Olive Mill Wastewater Valorization through Steam Reforming Using Multifunctional Reactors: Challenges of the Process Intensification

Olive oil mill wastewater (OMW) is a polluting stream derived from the production of olive oil and is a source of environmental pollution; this is relevant in many countries around the world, but particularly in all the Mediterranean region where major producers are located. In this effluent, several pollutants are present—namely, sugars, fatty acids, and polyphenols, among others. Nowadays, to reduce the pollutant load, several treatment techniques are applied, but these technologies have numerous cost and efficiency problems. For this reason, the steam reforming of the OMW (OMWSR) presents as a good alternative, because this process decreases the pollutant load of the OMW and simultaneously valorizes the waste with the production of green H2, which is consistent with the perspective of the circular economy. Currently, the OMWSR is an innovative treatment alternative in the scientific field and with high potential. In the last few years, some groups have studied the OMWSR and used innovative reactor configurations, aiming to improve the process’ effectiveness. In this review, the OMW treatment/valorization processes, the last developments on catalysis for OMWSR (or steam reforming of similar species present in the effluent), as well as the last advances on OMWSR performed in multi-functional reactors are addressed.


Introduction
The olive oil industry is very relevant in many countries around the world-particularly in the Mediterranean countries (cf. Figure 1), both economically and traditionally. Because of its excellent nutritional properties, olive oil contributes to cholesterol control and promotes cardiovascular health. This significance is explained in numbers: about 9 million hectares of olive trees have been cultivated worldwide during the 2019/2020 crop year [1]. For the 2019/2020 crop year, it was estimated that 70% of the production of olive oil around the world was performed in European Union countries, with the Iberian Peninsula (Portugal and Spain) being a leader zone (76% of the European Union's total)-see Figure 1 [1,2]. Nevertheless, there are other countries with high levels of olive oil production: Tunisia, Morocco, Syria, and the USA. Besides that, other countries-such as Australia, Argentina and South Africa-are also employing this type of crop [3].  [2]. Figure adapted from the site of International Olive Council [2].
In this way, it is observed that all these regions are significantly affected by the pollution resulting from olive oil production. The process yields massive amounts of liquid (olive oil mill wastewater-OMW) and solid waste, which have raised severe disposal difficulties for the producers. This agroindustry represents a huge environmental problem, where big quantities of effluents are produced in a small period. This environmental concern is emphasized by the large amount of word olive oil production/consumption (close to 3 million metric tons per year produced/consumed-this value has remained nearly constant in the last 10 years) [2,4,5]-see Figure 2. The aqueous residues related to this agro-industrial activity are produced using either the traditional pressing or the centrifugation methods (three-or two-phase systems) because, in all of them, water is used to generate the olive oil from the olive [1,6]. In Figure 3, a scheme of these processes of olive oil production can be seen [7,8].
The discontinuous pressing procedure (traditional method) is the oldest and most well-known method for processing olives. In the traditional pressing (nowadays adopted only in small mills/companies), the ground olives are pressed and then the liquid mixture is rested in a sequence of decanters to separate the olive oil. To separate the oil from the other phases, a small amount of water is added. This traditional process also produces a solid fraction denominated by olive pomace (or olive cake) constituted by water, olive stone, and olive pulp. This traditional method presents several advantages such as low-priced equipment and technological simplicity. Nevertheless, it also has handicaps, such as the fact of running discontinuity and high hand-labor costs [7]. Still, because it adds water in several steps of the production process (previous to the decantation step), a large quantity of OMW (approximately 40-60 L/100 kg olives) is produced [7]. The OMW generated has a higher chemical oxygen demand (COD) compared with the OMW produced by the other methods [9]. Until 40 years ago, this was the only process used, although financial problems pushed olive mills to gradually abandon the traditional process [10,11].
In the centrifugation systems, crushed olive fruits are fed into a two-phase or threephase decanter centrifuge and then the impure olive oil is centrifuged [12]. These processes are based on the differences in the density of the several components obtained (solids, olive oil, and water) and can produce three distinct fractions: pomace, olive oil, and OMW-see Figure 3.
Using the three-phase process (route a) in Figure 3, the olive paste is divided into three phases in the horizontal centrifugation (water added in this step): olive oil, OMW, and pomace [7,8]. Then, the liquid fraction generated (olive oil + OMW) is centrifuged (vertical centrifugation) with additional water to recover a high-quality olive oil, producing a high quantity of OMW-see Figure 3. Since the continuous three-phase centrifugation method adds water in the two centrifugation steps, a high volume of OMW is produced (about 100-120 L/100 kg olives-1.7-3 times more compared to the traditional method) [8]. Nevertheless, this three-phase system has various advantages, such as better quality of the olive oil and complete automation. Nonetheless, this system also presents several disadvantages, such as a higher level of water and energy consumption, higher volume of OMW produced (as previously mentioned), and more expensive installations [3]. Main processes of olive oil production-traditional (pressing) method, (a) three-phase centrifugation method and (b) two-phase centrifugation method [7,8]. Reprinted with permission from Ref. [7]. Copyright 2021 Elsevier.
To minimize the impact of OMW, the two-phase centrifugation process (route b) in Figure 3 was developed during the 1990s [7]. Using this two-phase process, the olive paste is divided only into two phases in the horizontal centrifugation (no water added in this step): wet pomace and olive oil, the first known as two-phase olive mill waste (TPOMW), a combination of pomace and OMW-see Figure 3 [7]. Then, the olive oil generated is washed (vertical centrifugation-oil washing) with a very small amount of water, producing only a small OMW stream-see Figure 3 [8]. Despite the lower volume of OMW produced in comparison with the three-phase process, the resulting TPOMW (10 L/100 kg olives) is harder to manage and treat, because its pollutant load is more concentrated [7]. Nevertheless, some quantity of OMW is drained from the TPOMW in the storage tanks and is collected using draining valves. Besides that, it is possible to extract some oil from TPOMW through several techniques (e.g., using solvents), valorizing, in this way, this by-product [7]. In this way, the two-phase system, besides presenting the same advantages of the three-phase system, also produces a lower quantity of OMW (about 8.5-11 L/100 kg olives) and reduces water consumption; thus, this centrifugation system has been called "ecological" [3,13]. Nevertheless, like the three-phase system, the two-phase system needs expensive installations. Figure 4 illustrates a scheme quantifying the OMW production (for all the extraction processes used in the countries of the Mediterranean area) in relation to the quantities of olives and extractive water used, and olive oil/olive husk (pomace) produced; one can see that big ratios between generated OMW and olive oil are obtained [14]. Moreover, it is estimated that 30 Mm 3 of OMW per year are produced around the world [6]. The OMW results from both the water present in the olives and the water used in the production process of the olive oil, especially with the three-phase centrifugation system and pressing method [1,6]. A quantitative relationship between the OMW production and the remaining components present in the olive oil extraction. * The olive husk mass presented in the two-phase centrifugation process does not include the mass lost in secondary drying processes. Data taken from Caputo et al. [14].
These big amounts of OMW cause large environmental problems [15][16][17][18]. It is reported that such residues have an effect on soil/air quality and on the aquatic ecosystems due to the deposition of toxic streams directly in the receiving bodies. Some of the main negative impacts are the increase of the organic charge concentration and the existence of phytotoxicity in the water systems. This happens due to the high values of COD (60-200 g·L −1 ), total organic carbon (TOC-> 9 g·L −1 ), and biochemical oxygen demand (BOD-14-100 g·L −1 ) present in the OMW [6]. Pollution from OWM is expected to be 200-400 times higher than from urban wastewater [1,19,20]. The uncontrolled disposal of OMW represents economic, social, and environmental problems that must be resolved.

OMW Composition
The composition of OMW is extremely variable and suggests that minor or major compounds can be present, depending on numerous factors, which include the age of the olive tree, region of cultivation, maturation level of the olive, weather conditions that the olive was subjected to in the ripening process, treatment of the tree, and method of extracting the oil [21]. El-Abbassi et al. [22] showed that OMW resulting from the traditional method had a higher phenolic content compared to that obtained from the three-phase centrifugal system. The investigation of these effects in the OMW composition is very important, in order to develop sustainable OMW management strategies.

