Small-to medium-scale deep syngas purification

The complete and economic removal of harmful components from biomass gasification-based syngas is a major challenge. A final gas cleaning concept for syngas purification to catalytic synthesis quality was developed as an alternative to organic-solvent scrubbing technologies. The motivation is to present smaller-scale BTL processes with a potentially lower-capex gas cleaning solution. The purpose of this study was to realize and validate the final gas cleaning concept in a real syngas environment and to study longer-term performance of deep gas cleaning. Two successful PDU-scale campaigns in complete biomass-to-liquids production chain were performed. A total on-stream time of 163 h was realized in syngas generated from residual woody-and agro-biomasses, with coupled gas feeding to Fischer-Tropsch synthesis. For S-species removal the final gas cleaning featured a novel hybrid of activated carbon-and ZnO-based bed materials. NH 3 and partial CO 2 removal was achieved by pressurized water scrubbing. The campaigns employed extensive continuous and non-continuous analysis techniques for the study of syngas impurities such as H 2 S, COS, NH 3 , HCN, HCl, benzene and tar. The final gas cleaning process demonstrated flexible deep removal of syngas contaminants of all tested biomass origins, thus achieving similar or better purification levels as conventional wet-scrubbing technologies. The cleaned gas was therefore suitable for catalytic synthesis purposes, demonstrating the technical feasibility of the new final cleaning process in conjunction with


Introduction
BTL-processes (biomass-to-liquids) are one of the most researched topics for the production of carbon-neutral synthetic transportation fuels or chemicals.Biomass as well as other carbonaceous feedstocks can be converted via thermochemical gasification route to synthesis gas and finally to liquid products [1].However, several techno-economic studies [2][3][4][5] have concluded that gasification based BTL concepts require very large scales in order to achieve positive economics.Consequently, small-to medium scale (<200 MW biomass input) facilities require reconceptualization of the design to be feasible.The control and management of syngas impurities has always been one of the biggest bottlenecks to commercial deployment of BTL processes.In the area of "final gas cleaning" i.e deep gas cleaning to meet strict end-use requirements for synthesis, most BTL concepts utilize existing technologies that originate from fossil fuel conversion.Significant cost saving benefits could be achieved by replacing these with optimized cleaning processes at relatively small biomass conversion plants.
The aim of the research was to develop a simplified "lower-capex" final gas cleaning process for trace impurities removal based on a hybrid of adsorption materials.The concept was constructed and evaluated at process development unit (PDU) -scale in a complete biomass-to-liquids process configuration.The intention of the experimental campaigns was to gain new knowledge of the challenges and opportunities of deep gas cleaning in a real syngas environment.

