Abstract
Dry-bed adsorptive desulfurization of biomass-based syngas with low to medium sulfur content using ZnO was investigated as an alternative to the conventional wet scrubbing processes. The technical feasibility of ZnO-based desulfurization was studied in laboratory-scale H2S breakthrough experiments. The experiments were set up to utilize realistic H2S concentrations from gasification and therefore long breakthrough times. Experiments were performed in a steam-rich model biosyngas in varying conditions. The long-term breakthrough experiments showed apparent ZnO utilization rates between 10 and 50% in the tested conditions, indicating intraparticle mass-transfer resistances partly due to space velocity and particle size constraints as well as the most likely product-layer resistances as evidenced by the large spent adsorbent surface area decrease. An empirical deactivation model to estimate full breakthrough curves was fitted to the laboratory-scale experimental data. Breakthrough experiment in tar-rich syngas was also performed with the conclusion that ZnO performance is not significantly affected by hydrocarbons despite carbon deposition on the particle surfaces.
1. Introduction
With the increasing need for renewable fuel and energy sources, gasification of woody and herbaceous biomass is of high interest. However, biomass gasification produces also unwanted byproducts such as particulates, organic tars, sulfur and nitrogen heteroatom species, and inorganic constituents containing sulfur (H2S and COS), chlorine (HCl), nitrogen (NH3 and HCN), and alkali metals.1 Consequently, cost-effective and efficient purification of syngas is one of the biggest bottlenecks of advanced biomass gasification applications such as fuel or chemicals production through synthesis routes such as Fischer–Tropsch, MeOH, or SNG. Furthermore, large-scale processes have high investment costs and create problems with procurement of biomass feedstock. Decentralized small-scale fuel and chemical production seems thus an attractive alternative approach, but this in turn further increases the need for low-cost purification solutions.
Hydrogen sulfide needs to be removed to prevent downstream catalyst poisoning and to protect equipment from corrosion.2 Biomass contains usually under 0.1 mass % sulfur, that is mostly converted to H2S in gasification conditions.3 Raw biosyngas, depending on the biomass feedstock, contains tens to hundreds of ppmv H2S.4 The estimates of the target content of sulfur impurities required by downstream synthesis vary by source. For Fischer–Tropsch synthesis (Co- or Fe-based catalyst) Tijmensen et al.5 report H2S + COS tolerance below 0.01 ppm, while Boerrigter et al.6 report a total sulfur tolerance below 1 ppm. Methanol synthesis catalysts (Cu-based) in general have slightly higher poison tolerance than FT catalysts.7
Hydrogen sulfide can readily be removed by cold wet scrubbing technologies along with other acid gases such as CO2. The driving force is the acid gas partial pressure difference between the gas phase and the liquid phase. The processes therefore benefit from higher pressures and lower temperatures, which increase the costs.8 The main examples of this technology include the physical-solvent-based Rectisol and Selexol processes.9 These processes are complex and present a high investment cost. Furthermore, the expected H2S content in effluent of physical absorption systems in general is in the range of 0.1–1 mol·ppm.9,10
Adsorptive removal of H2S using metal oxides is a medium- to high-temperature gas cleaning technology where noncatalytic gas–solid interaction occurs with species combining chemically to the solid material, the adsorbent.10 Mid- to high-temperature desulfurization avoids the thermodynamic penalty from cooling and heating the gas. It is especially favorable in, for example, IGCC power plants where the gas is fed to combustion turbines that do not need, for example, CO2 removal, which requires a scrubbing step.11 Metal oxide adsorbents have traditionally been used in industrial processes as a polishing step for sulfur gases where H2S levels do not exceed 10–20 ppmv. For streams with higher H2S concentrations, wet scrubbing processes have been employed.12
The desired properties for desulfurization adsorbents can be generalized as (1) high H2S adsorption capacity [high capacity reduces the required adsorbent quantity and process equipment size]; (2) fast adsorption kinetics and favorable reaction equilibrium; (3) good mechanical properties; (4) chemical properties [stable in reducing environments and not affected by steam and possibly hydrocarbons]; (5) cheap and readily available; (6) ideally the capacity to also remove other S-species, which would present advantages in terms of process simplification; and (7) Ideally regenerable.13 In Table 1. a comparison between common metal oxide adsorbents is presented.
Table 1. Comparison between Different Metal Oxides for Desulfurization at Mid to High Temperatures.
Zinc oxide is perhaps the most widely used adsorbent for H2S removal from gas streams such as natural gas or syngas. It has high capacity due to its fast intrinsic kinetics at mid to high temperatures (200–450 °C) and good stability under reducing conditions.17
The zinc oxide sulfidation reaction is reported to have first-order kinetics with respect to H2S.18 The exothermic sulfidation reaction for ZnO is depicted as
 |
1 |
For sulfidation, assuming there is excess metal oxide, the equilibrium constant can be expressed as the ratio of the gas phase species; Bakker et al.19 suggested an extension to this by assuming that mainly surface sites act as acceptor sites during the sulfidation.20 Surface site concentrations are not constant, and therefore the surface phase could be included in the equilibrium expression:
 |
2 |
Thus, for effective desulfurization, the equilibrium constant should be high or the sorbent should have a high surface area that results in a good surface capacity.19
In this study, experimental investigation into zinc oxide as a bulk desulfurization adsorbent for biomass-based gasification syngas was conducted. Long-term breakthrough tests in laboratory scale using model syngas with realistic hydrogen sulfide concentrations present in biosyngas were performed.