OMW Treatment/Valorization
As the substances present in OMW are very toxic, the effluent needs to be correctly treated. A great variety of treatment/valorization processes (see Figure 5) have been investigated in the past years to reduce OMW pollutant load or to allow its disposal, but research efforts turn nowadays toward other aspects. Still, those methods of OMW load reduction are not often used due to technical and economic reasons. The trivial management of the OMW consists of the direct discharge into sewer systems and rivers, disposal in evaporation lagoons or dilution with other effluents [4,39]. In addition, the latter practice requires large areas and produces a black foul-smelling sludge that is difficult to handle, with inherent pollutant infiltration into ground water and insect proliferation. Land disposal is also a common practice considering OMW as a fertilizer, if it is well managed. Nevertheless, to avoid the contamination of the soil, land disposal must be quantitatively controlled by releasing small amounts of wastewater. Thus, nowadays, an organized and controlled release of OMW into the environment is still permitted, according to specific regulations of the different countries.
Therefore, solutions for adding value to OMW or for extracting valuable compounds from it before treatment/discharge are required. The OMW valorization/treatment methods can be organized into several general categories (see Figure 5). Among all the existing sub-categories, the following stand out: • Thermal Processes [4]: • Distillation and Evaporation Processes: these processes concentrate the inorganic and organic contents of the OMW by water evaporation. These processes have high operation costs and the organic content must be treated afterwards; • Combustion: this process is very efficient but requires a large amount of energy.
Most of the methods studied aiming OMW valorization include the recovery of the polyphenols fraction owing to their interesting pharmacological properties [78][79][80][81]; besides that, OMW phenolic extracts can be used in the food industry, in order to improve the oxidative status of the products [77]. For instance, in recent years, membrane processes have been used for polyphenols recovery [26]-namely microfiltration, ultrafiltration, reverse osmosis, nanofiltration, membrane distillation, and osmotic distillation [4,40,41]. However, membrane-based processes have a range of limitations; e.g., may be limited by the OMW's high concentration of suspended solids. Adsorption processes [48,50] have been also applied to the recovery of phenolic compounds to valorize the OMW effluent.
Concerning the treatment processes, the most commonly utilized physicochemical treatments with this purpose include electrochemical oxidation [44,45], ozonation [46,47], electro-coagulation [66], adsorption [49], and advanced oxidation processes [43,82,83] (particularly Fenton oxidation [60,61]). Since the polyphenols have strong resistance to bacteria, the physicochemical methods can be extremely important, because they can reduce the organic content to make the effluent suitable for biological processes (two different techniques combined for OMW treatment/valorization). In general terms, several conventional valorization/treatment technologies with varying degrees of success have been proposed. However, most of them require considerable financial input and show low efficiency. In addition, they frequently focus on removing the polyphenols and end up producing a secondary waste that needs additional treatment.
Therefore, it is necessary to improve the scientific and technological solutions for the treatment and valorization of these wastes to decrease the environmental impacts and to allow sustainable use of resources, in line with international waste management practices. In this perspective, the possibility of waste valorization for green energy production has been considered [84], namely through OMW steam reforming (OMWSR) [85]. This technology would permit removing the high pollutant components present in the OMW while producing H 2 , therefore providing added-value. This biofuel is environmentally attractive once it is renewable and reduces the CO 2 emission into the atmosphere [84]. Unlike fossil fuels, H 2 burns cleanly, without producing any environmental pollutants. The OMWSR would help reducing the pollution load of the effluent as well as to economically valorize a by-product of the olive oil industry without any value so far [85]. The H 2 can be applied in several technologies like in combustion engines or electrochemical cells. In this perspective, it is worth mentioning that hydrogen technologies proved extremely resilient during the COVID-19 pandemic, with their momentum staying strong in 2020 [86]. Moreover, a recent publication of the International Energy Agency reported that the global H 2 demand in the Net Zero Emission Scenario (between 2020 and 2030) will increase significantly in the next years, in order to reach international climate and sustainable energy goals-see Figure 6 [86,87]. To meet this growing demand for H 2 , the environmentally friendly production of H 2 has become an important part of the H 2 market. However, this progress falls well short of what is needed in the Net Zero Emissions Scenario.

Olive Mill Wastewater Steam Reforming
The OMWSR (Equation (1)) consists of the sum of two independent reactions: decomposition of high molecular weight molecules in the presence of steam (Equation (2)) and the water-gas shift (WGS-Equation (3)) reaction. The values of ∆H 25 • C r were taken from the literature [84].
The reason why the OMWSR was chosen as a method of OMW valorization is mainly due to the fact that the steam reforming is extensively used in industry, apart from the reasons given above (i.e., environmental concern and production of green H 2 ). Besides that, syngas (H 2 and CO) produced can be applied as a resource for the fabrication of valuable chemicals and fuels through Fischer-Tropsch or dimethyl ether synthesis routes [88].
The OMWSR process (in a Traditional Reactor-TR-see Figure 7) has, however, some challenges that need to be overcome and optimized to be a valid approach for the OMW valorization, namely:

•
The OMWSR is an endothermic reaction, thus needing high temperatures and intrinsically high operating costs; moreover, producing H 2 by this process also yields CH 4 , CO 2 , and CO, whose production is a target of environmental alarm and limited by legislation; • The WGS reaction is thermodynamically limited (Equation Other products apart from the previous ones-such as butane, propane, acetone, methanol, benzene, ethanol, cyclopentadiene, ethylene, butanol, and acetaldehydecould be present in the reaction medium. These species are referred by many works as possible secondary products of some reactions (e.g., Equation (7)-cracking of oxygenates) or as intermediates of the reactions involved in this process [1,27,28]; • The formation of coke is a problem since it deactivates the catalysts (Equations (7) and (8)-cracking of hydrocarbons, Equation (9)-Boudouard reaction, Equation (10)methane cracking, Equation (11)-carbon monoxide reduction, and Equation (12)carbon dioxide reduction), thus affecting H 2 yield and purity and long-term operation of the process; With the aim of resolving these difficulties, new catalyst formulations must be developed. Furthermore, new reactor configurations merging the OMWSR catalytic process and H 2 and/or CO 2 selective removal should be addressed, due to their potential to solve the restrictions mentioned above (namely shifting in the desired direction the reversible reactions, like the WGS). In this work, some of the important progress regarding the OMWSR process (developed catalysts) and its combination with a H 2 perm-selective membrane (in a membrane reactor-MR), in the perspective of process intensification, are reviewed; still, the potential of the utilization of the sorption-enhanced reactor (SER with a CO 2 -selective sorbent) and the sorption-enhanced membrane reactor (SEMR-combining in the same device the separation of H 2 and capture of CO 2 ) are also addressed.

OMWSR Catalysts
Heterogeneous catalysis applied to the OMWSR is not very developed until now. The catalysts investigated so far have been Ni-, Pd-, Pt-, and Ru-based. However, several catalysts were extensively studied for the steam reforming of individual components that are similar to the components present in the OMWSR (like acetic acid, or similar molecules). Currently, the key goal is to prepare catalysts with high catalytic activity and high stability.