Final gas cleaning
Comprehensive bio-syngas clean-up presents several challenges.It requires treating a wide range of contaminants present in various concentrations.Cleaning is further complicated by the inherent inhomogeneity of biomass [6].These impurities can include particulates, organic tars, sulfur species (mainly hydrogen sulfide (H 2 S), but also carbonyl sulfide (COS), carbon disulfide (CS 2 ) and organic sulfur such as thiophene), halogens (mainly as chlorine), nitrogen-species (mostly ammonia (NH 3 ), but also hydrogen cyanide (HCN)) and alkali metals (mainly Na, K) [7].Table 1 presents estimates of impurities content after a hot gas cleaning section (hot filtration and tar reforming), adapted from previous studies of similar gasification facilities [8,9].Also included are literature estimates of the Fischer-Tropsch synthesis catalyst impurities tolerance.
For synthesis catalysts, the catalyst poisoning effect of sulfur species is perhaps best known, and it is reported that H 2 S adsorbs more rapidly to the surfaces and forms metal sulfides than COS and organic sulfurs.For Co-based FT catalysts an ex-situ catalyst poisoning study [12] showed that a 2000 cm 3 m − 3 sulfur loading caused almost full deactivation.However, the catalyst impurity tolerance is also an economic parameter where the investment in final gas cleaning versus catalyst lifetime need to be weighed [13].
Syngas cleaning can generally be divided into cold (where water vapour condensation occurs), medium (100-300 • C) and hot processes (>300 • C) [14].Hot processes generally involve catalytic routes for converting contaminants while cold processes often use techniques for separating species from the gas.The main drawbacks of cold gas cleaning methods in the BTL context is mainly 1) Thermal efficiency penalty due to syngas cooling 2) Cost incurred for treatment or disposal of contaminant streams, which in wet processes are solvent effluents and in dry processes spent adsorbents/catalysts [6].Cold gas cleanup methods are often considered more proven technology, thus the final gas cleaning in the BTL concept of this work is based on cold processes.
High acid-gas content streams are conventionally purified using organic solvent-based wet scrubbing technologies, either by physical absorption, such as Rectisol and Selexol processes, or chemical absorption, which are amine-based processes [15][16][17].The acid-gas scrubbing processes can however represent up to 20% (due to the process complexity, extreme process conditions and required gas treating units [17]) of the initial capital investments if applied to a medium-scale BTL concept, and consequently simpler alternatives to solvent-scrubbing methods are pursued [18].
Beyond conventional solvent-based solutions, several syngas purification processes for BTL demonstrations have been realized in research in the past years.The Bioliq® demonstration plant from Karlsruhe Institute of Technology (KIT) utilizes hot syngas cleaning concept where acid gases by alkali adsorbents and N-species removal by catalytic decomposition [19].The GoBiGas demonstration plant employed an RME scrubber step followed by several steam regenerated fixed activated carbon beds for removal of mainly BTX compounds, but also H 2 S and other trace contaminants.COS hydrolysis was performed separately [20,21].In a similar manner Güssing biomass DFB gasification demonstration plant the syngas cleaning for Fischer-Tropsch application was achieved by an atmospheric RME scrubbing step, followed by activated carbons.Pressurized fixed bed reactors involving Ni-based HDS catalyst (organic sulfur removal), ZnO and CuO adsorbents and Na-based adsorbents for HCl removal were utilized [22][23][24].The varying methods by which purified syngas is achieved highlights that each solution is tailored according to the specific gas impurity profile and content.

Final gas cleaning process concept
In this work we examine an integrated BTL concept, which coproduces FT syncrude and heat.Here primary tar control is carried out in the gasification and hot filtration units and finally by a simple, robust and highly effective catalytic reformer, as illustrated in the simplified BTL concept block diagram in Fig. 1 [9].The multifunctional hot gas filtration unit not only removes dust particles, condenses alkali-and heavy metals, but removes parts of the tars as well [25].The comprehensive hot gas cleaning section reduces the complexity of the final gas cleaning process by removing the need for a separate tar scrubbing steps.Calcium-containing gasifier bed materials such as limestone or dolomite act as tar decomposing catalysts and in-situ adsorbents for H 2 S removal, thus also simplifying downstream cleaning [26,27].Furthermore, Ni-based reforming catalysts have shown to promote NH 3 decomposition to N 2 and H 2 [28][29][30].

AR
Based on the estimated gas purity requirements for synthesis, a final gas cleaning process was designed that was optimized for the impurity target levels presented in Table 1, while retaining flexibility for variations in their concentration.The design is based on low-cost adsorption and organic solvent-free scrubbing removal methods that are simple and proven unit operations.Table 2 outlines the envisioned major units of the process and their primary function.
The process involves up to five steps, depending on the desired gas purity requirements, and are operated at low-to medium temperatures.Cold gas cleaning solutions for NH 3 typically involve absorption.NH 3 readily protonates in water, but removal is further improved in an acidic solution, which is why an acid washing step is included in the concept [6].
Activated carbon (AC) is a high surface-area potentially low cost (<1 $ kg − 1 ) adsorbent material that can be tailored by impregnation or surface modification to facilitate the removal of several types of impurities present in syngas [31,32].It can remove both inorganic and organic compounds at low temperatures, whereas ZnO is suitable for inorganic trace impurity removal at medium temperatures [11].In the BTL concept H 2 S is removed using activated carbons by physical adsorption or by a more effective selective oxidation route at low temperatures according to Reaction 1 [33,34].
In the latter case a small oxygen injection to the cooled gas is required, which results to an additional deoxygenation step in the gas cleaning process.A sufficient steam content in gas is also necessary for efficient removal of H 2 S on activated carbons [35].Complete carbonyl sulfide removal by activated carbons is not expected and thus residual COS removal is achieved by hydrolysis to H 2 S on metal oxide catalysts at medium temperatures in Guard Bed 1 according to Reaction 2.