2. Materials and Methods
The laboratory-scale experiments were conducted in a continuous atmospheric quartz packed bed reactor. The laboratory-scale experimental setup diagram is presented in Figure 1. The inner diameter of the quartz tube was 1.8 cm, with a thermocouple pocket of 0.4 cm in diameter in the bottom center of the bed. The bed length was up to 6 cm, and the reactor was placed in a 45 cm tall furnace. The gases were dosed using Bronkhorst mass flow controllers and water fed to an evaporator with a HPLC pump. The bottled gases were mixed with the vaporized water in a heated inlet line. The effluent gas was cooled in a condenser tube with a cooling water jacket. Tar containing gas was furthermore condensed in water wash bottles.
Figure 1.
(A) Laboratory-scale experimental setup schematic diagram. (B) Experimental setup photograph.
2.1. Methods
The experiments were conducted in a wet gas flow rate, V̇g, of 1.5–2.5 N dm3 min–1 and inlet H2S concentration, c0,H2S, of 100–120 ppmv, which resulted in up to multiweek breakthrough times. Before each run the GC was recalibrated, the system was leak-tested to an overpressure of 200 mbar, the feed H2S concentration, c0,H2S, was analyzed, and the Teflon tube analysis lines were saturated with H2S-rich gas.
The run was started by switching the gas stream from the reactor bypass to the preheated reactor line. The breakthrough curves were formed by non-interrupted operation. If any interruptions of the gas flow to the reactor occurred, the reactor was flushed with N2 and closed-off in the set point temperature.
After each run the reactor was cooled in nitrogen and the adsorbent material was visually inspected, photographed, and weighed.
The gas compositions used in the laboratory-scale experiments are presented in Table 2. Syngas 1 represents a gas that is obtained from a fluidized bed gasification process with almost complete reforming of the gas. Syngas 2 represents a pre-reformed gasification gas that includes a higher amount of CO2 and CH4 and C2H4 and the tar mixture. Syngas 2 as pre-reformed gas has also a slightly higher H2S concentration.
Table 2. Model Gas Compositions in the Laboratory-Scale Experiments.
| syngas 1 | syngas 2 | |
|---|---|---|
| H2 (vol %) | 38.01 | 31.17 |
| CO (vol %) | 19.00 | 13.94 |
| CO2 (vol %) | 13.50 | 16.43 |
| CH4 (vol %) | 0.50 | 6.47 |
| N2 (vol %) | 0.19 | 0.10 |
| H2S (ppmv) | 100 | 120 |
| H2O (vol %) | 28.79 | 30.03 |
| C2H4 (vol %) | 1.59 | |
| tars (g N m–3) | 14.81 |
The conditions listed in Table 3 are referred to as the standard conditions in this study, utilizing syngas 1. The breakthrough experimental data conducted in these conditions in Figure 3, Figure 4, and Figure 7 are depicted with blue dots.
Table 3. ZnO Laboratory-Scale Experimental Standard Conditions (Blue Dots).
| T (°C) | 400 |
| P (bara) | 1.0–1.2 |
| GHSV (h–1) | 21000 |
| V̇g (N dm3 min–1) | 2 |
| c0,H2S (wet gas) (ppmv) | 100 |
| bed L/D | 2.5 |
| Da (mm) | 1.0–1.25 |
D = particle size distribution.
Figure 3.
Laboratory-scale H2S (c0,H2S = 100 ppmv) breakthrough curves for ZnO with standard conditions according to Table 3 with variables (A) space velocity between 17000 and 26000 h–1 by varying gas feed rate between 1.5−2.5 N dm3 min−1. (B) temperature between 300 and 400 °C, (C) particle size between 1.0–1.25 and 2.0–2.38 mm, and (D) bed length between 2.5 and 4.5 cm.
Figure 4.
Laboratory-scale H2S breakthrough curve for syngas 1 (c0,H2S = 100 ppmv) and syngas 2 (c0,H2S = 120 ppmv) at 400 °C. The tests were performed in conditions according to Table 3.
Figure 7.
Linearized deactivation model equation of experimental (A) varying space velocities and (B) varying temperatures. Model breakthrough curves (lines) for (C) varying space velocities and (D) varying temperatures along with experimental breakthrough data (markers) from Figure 3A,B.
The standard reaction temperature is 400 °C, GHSV is 21000 h–1, bed length-to-diameter ratio is 2.5, and particle size distribution is 1.0–1.25 mm.
