likely yields in a typical commercial process. Lab-scale breeder reactors inoculated from CM digesters and fed on digestate from these were used to generate material for batch testing with H<sup>2</sup> addition [37]. The work was part of a study investigating the effect of inoculum source on metabolic pathway, and also tested wastewater biosolids digestate; but again, the aim was not to simulate performance in a full-scale process.

The results in Table S1 confirm that useful increases in SMP can be achieved by biomethanisation of CO<sup>2</sup> in CM digestion: this is of particular interest for this substrate, which is known for its low energy potential, making small-scale on-farm digestion economically challenging [38]. High rates of H<sup>2</sup> transfer and conversion were demonstrated in several of the studies. Membrane systems (ceramic and HFM) generally achieved good transfer without gas recirculation, while digester mixing also had an effect, although high mixing rates did not always improve methane yields [26,29,31]. Mixing and mass transfer performance are strongly affected by scale: effective mixing is more difficult to achieve in a full-scale plant, while depth and pressure effects may contribute to improved gas transfer [39–41]. Understanding of these effects and how they interact with system biology is still in its early stages, and more research at pilot and full scale is clearly needed if industry is to have confidence in adopting these technologies [3,4].

As can be seen from Table S1, the change in biogas methane content brought about by CO<sup>2</sup> biomethanisation is highly dependent on the applied H<sup>2</sup> loading and the effectiveness of the transfer method, as well as requiring the presence of an appropriate microbial population. Residual unconverted H<sup>2</sup> reduces the methane concentration in the output gas. Depending on the intended end-use, this may be a significant issue, while low transfer and conversion rates arguably represent a waste of H2. Two-stage systems were able to achieve a high methane content, but provision of a second digester is expensive, especially if heating and mixing are applied. In the systems described here, the primary digester generally received both organic feed and H2, with biogas and digestate then passed to the secondary digester to improve the H<sup>2</sup> conversion with or without further gas recirculation. This approach is already a step towards ex-situ biomethanisation, in which case, a reactor type with more efficient gas transfer may be preferred. Many manure digesters have unheated static tanks for digestate storage, however, so there may also be scope for research on whether and how these could be adapted to fulfil a similar role in increasing H<sup>2</sup> conversion.

Values for VMP and MER varied considerably between the studies in Table S1, reflecting differences in substrate properties and concentration/dilution, as well as the presence of residual unconverted H<sup>2</sup> and CO2. These parameters are commercially significant, however, and care is therefore needed when interpreting or reporting them. Co-digestion of Whey + CM may be attractive as a means of increasing the volumetric and specific methane productivity as well as reducing the pH, while both substrates may be locally available in dairy areas.

Conclusions for pig manure were broadly similar to those for CM, despite some differences in the typical characteristics for these substrates [16].

Most of the studies reported in Table S1 were carried out under thermophilic conditions, which generally give a higher SMP at a shorter HRT than mesophilic manure digestion [21]. Thermophilic digestion of CM is not widely practiced in the UK, however; and in combination with the different conditions applied in different studies, this makes it difficult to choose appropriate values for use in the current modelling. It was therefore decided to take typical baseline values for mesophilic CM digestion of SMP 0.190 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS, biogas composition 60:40% CH4: CO<sup>2</sup> and pH 7.5 [16,24], and apply the pH/pCO<sup>2</sup> relationship developed in Tao et al. [12]. Assuming stable mesophilic operation with H<sup>2</sup> addition is possible at pH 8.2 (Table S1, [21]), this equates to a biogas methane


content of 95% and SMP around 0.300 L CH<sup>4</sup> g

**Table 1.** Estimated maximum biogas methane content and SMP for use in modelling.

values are summarised in Table 1 for application in the modelling stage of this work.

<sup>−</sup><sup>1</sup> VS with CO<sup>2</sup> biomethanisation. These

<sup>a</sup> Based on literature values (see Tables S1–S4), cross-checked with ADAT [16] where possible. <sup>b</sup> Coefficient for equation relating pH and pCO<sup>2</sup> under mesophilic conditions [12].

#### 3.1.2. Crops and Agro-Wastes

Relatively few CO<sup>2</sup> biomethanisation studies to date have focused on these substrates. A comparison of different possible scenarios for in-situ and ex-situ CO<sup>2</sup> biomethanisation was carried out based on experimental data for grass silage [42]. Only in-situ results are included in Table S1, although ex-situ systems may correspond to the second stage of a two-stage process in some operating configurations. In this part of the study, the grass silage was digested thermophilically in a single-stage CSTR at an OLR of 4 g VS L−<sup>1</sup> day−<sup>1</sup> and HRT of 46 days using two types of diffuser. The findings demonstrated once again the importance of effective mass transfer, with the ceramic diffuser achieving a 68% increase in SMP compared to 19% for the low-capacity diffuser, and quite creditable VMP and MER values, although still with a significant residual H<sup>2</sup> content.

130-L anaerobic filters (AF) fed on maize silage hydrolysate were trialled at two H2/CO<sup>2</sup> ratios, using a venturi nozzle for gas injection and with both liquid and gas recirculation [43]. The AF acted as the methanogenic reactors in a 2-stage system, in which the first stage was a continuously stirred acidification reactor fed on maize silage with a small component of sugar beet silage effluent. The fixed bed in the AF consisted of high-density polyethylene bio-media with a specific surface area of 312 m<sup>2</sup> m−<sup>3</sup> . The AF configuration offers potential advantages in biomass retention and gas transfer, but previous studies have focused on ex-situ biomethanisation. It should be noted that some values presented in Table S1 differ from those reported in the paper due to differences in definition and calculation methods. Maize silage was also the feedstock in a 2-phase thermophilic system consisting of hydrolysis and fixed-bed methanogenic reactors [44]. The authors trialled a range of hydrolysis conditions, with H<sup>2</sup> added to the methanogenic reactor in some runs. Specific methane production is only reported for the whole system, and the control and experimental methanogenic reactors were, respectively, 145 and 180 L, making it difficult to compare performance on a volumetric basis; but methane concentrations were successfully increased to over 90%.