COS(g)
While HCN is soluble in water, for full removal a catalytic conversion might be required.HCN hydrolysis at medium temperatures proceeds according to Reaction 3 [36].
Simultaneous HCN and COS hydrolysis is possible, as catalysts are very similar in both cases [37].Due to the versatile nature of activated carbons, organic and inorganic trace impurities, such as BTX compounds, metals, halogens, can be removed in the guard bed steps.The adsorbents employed in the units are primarily non-regenerable.
For CO 2 removal, water scrubbing at elevated pressures is a simpler alternative to organic solvent-based processes, but only feasible for partial CO 2 removal [16].The investment costs for water-based purification processes are relatively low for a small plant, and its operation and maintenance is simple [38].Thus, if complete or almost complete CO 2 removal is not required, water scrubbing is a viable low-cost solution.

Experimental setup
A new PDU-scale final gas cleaning process was constructed for the two week-long tests, Campaign I and II.The process is related in design to the final gas cleaning in the BTL concept and involves all five units  C. Frilund et al. listed in Table 2.The purpose of the campaigns was to verify the role of each unit operation in the process train and to assess the suitability of chosen adsorbents and catalysts.An existing gasification test facility, consisting of a bubbling fluidized-bed (BFB) gasifier, a hot gas filter (HF) and a catalytic reformer (CR) was used for syngas generation, and the gasifier was operated in steam/oxygen-blown mode.A mixture of silica sand and dolomite was used as bed material.The filter unit involved two sintered metallic filter elements.The catalytic reformer involved commercial nickel catalysts.The reformer was autothermally operated using a mixture of oxygen and CO 2 as feed gas that was injected on top of the catalyst bed.Three different feedstocks were selected for the test campaigns: bark, forest waste residues (FWR) and straw.The feedstock analyses are given in the Appendix.The final gas cleaning process, called "UC5", is schematically represented in Fig. 2. The final gas cleaning process, UC5, was constructed to match the scale of the synthesis process, with ca. 5 m 3 h − 1 (at standard conditions) dry syngas output as design basis.Standard conditions is defined as 101 325 kPa and 273.15 K, and hereafter all flowrates are normalized to standard conditions.In section A (atmospheric pressure) the Acid Wash Condenser (AWC) consisted of a counter-current acid wash column (i.d 0.16 m).The condenser step was followed by an Adsorbent Reactor (AR) involving two-stage fixed bed (i.d 0.25 m).The AR included an air injection, and gas relative humidity (RH) adjustment.The two compressors, CP1 and CP2, were of metal diaphragm type by Sera ComPress GmbH.Section B featured the Warm Guard Bed 1 (WGB1) step which consisted of three-stage fixed bed (i.d 0.08 m), placed in a furnace.The Pressurized Water Scrubber (PWS), which consisted of a pressurized counter-current absorption column (PWS AC) and an atmospheric desorption column (PWS DC) (both i.d 0.16 m) employed N 2 as stripping gas.The final unit in section B was the Cold Guard Bed 2 (CGB2), which was a two-stage fixed bed unit (i.d 0.08 m).A FT synthesis step, not described in Fig. 2, was connected downstream to the final gas cleaning for the entire duration of the campaigns.
Table 3 presents the packed materials.The primary function describes the intended use, however none of the materials are limited to only their primary function.
Four different activated carbon types were utilized in the process, manufactured by Jacobi Carbons.Two of these were non-impregnated virgin carbons, VAC1 and VAC2.Impregnated carbons involved a caustic impregnated carbon, CaAC and acid impregnated carbon AcAC.Literature suggests Al 2 O 3 and ZnO having COS hydrolysis promoting effects [39,40] A commercial ZnO adsorbent with alumina, Actisorb S2 by Clariant was therefore used to hydrolyse COS and simultaneously capture the formed H 2 S. A Cu/Zn catalyst, CuZn1, Research Catalysts Inc GetterMax® 133, was utilized for the deoxygenation of the syngas.
The bed materials and volumes are presented in detail in the Appendix.Since the final gas cleaning process was first time operated coupled to a FT synthesis process, the UC5 beds were packed to maximum capacity to avoid unwanted breakthrough of impurities.All beds were fresh packed for each campaign.AR was packed with a small top bed of caustic activated carbon, CaAC, followed by a large bed with top layer of VAC1 and bottom layer VAC2.For Campaign II the AR Bed 2 VAC2 packing volume was reduced by 25% from Campaign I. WGB1 two top beds were filled with ZnO1, and the bottom bed was reserved for CuZn1.CGB2 was packed with impregnated carbons.