The system pressure was a few hundred millibars above atmospheric pressure in the inlet side, which is taken into account in calculating the real space velocity GHSV and flow rate. The real GHSV converted to nominal GHSV (at STP) in reference conditions using ideal gas laws was around 10000 h–1. The reported sieved particle size distribution diameter is considered roughly equal to the equivalent spherical diameter, Deqv.
2.2. Materials
The gases were supplied by AGA and the gas purities were as follows: H2, 99.999%; CO, 99.97%; CH4, 99.995%; CO2, 99.99%; C2H4, 99.95%; H2S, 0.500 mol % in N2; N2, 99.999%. Kurkela3 identified that in fluid bed biomass gasification the main constituents of the volatile hydrocarbons are mono- and polyaromatic compounds, benzene being the main component. Thus, a tar model mixture that consisted of 10 wt % benzene (Merck, >99.7%), 80 wt % toluene (Merck, 99.9%), and 10 wt % naphthalene (Merck, 99%) was used. The tar mixture was fed with a HPLC pump to an evaporator and mixed with the other gases in the reactor inlet. Washing bottles were used for condensing the water and tars from the effluent gas.
The tested ZnO adsorbent was commercial adsorbent of the type Actisorb S2 manufactured by Clariant, hereafter called adsorbent 1. According to the manufacturer, it consists of 90% ZnO and 10% alumina binder. The full-sized pellets of 4.5 mm D were crushed and sieved to the desired particle size distribution of either 1–1.25 mm or 2–2.38 mm.
2.3. Analytics
An online gas analyzer (Sick Maihak S710) was used for continuous analysis of the effluent dry gas composition. The analyzer included a TCD for H2 and an oxygen cell and a nondispersive infrared adsorption detector (NDIR) for CO, CO2, and CH4. The online gas analyzer was used to verify steady gas feeding and that no permanent gases were being consumed or formed in the reactor.
H2S was continuously measured from the wet effluent with a gas chromatograph Agilent 7890B with a FPD detector and a GS-GASPRO 30 m × 0.53 mm D column using carrier gas He. The sample line from the setup to the GC was heated. The GC calibration was conducted for H2S in the range between 7 and 71 ppmv dry gas. Calibration was conducted by feeding gas to the GC from an AGA calibration gas bottle containing 200 ppmv H2S and 20 ppmv COS with relative error of ±2% through gas diluter Pierburg 2000B with N2 as dilutant. As a reference, Dräger test tubes of type H2S 2/A were used for manually measuring the H2S concentration of the condensed gas. Dräger tubes are quantitative colorimetric chemical sensors, which react with H2S to form mercury sulfide, HgS. The reported standard deviation of this analysis method is ±5–10%. Dräger tubes were used to determine concentrations below 7 ppmv. Since the breakthrough was determined at 2 ppmv outlet H2S concentration, Dräger tube results were used to determine the breakthrough time, tb. Breakthrough curves were presented in this study using the FPD-GC data. For the experiment with syngas 2, only Dräger tubes were used for H2S analysis due to the limitations of the GC analytics with tar-rich gas. As the Dräger samples were taken from dry gas, the results were corrected for wet gas H2S composition to be comparable with FPD-GC results.
Adsorbent BET surface areas were measured using a Micrometric Tristar 3000 analyzer using N2 adsorption isotherms at 77.3 K. For spent adsorbent analysis the samples were taken from the upper first half of the bed. Scanning electron microscopy (SEM) was performed with a Carl-Zeiss Merlin scanning electron microscope. Chemical compositions of the sample were determined by energy dispersive X-ray spectroscopy (EDS). The microscope was equipped with a Thermo-Fischer UltraDry energy dispersive X-ray spectrometer (Silicon drift detector). Acceleration voltages used for imaging varied from 3 to 7 kV; for EDS 5, 7, and 10 kV were used. Adsorbent 1 was sampled before crushing and sieving in its original form of 4.5 mm D. The sample was placed on an Al stub for imaging.
2.4. Calculation Methods
Outotec HSC 8 software was used for the chemical equilibrium calculations. The software-derived equilibrium constant data were used to graph the equilibrium concentrations of reaction products as a function of temperature.
Generally, the allowed breakthrough concentration is defined to be usually between 1 and 5% of the inlet adsorbate concentration.21 In this study, 2 ppm H2S in the effluent was chosen as the breakthrough concentration, roughly equaling a 98% removal rate.
The sulfur adsorption capacity, Scap, is the sulfur captured per mass of adsorbent at breakthrough. It is a useful metric for determining the desulfurization performance of a given adsorbent. It can be calculated according to eq 3.
 |
3 |
where Scap is the sulfur adsorption capacity
(
), tb the breakthrough
time (min), V̇g the total gas flow
rate (N dm3 min–1), c0,H2S the wet gas H2S concentration
(ppmv), MS the molar mass of sulfur, Vmol the molar volume, and mads adsorbent weight (g).
The adsorbent utilization
percentage describes the amount of sulfur
captured relative to the adsorbent theoretical maximum. Pure ZnO has
a maximum Scap of 394 
based on the stoichiometry
of reaction 1. A 100
wt % adsorbent ZnO
content was assumed in the utilization rate calculations.