Catalysts Used for Steam Reforming of OMW
Firstly, this reaction was studied by Tosti et al. [1,6,89] and by Llorca et al. [28]. Recently, Alique et al. [89] also studied the steam reforming of the OMW in an MR, using a catalyst developed and tested in a previous work of Tosti et al. [89].
In the first study, Tosti et al. [6] used a commercial Pt-based catalyst (1 wt.% Pt on Al 2 O 3 -BASF SP-01 T) in an MR (150 µm thick Pd-Ag dense tubes). The experimental tests were carried out at 450 • C and in the range of 1-5 bar, at different space velocities (Weight Hourly Space Velocity-WHSV = 2.78 × 10 −3 -8.33 × 10 −3 mol·h −1 ·g cat −1 . The gas chromatographic analysis of the retentate side revealed the existence of H 2 , CO 2 , and CH 4 , thus showing the presence of side reactions such as the hydrogenolysis and cracking of high weight molecules. Additionally, catalyst deactivation due to coke formation was detected. Then, this research group prepared a new noble metals-based catalyst [89] and compared it with the commercial Pt-based catalyst used in the previous work under similar operation conditions (using the same reactor configuration-MR). The new catalyst consists of Al 2 O 3 pellets coated with a CeO 2 -ZrO 2 layer. The active species (Pd, Rh, and Pt) loading was lower than 3 wt.%. CeO 2 -ZrO 2 presents several benefits in the steam reforming reactions: oxidation reactions and WGS reaction are enhanced by the high oxygen mobility, which is also related to the decrease of the coke formation and sintering inhibition. Moreover, when these supports are employed, Rh can exhibit high catalytic stability during the reaction, low Pt loadings (<1 %) are enough for limiting deactivation phenomena and Pd can show good stability even at very low steam-to-carbon feed ratios (SCFRs). This catalyst exhibited no substantial loss of activity after 6-8 h of operation, in contrast with the commercial Pt-based catalyst whose activity reduced after 1-2 h of experimental test. When compared to the commercial material, the new catalyst exhibited higher selectivity, allowing it to significantly reduce the production of CH 4 and coke; in this way, it is obtained a higher H 2 production and higher durability and stability. This new catalyst was also used in the recent work of Alique et al. [90], who reported that, in a MR, higher pressures increased the permeation of H 2 , prevailing the membrane shift effect against the negative effect on thermodynamics of the reactions involved in the steam reforming process. In this way, an increase in H 2 production was obtained as the pressure increased.
Llorca and co-workers are another research group that studied this reaction [28] in the last few years in a TR. Several catalytic honeycombs loaded with Ni or noble metals (Ru, Pt, Rh, or Pd) over La-stabilized CeO 2 were prepared. The experiments were carried out at 1 bar and in the range of 600-750 • C, at different space velocities (Gas Hourly Space Velocity-GHSV = 4500-16,000 h −1 ). Their catalytic performance in terms of production of H 2 , stability and selectivity towards reforming products followed this trend: Pt/CeLa > Rh/CeLa > Ru/CeLa >> Ni/CeLa. The trend of coke formation on the surface of these catalysts was: Ru/CeLa > Rh/CeLa > Pt/CeLa, which is contrary to that of the catalytic performance, strongly suggesting that coke deposition was the cause of catalyst deactivation during OMWSR-see Figure 8. Besides, it was found that the selectivity for secondary reactions increases with time, for all the catalysts, and the catalyst with higher selectivity for H 2 formation was Rh/CeLa. In addition, Ca deposition was also observed after the stability tests in all samples; seemingly, calcium is present in the OMW and its possible influence on the long-term catalytic activity must be further studied. The 1 wt.% Pt/CeLa catalyst showed a very stable distribution of products in long term runs. An increase of the Pt load up to 2 wt.% had a positive effect on the product distribution, particularly on the suppression of CH 4 . Reprinted with permission from Ref. [28]. Copyright 2021 Elsevier.
Finally, Rocha et al. [85,91] tested several commercial and prepared catalysts (Cu-Zn-, Ni-, and noble metal-based) for the OMWSR process (using a synthetic OMW effluent-a complex mixture of model oxygenates, representative of a real effluent) to compare their catalytic performances. A catalytic screening study with all the materials was performed (@ 1 bar and 350/400 • C), and stability tests (@ 1 bar and 400 • C) were conducted with the most promising samples. Although there are a few materials with good catalytic performance, the commercial Rh-based catalyst stood out, exhibiting high performance (catalytic activity and stability): the H 2 yield (over 9 mol H2 ·mol −1 OMW ) and total TOC conversion (>98%) were high during the long-term experiment (24 h of the stability test)-see Figure 9. Besides that, several studies regarding the steam reforming of guaiacol (component typically present in the composition of the OMW) using Ni-based catalysts were also reported [92,93]. Finally, it is important to emphasize that several works have already studied the steam reforming of effluents (e.g., bio-oil and palm oil mill) with compositions very similar to OMW (oxygenated species), using Ni and Co-based catalysts [94][95][96][97][98][99][100][101][102][103].
A summary of some results obtained with the more relevant catalysts reviewed in this section is presented in Table 3.
where F OS is the molar flow rate of organic species present in the OMW fed to the reactor.
New catalysts must be developed for future catalytic screening and improvement. These catalysts will need to maximize the reaction conversion, decrease the coke formation, and increase the H 2 selectivity under milder operating conditions.

Catalysts Used for Steam Reforming of Model Molecules of OMW
Among the materials above-mentioned, numerous catalysts (Ni-, Pt-, Ru-, Co-, and Rh-based) have been extensively studied for the steam reforming of individual components similar to the species present in OMW. Thus, in this work, three different molecules were defined as OMW model compounds: acetic acid, phenol, and toluene. The catalytic performances of the catalysts used for the steam reforming of these species are affected by the active phase, support, and promoter agents. These topics are discussed in the following sections.

Effect of the Active Phase
The active phase is the metal element of the catalyst that catalyzes the reaction. Numerous catalysts (Ni-, Pt-, Ru-, Co-, and Rh-based) have been extensively studied for the steam reforming of oxygenates.
The Ni-based catalysts are the most used in the steam reforming processes because of their effectiveness (they have the potential to be highly active and stable), being commercially available and relatively cheap [104][105][106][107][108]. Nevertheless, the Ni loading, support, promoter(s), and synthesis method used in the catalyst preparation have a crucial impact on both catalytic activity and price [109]. In the study of Zhang et al. [110], it was concluded that the increase of Ni loading from 10 wt.% did not significantly increase the activity of the catalysts, but enhanced the stability, and especially the resistivity towards coking in the steam reforming of acetic acid (@ 1 bar, Liquid Hourly Space Velocity-LHSV = 4.2 h −1 , SCFR = 5 and 300-600 • C). The coke formed over the catalyst with the lower Ni loading was more amorphous and the coke formed over higher Ni loading was more fibrous [110]. It was verified by Borges et al. [111] that the higher the Ni loading is, the lower the temperature and the time necessary for the catalyst reduction. The catalyst with a molar ratio of Ni/Co = 3 exhibited the highest activity and the best carbon deposition resistance due to the high density of basic sites for the steam reforming of phenol. In a study by Yu et al. [112], the effect of the type of anion in the Ni precursors in the catalytic behavior of Ni/Al 2 O 3 catalysts in steam reforming of acetic acid was studied. It was found that the NO 3 − and CH 3 COO − as the anions of the Ni precursors did not cause negative effects on the catalytic performance of the catalyst.
Güell et al. [113] reported that the Pt catalyst was active for the steam reforming of oxygenates compounds, as shown in the study of de Castro et al. [114]. Pant et al. [115] reported that Ni-Co, Ni-Co/CeO 2 -ZrO 2 , and Ni/La 2 O 3 -Al 2 O 3 catalysts can catalyze the steam reforming reaction of acetic acid. Among the three catalysts investigated, the Ni-Co catalyst was found to be more effective for this reaction. Several authors [116][117][118] found that Rh-based catalysts have a high performance in the steam reforming of oxygenates and several of these materials exhibit superior performance when compared to a commercial Nibased sample. A series of Ni-Cu bimetallic catalysts supported on sepiolite (Ni x -Cu y /SEP) were prepared by Liang et al. [119] for the steam reforming of phenol. The results show that the Ni-Cu alloys were successfully produced in bimetallic catalysts, and the addition of Cu decreased the size of the Ni particles, and improved the redox ability and metal dispersion of bimetallic materials (in this case, Cu can also be considered as a promoter of the catalyst).
The Fe 2 O 3 /SiO 2 −Al 2 O 3 catalyst was used by Adnan et al. [120] for the steam reforming of toluene. The results indicate that Fe 2 O 3 /SiO 2 −Al 2 O 3 is a promising catalyst (in terms of activity and stability, @ 500 • C, 3.8 bar and SCFR = 0.14) since it has a large quantity of acidic active sites. However, the study of Wang et al. [121] (steam reforming of acetic acid) reported that, in the Co-Fe catalysts, the more active component is Co rather than Fe.
Furthermore, several authors [122][123][124] show that multi-metallic catalysts show high activity and improve the H 2 yield due to synergistic effects of the multiple components of the active phase.
Finally, mixed oxides such as hydrotalcite-type oxides have been reported as promising catalyst precursors for the steam reforming of these types of oxygenates [111,125].

Effect of the Support
In several studies, it was verified that the support has a significant effect on the performance of the catalysts. Different catalyst supports significantly influenced the reforming reaction.
A series of Ni core-shell catalysts with various shell species (SiO 2 , Al 2 O 3 , CeO 2 , and TiO 2 ) were prepared by Pu et al. [126]. By comparing the catalytic activities of the catalysts with various shell materials, it was concluded that the Ni/Al 2 O 3 catalyst was the most suitable material for the steam reforming of acetic acid, showing much higher catalytic activity than the other catalysts. It was also observed in the work of Li et al. [127] that Al 2 O 3 support has a high specific area that can enhance the catalytic activity, but the acidity of Al 2 O 3 promotes the carbon deposition, reducing the stability of the materials.
The nature of the support was also studied by Zhang et al. [128], using attapulgite (ATTP) and Al 2 O 3 as support of Ni-based catalysts. It was verified that ATTP has a lower specific surface area and lower thermal stability than Al 2 O 3 . The stability of the Ni/Al 2 O 3 catalyst was higher than that of the Ni/ATTP catalyst, not only due to the higher surface area of Al 2 O 3 , but also due to the type of coke formed on the surface of the Ni/Al 2 O 3 catalysts (fibrous) instead of the amorphous coke formed on Ni/ATTP catalyst.
Chen et al. [129] studied the effect of biochar as support in Ni-based catalysts for the steam reforming of acetic acid. The results of the catalyst characterization showed that the porosity of biochar was increased significantly after activation, increasing the surface area of the materials and the dispersion of the Ni particles on the catalyst's surface. Besides that, in the work of Wang et al. [130]-study of the steam reforming of acetic acid-it was found that the treatment of biochar support with HNO 3 enhanced the quantity and the strength of the acidic active sites as well as the catalytic activity of the Ni/biochar catalysts.
Additionally, Yang et al. [131] using a NiO/MgO catalyst have concluded that the interaction between the support and active phase can prevent the material from sintering.
Finally, silicate structured mesoporous materials were reported to be less vulnerable to deactivation due to carbon formation than the traditional microporous catalyst supports. They also cause less resistance to the diffusion of reactants to the active sites [88].