Analytical methods
The analytical instruments were calibrated for small impurities concentrations.The analytical emphasis was in continuous monitoring of gas quality in the cleaned gas for breakthrough of impurities.Small gas quantities are presented as parts per million by volume (cm 3 m − 3 ).

Continuous analytical methods
The gases CO, H 2 , CH 4 , O 2 , N 2 and CO 2 were continuously monitored by ABB manufactured on-line analysers from sampling points after the hot filter S-BFB-1 (OA1), after the reformer S-BFB-2 (OA2) and after UC5 S-UC5-4 (OA3).These were used for real time process monitoring and control.Varian CP-4900 micro gas chromatographs (μGC) with thermal conductivity detectors (TCD) were also utilized and samples were taken from the same points as the OA, and were used in the results calculations for this study.The three micro-GCs (μGC1-3) along with the online analysers were calibrated using a calibration gas mixture in volume fractions of 15% H 2 , 15% CO, 15% CO 2 , 15% CH 4 and N 2 balance with a relative error of ±2%.
Oxygen breakthrough after final gas cleaning was monitored using the micro gas chromatograph (μGC3) with a specific method for low oxygen concentrations and an estimated limit of detection (LoD) of 0.001 vol % O 2 .
For sulfur compound detection, an Agilent 7890A gas chromatograph with a flame photometric detector (FPD-GC) and a GS-GASPRO 30 m × 0.32 mm i.d column with carrier gas He was used.The GC was calibrated using a H 2 S and COS containing calibration gas with concentrations 200 cm 3 m − 3 and 20.1 cm 3 m − 3 respectively, with relative error ±2%.The calibration gas was diluted using N 2 to achieve calibration minimum of 6 cm 3 m − 3 H 2 S and 0.61 cm 3 m − 3 COS.Other sulfur compounds were qualitatively analysed.The LoD for H 2 S was estimated at 0.1 cm 3 m − 3 and for COS 0.2 cm 3 m − 3 .
A Fourier transform infrared Spectroscopy (FTIR) of model Gasmet DX4000 was used to measure NH 3 , benzene and H 2 O content in the cleaned syngas.The component reference ranges were the following: NH 3 20-120 cm 3 m − 3 , benzene 50-2000 cm 3 m − 3 and H 2 O 0.1-50%.The limits of detection of the compounds were not separately tested in the syngas matrix.

Offline analytical methods
The concentrations of volatile organic compounds from benzene to higher molecular weight components up to pyrene were offline sampled and analysed following the European tar protocol [41].Tars were sampled in each test setpoint after the hot filtration, S-BFB-1, and after the reformer, S-BFB-1.Nitrogen compounds (NH 3 and HCN) were sampled and analysed after the reformer S-BFB-2.A known gas quantity was injected to a water sample and titrated with HCl for NH 3 determination.For HCN determination the water sample was analysed with a gas chromatograph.

Syngas generation and composition
The gasification temperature was maintained at around 820 • C with woody biomass feeds, while it was lowered to 710 • C with straw feedstock in order to prevent ash sintering and agglomeration in the gasifier bed.Filtration was conducted in the temperature range of 522-773  [42,43].Full conversion of tars and C 2 hydrocarbons was achieved with all the tested feedstocks.Benzene was the only detected residual hydrocarbon in addition to methane.
Benzene concentrations measured at reformer outlet were in the range of 0.2-0.4g m − 3 corresponding to benzene conversions of 96% with bark and forest residues and 92% with straw.The H 2 :CO mol-ratio of the reformed gas was maintained at around 1.8.
Compiled in Table 4 are the main gas analysis results after gasification and the hot gas cleaning section.