Deactivation models, which predict the decrease of activity of the solid reactant in gas–solid reactions, can be used to estimate the breakthrough curves in desulfurization. These models are used in literature by several authors including Yaşyerli et al.,22 Suyadal et al.,23 and Garces et al.,24 and they have been found to be more accurate than, for example, the theoretical unreacted shrinking core and the adsorption isotherm models. The deactivation model assumes isothermal conditions and neglects external mass-transfer limitations and that the deactivation of the adsorbent is first-order with respect to the solid surface. It can be described in terms of an exponential decrease with time in its available surface as2324
 |
4 |
where 
is the rate of change of activity of the
adsorbent, kd is the deactivation rate
constant, CA is the concentration of H2S in the gas phase, and a is the solid active
sites. Deactivation of the adsorbent is first-order with respect to
the solid active sites, n = 1, and zero-order for
the concentration of H2S, m = 0.
The equation is integrated with a0 being 1 at time zero, and pseudo-steady state is assumed. After some rearranging the following equation in linearized form is obtained:24
 |
5 |
The left side of the equation
can be plotted vs time using experimental breakthrough data, and a
straight line should be obtained with a slope of – kd and an intercept equaling 
. Thus, the adsorption rate constant and
deactivation rate constants are obtained.
3. Results
3.1. Chemical Equilibrium Calculations
Equilibrium calculations for the main sulfidation reaction and H2S reactions with syngas components to form COS were performed with equilibrium constant data from HSC 8 software. Figure 2 graphs the results as a function of temperature in biomass-derived syngas per Table 2, syngas 1.
Figure 2.
(A) Sulfidation equilibrium H2S concentration in excess ZnO as a function of temperature for syngas 1 (Table 2). (B) Equilibrium COS concentration as a function of temperature for two H2S reactions with syngas components for syngas 1 (Table 2).
The sulfidation equilibrium favors low temperatures. At 400 °C with the model syngas 1 composition, the equilibrium H2S concentration is 400 ppb (mole basis). Lower steam content in the gas favors desulfurization. In addition, steam could compete for adsorption on the ZnO surface, as was hypothesized by Novochinskii et al.,2 further negatively affecting desulfurization performance.
Equilibrium calculations for the COS forming reactions, the reverse COS hydrolysis and reverse COS hydrogenation, suggest that the presence of COS is possible. The equilibrium calculations suggest that the reverse COS hydrogenation is more significant compared to the reverse COS hydrolysis in the desulfurization conditions.
3.2. Breakthrough Experiment Results
The reference conditions for the long-term breakthrough experiments with ZnO in laboratory scale are listed in Table 3. The gas composition was syngas 1 according to Table 2.
The effects of space velocity, reaction conditions, particle size, and bed length on the H2S breakthrough are presented in Figure 3. The figures present the effluent H2S concentration as a function of time. The experiments were conducted up to around 30–60% H2S breakthrough completion with focus on initial breakthrough determination. The reference condition run according to Table 3 is indicated with blue dots. The breakthrough capacity of 2 ppmv is indicated as a horizontal red line as the basis for calculating the sulfidation capacity. The calculated sulfidation capacity is layered in text format on top of the indicated breakthrough curve.
An increase in the space velocity decreases the
sulfidation capacity
as indicated in Figure 3 A). The sulfidation capacity, Scap(2 ppm), drops from the low space velocity experiment of 176 
to 67 
, most likely indicating
significant intraparticle
mass-transfer limitations in the tested range. Gas–solid mass-transfer
limitations in the reactor scale cannot completely be ruled out, as
the space velocities in these experiments were high. Even though the
experiments were performed with long breakthrough times to simulate
more realistic adsorption conditions of industrial scale, the space
velocity is still much higher in all experiments than is generally
recommended as a design basis for large adsorption reactors. The reference
run 2 ppmv breakthrough capacity was 124 
, which
assuming a 100 wt % adsorbent ZnO
concentration equals to 32% utilization rate. In the runs presented
in Figure 3 D) the
space velocity was varied by changing bed length, and the effect of
the small bed length was not only evident in the early breakthrough
but also in the high subsequent H2S breakthrough concentrations.
With the bed length of 2.5 cm the reactor length to diameter-ratio
was 1.4 and the Scap (2
ppm) was 45 
. As the mass-transfer
zone and the residence
time is shorter, the breakthrough will be more intense. The sharp
drop in capturing capacity with the 2.5 cm bed suggest strong outlet
effects due to the mass-transfer zone being longer than the bed. Since
the 4.5 cm bed was the longest bed length tested, it is unknown if
this bed length suffers from the same problem. From the results presented
in In Figure 3 B) it
is evident that the effect of reaction temperature between 300–400
°C was relatively small on the sulfidation capacity. The best
performance at 2 ppmv H2S breakthrough level was found
to be at 300 °C, with Scap of 159 
. The effect of temperature
can especially
be seen in the slope of the breakthrough curve, with the lower temperatures
showing a steeper slope. At a higher breakthrough level of 7 ppmv
H2S, the best desulfurization performance was at 400 °C.