Digestate from an AD plant processing mixed agro-wastes was used as inoculum and feed in a single-stage mesophilic CSTR trial at different H<sup>2</sup> loadings [45]. The feedstock of the main AD plant is not described in detail, but is believed to include pig manure, deep litter, slaughterhouse residues and some high-lipid wastes [46]. H<sup>2</sup> was added to the headspace for up to 5 consecutive days, followed by a pause of 10–21 days, in a process described as pulsed injection. While the use of digestate in this way may be viewed as replenishing an inoculum rather than adding a substrate, the slow anaerobic biodegradation rates of some of the original feedstock components mean there is likely to be residual methane potential even after conventional digestion: in this case, the control reactor without H<sup>2</sup> addition had an SMP of 0.293 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS, higher than typical values for CM. This increased to 0.571 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS at the best-performing H2/CO<sup>2</sup> ratio

tested. Although the experimental method did not simulate a conventional AD process, it confirmed the potential for CO<sup>2</sup> biomethanisation to raise methane yields for this material, and provided some insights on process and microbial community parameters. The same set-up was used to explore the effect of H<sup>2</sup> additions at higher and lower headspace CO<sup>2</sup> concentrations across a range of OLR [47], with the results also showing acclimatisation over successive pulsed injection cycles (Table S2).

Two important studies using a similar mixed-agrowaste feedstock containing manure, straw, deep litter, grass and maize silage were carried out in a 1110 m<sup>3</sup> CSTR operating at 52 ◦C with a HRT of 13 days [39,40]. Some details of the substrate components, which varied over the experimental period, are provided in the papers. H<sup>2</sup> was injected using a venturi system over relatively short periods in order to allow evaluation of the mass transfer potential at different flow rates and with/without gas recirculation. These studies are not included in Table S2 as they represent short-term tests without equilibration and acclimatisation of the system; but the very large size supported by detailed analyses of performance provides valuable reassurance on the potential for full-scale application.

A full-scale thermophilic digester fed on a mixture of pig and cattle manure, maize silage and deep litter was used to provide inoculum for batch reactors. These were fed once only on maize leaf then periodically supplemented with H2, in an experiment designed to investigate metabolic pathways and microbial community structure [48]. Fed batch operation was also employed in a serum-bottle trial using inoculum from a mesophilic AD plant fed on maize and sweet sorghum silage and pig manure, with cellulose as the trial feedstock [49]. Sarker et al. [50] operated a 5-L CSTR at 39 ◦C, using inoculum from a food waste digester with a 0.55% *w*/*w* admixture of CM. From day 39–71 the digester was fed intermittently on CM without digestate removal. Increasing volumes of H<sup>2</sup> were injected sequentially between days 40 and 71 into the headspace without gas recirculation. VFA concentrations increased and effective CO<sup>2</sup> biomethanisation was not achieved, which the authors attributed to mass transfer limitations. Although these studies provide insights on the respective feedstocks and inoculums, they did not attempt to simulate conventional digestion and thus do not appear in Table S2.

Details for the two types of crop material investigated are summarised in Table S2. Digestion of grass silage in a conventional CSTR with H<sup>2</sup> addition via a ceramic diffuser was an effective means of increasing SMP [42]: values for VMP and MER were also relatively high, thanks to the good baseline SMP and applied OLR for this substrate. The output gas contained a significant percentage of residual unconverted H2; but ratios between H<sup>2</sup> consumed and methane produced were close to stoichiometric, giving confidence in the experimental results. In situ CO<sup>2</sup> biomethanisation in fixed bed reactors processing maize hydrolysates was also effective at increasing SMP [43,44], though here too, there was relatively little improvement in biogas methane content under the operating modes tested. CH<sup>4</sup> production and CO<sup>2</sup> consumption ratios in Illi et al. [43] were lower than would be expected based on H<sup>2</sup> transferred, possibly indicating dissolution in the liquid phase; no detailed analysis was possible for Schönberg and Busch [44] as results for the two stages could not be disaggregated.

Studies on CO<sup>2</sup> biomethanisation of digestates from mixed agro-wastes clearly demonstrated the potential for increasing methane yields towards those typical of commerciallyinteresting substrates (Table S2, [45,47]). The high biogas methane concentrations achieved here reflected the pulsed addition mode of H<sup>2</sup> operation, however, and cannot be transferred directly to conventionally operated full-scale digestion.

Based on the results in Table S2 and the ADAT database of feedstock properties [16] a baseline SMP of 0.35 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS with a biogas methane content of 55% was selected for both crops and mixed agro-wastes, giving an assumed maximum methane content of 95% (Table 1).

#### 3.1.3. Food Wastes

Mixed food waste from a commercial AD plant, including wastes from food processing, catering, dairy production and selected abattoir waste fractions, was the substrate for a mesophilic trial with biomethanisation of both endogenous and exogenous CO<sup>2</sup> [12]. Conversion of the internally produced CO<sup>2</sup> increased the SMP from 0.561 to 0.776 L CH<sup>4</sup> g −1 VS with a methane concentration of around 90% and no residual H2. Attempts to further increase the methane content led to VFA accumulation, and it was concluded that stable operation was possible up to pH 8.5. Addition of exogenous CO<sup>2</sup> with H<sup>2</sup> enabled a further increase in SMP to 1.215 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS with the same biogas composition. The system achieved an impressive VMP of 5 L CH<sup>4</sup> L <sup>−</sup><sup>1</sup> day−<sup>1</sup> , and the authors commented that the maximum exogenous CO<sup>2</sup> addition was determined by the capacity of the experimental equipment rather than by any biological limitation in this case.

Source-separated domestic food waste has been widely studied because of its rising popularity as an AD feedstock. A mesophilic trial carried out using FW from a source characterised in previous work [51,52] demonstrated an increase in SMP from 0.446 to 0.719 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS and stable operation at pH 8.1. Around 45% of CO<sup>2</sup> was converted, but the methane content of the output gas was only around 74% due to the presence of residual H<sup>2</sup> [53]. A combination of H<sup>2</sup> addition and auto-generated pressure was tested as a means of increasing biogas methane concentration using a synthetic FW composed of tinned pork and beans with white bread [54]. In the example in Table S3, the quantity of H<sup>2</sup> added was only sufficient to convert around 9% of the endogenous CO<sup>2</sup> on a stoichiometric basis; but when combined with additional dissolution of CO<sup>2</sup> caused by the 5-bar operating pressure, the biogas methane concentration increased to 90.6%.