Effect of Promoters
In several studies of the steam reforming of oxygenates, the catalysts were doped with promoters. It was reported that Cr inhibited the formation of Ni 3 C and can modify the metal sites forming alloys with Ni [132,133], La 2 O 3 led to a decrease in coke formation [105,[132][133][134][135][136], K and MgO improved the stability [105,117,132,133,[136][137][138][139], and TiO 2 decreased the conversion [132]. It was also observed that the addition of CeO 2 improved the activity and stability of the catalyst [106,108,114,135,[140][141][142].
The work of Li et al. [143] reported that the presence of CeO 2 in the catalyst increased the support-metal interaction, restricting the growth of metal particles. Besides that, it was found that the oxygen vacancies of the CeO 2 are responsible for the activation of the water to generate OHgroups, which can react with intermediate products to produce CO 2 and H 2 . In the study of Iiada et al. [118], Ru/SrCO 3 -Al 2 O 3 catalysts with a low Ru loading exhibited higher catalytic activity for steam reforming of toluene compared to a commercial Ru/Al 2 O 3 catalyst (@ 600 • C and SCFR = 2). In this way, it was possible to see that the SrCO 3 increases the performance of the catalyst. Additionally, in the study of Zhang et al. [144], it was found that the addition of KOH to Ni/Al 2 O 3 with the lowest Ni loading significantly enhances the catalytic activity and promotes gasification of the reactive intermediates such as methyl group, carbonyl group, etc.
Choi et al. [145] studied the steam reforming of acetic acid by using catalysts modified by MgO, LaO 3 , Cu, and KOH. Ni/Al 2 O 3 modified with Mg showed the best performance at low temperatures. It was possible to verify that a catalyst with a large quantity of weak basic sites and few middle/strong basic sites is necessary to improve the catalytic performance and minimize the coke formation. Finally, it was observed that the toluene conversion and H 2 yields increased with the addition of alkaline-earth metals [146]. In a general way, the addition of basic oxides (e.g., MgO, CeO 2 , La 2 O 3 , etc.) enhanced the coking resistance, thus improving the catalytic performance. These oxides aim to enhance the redox properties of the catalyst [143,[147][148][149][150], which increases the oxidation of carbon deposits. For instance, Kumar et al. [151] reported that the coke formation was also significantly suppressed by the γ- Finally, Yu et al. [152] reported that the addition of Mn enhanced the specific area of the catalyst, by the formation of more pores and the addition of Fe, Mn, La, and Ce could enhance the dispersion of Ni particles in the support. Besides that, Co, K, La, Na, or Ce enhance the activity of Ni-based catalysts and K or Na addition enhanced the reduction of Ni oxides.
However, Galdamez et al. [153] concluded, in the steam reforming of bio-oil, that the addition of La 2 O 3 in the Ni catalyst does not increase the H 2 yield. It was also found that ZrO 2 had a negligible effect on the performance of the catalyst [138].
A summary of the more relevant catalysts reviewed in this section for the steam reforming of individual components (or similar) of the OMW and their best catalytic performances are presented in Table 4. It is possible to verify that Ni-based catalysts are effective for these reactions (they exhibit similar catalytic performance compared to noble metal-based catalysts) and are relatively cheaper. Therefore, more detailed information about the catalytic stability and preparation method of these catalysts is reported in Table 5. It is possible to see that several Ni-based catalysts present no deactivation until the end of the experimental test, even though noble metal-based catalysts are less susceptible to coke formation [94,154,155]. Furthermore, the most used preparation method of these catalysts is the impregnation of the Ni on the support; however, there are only a few studies about the effect of the preparation method in the catalytic behavior of the materials for the steam reforming of long-chain oxygenates.

Summary
Until now, only a few catalysts have been developed for the OMWSR process: Ni-, Pd-, Pt-, Rh-, and Ru-based. However, several catalysts were studied for the steam reforming of components that are similar to species present in the OMW.
It is possible to conclude that Ni-based catalysts are promising for the steam reforming of oxygenated species. However, almost all these catalysts showed conversions close to 100% and high H 2 yields and selectivities, making it difficult to choose the best one. These results show how attractive Ni-based catalysts are for the OMWSR process, in comparison with the noble metal-based catalysts, due to their lower costs.
Almost all Ni-based catalysts were prepared by impregnation, appearing to be the best method for the preparation of these materials. It is still necessary to study the effect of the preparation method on the catalytic behavior of the materials.
A bi-metallic catalyst with Ni and one noble metal (e.g., Ru) as active phases, supported in an adequate material (e.g., Al 2 O 3 or SiO 2 ) and doped with a promotor (e.g., CeO 2 ) is a potential catalyst to achieve high catalytic performance (H 2 yield and OMW conversion) for OMWSR during long reaction times.
In future studies, it is necessary to rigorously assess the costs associated with the preparation of the different catalysts (with different preparation methods, reagents, active phases, etc.), to better benchmark the numerous catalytic formulations available, i.e., make a selection (for industrial applications) based on both techno and economic criteria.

Multifunctional Reactors
In the last few years, different reactor configurations for improving the catalytic steam reforming processes have been investigated and debated, since this process involves reversible reactions, and thus equilibrium limitations. Thus, the utilization of multifunctional reactors in the OMWSR process can become a very promising approach [160]. Even so, and despite the significant number of studies addressing such devices for other applications, in future works more attention should be paid to detailed techno-economic analyses to assess the economic feasibility of this process, as well as the difference, in financial terms, resulting from the application of different reactor configurations (traditional vs. multifunctional reactors). To the authors' knowledge, this type of study has not yet been developed for the OMWSR process.

Sorption-Enhanced Reactor
One of those alternatives for the process intensification of the OMWSR is the utilization of a SER (see Figure 10), which consists of combining the TR and CO 2 sorption in the same reactor. This innovative configuration allows shifting the thermodynamic equilibrium of the steam reforming reaction by removing one of the products (in this case CO 2 ) from the reaction medium-Le Chatelier principle [160,161]. This reactor configuration was already used in multiple applications [109, in the last decade. The SER can shift the equilibrium of the reversible reactions to higher conversions of the reagents. In this way, this reactor configuration will enhance the production of H 2 , since the removal of CO 2 shifts the WGS reaction (Equation (3)) to the forward direction and inhibits CH 4 and coke production (Equations (5) and (12)). However, the drawback of this technology is that after some time the sorbent gets saturated with CO 2 and then regeneration is required. To make it possible to run continuously, the system must have at least two parallel devices, so that while one of the reactors is utilized as SER (reaction stage), the other is being regenerated. It is possible to distinguish three regimes along time: pre-breakthrough, breakthrough, and post-breakthrough. In the pre-breakthrough region, the CO 2 is being removed, thus enhancing H 2 production. In the breakthrough region, the H 2 concentration at the reactor outlet begins decreasing. In the pos-breakthrough region, the sorbent is already saturated and so the improvement of the H 2 production due to CO 2 capture disappears- Figure 11.
The effect of the capture of CO 2 from the reaction medium was already discussed for the OMWSR from the thermodynamic point of view [194], though this effect has already been analyzed for other similar reactions [84,195,196]. In a previous work of Rocha et al. [197], a comparison of the performance between a SER and a TR was already performed for this process. This reactor configuration intensifies the performance of the process by combining the reaction and CO 2 capture in a single device. In this way, a SER allows the production of more high-purity products at the outlet (in this case, pure H 2 before the breakthrough time).
In this reactor configuration, the decrease of the CO 2 emissions also presents a very interesting environmental advantage. Since the industrial revolution, the global CO 2 concentration increased from about 280 ppm to 415 ppm in 150 years. The higher concentration of pollutants in the atmosphere increases the greenhouse effect with a huge environmental impact. To avoid irreversible changes to the global climate, the United Nations in the COP26 have agreed to establish goals to reduce the CO 2 emissions to limit its concentration in the atmosphere [198]. Therefore, this reactor configuration is also in line with this premise, in order to decrease the environmental impact of CO 2 .