The tests presented in Figure 3 C) indicated substantial decrease of adsorption capacity
with increased particle size. With the 2.0–2.38 mm particle
size Scap was 29 
. Channeling effect cannot
be ruled out
due to the large particle size to reactor diameter- ratio. Despite
this, strong mass-transfer resistances for gas diffusion in to the
particle can explain the results.
The effect of tars on desulfurization performance was studied by feeding Syngas 2 (Table 2). Figure 4 presents the breakthrough curves for the experiments with Syngas 1 and Syngas 2; conditions are found in Table 3. For Syngas 1 in addition to the FPD-GC results, the Dräger H2S tube results are presented.
For the reference run, the FPD-GC and Dräger tube breakthrough
results agree well. The main difference is the sampling interval,
which for Dräger test tubes is more irregular and longer. Dräger
results for both experiments were used for breakthrough capacity determination.
The breakthrough capacity for the tar-rich gas syngas 2 was 126 
, which is virtually
equal to the reference
run with syngas 1. The breakthrough curve for syngas 2 is steeper
than for syngas 1.
3.4. Adsorbent Characterization
The SEM micrographs of the fresh adsorbent sample were used to investigate the zinc surface morphology and are presented in Figure 5.
Figure 5.
SEM micrographs of fresh adsorbent 1, with magnifications of (A) 5000 and (B) 25000.
The images reveal a fine surface structure of adsorbent 1. Aluminum is visible, but minimally present. The fresh adsorbent EDS analysis is presented in Table 4.
Table 4. EDS Elemental Analysis Results for Fresh Adsorbent 1.
| composition |
|||
|---|---|---|---|
| line | wt % | at. % | |
| Zn | L | 82.9 | 55.1 |
| O | K | 15.7 | 42.6 |
| Al | K | 0.9 | 1.5 |
| Si | K | 0.5 | 0.8 |
| total | 100 | 100 | |
The fresh adsorbent 1 analysis indicated in addition to Zn and O around 0.9 wt % Al and 0.5 wt % Si. No other species were detected by the EDS analysis.
The presence of hydrocarbons had a clear visual effect on the adsorbents, which is evident from Figure 6 showing the fresh and spent adsorbents.
Figure 6.
Photographs of (A) fresh adsorbent 1, (B) spent adsorbent 1 used in syngas 1 at 400 °C, and (C) spent adsorbent 1 used in syngas 2 at 400 °C.
The syngas 2 run adsorbent particles have a darker surface coloring, while the syngas 1 spent adsorbent is only slightly darker than the fresh sample. The BET results in Table 5 reveal a slight decrease in surface area compared to the syngas 1 run, which, however, is not significant compared to the general spent vs fresh surface area difference.
Table 5. Laboratory-Scale Experiment BET Surface Area and Bed Weight Change for Samples 1–3 from Figure 6.
| sample | 1 | 2 | 3 |
|---|---|---|---|
| sample name | fresh adsorbent | spent, 400 °C syngas 1 | spent, 400 °C syngas 2 |
| BET surface area (m2 g–1) | 42.6 | 18.2 | 16.9 |
| BET surface-area change vs fresh (%) | –57 | –61 | |
| spent bed weight change (%) | 13 | 12 |
The main contributor to surface area decrease according to these results is thus the sulfidation reaction. Fresh sample had a surface area of 42.6 m2 g–1, and the spent syngas 2 sample 16.9 m2 g–1 and syngas 1 sample 18.2 m2 g–1. It is likely that a carbon layer from the syngas 2 tars form the deposits to the adsorbent surface. The bed mass increase is similar after both experiments at around 12–13%, suggesting carbon deposition is not significant.
3.3. Deactivation Model
A deactivation
model of the first-order with respect to the solid surface was used
to predict the breakthrough curves to the saturation point. The breakthrough
curve can be thought of as a representation of the non-ideality of
the system and shows the global adsorption kinetics, and is thus of
interest to model. The FPD-GC breakthrough data from Figure 3 (A) GHSV runs and (B) temperature
runs were used. Only data from the GC calibrated H2S concentration
range from 7 to 70 ppmv were used to derive the rate constants by
linearizing the breakthrough data. Panels A and B of Figure 7 present the linearized equation according to eq 5. In Figure 7C,D, the parameters were used to calculate the model
breakthrough curve to saturation point 
= 1.
In Figure 7A,B straight
lines were obtained using the breakthrough data with a slope of −kd and intercept 
.
The linear regression, R2, coefficient
of determination is between 0.96 and 0.99.
In Figure 7C,D using
the obtained constants, the full breakthrough curves were plotted
against the real breakthrough curves. The derived deactivation rate
and initial adsorption rate constants (real basis) are presented in Table 6.