Food waste from a University cafeteria was the organic substrate in a trial involving injection of a synthetic syngas consisting of H<sup>2</sup> and carbon monoxide (CO) at a 5:4 *v*/*v* ratio [55]. Both mesophilic and thermophilic conditions were tested and the digestates were also pyrolysed to determine the properties of the resulting syngas, with a view to creating an integrated process. The authors noted that methane productivity was higher than expected from stochiometric conversion of syngas, and attributed this to synergistic effects; although the difference decreased as the trial progressed. There is some inconsistency, however, between the reported feedstock properties, OLR and HRT: a feedstock VS of 25.1 g L−<sup>1</sup> would give an HRT of 7 days at the reported OLR. A value of 251 g VS L−<sup>1</sup> is more typical for FW, and without dilution, would correspond to an HRT of around 71 days: the reported HRT is 20 days, however, so feedstock dilution may have been carried out to facilitate thermophilic operation without ammonia inhibition [56]. Digesters typically require around 3 HRT to approach steady-state conditions with regard to organic loading, and this washout process might offer an additional explanation for observed changes in VS and COD content if the HRT was set at 20 days. In any case, stable operation was achieved with high syngas utilisation, a methane concentration of around 64%, and an increase in SMP of around 31% and 33% in mesophilic and thermophilic conditions, respectively.

CO<sup>2</sup> biomethanisation of thermally-treated FW digestate from a laboratory-scale digester was carried out in a trickle-bed reactor operating at 37 ◦C with a 10-day HRT, at different H<sup>2</sup> loadings and recirculation rates [57]. The SMP increased from 0.248 to 0.450 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS under the best conditions trialled, but biogas methane content was limited by the presence of residual H2, and dosing with HCl was introduced after the pH reached 8.45.

Batch experiments using food waste from a university dining hall were conducted to investigate the effect of acclimatisation and other parameters [58–60]. The trials used headspace injection over consecutive periods and were not designed to simulate conventional AD. There were some errors in the mass balance calculations and the H<sup>2</sup> addition proposed is far from stochiometric needs or typical literature values (Tables S1–S4) [59]; but inoculum acclimatisation was clearly demonstrated.

Other CO<sup>2</sup> biomethanisation studies have used specific waste streams from food processing industries. Cheese whey and related dairy wastes were used as the sole or main

substrate by several researchers. Treu et al. [61] attempted to digest whey at mesophilic and thermophilic temperatures with and without CO<sup>2</sup> biomethanisation, but were unable to achieve stable operation in thermophilic conditions. NaHCO<sup>3</sup> was added to provide buffering during mesophilic digestion without H<sup>2</sup> addition; and consumption of accumulated VFA may have affected methane production values in some periods. A feedstock of cheese whey permeate and cheese powder was successfully digested at 54 ◦C using NaOH for pH control [62]. In both studies, SMP increased with biomethanisation, but residual H<sup>2</sup> reduced the biogas methane concentration. Co-digestion of cheese whey with CM was successful as a means of improving process stability in thermophilic conditions [61,63], though increases in methane content and SMP were relatively small under the conditions applied. The co-digestion trial begun in Treu et al. [61] was continued in a study focusing on the effects of bioaugmentation [64]: no control digester without H<sup>2</sup> addition was run and the results are therefore not presented in Table S3.

Other food industry substrates trialled include potato starch wastewaters [65] and bioethanol distillery effluents [66]. The former was tested in a thermophilic UASB reactor with different diffuser types and gas and liquid recirculation rates. As can be seen from the results in Table S3, H<sup>2</sup> transfer efficiency was higher with a ceramic diffuser and gas recirculation but some residual H<sup>2</sup> was still present in the output gas. The latter study used a 148-L mesophilic anaerobic membrane bioreactor, and achieved a biogas methane content around 98% with minimal residual CO<sup>2</sup> or H2. Both studies contained minor inconsistencies between reported values for some parameters, but clearly demonstrated that CO<sup>2</sup> biomethanisation could successfully enhance SMP from organic wastes in these systems.

A synthetic feed containing yeast extract, sucrose and a range of nutrients was used in trials to assess the influence of OLR [67] and of total ammonia nitrogen (TAN) concentration [12] on CO<sup>2</sup> biomethanisation performance. Mesophilic digesters at OLR of 2 and 3 g COD L−<sup>1</sup> day−<sup>1</sup> also received additional exogenous CO<sup>2</sup> and H2, increasing the original SMP by more than 2-fold and giving biogas methane contents of around 90%. The limiting pH for this substrate and inoculum appeared to be around 7.9 at lower TAN concentrations, and 8.2 at higher TAN. The authors noted that further studies are needed to identify the factors determining maximum operating pH in different systems [12].

Glucose was used as a model feedstock in mesophilic reactors in a trial investigating the effect of different H<sup>2</sup> loadings [68]. H<sup>2</sup> was injected once per day into the headspace, which was sealed and allowed to pressurise until venting before the next injection. The inoculum used came from a digester fed primarily on cattle manure, and nutrients were provided by occasional dosing with diluted inoculum. The applied OLR was very low and pH control was required, but the mode of operation enabled a reported increase in biogas methane content from 66% to 94%. A two-stage mesophilic UASB system with glucose as the sole carbon source was tested at different OLR, H<sup>2</sup> loadings and gas recirculation rates [69], and achieved H<sup>2</sup> transfer efficiencies of up to 98.8% with biogas methane contents between 92–94%. Glucose was also used as the organic substrate in the trial of CO injection into a mesophilic UASB reactor [70]. The SMP increased from 0.312 to 0.536 L CH<sup>4</sup> g −1 COD added but biogas methane content fell due to additional CO<sup>2</sup> production, as indicated by the stoichiometric relationship shown in Equation (2):

$$\text{4CO} + 2\,\text{H}\_2\text{O} \to \text{CH}\_4 + \text{3CO}\_2\tag{2}$$

Synthetic substrates are generally adopted to provide controlled conditions for laboratory studies. Where the substrate is chemically defined, the theoretical SMP and expected biogas methane content are known or can be estimated; while parameters such as operating pH may be determined by the substrate composition or be controlled for experimental purposes. In any case these materials are rarely encountered as real-world feedstocks, and thus, values for modelling purposes were not required.

The characteristics of individual waste streams from food processing industries vary widely and it is clearly not possible to choose representative values; but information on baseline digestion conditions is often available from other studies. The minimum CO<sup>2</sup> and corresponding maximum achievable methane content for stable operation is more difficult to determine, and may depend in part on reactor type and operating conditions. In CO<sup>2</sup> biomethanisation trials with potato starch wastewater in UASB reactors (Bassani et al., 2016), stable operation was reported at a pH 8.38 with a CO<sup>2</sup> content 10% in the output gas. For ethanol distillery wastewater treated in an anaerobic membrane bioreactor (AnMBR) [66], the pH had reached 7.9 at a biogas methane content of 97.9% CH<sup>4</sup> and 1.4% CO2, with a corresponding increase in SMP from 0.297 to 0.389 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS: values close to these can therefore be taken as the maximum for this substrate for modelling purposes.