CO 2 Sorbents
Sorption-enhanced technologies require the mixture of a catalyst, that is stable and with high catalytic activity in the temperature range considered (in this case 300-500 • C, or even higher), and a solid sorbent, in this case with high CO 2 sorption capacity and with adequate sorption-desorption kinetics.
Nevertheless, some research was performed with other sorbents to compete with the HTCs, which is shortly described herein below.
Akgsornpeak et al. [245] avoided the sintering phenomena by preparing CaO materials through the sol-gel synthesis with cetyltrimethyl ammonium bromide. Besides that, the incorporation of Y 2 O 3 has been lately stated to significantly enhance the carbonation (sorption) rate and adsorption capacity of CaO-based materials [206]. The addition of optimized quantities of K to Li 2 ZrO 3 sorbents was reported to enhance their CO 2 sorption capacity, sorption rate, and stability in sorption-desorption cycles [226]. Seggiani et al. [226] examined the performance of several doped-Li 4 SO 4 samples, having noticed that the Kdoped and Na-doped sorbents presented the highest sorption capacities and sorption rates. In terms of stability, only the K-doped sorbent was capable of maintaining its properties after 25 sorption/desorption cyclic runs. However, Iwand et al. [240] concluded that the traditional preparation method of K-doped materials results in sorbents with slow regeneration step.
The CO 2 sorption capacities and the sorption and regeneration temperatures of several materials are summarized in Table 6.
HTC is a layered double hydroxide. This natural or synthetic lamellar hydroxide has two kinds of metallic cations (divalent and trivalent) in the main layers and interlayer space containing anionic species (compensating/interlayer anions)- Figure 12 [248,249]. These materials are used in a wide range of applications as catalysts, precursors, and sorbents [214]. Several thousands of tons of HTCs are produced annually by different chemical companies-such as BASF, SASOL, Clariant, Kisuma Chemicals, and Sakai Chemical [250].
The general formula is M 2+ 1 , where M 2+ is a divalent metal cation and M 3+ is a trivalent metal cation. A nis a charge balancing anion (usually CO 3 2− ) located in the interlayer volume and x is generally between 0.2 and 0.4, while y is the number of moles of water located in the interlayer space [209,210]. Similar layered structures of HTCs could be produced with several interlayer anions [222,224,246,247,[250][251][252][253][254][255][256][257].  (a) Sorption capacity of the fresh sorbent. (b) Not specified (or regeneration step not performed). Co-precipitation is the most used method for the preparation of this type of material. It is based on the slow addition of a mixed solution of divalent and trivalent metals salts in adequate proportions into a reactor. An alkaline solution (e.g., NaOH) is added to the reactor in order to maintain the pH at a selected value, leading to the co-precipitation of the two metallic salts [248]. Therefore, there is often competition between some anionic species, and the tuning of the synthesis experimental conditions can lead to the selective intercalation of one of them as the primary interlamellar anion. Secondary interlamellar anions can replace them with further treatments such as an anionic exchange. In addition, it is not clear that a pure HTC phase is always obtained: for instance, although the overall composition of a certain precipitate corresponds to the estimated value, secondary phases such as hydroxides or basic salts of the divalent or trivalent metal can be present in the material [248]. HTCs can, however, be synthesized by various techniques depending on the specific requirement and properties of the compounds. Other methods include combustion, recrystallization, urea hydrolysis, ion exchange, reverse-microemulsion, solgel, and hydrothermal [209,[258][259][260][261][262].
Fresh HTCs as such are not excellent CO 2 sorbents due to poor basic properties and the presence of species that hinder CO 2 adsorption. Therefore, they are subjected to thermal treatment (calcination) to obtain a nearly amorphous mixed solid solution-with a poorly defined 3D network-with more favorable and stable properties for CO 2 sorption [248]. HTCs undergo different stages of transition during the calcination. These transitions depend on many aspects such as the nature and relative quantities of cations, type of anions, method of synthesis, and crystallinity. Several authors have methodically investigated the thermal decomposition process [214,261,263,265,269]. The optimum temperature of calcination is usually 400 • C, since it allows the decomposition of the interlayer anions (and then, the active sites can be occupied by the CO 2 ) and the formation of the amorphous phase (mixed oxides) with sorption properties, without the complete and irreversible destruction of the structure. Apparently, such optimum temperature results from the balance between surface area and the number of basic active sites [214,263]. In the calcination program, the first weight loss is detected between 70 • C and 190 • C, attributed to the loss of interlayer water; then, in the temperature range of 190 • C to 280 • C, the OH − group bonded with Al 3+ disappears [270]. The OH − group bonded with Mg 2+ disappears from the structure of the material in the temperature range 280 • C to 405 • C and the interlayer anions are usually released between 405 • C and 580 • C [270]; in this way, the material becomes an amorphous mixed oxide solution, which is reversible if the calcination temperature is not higher than 550-600 • C [270,271]-see Figure 13. The use of very high calcination temperatures leads to the development of spinel phases (e.g., MgAl 2 O 4 ), which should be prevented in order to conserve the platelet morphology of HTCs and an acceptable basicity [244,246]. The original structures of the calcined samples can be recovered again by exposure to steam and atmospheric CO 2 (or through anion intercalation)-"memory effect" [264,[272][273][274]. Different optimum Mg 2+ /Al 3+ ratios have been reported in the literature, which varies from 1:1 to 3:1-the optimum ratio depends on the type of interlayer anion used, preparation method, etc. [212,276,277]. Although high Mg 2+ content allows the formation of a higher number of basic active sites in the material, at higher Mg 2+ /Al 3+ ratios MgCO 3 may also form at high partial pressures of water (steam) and CO 2 , which can lead to mechanical stability problems [278,279]. In the work of Yong et al. [212], the authors studied the effect of different Mg 2+ /Al 3+ ratios (0.5, 1.3, and 3.0) on the sorption capacity of three commercial HTCs from SASOL. It was verified that the HTC with Mg 2+ /Al 3+ ratio of 1.3 had the highest performance. However, there is a lowest limit of the Mg 2+ /Al 3+ ratio (=2.0) to obtain a crystallographically pure HTC, since this ratio influences the electrostatic repulsion between neighboring trivalent metals in the layers [248,280]. The optimum Mg 2+ /Al 3+ molar ratio reported in several studies is often equal to 2 [209,210,212,255] for the sorbents impregnated with potassium.
As previously mentioned in this section, the CO 2 sorption capacity of the HTCs can be enhanced by impregnating with an alkali metal such as Na, K, or Cs [281,282]. It was verified by several works that HTCs modified with alkali promoters like K 2 CO 3 (or Na/KNO 3 ) showed best CO 2 sorption performance than the base HTCs [209,210,218,243,[281][282][283][284][285][286]]see, for instance, Figure 14. Besides that, several authors [209,210,218,287] have also shown that the simultaneous impregnation with K and substitution of Al 3+ with Ga 3+ markedly improved the sorption capacity of HTCs, the first modification acting as a chemical promoter and the latter providing a superstructure, robust and stable at elevated temperatures. For K 2 CO 3 -modified HTCs with high Mg 2+ /Al 3+ ratios, the doping is mostly concentrated in the bulk phase and behaves as a reactant to form high stable K-Mg double carbonates after sorbed CO 2 . With the increase of the Al 3+ content, surface modifications occur and become the dominant improvement mechanism via the interaction between the K + ions and unsaturated oxygen sites [283]. It has also been found that the physicochemical properties of HTCs are highly influenced by the interlayer anions used. The anion influences the crystallinity and layer spacing in HTCs. In this way, materials with different interlayer anions present different morphologies and chemical surfaces, influencing the sorption performance of the HTCs-derived mixed oxides. For instance, Yong and Rodrigues [288] reported that the sorption capacity of HTC containing Fe(CN) 6 4was higher(1.3 mmol·g −1 ) than the one containing NO 3 − (0.3 mmol·g −1 ) at 25 • C and 1 bar of CO 2 . However, Wang et al. [222] analyzed numerous charge interlayer anions (SO 4 2− , Cl − , CO 3 2− , HCO 3 − , and NO 3 − ) having noticed that when using CO 3 2− , a spheroidal "sand rose" type of HTC with a very high surface area was produced. On the other hand, the other compensating anions led to the formation of "stone" type HTCs with very low surface areas. For this reason, the HTC with CO 3 2− as interlayer anion presented the highest CO 2 capture capacity. However, larger size anions tend to increase the interlayer spacing; this space can help the K ions to disperse into the HTC interlayer and, thus, create more Al(Mg)−O−K bonds to enhance the performance of the sorbent [247].
Several works have shown that the existence of steam in the feed gas increases the CO 2 sorption capacity of potassium doped-HTCs by 10-17 wt.% and enhances the stability of these materials (in the temperature range of 200-400 • C and in the pressure range of 1-3.5 bar) [193,216,218,221,279,[289][290][291]. It was observed that at least four different adsorption sites take part in the sorption/desorption of CO 2 and H 2 O and the presence of steam activates some active sites for the CO 2 sorption. Two adsorption sites can be restored with N 2 (or other inert gas), whereas the other adsorption sites require the presence of H 2 O to be activated/regenerated [279,289,290]. These features are particularly important for the application envisaged herein, once the OMWSR occurs in the presence of large excess of water.
The CO 2 sorption kinetics in dry conditions for these materials was studied by Miguel et al. [210] and Silva et al. [209] and it was noticed that the kinetics is dependent on the morphology/composition of the material, with the physical sorption being faster than the chemical one.
The conclusion on which type of CO 2 sorbent is more suitable to be used in the SER strongly depends on the operating temperature at which it will be used. Since one of the main goals is to reduce the temperature of the steam reformer to reduce the operation costs, it is attractive to work at relatively low temperatures (300-500 • C). For such temperatures, HTCs have been reported to present higher CO 2 sorption capacities and faster sorptiondesorption kinetics, as well as easier regeneration, and lower loss of sorption capacity along consecutive cycles. HTCs are also better than other sorbents due to nearly infinite selectivity for CO 2 sorption over gases like CH 4 , CO, and N 2 [250]. Moreover, the fact that the CO 2 sorption capacity of HTCs is strongly improved under wet conditions (steam reforming process conditions) makes them an even better option for use in the OMWSR process using a SER (for instance, Lu et al. [207] reported that the sorption capacity of a CaO sorbent over the carbonation/decarbonation cycles decreased faster when steam was present in the system).