Table 6. Derived Constants from Experimental Data for the Deactivation Model.
| kd (h–1) | ko (dm3 g–1 h–1) | |
|---|---|---|
| set point variable: GHSV (h–1) | ||
| 17000 | 0.0128 | 808.0 |
| 21000 | 0.0117 | 341.2 |
| 26000 | 0.0241 | 1091.6 |
| st point variable: T (°C) | ||
| 300 | 0.0262 | 2779.3 |
| 350 | 0.0162 | 483.4 |
| 400 | 0.0117 | 341.2 |
4. Discussion
The equilibrium calculations suggest that lower temperatures are preferred for deep H2S removal and that ZnO has the potential for deep H2S removal in biosyngas. In a H2S removal process where bulk desulfurization is conducted at high temperature to encourage maximum mass-transfer rate, an additional ZnO guard bed could be used as a polishing step to achieve deep desulfurization. The guard bed should be operated at low temperatures of 200–300 °C to ensure deep sulfur removal according to the equilibrium calculations.
The reactions between H2S and syngas components CO2 and CO are harmful since COS is hardly captured or converted by zinc-based adsorbents25.18 Sasaoka et al.,26 however, suggest ZnS is active for the catalytic hydrolysis and hydrogenation of COS; thus the spent adsorbent bed part might aid total sulfur removal. Nevertheless, COS formation should not be neglected in the presence of CO and CO2. Dedicated COS conversion is suggested and usually involves the hydrolysis route and catalysts that include metallic oxides with basic active sites.27
Most industrial gas–solid adsorption processes employ a continuously operated vertical fixed bed reactor design where the bed L/D is <5. This was the basis for the experiments in study as well. For typical desulfurization breakthrough capacity studies in literature, high inlet H2S concentrations to achieve fast breakthrough times have been employed.172829.30 For this study, a realistic H2S concentration was used and thus longer breakthrough times with the benefit of (A) more accurately representing the breakthrough behavior in a full-scale process and (B) more reliable breakthrough results due to less error in the measurement of the breakthrough time, tb.
Adsorption processes are commonly sized based on the “mass-transfer zone” (MTZ) concept using experimental data. The adsorbent closer to the inlet of the fluid becomes saturated more quickly and develops a concentration gradient beyond the saturated zone called the “equilibrium zone”. The region where the concentration profile with the typical S-shape occurs is called the “mass-transfer zone”, MTZ, and is the sum of bulk and pore diffusional resistances and adsorption/reaction kinetics.21Figure 8 displays a symmetrical mass-transfer front along the length of the bed.
Figure 8.
Progress of a stable mass-transfer front through an adsorption bed of length L at breakthrough. C1,S is the adsorbed sulfur equilibrium concentration, and Ls is the saturated bed length.
LUB is the length of the unused bed at breakthrough. LUB can be calculated by
 |
6 |
where ts is stoichiometric
time; breakthrough time when 
= 0.5. In theory, the slower the mass-transfer
process and the adsorption kinetics are, the longer the LUB and MTZ
will be. The stoichiometric front travels with the same velocity as
the real front, and the saturation velocity, us, can be expressed as
 |
7 |
Now the new bed length of the scaled-up adsorber can be calculated by
 |
8 |
Assuming the wavefront to be symmetric gives MTZ = 2LUB.
However, scaleup using the LUB concept can only be used when the length of the overall bed is large relative to the mass-transfer zone, since short beds may have unstable mass-transfer zones.31 Estimating the LUB and MTZ from the breakthrough capacity results and the modeled breakthrough curves from this study shows that the mass-transfer zone is between 3.7 and 6.9 cm (5.9 cm in standard conditions according to Table 3) in the tested conditions, and thus the bed length is not sufficient for the LUB sizing concept to be applied to the results in this study. Nevertheless, the mass-transfer zone concept can be useful in interpreting the experimental results in this study.
The experimental results display incomplete sulfidation, below 50% utilization rate, at the tested conditions. The short bed length in all experiments relative to the mass-transfer zone caused diminishing of the utilization rates. This effect is evident from Figure 3D), where the smaller bed length 2.5 cm exhibited an overproportional loss of capacity over the 4.5 cm bed. A significantly taller bed than the current mass-transfer zone in the experiments would have neglected the strong breakthrough effects and increased utilization rates. The sulfidation capacities and utilization rates in this study do not therefore likely represent the equilibrium values and should be regarded as apparent capacities and rates.
The effect of GHSV and especially particle size on desulfurization performance suggested significant intraparticle diffusional limitations. As a comparison, Kim and Park30 studied the effect of GHSV in desulfurization and almost full utilization rate (96%) was achieved at 8000 h–1 (particle size distribution of 150–250 μm) and at above 8000 h–1 the intraparticle mass-transfer limitations even with this low particle size distribution were significant. In the same study the effect of particle size on desulfurization performance was studied. The intraparticle diffusion path is dependent on the particle size, and in a 450–1000 μm particle size distribution range the performance dropped significantly. Since commercial-scale adsorption processes utilize significantly larger particle sizes than 1000 μm due to pressure drop constraints, there can always be assumed to be significant intraparticle diffusional resistances.