For post-consumer domestic food wastes in Europe, the SMP value is typically around 0.450 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS [71] with a methane content of around 54%. There are no published studies to confirm the minimum biogas CO<sup>2</sup> content or maximum pH for stable operation, so for the purposes of modelling, a maximum pH of 8.2 was conservatively assumed [53], corresponding to a biogas methane content of 79% (Table 1).

#### 3.1.4. OFMSW

Very little work has been done on CO<sup>2</sup> biomethanisation of OFMSW. When H<sup>2</sup> and CO<sup>2</sup> were batch-fed to samples from real and simulated landfill wastes, methane production was observed but homoacetogenesis was found to be the dominant pathway under the conditions used [72]. Mixtures of CM and synthetic OFMSW (composed of paper, bread and fruit and vegetable wastes) were batch digested at 55 ◦C at five CM:OFMSW ratios from 100:0 to 50:50 on a mass basis [73]. A nutrient medium containing glucose was added but the only source of inoculum was the fresh CM. After 24 h of fermentation, H<sup>2</sup> and CO<sup>2</sup> were injected continuously at a 4/1 *v*/*v* ratio for 20 days. The results confirmed that increases in VMP could be achieved using this approach, with biogas methane contents up to 97%; the values are not included in Table S2 as the study was not designed to simulate a conventional AD process. In the absence of suitable experimental findings, a baseline SMP of 0.35 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS with a biogas methane content of 60% was adopted [16] for the purposes of modelling, with assumed baseline and maximum pH values of 7.5 and 8.2, respectively, giving a maximum methane content of 92% [53].

#### 3.1.5. Sewage Sludges

AD is a well-established treatment technology for sewage sludges, with existing infrastructure across the water industry, making CO<sup>2</sup> biomethanisation of these substrates an interesting prospect. Alfaro et al. [74] carried out an extended trial using two 20-L mesophilic CSTRs fed on thickened primary and secondary wastewater biosolids from a full-scale wastewater treatment plant (WWTP). H<sup>2</sup> was injected via a HFM and the system was tested at different gas recirculation rates. The baseline SMP varied slightly with different batches of feedstock, but SMP was increased by more than 40% with a biogas methane content of 70–73% and good H<sup>2</sup> utilisation at higher recirculation rates. A study in a mesophilic fermenter with H<sup>2</sup> addition via bubbling and gas recirculation produced a similar improvement in SMP, and demonstrated stable operation at a biogas methane content of 90% with a maximum pH in the range 7.9–8.0 [12]. A trial using the same equipment and a similar sludge from a different wastewater treatment plant achieved a biogas methane content of 85% at pH 7.9. Addition of exogenous CO<sup>2</sup> and H<sup>2</sup> enabled a four-fold increase in the original SMP, but the biogas methane content fell due to the presence of residual CO<sup>2</sup> and H<sup>2</sup> [53].

An enrichment trial was carried out in three mesophilic CSTR digesters fed on mixed primary and secondary sludge [75]. The low feedstock solids content led to a relatively low OLR of around 1 g VS L−<sup>1</sup> day−<sup>1</sup> at an HRT of 15 days. Batchwise addition of H<sup>2</sup> to the headspace produced a maximum biogas methane content of 80% with a near-stoichiometric ratio between H<sup>2</sup> consumption and CH<sup>4</sup> production. Another trial using sewage sludge from the same source and continuous H<sup>2</sup> addition achieved a biogas methane content of 90% at a H2/CO<sup>2</sup> ratio of 7/1 without any adverse effect on VS degradation or significant VFA accumulation [76].

A study combining H<sup>2</sup> addition and pressurisation was carried out in a 35-L mesophilic digester treating mixed wastewater biosolids under a range of operating pressures and at two H<sup>2</sup> loadings [77]. The experimental design did not include a control without H<sup>2</sup> addition, but the best performance was achieved at the highest pressure and H<sup>2</sup> loading tested, with a SMP of 0.418 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS and a biogas methane content of 92.9%. Based on VS removal, it was estimated that around 0.37 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS of the SMP was due to organic load and 0.13 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS to CO<sup>2</sup> biomethanisation. Digester pH remained below 7.2 presumably due to CO<sup>2</sup> dissolution, and there were no signs of process instability or VFA accumulation. Pressurisation is likely to require more extensive modifications to reactor design and operation than H<sup>2</sup> addition under ambient pressures, but the authors noted that process efficiency in terms of VS removal was unaffected by the changing conditions and that a pressure increase could be an effective way to enhance H<sup>2</sup> mass transfer without incurring high energy costs.

Addition of CO was trialled in a thermophilic digester fed on mixed primary and secondary wastewater biosolids [78]. The highest SMP achieved was more than twice the value without CO addition, but the biogas methane content was low (<20%) due to the presence of both residual CO and additional CO<sup>2</sup> generated in accordance with Equation (2). The highest SMP without residual CO was around 1.7 times the baseline value with a biogas methane content of around 30%. The CO was added via a HFM module, ensuring good dissolution despite its rather low solubility, and no signs of process inhibition were observed. A HFM module was also used to inject simulated coke-oven gas (SCOG) consisting of 92% H<sup>2</sup> and 8% CO into a mesophilic CSTR digester fed on mixed primary and secondary sludge [79]. The maximum SMP achieved was 0.604 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS, compared to a baseline value of 0.256 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS without SCOG addition. Maximum biogas methane content was 98.8% with 0.3% CO2, and pH controlled to 8.0.

In other work with wastewater biosolids, Hu et al. [80] carried out batch tests using nano-scale zero-valent iron and waste iron scraps as a means of generating H2, with waste activated sludge (WAS) as an organic substrate in some set-ups. Vechi et al. [37] used lab-scale reactors inoculated with sewage sludge digestate and fed on primary and secondary sewage sludge to produce material for batch testing of H<sup>2</sup> additions, for comparison with inoculum from CM digestion. Inoculum from a range of sources including digestion of sewage sludge, paper mill sludge, cattle and poultry manures and FW, plus aerobic wastewater sludge, was tested to investigate the immediate response of different microbial communities to exposure to high H<sup>2</sup> partial pressures [81]. These studies did not attempt to simulate typical operating regimes but provide a variety of insights on microbial populations and metabolic pathways.