OMWSR in Sorption-Enhanced Reactors
As already mentioned, in a previous work of Rocha et al. [197], a comparison of the performance between different reactor configurations was performed for the OMWSR process. In a general way, it was verified that the conversion of the organic carbon (above 99%) remains almost complete along all the experimental tests when using a SER (@ 1 bar and 350/400 • C), and it was verified an improvement of the H 2 production during the pre-breakthrough of CO 2 as compared to the conventional reactor (TR)-see Figure 15.
In this work [197], the experimental tests with the SER configuration were separated into two different sets: a first set of experimental tests using a commercial sorbent, and a second set with a sorbent prepared in the laboratory. It was observed that, at both operating conditions and using the commercial sorbent (Figure 15), the H 2 yields obtained in the first cycle were higher than those reached for the first hour in the TR. Furthermore, it was noticed that the utilization of the commercial sorbent (that promotes the removal of CO 2 from the reaction medium) increased the stability of the catalyst in comparison with the performance of the TR-lower formation of coke.

Summary
The OMWSR through a SER consists of combining the reaction and CO 2 separation in the same device. This innovative reactor configuration requires, besides a catalyst, a solid sorbent with high CO 2 sorption capacity, which must be simple to regenerate and have good performance, particularly in the presence of steam. Several materials-such as CaO-based materials, lithium cuprate, lithium silicates, and lithium zirconates, among others-have been reported in the literature as high-temperature CO 2 sorbents. However, it is possible to conclude that HTCs have adequate properties to be used in a SER for OMWSR: practically constant and high CO 2 sorption capacity over the cycles at moderated temperatures, increased performance in a wet environment, adequate sorption-desorption kinetics, and great thermal and mechanical stability.
In this way, an HTC impregnated with potassium is a potential sorbent to be used to capture from the reaction medium the CO 2 produced in the OMWSR (sorption-enhanced process), increasing the purity of the H 2 and shifting the thermodynamic equilibrium to the production of more H 2 (increasing the H 2 yield).  [197]. Copyright 2021 Elsevier.

Membrane Reactor
Another alternative to enhance the OMWSR performance is the utilization of a MR (see Figure 16), which consists of combining the TR and H 2 separation in the same device. In such an MR, the membrane selectively removes one or more species, thus allowing to overcome the thermodynamic boundaries of equilibrium-limited reactions-Le Châtelier principle [160,-as is the case of the reversible WGS, which is, therefore, shifted forward. Regarding its advantages compared to TRs, they are the same as SERs, except for the reduction of the emissions of CO 2 in the pre-breakthrough zone, being that the capital costs reduction can be higher in the case of MRs since only one device is needed (instead of the parallel reactors required for the SER), as long as low-cost and long-lasting H 2 perm-selective membranes can be obtained. This reactor configuration was already used in other applications [173,[293][294][295][296][297][298][299][300][301][302][303][304] and for the OMWSR [1,6,89,90,197] in the last decade.
One of the advantages of the MR for this particular application is that it is possible to produce ultra-pure H 2 streams if membranes with a nearly infinite H 2 perm-selectivity are used (e.g., dense Pd-based membrane). Thus, a MR is an important technology to increase the energy performance of the process by combining the reaction and recovery stages in a single device (avoiding the construction of two unit operations), leading to considerably smaller dimensions of plant layout, producing one stream with highpurity H 2 (in the permeate side-see a scheme of a MR in Figure 16). The effect of the removal of H 2 from the reaction medium, from a thermodynamic point of view, was already studied for the OMWSR [194], though it was already analyzed for other applications [84,196,293,299,[305][306][307].
In order to use a MR in the OMWSR process, it is fundamental to select a suitable H 2 perm-selective membrane. Consequently, the next sections comprise a short overview of H 2 perm-selective membranes described in the literature, as well as an evaluation of the experimental tests performed in MRs for OMWSR. The performance of a MR depends considerably on the flux of permeated H 2 obtained and on its purity. The stability of a membrane is here considered as the capability of a membrane to maintain a constant flux of H 2 , as well as a constant perm-selectivity over time.