Final gas cleaning
The final gas cleaning process UC5 was operated in upstream coupled mode for a total of 87 h in in each campaign.The longest noninterrupted run was achieved in Campaign I, 43 h.The issues were related to short upstream interruptions in gas feeding.During the downtime the system was N 2 inertized.
The main gas cleaning and process measurement results are summarized in Table 5, in time-weighted average setpoint format, while Fig. 3 illustrates the overall measurement results as a time series.Only values from the specific experimental setpoints were included.
The formic acid:ammonia-ratio gives the consumption of acid in relation to the quantity of gaseous ammonia.The results suggest that in the two measured setpoints the ratio is similar, around 4. FTIR results measured from S-UC5-1 suggest complete or almost complete removal of NH 3 in the AWC step.Ammonia is the main basic syngas impurity that is absorbed at the AWC, and therefore acid is fed in a fixed ratio to keep the circulating water pH constant.
The AR air feeding was maintained fixed at 0.3 dm 3 min − 1 in both campaigns resulting in fluctuating O 2 :H 2 S-ratio, with minimimum as low as 1.8.The gas moisture content, expressed as relative humidity (RH), remained between 60 and 80% at the AR.
The exothermic deoxygenation occurring in WGB1 Bed 3 resulted to autothermal operation.The bed temperature was around 10-15 K warmer than the upper ZnO1 beds 1 and 2. The heating of these beds was increased for Campaign II to 205-210 • C from 200 • C in Campaign I.
The results for the continuous gas analytics and select process measurements, pressure and flowrates, are illustrated as a time series in Fig. 3.
The average pressure level in section A was ca. 4 kPa above atmospheric pressure, and the flowrate in the same section was ca.110-115 dm 3 min − 1 .Fig. 3a) and b) top chart area shows the relatively stable gas composition during the gasification setpoints, with only small fluctuations for the major gas components H 2 , CO, CO 2 and N 2 (TCD-GC analysed).NH 3 and benzene were not detected in any analysis samples in either campaigns.The bottom chart area in Fig. 3a) and b) shows that a small, but non-continuous, breakthrough of COS occurred.This was evident especially in Campaign 1, where concentrations up to 0.3 cm 3 m − 3 were detected, especially during the first tens of hours on stream.Breakthrough is believed to be caused by too low WGB1 ZnO1 bed temperature for COS hydrolysis.For Campaign 2 the bed temperatures were increased to above 205 • C, as reported in Table 5, and the breakthrough was consequently almost non-existent.H 2 S was not detected in any samples.Other sulfur compounds, such as sulfur oxides and organic sulfurs were qualitatively analysed, and none were detected (within the detection limits of the GC), which is expected since no other sulfur compounds were detected after the hot gas section.Yet it indicates that the oxygen feeding to the cold AR does not cause for example SO x formation.Fig. 3a) and b) also indicate that pressure fluctuations during the run were minor.As the gas pressure (flow) varies in section A, the majority of the fluctuation is dampened by OGP1, where off-gas flowrate increased or decreases.
The average gas analysis results after the final gas cleaning is given in Table 6.The results are averaged from continuous measurement, using TCD-GC, FTIR and FPD-GC, with the exception of offline colorimetric analysis samples of HCN, HCl and SO 2 .
The only confirmed impurity breakthrough was COS, as evident by Fig. 3a) and b).The time weighted average during the course of the campaigns was significantly below 0.1 cm 3 m − 3 , while highest breakthroughs were up to 0.3 cm 3 m − 3 .It is noteworthy that the higher COS inlet concentration level in Campaign II SP2 (29 cm 3 m − 3 versus 5-6 cm 3 m − 3 in the other setpoints) did not cause a COS breakthrough, suggesting the reaction temperature was adequate also for the higher Ammonia and benzene were not detected at the outlet, thus high removal efficiency in the AWC step was achieved.NH 3 is very water soluble, but it was enhanced by the addition of acid.The full removal of benzene (ca.60-120 cm 3 m − 3 in raw syngas) meant that tar compounds were also fully removed.Hydrocarbons are commonly removed by active carbons from gases for example in odour control applications, and can therefore be considered proven technology [45].Hydrocarbons are primarily reversibly adsorbed and therefore spent activated carbon beds could be regenerated relatively easily.The offline colorimetric analysis samples taken at 1-3 occasions during setpoints indicated that HCN, HCl or SO 2 impurity breakthrough did not occur.Since SO 2 was known not exist in the raw syngas, and HCl concentrations were non-existent or very low (0.2-2 cm 3 m − 3 from experience of earlier experiments with the same gasification facility).HCN was also totally removed in either the adsorption or water-based scrubbing steps.It was also fully removed in Campaign II SP2 with up to 6 cm 3 m − 3 inlet concentration.