Apart from one study of with CO addition [78], all of the work presented in Table S4 was carried out under mesophilic conditions, reflecting the current widespread use of this temperature range by the water industry. CO<sup>2</sup> biomethanisation gave useful increases of 40–50% in SMP. VMP remained low, reflecting the dilute nature of this substrate, but high values were achieved when additional CO<sup>2</sup> and H<sup>2</sup> were added [53], reinforcing the observation that at sites with multiple digesters a single digester retrofitted for H<sup>2</sup> injection should be capable of processing biogas from several others [12]. Good H<sup>2</sup> transfer was achieved at the scales tested even without membrane diffusers, and ratios between H<sup>2</sup> consumed and methane produced were close to stoichiometric values. The results indicate that CO<sup>2</sup> biomethanisation of this type of substrate can operate stably in a pH range of 7.9–8 [12,74] corresponding to a max biogas methane content around 90% [12]. Baseline SMP values without CO<sup>2</sup> biomethanisation vary depending on factors such as HRT, OLR and the proportion of primary and secondary wastewater biosolids: for the conventional CSTR trials reported here, they ranged from 0.21 to 0.3 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS with a biogas methane content of 60–65%. These correspond well to typical values of 0.260 L CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS and 60% CH<sup>4</sup> given by ADAT [16], and were therefore taken forward into Table 1.

#### 3.1.6. Conclusions from Performance Analysis for CO<sup>2</sup> Biomethanisation of Organic Feedstocks

Consideration of the literature shows a wide range of study types, though there are also obviously many gaps. Table 1 contains a group of parameters for each feedstock type, selected specifically for the purposes of the current modelling work. In Tables S1–S4, a wider selection of parameters and indices to facilitate comparison between results is presented: the significance of these is briefly discussed in the Supplementary Materials. Tables S1–S4 focus on the most successful outcomes in which stable operation was apparently achieved: the studies considered cover many other sets of operating conditions, including some where failure occurred as evidenced, e.g., by irreversible VFA accumulation. Many studies also now include analysis of microbial populations, which when combined with operational data contributes to a growing understanding of the links between community structure, functionality and performance. The majority of studies reported are still at laboratory scale, however, and more work is needed on the effects of scale-up and the interaction of system biology and engineering parameters such as mixing and mass transfer behaviour. For industry to be able to adopt this technology with confidence, a better understanding of the mechanisms affecting performance is needed, together with the development of simple and robust control strategies to avoid instability or suboptimal conditions.

While there are issues in attempting to compare data from different studies, Tables S1–S4 provide some useful insights and parameter values. VMP, MER and H<sup>2</sup> transfer efficiency are important for techno-economic assessment, although values are likely to vary considerably with operating scale and system configuration. SMP is an indicator of the efficiency of conversion for a particular feedstock, while CH<sup>4</sup> and H<sup>2</sup> content of the output gas affect potential end uses.

In many studies, though not all, the ratio of H<sup>2</sup> transferred to CH<sup>4</sup> produced and to CO<sup>2</sup> removed settled at close to the stoichiometric value of 4. Values diverging from this can be explained by utilisation of transferred H<sup>2</sup> for other purposes such as VFA production or biomass growth. Most studies that calculate ratios of this type also choose to ignore the very small amount of dissolved H<sup>2</sup> leaving in the digestate [1,77,82]. Intriguingly, in several cases, the introduction of CO<sup>2</sup> biomethanisation is associated with an increase in VBP [21,22,24,27,29,33,43,45,47,57,61,66,68,74], meaning a larger amount of CH<sup>4</sup> is being produced and/or there is more CO<sup>2</sup> to biomethanise, per unit of organic feed.

It is difficult to assess the significance of this as other factors can also vary: since these studies use real organic feedstocks, there may be minor changes in feedstock properties both day-to-day and between batches in longer studies. HRT are relatively long and not all reported values are from periods where stable steady state operation (often defined as 3 HRT under the same conditions) has been achieved. Accurate determination of H<sup>2</sup> concentrations can also be problematic and as noted, there are different ways of dealing with variability in experimental measurements, with some studies normalising gas composition to 100% and others not. In some cases, however, the observed MER is greater than the volumetric CO<sup>2</sup> production without CO<sup>2</sup> biomethanisation [21,30,33,45,66,68], strongly supporting an increase in VBP. A similar phenomenon of increases in VBP or VMP has been observed in other work on gas recirculation [83], though no clear mechanisms have been confirmed. Many studies are not designed to separate the impacts of gas recirculation and H<sup>2</sup> addition: but Khan et al. [33] found that the increase from H<sup>2</sup> addition and recirculation was greater than from recirculation only. The simple method adopted in the current work, of calculating the MER and the increase in SMP from the difference between experimental and control or baseline values, conflates these effects as well as other factors; but several authors have suggested that addition of H<sup>2</sup> or syngas may have synergistic effects beyond stoichiometric CO<sup>2</sup> conversion [25,33], and more research is clearly needed in this area.

### *3.2. UK Feedstock Data Extraction and Analysis of CO<sup>2</sup> Biomethanisation Potential*

Several data sources, in addition to those used, were considered as a basis for assessing the potential for CO<sup>2</sup> biomethanisation from organic wastes in the UK. This included three UK government sources based on specific incentive schemes and expressed in terms of energy generation capacity: the FIT, RHI and Renewable Transport Fuel Obligation (RTFO). In addition, several other organisations collate data on AD plant operations, including the Anaerobic Digestion and Bioresources Association (ADBA), the Renewable Energy Foundation (REF) and the National Non-Food Crops Centre (NNFCC). The different datasets are collected for different purposes, do not cover all materials and use different reporting formats and assumptions, making comparison and generalisation difficult. Brief details on these sources and their advantages and disadvantages are presented in Section S2 of Supplementary Materials.

#### 3.2.1. Estimation of CO<sup>2</sup> Biomethanisation Based on Ofgem RO Data

Although only a proportion of UK AD plants generating electricity are covered by the RO, the RO dataset on feedstock and biogas production is arguably the most detailed freely available. The Ofgem 2019–2020 RO Annual Report [13] aggregates the feedstock consignment data provided by 157 AD plants operating under the RO scheme into 10 broad feedstock categories: Silage; Food, garden and plant waste; Manures and slurries; Distillery waste; DAF sludge/wastewater; Crops; Glycerol; Dairy waste; Municipal waste; and Other. For this part of the study, it was decided to use the aggregated categories and not the original RO-SUS data [15] because it was not clear how the consignment data had been mapped to the categories and because some of the individual RO-SUS consignment data were coded, so the organic source could not be ascertained. The Ofgem feedstock categories also did not map precisely onto those used in the literature survey. The following assumptions were therefore made and confirmed where possible by cross-checking with uncoded RO-SUS data.