In theory, higher reaction temperature improves mass-transfer; however this is not fully evident in the results in Figure 3B). The effect is visible from the breakthrough curve after tb in the highest temperature run at 400 °C, which has the lowest gradient S-curve and thus longest MTZ. The 2 ppmv H2S breakthrough is still dominated by factors other than the product-layer diffusion, and thus the 400 °C had the lowest breakthrough capacity. Since the process is pore diffusion limited, the surface S-species concentration is high and this decreases the driving force for mass transfer from the bulk to the surface. As a result, higher temperatures improve adsorption, while higher space velocities decrease it. Since the inlet gas feed rate was kept constant, the space velocities varied, and the 400 °C run had the highest space velocity, negatively affecting performance.
It can be concluded that in this study the particle size was the limiting factor as also evident from the test with the larger particle size distribution in Figure 3C). Furthermore, during the reaction between ZnO and H2S, S2– ions substitute O2– in the ZnO lattice.32 Since the formed ZnS occupies a larger volume, sulfidation causes a lack of porosity, as evident by the lowering of the BET surface area found in this study. This also causes product-layer diffusion resistance, and several studies333435 have attempted to model these diffusional resistances. The fresh adsorbent had a surface area of 46 m2 g–1 and spent adsorbents in the range 17–18 m2 g–1, Table 5.
The empirical deactivation model was used to model the laboratory-scale experimental breakthrough curves in this study. FPD-GC data from the calibrated H2S range from 7 to 70 ppmv could be used to derive the constants. The deactivation rate term, kd, has to take into account the changes in surface area, pore volume, and activity of the solid reactant due to sulfidation. Despite these challenges, the deactivation model represents the experimental data well. However, for the sizing of larger adsorbent beds additional experimental data with bigger bed length is required.
Adsorbent tests found in literature have mostly been performed in the absence of tars and other hydrocarbon contaminants, using model process gas or H2S with an inert gas. Wakker et al.,20 however, performed deactivation tests for MnO/Al2O3 and FeO/Al2O3 adsorbents in the presence of hydrocarbons CH4, C2H6 and C4H10. At the lowest test temperature of 350 °C there was no deactivation detected due to carbon deposition, but at 600 °C there was brown deposition on the reactor tube that was polymerized or vulcanized H2S with hydrocarbons. The authors note that also CO can cause carbon deposition, but such phenomena were not observed in the conducted tests. Water has also been reported to prevent carbon formation, which further prevents deactivation problems by hydrocarbons in wet syngas. In this work, the effect of biosyngas tars on desulfurization at 400 °C was investigated in wet syngas. The sulfidation capacity presented in Figure 4 was very similar to the reference run without tars. The breakthrough curve for the adsorbent tested in syngas 2 was steeper than in syngas 1, which might be attributed to either the 20% higher initial H2S concentration or competing adsorption of other species onto the surface. However, syngas desulfurization requires a low H2S outlet concentration, and thus breakthrough time and capacity are more important parameters in evaluating performance than the length of the mass-transfer zone. The spent adsorbent particulates from the syngas 2 run showed some carbon deposition on the surface, but the BET surface area analysis indicated only a slight loss of surface area compared to the reference. Thus, the effect of hydrocarbons was minor and desulfurization with ZnO adsorbents in such gas is feasible.
5. Conclusion
In this work, the technical feasibility of non-regenerable ZnO-based adsorbents for bulk biosyngas desulfurization was studied. Chemical equilibrium calculations predicted good desulfurization performance of ZnO but also the possible presence of COS, for which ZnO has no adsorptive affinity. Long-term breakthrough capacities of a commercial ZnO adsorbent were determined in laboratory scale using model biosyngas. The results indicated heavy intraparticle diffusional resistances within the tested conditions, with <50% apparent adsorbent utilization rates at 2 ppmv H2S breakthrough. The low utilization rate is partly attributed to the low bed length relative to the mass-transfer zone in the experiments, overstating the breakthrough effects The BET SA results showed significant loss of porosity due to the sulfidation product ZnS blocking the fine pores. A first-order deactivation model to estimate the full breakthrough curves at specific conditions was applied and found to represent the laboratory-scale results well. The effect of tars on desulfurization performance was found to be small, and the spent adsorbent surface areas were similar to reference adsorbents run in nontar gas even though heavy dark coloring of the particles occurred.
Acknowledgments
We thank Katja Heiskanen, Päivi Jokimies, and Patrik Eskelinen for the experimental and analytical work. We acknowledge Business Finland, VTT and the Finnish Industries for the financial support through the BTL2030 project (Financing Decision No. 4061/31/2015). We also thank the “COMSYN” project for the financial support. COMSYN has received funding from the European Union’s Horizon 2020 research and innovation program under Grant Agreement No. 727476.