Hydrogen sulfide removal
Manual H 2 S measurements during setpoints from various sampling locations after AR Bed 1 (S-UC5-2) and after AR (S-UC5-3), along with the spent adsorbent characterisation were used to gain insights into the stepwise removal of H 2 S in the process.Table 7 summarizes these

results.
AR Bed 1 was filled with caustic activated carbon CaAC1.The bed heights were 5 cm and 2.5 cm for Campaign I and Campaign II respectively.The removal rate amounted to over 50% with the low sulfur feedstock setpoints (Campaign I SP1 and Campaign II SP2).This, despite competing adsorption onto the surface by other species, especially benzene.In the high sulfur setpoint (Campaign II, SP2) almost full breakthrough occurred.Köchermann et al. [46] showed that oxygen and steam-free syngas cannot feasibly be desulfurized by activated carbons.Therefore the high removal rate by AR Bed 1 shows that air injection likely promoted H 2 S removal.AR Bed 2 contained both VAC1 and VAC2, and H 2 S was removed below analysis detection limit (~<0.1 cm 3 m − 3 ) in the low sulfur setpoints.In the high sulfur setpoint a breakthrough of ca.0.5 cm 3 m − 3 , was observed and was monitored for 8 h (5 samples) during which the breakthrough did not grow.It suggests that the equilibrium removal rate for the oxidative removal route was reached.The possible reason is that the average O 2 :H 2 S-ratio was only 1.8, as reported in Table 5, which was not sufficient for complete removal of H 2 S.
The post-campaign adsorbent characterisation results show decreased specific surface areas relative to fresh samples.As expected, surface sample available specific surface area and consequently available pore volume decreased most, while the bed samples (− 15 cm and − 30 cm) showed less decrease.The AR Bed 2 bottom sample BET-SA was 930 m 2 g − 1 which is almost equal to fresh sample surface area, suggesting unused bed volume.

Pressurized water scrubber
The average CO 2 removal rates calculated using continuous μGC3 data for the two campaigns are presented in Table 8.
Table 8 shows that the Campaign II removal rates are higher than in Campaign I.This is due to higher N 2 feed rate and lower circulating water temperature, and thus up to 53% removal rate was achieved.The liquid-to-gas ratio was kept constant during both campaigns at around 220-250 kg of water per m 3 of syngas at standard conditions.Fig. 4 visualizes the CO 2 removal as a time series for Campaign I along with operating conditions: CO 2 partial pressure, temperature and flowrates.
Fig. 4 shows that the RR CO2 increased from 46% to almost 52% during the campaign.This was mainly due to manual tuning of PWS conditions in an effort to improve removal.N 2 flow was systematically increased, which directly correlates with the CO 2 removal rate.CO 2 stripping from the gas is thus enhanced by increased N 2 flow as a consequence of improved solubility of CO 2 at the pressure side.The data suggests only marginal benefits of N 2 flowrate above 17.5 dm 3 min − 1 (The mol ratio of water to N 2 was ca.1300).The improved CO 2 removal at the end of the campaign is caused by adjusting the circulation water cooling and the subsequent 2 K drop in water temperature.The results illustrate that the CO 2 removal worked well with no drop in performance, thus showing that no significant contamination of the circulating water occurred during the campaigns.The PWS DC outlet gas was analysed during Campaign II SP2.To demonstrate the scrubber operational efficiency, the major gas component saturation solubilities in water at Campaign II SP2 PWS AC exit conditions were calculated using Henry's law with experimental parameters obtained from a compilation by Rolf Sander [47].The results are compiled in Table 9.
From the estimated N 2 volumetric feedrate to the stripper, the PWS component mol balances were calculated to yield removal rates.The removal rate for CO and H 2 was around 3% during the measurement period.The estimated syngas feedrate to PWS DC was 80 dm 3 min − 1 and the CO 2 removal rate was 10 dm 3 min − 1 .If the CO 2 solubility in water had reached saturation, the water circulating rate could have been 36% lower, i.e. 12 dm 3 min − 1 or 145 kg water per m 3 syngas.The CO 2 solubility is a bit lower than the saturation prediction when non-idealities of experimental column gas-liquid mass-transfer are factored in.