Tables S1–S4 demonstrate that, under suitable conditions, biogas methane contents of 90% and above can be achieved for a range of feedstocks commonly used in commercial AD plants. Where possible, expected values for methane content after CO<sup>2</sup> biomethanisation of the Ofgem feedstock categories were therefore taken from Table 1; where no values were available (i.e., Distillery waste, Glycerol, Dairy waste and Other, detailed below), a methane content of 90% was assumed.

The majority of biogas in the 'Silage' category derives from maize, although RO-SUS data also shows the presence of grass silage, and crop silages such as those from rye and wheat. This category was therefore assumed to map onto 'Crop and agro-wastes' in Table 1 and to have the TS and VS content of maize silage in the ADAT database [16].

'Food, garden and plant waste' likely included post-consumer food waste (e.g., from households and restaurants), as well as food production and processing wastes such as reject potatoes, fruit and vegetable waste. It could also include garden and plant materials such as 'Food/garden waste', bulbs, grass and sugar beet pulp. The majority of named plant materials were high energy (e.g., sweetcorn and beet), however, and consignments mentioning 'Food/garden waste' only appeared to make up around 5% of the total for this category. 'Post-consumer food waste' values from Table 1 and source separated food waste from ADAT [16] were therefore adopted for this category.

The 'Manures and slurries' category was conservatively assumed to have the characteristics of cattle slurry, since no indication was given of the proportion of manure to slurry, or of different slurry types. Data for 'Livestock manures' from Table 1 were used, with feedstock characteristics for 'Cattle Slurry' from ADAT [16].

'Distillery waste' may include feedstocks such as draff, pot ale syrup and malt effluent. Again, a conservative approach was taken and, as aggregates, these were assumed to be low TS wastes. There was no directly applicable category in Table 1, so a default methane content of 90% was used. Feedstock characteristics were not available from ADAT [16], so data from the literature were used [84].

The 'DAF sludge/wastewater' category was assumed to equate to the feedstock characterisation for 'Sewage sludge' in Table 1 and in ADAT [16].

The definition used by Ofgem [13] for the 'Crop' category is unclear, but might include feedstocks such as whole crop and maize meal. For these, the maximum methane content was taken from 'Crop and agro-wastes' in Table 1 with TS and VS values based on an average for maize silage and maize corn in the ADAT database [16].

'Glycerol' could be clearly identified in the RO-SUS data [15] and was characterised using standard ADAT values [16], with a default maximum methane content of 90%, as Table 1 had no direct equivalent.

The 'Dairy waste' category could include RO-SUS data for milk whey, dairy waste, liquid food/dairy waste, AD whey permeate and dairy sludge, and was assigned the characteristics of 'Whey' in the ADAT database [16]. With no direct equivalent in Table 1, a default maximum methane content of 90% was assumed.

'Municipal waste' was characterised as 'OFMSW' as this category appears in both Table 1 and ADAT [16].

The narrative in the Ofgem annual report identified 'Other' consignments as including blood, viscera, tallow, fishery waste and plant oils [13] which are all high-strength wastes, and were assumed to have the same properties as the Ofgem 'Food, garden and plant waste' category.

The dataset accompanying the Ofgem report [13] lists 'Quantity [of biogas] burnt (million m<sup>3</sup> )'. Biogas volumes for individual feedstocks in the dataset are reported in million m<sup>3</sup> , to a numerical total of 525,191,780. This does not match the report narrative, which notes production of 525 million m<sup>3</sup> ; and is also three orders of magnitude larger than the 364.06 million m<sup>3</sup> biogas in the 2018–2019 Annual Report and dataset [85]. The units in the database were therefore corrected from 'million m<sup>3</sup> ' to 'm<sup>3</sup> '.

To calculate the gross energetic value of the biogas, biogas volume was multiplied by the estimated methane content and the calorific value (lower heating value) of methane, taken as 35.82 MJ m−<sup>3</sup> CH<sup>4</sup> at STP, and converted to GWh as shown in Table 2. The gross energetic value with CO<sup>2</sup> biomethanisation was then calculated by multiplying the original gross energetic value by the ratio of biogas methane contents with and without CO<sup>2</sup> biomethanisation for the given feedstock, i.e., assuming no change in biogas volume. Table 2 shows the resulting improvement in gross energetic values through the increased methane production due to the addition of H2. This equates to a 57% uplift in the overall energy value of the biogas. While this assumes CO<sup>2</sup> biomethanisation could be applied in every participating AD plant, which is clearly an over-optimistic scenario, it nevertheless represents a very significant potential uplift that warrants further investigation.

In order to estimate the tonnages of fresh feedstock giving rise to the produced biogas, typical values for the SMP, total solids (TS) and volatile solids (as a percentage of TS) for each feedstock were taken from Table 1 and from the ADAT database [16]. Tonnages for each feedstock category were then calculated from these values as shown in Table 2. The assumptions used to derive the feedstock parameters are provided in Table 2, and result in a total estimated tonnage of 6.8 million tonnes of FM, of which 1.3 million tonnes consists of silage and crops. Together, silage (18.1%), food/garden/plant waste (31.5%) and manures/slurries (12.3%) produce 85.6% of the biogas shown in Table 2.

*Processes* **2022**, *10*, 1202


**Table 2.** RO AD plant feedstock characteristics with and without CO2 biomethanisation.

(h) = (e) ∗ (f)/(c). 7

(k) = (b) ∗ (c) ∗ 100/((d) ∗ (i) ∗ (j)).

#### 3.2.2. Estimation of CO<sup>2</sup> Biomethanisation Uplift Based on Defra's UK AD Feedstock Data

The 2019–2020 Defra feedstock data for anaerobic digestion [17] are derived from the (paywalled) NNFCC report on Anaerobic Digestion Deployment in the UK [86] which records annual feedstock usage in AD plants operating under a range of AD incentives. It therefore covers a wider range than the Ofgem data [13], which only include plants covered by the Renewables Obligation, in which the biogas produced is burned to generate electricity for on-site use and/or feeding into the national electricity grid. The values exclude feedstock data from the wastewater sector (i.e., sludges from WWTP).