Hydrogen Perm-Selective Membranes
H 2 perm-selective membranes may be categorized as dense metallic membranes, proton-conducting dense ceramic membranes, dense polymeric membranes, microporous ceramic membranes, and porous carbon membranes [308,309]. The main benefit of dense metallic membranes is that typically they tolerate high temperatures (e.g., as compared to polymeric ones) and present high selectivity towards H 2 , thus being possible to obtain high purity H 2 -permeate streams [308]. Thus, and for the application envisaged herein, dense metallic membranes are preferable for ultra-pure H 2 production and will be the main target of focus. Nevertheless, dense ceramic membranes (proton conducting membranes) have also achieved high performance at high operation temperatures [310,311]. Although Pd-based membranes have been more used in the past for H 2 permeation, the advances in dense ceramic membranes in the last few years have been outstanding.
When selecting a membrane to be used in a MR, it is required to take into consideration the operating conditions under which the membrane will be applied. In the last years, there has been an incredible effort in order to develop appropriate membranes for MR applications. Pd-based membranes are presently the most promising for high purity H 2 production in MRs, especially Pd alloys (with transition metals-e.g., Pd-Ag, Pd-Ru, Pd-Cu, and Pd-Au), which are less sensitive than pure Pd to embrittlement (caused by the presence of H 2 at temperatures below 300 • C and H 2 pressures below 20 bar) and poisoning (caused by contact with CO, CO 2 , or H 2 S) [308,[312][313][314][315][316][317]. The embrittlement phenomenon may produce pinholes/cracks on the membrane, which will negatively affect the H 2 perm-selectivity of the membrane. In addition, one of the biggest inhibitory factors for H 2 permeation across a Pd membrane is concentration polarization and it is necessary to avoid that (this topic will be more deeply discussed later in this section) [318]. However, and in spite of these negative aspects, the ability of Pd to dissociate H 2 to its atomic form makes this metal ideal for H 2 separation applications [319]. Pd-Ag membranes show good stability, relatively lower material costs, and better mechanical properties than pure Pd membranes. Besides, it has been shown that the addition of a third element to the Pd-Ag alloy (e.g., Au) improves the tolerance to poisoning with H 2 S [320].
Several studies reported that the Pd alloys can obtain high permeances [321][322][323][324][325]. Moreover, their H 2 permeability increases with the percentage of Ag in the alloy up to a maximum, which occurs at around 23 wt.% Ag [326]. In previous studies, it was found that the maximum recommended operating temperature for long-term stable tests with Pd-Ag membranes was close to 500 • C [297,[327][328][329][330]. Above this temperature, besides the existence of leakages through the sealings, it is observed the formation of surface defects (small pinholes) that leads to the decrease of the membrane performance [327]. Tosti et al. [331] demonstrated that self-supported dense 23 wt.% Pd-Ag tubular membranes with fingertype configuration are extremely durable and reliable since these membranes allow to attain complete H 2 selectivity and no malfunctions were observed after one year of hydrogenation and thermal cycles. These properties, together with high permeability, make such materials very promising for ultra-pure H 2 production in energetic and industrial applications [331]. Nevertheless, several works reported that the H 2 removal fraction never reaches a value higher than 0.80 [305,326,332] for low-to-medium pressures. Besides that, the application of these dense membranes is still limited; for instance, self-supported Pd membranes are considered by several authors not being suitable in MR processes because of their large thickness required for reaching mechanical stability and corresponding low fluxes across the membrane and high costs. Due to these reasons, a strong effort has been put in the preparation of thinner Pd films over different supports and in the development of new membranes with less expensive metals. For instance, group V metals (such as V, Nb, and Ta) are promising alternative materials to Pd because of their lower costs and higher H 2 permeabilities. However, the main problem of these elements as membrane materials is their too high solubility to H 2 , making the membranes susceptible to severe H 2 embrittlement, which can lead to the formation of cracks [333]. Still, direct use of these materials as a membrane is hindered due to the formation of oxide layers and the occurrence of surface reactions, which promote the reduction of the H 2 permeability through the membrane. Thus, the main criteria for membrane selection are high permeability, high selectivity, resistance to reactive gases-particularly CO, CO 2 , CH 4 and H 2 O-and resistance to coking [334].
Nowadays, many researchers are pursuing the possibility of using supported membranes with as thin as possible Pd-alloy layers [335], thus reducing costs and increasing hydrogen permeation fluxes. Concerning these supported membranes, it is important to emphasize that metallic supports are mechanically stronger and can be more easily assembled into metallic reactors than ceramics supports (lower probability of leakages). However, the metallic supports have low porosity, their pores are larger, the surface is rough and, at this moment, they are much more costly than ceramic supports [333]. Several works [329,336] reported that the materials (namely, the support) that contact with the membrane must be meticulously chosen. For instance, Okazaki et al. [329] stated that the decrease of the H 2 permeation observed in their work was related to the interaction between the Al 2 O 3 support and the Pd layer of the membrane.
In terms of preparation methods of these H 2 -selective membranes, namely supported ones, chemical vapor deposition (CVD) and electroless plating (EP) and are the most used ones. The last method (EP) normally consists of the production of Pd particles through the reduction of a plating solution containing Pd-amine complexes. This method can provide high coating adhesion, requires low operation costs and allows simple operation. Though, the impurities present in the plating solution may lead to the presence of defects on the Pd layer. The CVD technique can very easily allow to deposit a metal film on a support with the utilization of several reactions with water [293].
Generally, the permeation of H 2 through a Pd-based membrane is described by a solution-diffusion mechanism composed by several steps, as shown in Figure 17.
The simplest model used to explain the mass transport of H 2 permeating through a dense membrane assumes that the H 2 diffuses through the membrane according to Fick's first law of diffusion for a pure component [338]. Still, in the permeation of H 2 through Pd-based membranes is commonly assumed that the diffusion coefficient is constant along the membrane and the mass transport kinetics is controlled by the diffusion step; this implies that the sorption and desorption process should be considered always to be at equilibrium (concentration of H 2 in the surface of the membrane in equilibrium with the gas phase-in the retentate and permeate sides); in the case of Pd-based membranes, the sorption isotherm is generally described by the Sieverts' law [339]. Taking into account all these considerations, the flux of H 2 that permeates through the membrane (J H 2 ) can be determined by the integration of Fick's first law of diffusion along the thickness of the membrane (Equation (13)) [339][340][341]. Figure 17. Solution-diffusion mechanism of the permeation of H 2 through a Pd-based membrane [337].
where L H 2 is the permeability of the membrane, δ is the membrane thickness, p H 2 is the partial pressures of H 2 in the retentate (inlet-feed side) or in the permeate (outlet) side.

The ratio
L H 2 δ is normally called permeance. The permeability is described by an Arrhenius-type dependency on the temperature (T)-Equation (14) where L H 2 0 is the pre-exponential factor, E a is the activation energy of the membrane permeation process, and R is the universal gas constant.
Despite the extensive use of Equation (13), it is necessary to emphasize that this equation is valid only for certain operation conditions: for instance, the expression is only applicable to ideal H 2 /metal mixtures and when the resistance in the diffusion of the H 2 along the thickness of the membrane is the rate-dominant process [323]. In general, the flux of H 2 can be influenced by several factors such as temperature, partial pressure difference, membrane thickness, membrane diffusivity, adsorption/desorption kinetics, and concentration polarization at the surface of the membrane. Ward and Dao [342] reported that for a clean Pd membrane, in the absence of external mass transfer resistance, the diffusion is the dominating step at moderately high temperatures, even for membrane thicknesses close to 1 µm.
However, as also reported by Ward and Dao [342] (and shown in Figure 17), beyond the H 2 diffusion through the membrane, other phenomena occur in the permeation process and the resistances to mass transport in the external gas phase may also be significant in some cases. The concentration polarization phenomenon is related to a decreased H 2 concentration in the feed side of the membrane in a multi-component system. In certain cases, due to the fluid dynamics inside the device, there is an accumulation of several species in the adjacent zone to the membrane surface, increasing the concentration of these species in the boundary film nearby to the membrane surface [343]. This causes a H 2 partial pressure gradient in the boundary layer and prevents the H 2 adsorption on the membrane surface (increase of the mass transport resistance). Thus, the H 2 concentration adjacent to the surface of the membrane is lower than that in the bulk gas and thus decreases the flux of H 2 permeated through the membrane [344][345][346][347][348]. The concentration polarization is affected not only by the external hydrodynamic conditions of the module, but also by the mass transport process inside of the membrane layer [349]. While most of the existing permeation results for thick Pd membranes are consistent with the considerations for the diffusion-limited permeation, significant differences exist for membranes with less than 10 µm thickness, since the very high fluxes made possible with such thin membranes makes the external mass transfer resistances relevant [342].
In this manner, due to the several possible phenomena that occur in this separation process, it is typical to use the empirical Equation (15) to describe the permeating H 2 flux [340,341]. It was reported in the literature that the pressure exponent n, in a general way, can provide insight regarding which transport step(s) control the overall permeation process. Several works studied and evaluated the physical meaning of the exponent n [342,350,351] for H 2 permeation through Pd-based membranes.
As already mentioned, the parameter n obtained is near 0.5 for a situation of no external mass transfer resistance (and the diffusion through the membrane controls the permeation process) [342,351,352]. It was also observed that the required value of the exponent n usually varies between 0.5 and 1 in the presence of external mass transfer resistances (effects are particularly strong on the permeate side) [342]. Moreover, it was reported that, in the absence of external mass transfer resistances, the n value approaches zero when the H 2 permeation is desorption-limited and varies between 0 and 0.5 when the desorption and diffusion are both influencing the H 2 flux [342]. It was also stated that (again in the absence of external resistances) the value of the exponent n is equal to the unity if only the adsorption process in the feed side limits the H 2 flux, and varies between 0.5 and 1 when adsorption and diffusion in the metal bulk are both limiting the permeation [352]. In line with the aforementioned, the study of Caravella et al. [351] verified that, at low temperature and using a thin membrane, the value of the exponent n decreases toward values lower than 0.5-desorption-limited process. However, higher values (>1) for the pressure exponent have been determined in previous works [352,353]. Besides that, the H 2 diffusion coefficient may also exhibit a concentration dependence and the membrane may present defects on the membrane films which may lead, for instance, to Knudsen diffusion. Tables 7 and 8 show a summary of different Pd-based membranes that have been described in the last few years with different parameters and performances.