Comparison to wet scrubbing technologies
Conventional organic-solvent scrubbing technologies, like Rectisol and Selexol, achieve removal efficiencies proportional to the impurity solubility to the solvent.Removal of H 2 S down to <0.1 cm 3 m − 3 is possible, thus potentially allowing direct feeding to synthesis, though at the additional cost of a fine wash step.The Rectisol process consists of a prewash step (hydrocarbon removal), main wash (bulk acid gas removal) and often a fine wash step (deep sulfur removal), and utilizes cryogenic methanol [48].At these extreme conditions CO 2 is mostly removed (95+ %) as well [13].The Selexol utilizes dimethyl ether/polyethylene glycol solvent, and is operated at less energy consuming conditions [48].Thus, the removal limit is higher, likely requiring a further H 2 S control step before synthesis.At large scales, the separated acid gas streams are treated by standard sulfur removal technologies such as the Claus process.However, as co-removal of CO 2 is occurring from a syngas stream, sulfur recovery becomes more expensive [13].For methanol solvent at the Rectisol working conditions, the relative solubility compared to H 2 S is higher for NH 3 but lower for COS and HCN [49,50].For the Selexol, the relative solubility of all the aforementioned compounds is significantly lower than for H 2 S, thus requiring additional processing steps for especially COS and HCN.Residual tar removal prior to all acid gas wash processes is required to avoid accumulation to the solvent streams [51].The final gas cleaning process general technical characteristics are compared to existing wet scrubbing gas cleaning solutions in Table 10.
Both the final gas cleaning process and conventional wet scrubbing processes incur a thermal efficiency penalty for operating at low temperatures and cooling down the syngas, although this heat is partially recoverable.Generally, the wet scrubbing working pressures are higher than used in the process of this study, which in-part operates at atmospheric pressure and medium-pressure range for the PWS.The wet scrubbing processes consume energy for cooling and regenerating the solvent, which contribute to the majority of the operating costs, while majority of the costs of the studied process costs are related to adsorbent replacement, gas compression and PWS water pumping.Adsorptionbased cleaning has an advantage at smaller scale or lower impurities concentrations, since the unit operations are simpler and adsorbent replacement quantity small, thus the operating cost is likely more competitive than for the wet scrubbing technologies, which require efficient heat integration regardless of scale.Comparing the cleaning results of the process in this study to alternatives, the achieved purity levels are comparable to the best available wet scrubbing technologies.Summarizing, it can be said that the two technologies are both flexible syngas purification technologies, but they serve different scale and impurity profile purposes with trade-offs that should be weighed caseby-case.

Conclusions
A simplified deep gas cleaning process, based on adsorbent-and organic solvent-free absorption steps, was successfully piloted and evaluated as an alternative to conventional wet scrubbing technologies for smaller-scale multi-contaminant syngas applications.Gas cleaning performance was assessed in two week-long PDU-scale biomass-toliquids (BTL) experiments with gasification-generated syngas by an extensive analytical setup for catalyst poison breakthrough.It featured a novel combination of packed bed adsorbents and catalysts for bulk impurities removal mainly facilitated by activated carbons.The system also included a pressurized water scrubber for partial CO 2 removal.
H 2 S removal is achieved by the efficient oxidative route on activated carbons with air injection and subsequent deoxygenation.H 2 S was completely removed by activated carbons in the first adsorption step (AR).A ZnO-based adsorbent was validated for combined COS hydrolysis and H 2 S removal.A small COS breakthrough was detected, which was mitigated by reaction temperature increase.NH 3 was likely completely removed in the first acid wash step, and the low concentration impurity HCN removal was complete.Other minor impurities eg.Benzene and HCl were all were below detection limit after the gas cleaning process.CO 2 removal by water scrubbing achieved a removal rate of 50% with no decrease in performance.The cold guard bed was essentially considered redundant due to the effectiveness of the prior steps.The final gas cleaning process demonstrated the flexible removal of syngas impurities of residual woody-and agricultural-biomass origin.Most notably, the syngas of purity levels suitable for the coupled catalytic Fischer-Tropsch synthesis unit was produced.Hence it was established that the simplified final cleaning concept is sufficient for biomassderived syngas when combined with the optimized hot gas cleaning process.