The Defra data [17] are presented as percentages of a total tonnage across five feedstock categories: Crops, Food waste, Manures and slurries, Crop wastes and Others. These were mapped using the same assumptions on feedstock characteristics as in Table 2. Specifically, it was assumed that 'Crops' were likely to be silage so this category was treated as maize silage, even though there will be a proportion of other silages and crops. As with the Ofgem data [13], 'Manures and slurries' were conservatively characterised as slurry, as the proportions were unknown. TS, VS and methane production values for 'Crop waste' were aligned with those for 'Crops' in Table 2. Feedstock tonnages shown in bold in Table 3 were taken directly from the Defra report [17].

**Table 3.** Potential uplift in gross energetic value through CO<sup>2</sup> biomethanisation of reported UK feedstocks (Defra data).


<sup>1</sup> Based on Defra data [17]. <sup>2</sup> Tonnage FM basis. <sup>3</sup> From ADAT [16] using the assumptions described above. <sup>4</sup> From Table 2. 5 (g) = (b) <sup>∗</sup> (c) <sup>∗</sup> (d) <sup>∗</sup> (f) <sup>∗</sup> 35.82 MJ m−3/(100% <sup>∗</sup> 100% <sup>∗</sup> 3.6 <sup>∗</sup> <sup>10</sup><sup>6</sup> MJ GWh−<sup>1</sup> ). <sup>6</sup> (j) = (g) ∗ (h)/(e)

Table 3 shows the estimated gross energy value with and without CO<sup>2</sup> biomethanisation. The results indicate an overall energy uplift of 61% for this set of feedstocks-further indicating that CO<sup>2</sup> biomethanisation has the potential to significantly increase the energy value of existing feedstocks, as well as the viability of the digesters that produce the gas.

The overall gross energetic value is significantly larger than that derived from the Ofgem RO data as the Defra data covers more than twice the total tonnage of feedstock (Table 3). Additionally, the derived uplift value of 61% differs slightly from the Ofgem RO value of 57% due to differences in the relative proportion of feedstocks.

To estimate the potential biomethane increase from CO<sup>2</sup> biomethanisation, analysis of the available feedstock data in conjunction with data from the scientific literature is a logical approach. It is also possible to apply the overall feedstock-derived energy uplift to generation data when the feedstock underlying it is unknown.

The 2021 DUKES data [18] show energy production from UK AD plants for the year 2020 of 953, 68 and 490 thousand tonnes of oil equivalent (ktoe) for electricity generation, heat generation and grid injection of methane, respectively; no figures are specifically reported for use of biomethane as a vehicle fuel. Applying the DUKES data conversion factor of 11.63 GWh ktoe−<sup>1</sup> to this total of 1511 ktoe gives a total gross energy production from anaerobic digestion of 17,572 GWh. This is considerably more than the 11,576 GWh shown in Table 3, reflecting the degree of uncertainty associated even with official data. Timing differences could account for some part of this, e.g., the reporting year for incentives tends to run from 1 April to 31 March whereas DUKES [18] reports by calendar year. Further

differences could be due to the conservative feedstock assumptions made above; to DUKES load factor (LF) assumptions when deriving gross energy values from net electricity/heat generation; and/or to variations in data sources and rounding errors. Differences on this scale, however, strongly support the need for more unified and consistent data collection.

Nevertheless, applying the overall energy uplift of 61% derived from Table 3 would result in a gross energy value for AD of 28,317 GWh based on DUKES data [18]. It should be noted this value does not include biomethane used in vehicle fuel.

#### 3.2.3. Potential for CO<sup>2</sup> Biomethanisation Using Generation Derived from Wastewater Treatment

Neither the Ofgem/RO [13] nor the Defra [17] feedstock data shown above include the energy contribution from the AD of sludges from municipal wastewater treatment.

This approach could be used to ascertain improved energy production from DUKES data on sewage gas [18] which show electricity, heat and gas grid injection energy figures of 350, 90 and 54 ktoe, respectively, equating to a total of 5744 GWh. If the maximum methane content from Table 1 is applied, equating to a 38% uplift in gross energy output, generation of 7953 GWh could potentially be achieved if CO<sup>2</sup> biomethanisation were introduced across the wastewater sector.

In the UK, 93% of sewage sludge is treated by AD or advanced AD [87] at approximately 170 AD plants [88]. Unlike commercial and industrial plants where feedstocks are often mixed, and thus, characteristics can be variable, sewage sludge is a relatively consistent material. Additionally, the water industry has ambitious decarbonisation plans [89], as its energy use is considerable (e.g., in 2009 this sector accounted for up to 3% of total energy use in the UK [90]). A relatively small number of AD plants treating a large quantity of a minimally variable feedstock in a sector that has high energy use, relevant technical expertise and ambitious decarbonisation plans makes it relatively feasible to introduce this technology and achieve a significant proportion of the theoretical uplift.

### *3.3. Advantages of CO<sup>2</sup> Biomethanisation at Scale*

Despite some variability between sources, the existing feedstock and energy data, supported by the scientific literature, indicate that CO<sup>2</sup> biomethanisation has significant potential to increase the energy contribution of AD to the UK's energy mix, while also increasing the efficiency of utilisation of carbon from organic feedstocks. Applying biomethanisation technology to current DUKES [18] figures of 23,316 GWh (sewage gas and AD) could produce 36,270 GWh, an uplift of 55%. AD and sewage gas currently account for 15% of the UK's bioenergy contribution and this could increase to ~22%. This potential uplift of 12,954 GWh is equivalent to the annual energy use of 858,000 households [91], e.g., the domestic properties in a city the size of Leeds.

To effect a similar increase in current gas grid injection alone would require the growth of another 251,000 ha of maize (~4.4% of UK arable land), more than double the 121,000 ha used for all UK bioenergy and more than three times the 75,000 ha used for AD maize in 2020 [92]. Maximizing the carbon utilisation efficiency from existing feedstocks thus makes sense in terms of land use and the food/energy/climate nexus.

Although not explicitly included in the 2021 DUKES [18] figures, the role of biomethane for transport also is growing, particularly in the heavy goods vehicle (HGV) sector where it is currently the only commercial decarbonisation option, until suitable fuel current battery and hydrogen technologies can be developed. Transport is difficult to decarbonise: as energy systems have become less carbon-intensive, transport became the UK's largest emitting sector of GHG emissions in 2016 [93]. If the uplift were used for transport, it could replace nearly 1.1 million litres of diesel, with a fossil CO<sup>2</sup> equivalent of 3.4 million tonnes [94].