OMWSR in Membrane Reactors
Just a few experimental works regarding the OMWSR in MRs are reported in the literature [1,6,89,90]; in such studies, the authors assessed the effects of pressure and gas/weight hourly space velocity (GHSV/WHSV) and will be summarized in this section.
In the first work of Tosti et al. [6], filtered and concentrated OMW was used for feeding a MR consisting of a dense Pd-Ag permeator tube packed with a Pt-based catalyst. Reforming tests were performed at 450 • C in the range of 1-5 bar total pressure. The experimental tests have shown the ability of the membrane device to selectively separate the H 2 produced and to provide a retentate stream rich in non-permeated H 2 , CO 2 , and CH 4 . A maximum H 2 yield equivalent to 2 kg of H 2 permeated per ton of OMW has been measured, when operating at 3 bar with a space velocity of 5.56 × 10 3 mol·h −1 ·g −1 cat . The gas chromatographic analysis of the retentate stream showed the presence of H 2 , CO 2 , and CH 4 , thus proving evidence of the presence of side reactions such as the hydrogenolysis and cracking of alcohols. No detectable quantities of CO were determined due to the completion of the WGS promoted by the presence of the membrane. Additionally, a deactivation of the catalyst due to coke deposits was observed.  In the second work, Tosti et al. [89] conducted again the OMWSR process in a MR (using a Pd membrane) to investigate the capability to recuperate pure H 2 and H 2 -rich gas mixtures from OMW. Specifically, a Pt-Rh-Pd-based catalyst supported on rare earth oxides was used in order to validate the capability of this reactor configuration of maximizing the H 2 yield and decreasing side reactions. As a main result, this work permitted verifying the following capabilities of the studied membrane process: the production of H 2 from waste (up to 3.25 kg of H 2 per ton of OMW) and a reduction in the pollution potential of the OMW (phenols and total carbon concentrations are reduced by ≈90%). Compared to a commercial Pt-based material that was used in the previous work, the new catalyst demonstrated higher selectivity toward the steam reforming reaction, permitting a substantial decrease in the formation of CH 4 and coke and thus exhibiting a higher H 2 yield and higher stability of the catalyst. In fact, in all experiments, low CH 4 concentrations were found in the retentate stream, while no significant loss of catalyst activity was detected after 6-8 h of operation. The low selectivity of the noble metals-based catalyst toward CH 4 formation was related to its intrinsic high reforming activity.
In the last work of the same research group [1], they tested the reforming of OMW and CH 4 in a Pd-membrane reformer and demonstrated that the steam reforming of CH 4 + OMW exhibits about the same H 2 yield of the reforming of the OMW alone. Besides that, the addition of CH 4 to OMW does not affect the ability of the steam reforming process to reduce pollution. Finally, it was possible to verify that the H 2 yield reached in the steam reforming of OMW together with CH 4 was higher than the H 2 yield obtained for the steam reforming of CH 4 alone.
More recently, Alique et al. [90], using a catalytic material prepared by Tosti et al. in a previous study [89] reported that, in a MR, higher pressures were beneficial, dominating the membrane shift effect (with enhanced H 2 permeating fluxes) against the negative influence of thermodynamics on the steam reforming reactions-see Figure 18. In addition, and as previously mentioned, in the work of Rocha et al. [197], several reactor configurations were tested for the OMWSR process-namely the SER and the SEMR (@ 1 bar and 350/400 • C). The difference between the H 2 yields obtained in the SER and SEMR is considerable, even after saturation of the sorbent (instant after which the system began to behave like a MR), showing that the impact of the membrane is very critical in this multifunctional system.

Summary
The OMWSR through a MR consists of combining the reaction and H 2 separation in the same device. This innovative reactor configuration requires a H 2 perm-selective membrane.
Pd-based membranes are currently promising for high purity H 2 production in MRs. Since the price of these dense membranes is still a restricting factor, a major effort has been put into the preparation of thinner Pd films over different supports, by several techniques. This decrease of the thickness of the membrane will allow the increase of the permeation flux, but the selectivity will be lower.
In this way, a thin supported Pd membrane is a potential membrane to separate from the reaction medium the H 2 produced in the OMWSR, in order to increase the H 2 yield (since it shifts the thermodynamic equilibrium to the production of more H 2 ) and produce an ultra-pure H 2 stream. However, it is necessary to further study the effect of the species present in the OMW on the membrane performance.

Sorption-Enhanced Membrane Reactor
Another technological alternative is the multifunctional hybrid reactor combining capture of CO 2 and H 2 separation: the SEMR (see Figure 19). This hybrid reactor configuration allows shifting the thermodynamic equilibrium of the steam reforming reaction by removing two of the reaction products from the reaction medium: hydrogen and carbon dioxide. This reactor configuration, intensifies the process even further by combining the reaction, H 2 separation, and CO 2 capture in a single device. This reactor configuration was not studied in detail until now for the OMWSR with the exception of a recent experimental work [197] and one (theoretical) thermodynamic study [194]. The SEMR combines the advantages of a SER and a MR; however, like in a SER, the CO 2 sorbent gets saturated at a certain point and then regeneration is needed (see details in Section 3.1). At this point, this reactor configuration is similar to a MR, since the CO 2 is not captured, but the H 2 is continuously separated from the reaction medium to the permeate side.

OMWSR in Sorption-Enhanced Membrane Reactors
As already stated, in the study of Rocha et al. [197], a comparison of the performance between different reactor configurations was performed for the OMWSR process (TR, SER, MR, and SEMR) @ 1 bar and 350/400 • C.
It was observed that the removal of H 2 and capture of CO 2 could provide higher H 2 yields in the SEMR during the pre-breakthrough region in comparison with the other hybrid multifunctional reactors. This reactor configuration allows the simultaneous production of highly pure H 2 in both the retentate and permeate sides. An H 2 yield improvement of about 44% was seen compared to the TR, approximately 39% regarding the SER and about 16% compared to the MR.
The H 2 yield observed in the SEMR at 350 • C and 1 bar was permanently between 8.0 and 6.0 mol H2 ·mol OMW −1 , and always between 11.5 and 9.0 mol H2 ·mol OMW −1 at 400 • C and 4 bar. This way, during the pre-breakthrough region, the H 2 yield reached practically the maximum theoretical yield (11.84 mol H2 ·mol OMW −1 ) in all cycles at 400 • C and 4 barsee Figure 20. In the first cycle, the H 2 yield inclusively goes over the thermodynamic equilibrium yield calculated for the TR in these operating conditions (cf. Figure 20). Besides that, CH 4 , CO, and CO 2 are not detected in the pre-breakthrough region, thus preventing the poisoning of the membrane and minimizing the coke deposits formation (particularly via CH 4 cracking). Reprinted with permission from Ref. [197]. Copyright 2021 Elsevier. Finally, with these experimental tests, it was observed that the OMWSR, when carried in a SEMR, can competently treat real OMW effluents, with high productions of H 2 , with complete decrease of the pollutant load.

Conclusions
OMW is a polluting stream derived from the olive oil industry and is an important source of environmental pollution, particularly in the Mediterranean countries. This stream is typically constituted by polyphenols, sugars, fatty acids, and water. Nowadays, and in order to reduce the pollutant load, several treatment/valorization techniques are applied, but these technologies have large cost and efficiency problems. Thus, OMWSR presents as a good alternative, since this process decreases the pollutant load of the OMW and simultaneously the waste is valorized with the production of green H 2 .
Currently, the OMWSR is an innovative treatment with high potential; however, alternatives for improving the process have been studied, since this process involves reversible reactions. In this way, it is expected that several reactor configurations (SER, MR, and SEMR) could work at lower temperatures than the conventional steam reformers and attain similar or even better performances than TRs operating at higher temperatures. Nevertheless, for the successful implementation of these reactor configurations, CO 2 sorbents with high sorption capacity, easy regeneration, and high stability-as well as H 2 permselective membranes with high H 2 permeance and selectivity-have to be used. Nowadays, HTCs and Pd-based membranes are seen as promising systems for use in the SEMR at temperatures in the range of 300-400 • C. Funding: This work was financially supported by the Base Funding-UIDB/00511/2020-of the Laboratory for Process Engineering, Environment, Biotechnology and Energy (LEPABE) funded by national funds through the FCT/MCTES (PIDDAC), and by the Project NORTE-01-0247-FEDER-39789, funded by European Regional Development Fund (ERDF) through the Programa Operacional Regional do Norte (NORTE 2020). This work was also financially supported by: Base Funding-UIDB/50020/2020-of the Associate Laboratory LSRE-LCM, funded by national funds through FCT/MCTES (PIDDAC); and project HyGreen&LowEmissions, with reference NORTE-01-0145-FEDER-000077, through the Programa Operacional Regional do Norte (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, funded by the European Regional Development Fund (ERDF). M.A. Soria thanks the Portuguese Foundation for Science and Technology (FCT) for the financial support of his work contract through the Scientific Employment Support Program (Norma Transitória DL 57/2017).

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