Table 10
Generalized comparison between the final gas cleaning process presented in this study and conventional organic solvent-scrubbing methods [13,48,49].

Fig. 3 .
Fig. 3.For a) Campaign I b) Campaign II, continuous gas analysis results after UC5 final gas cleaning from sampling point S-UC5-4.For c) Campaign I d) Campaign II, continuous process measurement results from top chart to bottom chart: section C, B, A pressures and bottom chart for flowrates.Final gas cleaning upstream coupling is indicated with dashed black vertical lines.Setpoint start and stop times indicated with solid black vertical lines.Gas analysis results are indicated with the following markers: dots TCD-GC, triangles FTIR, squares FPD-GC.Colours represent gas components.Solid filled markers utilize left y-axis units and non-filled markers right y-axis units.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) GC, dry basis.b FTIR, dry basis.c FTIR, wet basis.d FPD-GC, dry basis.e Offline sampling (average) with colorimetric tubes, dry basis.f Measured using a TCD-GC from S-UC5-4 calibrated for low O 2 concentrations.A 0.01% O 2 base level concentration was subtracted from the results.C.Frilund et al.

Fig. 4 .
Fig. 4. Campaign I pressurized water scrubber CO 2 removal performance.RR CO2 indicated as blue dots.Green line represents CO 2 partial pressure before PWS (bar), blue line circulating water temperature ( • C), orange line PWS DC nitrogen feed (dm 3 min − 1 ), yellow line circulating water flowrate (dm 3 min − 1 ).Final gas cleaning upstream coupling and decoupling is indicated with dashed black vertical lines.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Table 1
Estimates of steam fluidized bed gasification gas impurities concentrations after hot gas cleaning section, and gas purification requirements for Fischer-Tropsch application.

Table 2
Final gas cleaning concept unit operations and function.

Table 3
PDU-scale final gas cleaning process UC5 packed materials and volumes for Campaign I and II. the baseline pressure drop across the filter remained stable throughout the test campaigns.The catalytic reformer was operated at maximum ca.1000 • C and reformer outlet at 900 • C. The well-known poisoning effect of reformer nickel catalysts by H 2 S can be largely avoided by operating at 900 • C or above • C and

Table 4
Setpoint average analysis results after gasification and hot gas cleaning sections.

Table 5
Setpoint average process measurement results for final gas cleaning process UC5.
a 100% formic acid.b Based on setpoint average H 2 S concentration from Table 4 and measured air feed rate.C. Frilund et al. concentrations.ZnO/alumina acted as a dual functioning COS hydrolysis and H 2 S sulfiding adsorbent, similar to what Spies et al. [44] have reported.Almost full COS conversion was achieved by operating the reactor at sufficiently high temperatures.

Table 6
Setpoint average gas compositions of cleaned gas, sampled from S-UC5-4.

Table 7 H
2 S gas analysis results from multiple sampling locations.Spent adsorbent chracterization: BET-SA and sulfur concentration.

Table 8
Setpoint average pressurized water scrubber CO 2 removal rates.Removal rate in fractions is calculated as RR CO2 = y CO2,OUT − y CO2,IN y CO2,OUT − y CO2,OUT ⋅y CO2,IN (4) where y CO2,IN/OUT denotes CO 2 fraction in the PWS AC.IN and OUT gas composition offset by final gas cleaning process estimated average residence time of 10 min. a

Table 9
Pressurized water scrubber saturation concentrations c * of syngas components in water and realized concentrations c from Campaign II SP2 S-UC5-5 stripper gas analysis (PWS DC exit).
a By volume, dry basis.Low concentration species, CH 4 , was not included.bAt the time of measurement, N 2 feed to PWS DC was ca.20 dm 3 min − 1 C.Frilund et al.