This approach could eliminate the energy and capital expenditure required to retrofit CO<sup>2</sup> carbon capture and storage (CCS), allowing more effective utilisation of short-cycle carbon rather than, for example, pumping it into permanent underground storage. This would have the added advantage of displacing its equivalent in fossil gas and associated imports: the Sixth Carbon Budget accounts for abatement from the additional use of biomethane to displace fossil gas [9] and CO<sup>2</sup> biomethanisation could offer an effective way to achieve this.

The longer-term value of incorporating CO<sup>2</sup> biomethanisation into AD systems is further strengthened by the CCC's projected growth in the hydrogen economy, in particular, green hydrogen from water electrolysis powered by renewables, since equipment and installation costs tend to fall as deployment increases. As noted earlier, achieving the benefits of CO<sup>2</sup> biomethanisation requires the use of renewables-based H2, and this may be seen as a barrier in the short term since competition for H<sup>2</sup> from grid-based resources is likely to intensify [95]. AD is used at a wide range of scales, however, from on-site slurry treatment at a single dairy farm to processing the municipal wastes of a city. It is therefore flexible in its needs and could be coupled with electrolytic H<sup>2</sup> production across a similar range. In their simplest form, small-scale AD applications can provide an opportunity for hybridisation of on-site renewable power with waste processing, to provide short carbon-cycle biomethane for local or on-site use in locations where export of electricity to grid is technically challenging or economically unattractive. Larger AD installations could be used, with relatively low capital expenditure [2,5,10], to transform excess intermittent power production from large-scale renewable facilities into biomethane for gas grid injection, hence providing buffer storage and allowing a more rapid expansion in the renewables sector. By taking electricity from multiple sources and across different time spans, the technology could thus make a major contribution in supporting the energy transition from carbon-based to non-carbon-based gaseous fuels [7,20].

Given the potential contribution of CO<sup>2</sup> biomethanisation, it is surprising that it does not appear to be mentioned in policy documents and remains under the policy radar, even where other technologies-some with technology readiness levels that are similar or lower-are actively discussed. For example, the BEIS November 2021 Biomass Policy Statement [96] which 'provides a strategic view on the role of biomass across the economy in the medium- to long-term' mentions the 'clear opportunity' for 'material processing of biomass into high value products'; but does not consider CO<sup>2</sup> biomethanisation as either an interim or long-term strategy. Indeed, although it identifies AD as 'the only commercially scalable technology currently available for greening the gas grid', the sole approach for AD's biogenic carbon is Bioenergy Carbon Capture and Storage (BECCs), which will require the development of an effective market for greenhouse gas removals. The UK market for biomethane-to-grid is already well-developed and growing, so maximizing the carbon utilisation in such plants instead of permanently storing it underground could potentially be a better option in the decarbonisation pathway. This, of course, would need to happen alongside the increase in hydrogen production outlined by the CCC.

#### **4. Conclusions**

Biomethane has an important role to play in the UK's energy decarbonisation plans, due to its flexible use in transport, heat and electricity production. A considerable amount of work has been conducted at laboratory scale on CO<sup>2</sup> biomethanisation of organic wastes. The results show clearly that significant improvements can be achieved in methane yield per unit of organic feed and in biogas methane content. Interestingly, the overall survey also appears to indicate that underlying biogas productivity can be increased in some systems. Many scientific and technical questions remain to be addressed, particularly those associated with the effects of scale-up and of differences in operating practice, which need to be resolved if industry is to adopt this technology with confidence. While approaches such as ex-situ CO<sup>2</sup> biomethanisation may be more suitable in some applications, however, it appears there is clear potential for in situ or hybrid conversion of CO<sup>2</sup> generated by anaerobic digestion of organic materials.

Data on existing UK AD feedstocks are highly fragmented, and it is not always clear how values in different sources have been arrived at or how they relate to one another. In the absence of coherent information, it is difficult for government bodies to reach conclusions on overall policy, especially when individual sectors and technologies are lobbying for their own interests. The availability of better and more consistent data sources as a basis for evaluation and policy making is thus a clear priority. Based on the available UK data, however, significant increases in biomethane productivity could potentially be achieved, ranging from 38–68% for different feedstock types and equivalent to an overall uplift in the contribution of AD to UK bioenergy from 15 to 22%. Again, there are many issues to consider: the current survey only looked at data on the highest level, and for realistic assessments of the potential scale and impact of technology application it will be necessary to take into account both the end-uses of biomethane and techno-economic viability on individual sites. The potential contribution from CO<sup>2</sup> biomethanisation of organic wastes is large enough, however, to warrant consideration in both short and long-term planning. There is thus a clear need for more work looking at both the research issues and the policy needs to optimise the contribution of this approach and integrate it with national energy and sustainability strategies.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/pr10061202/s1. Text S1: Assessment of performance data for CO<sup>2</sup> biomethanisation of organic feedstocks (including nomenclature); Table S1: Performance data for CO<sup>2</sup> biomethanisation of livestock manures; Table S2: Performance data for CO<sup>2</sup> biomethanisation of crops and agro-wastes; Table S3: Performance data for CO<sup>2</sup> biomethanisation of food wastes; Table S4: Performance data for CO<sup>2</sup> biomethanisation of sewage sludges; Text S2: Sources of Anaerobic digestion and Feedstock data.

**Author Contributions:** Conceptualization, A.B. and S.H.; methodology, A.B. and S.H.; validation, A.B., S.H., Y.Z. and C.J.B.; formal analysis, A.B., S.H. and Y.Z.; investigation, A.B., S.H. and Y.Z.; data curation, A.B. and S.H.; writing—original draft preparation, A.B.; writing—review and editing, A.B., S.H., C.J.B. and Y.Z.; funding acquisition: C.J.B., S.H. and Y.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the IBCat H2AD project (EP/M028208/1) funded by the Engineering and Physical Sciences Research Council (EPSRC), by a grant from the Carbon Recycling Network (POC-02-ZHANG-CCnet) funded by the Biotechnology and Biological Sciences Research Council (BBSRC), and by in-kind support from the Environmental Biotechnology Network (BB/S009795/1) funded by BBSRC and EPSRC.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data supporting this study are openly available from the University of Southampton repository at http://doi.org/10.5258/SOTON/D2241 (accessed on 30 May 2022).

**Acknowledgments:** The authors wish to acknowledge the help of Alam Khan of the Sustainable Bioenergy and Biorefinery Laboratory, Department of Microbiology, Quaid-i-Azam University, Pakistan for provision of additional information on his work, and Bing Tao, formerly of the University of Southampton, for additional data. Thanks are also due to the AD Working Group of the Environmental Biotechnology Network (BB/S009795/1) for comments on the text.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**