Glossary
Abbreviations
- c
concentration
- CW
cool water
- D
diameter
- Deqv
equivalent spherical diameter, mm
- EDS
energy dispersive X-ray spectroscopy
- FPD
flame photometric detector
- FT
Fischer–Tropsch (synthesis)
- GHSV
gas hourly space velocity real basis, h–1
- HW
hot water
- K
equilibrium constant
- kd
deactivation rate constant, h–1
- ko
initial adsorption rate constant, dm3 g–1 h–1
- L
bed length, cm
- L/D
length-to-(inner) diameter ratio
- LUB
length of unutilized bed
- MTZ
mass-transfer zone
- N
nominal (at 273.15 °C and 101 325 kPa)
- n.d.
not detected
- NDIR
nondispersive infrared adsorption detector
- Scap (x ppm)
desulfurization capacity where x is the breakthrough concentration
- SEM
scanning electron microscope
- SNG
synthetic natural gas
- tb
breakthrough time, h
- ts
saturation time, h
- TCD
thermal conductivity detector
- TOS
time on stream, h
- V
gas flow rate, N dm3 min–1
The authors declare no competing financial interest.
References
- Chiche D.; Diverchy C.; Lucquin A.-C.; Porcheron F.; Defoort F. Synthesis Gas Purification. Oil Gas Sci. Technol. 2013, 68 (4), 707–723. 10.2516/ogst/2013175. [DOI] [Google Scholar]
- Novochinskii I. I.; Song C.; Ma X.; Liu X.; Shore L.; Lampert J.; Farrauto R. J. Low-Temperature H2S Removal from Steam-Containing Gas Mixtures with ZnO for Fuel Cell Application. 1. ZnO Particles and Extrudates. Energy Fuels 2004, 18 (2), 576–583. 10.1021/ef030137l. [DOI] [Google Scholar]
- Kurkela E.Formation and Removal of Biomass-Derived Contaminants in Fluidized-Bed Gasification Processes, VTT Publications 287; VTT Energy: Espoo, Finland, 1996; pp 47–87.
- Van Der Drift A.; Van Doorn J.; Vermeulen J. W. Ten Residual Biomass Fuels for Circulating Fluidized-Bed Gasification. Biomass Bioenergy 2001, 20 (1), 45–56. 10.1016/S0961-9534(00)00045-3. [DOI] [Google Scholar]
- Tijmensen M. J. A.; Faaij A. P. C.; Hamelinck C. N.; Van Hardeveld M. R. M. Exploration of the Possibilities for Production of Fischer–Tropsch Liquids and Power via Biomass Gasification. Biomass Bioenergy 2002, 23 (2), 129–152. 10.1016/S0961-9534(02)00037-5. [DOI] [Google Scholar]
- Boerrigter H.; den Uil H.; Calis H.-P.. Green Diesel from Biomass via Fischer–Tropsch Synthesis: New Insights in Gas Cleaning and Process Design. Pyrolysis and Gasification of Biomass and Waste, Expert Meeting, Oct. 1–13, 2002, Strasbourg, France; CPL: Newbery, U.K., 2003; pp 371–384.
- Kung H. Deactivation of Methanol Synthesis Catalysts - a Review. Catal. Today 1992, 11 (4), 443–453. 10.1016/0920-5861(92)80037-N. [DOI] [Google Scholar]
- Mohammed I. Y.; Samah M.; Mohamed A.; Sabina G. Comparison of Selexol and Rectisol Technologies in an Integrated Gasification Combined Cycle (IGCC) Plant for Clean Energy Production. Int. J. Eng. Res. 2014, 3 (12), 742–744. 10.17950/ijer/v3s12/1207. [DOI] [Google Scholar]
- Bertau M.; Wernicke J.; Schmidt F.. Methanol Utilisation Technologies. Methanol: The Basic Chemical and Energy Feedstock of the Future; Springer: Berlin, Heidelberg, 2014; pp 327–601, 10.1007/978-3-642-39709-7_6. [DOI] [Google Scholar]
- Kohl A.; Nielsen R. B.. Miscellaneous Gas Purification Techniques. Gas Purification, 5th ed.; Elsevier, 1997; pp 1296–1373, 10.1016/B978-0-88415-220-0.X5000-9. [DOI] [Google Scholar]
- Giuffrida A.; Romano M. C.; Lozza G. G. Thermodynamic Assessment of IGCC Power Plants with Hot Fuel Gas Desulfurization. Appl. Energy 2010, 87 (11), 3374–3383. 10.1016/j.apenergy.2010.05.020. [DOI] [Google Scholar]
- Ibsen K.Equipment Design and Cost Estimation for Small Modular Biomass Systems, Synthesis Gas Cleanup, and Oxygen Separation Equipment Task 2: Gas Cleanup Design and Cost Estimates - Wood Feedstock, Subcontract Report No. NREL/SR-510-39945; National Renewable Energy Laboratory (NREL): Golden, CO, USA, 2006.
- Cheah S.; Carpenter D. L.; Magrini-Bair K. A. Review of Mid- to High-Temperature Sulfur Sorbents for Desulfurization of Biomass- and Coal-Derived Syngas. Energy Fuels 2009, 23 (11), 5291–5307. 10.1021/ef900714q. [DOI] [Google Scholar]
- van der Ham A. G. J.; Venderbosch R. H.; Prins W.; van Swaaij W. P. M.. Survey of Desulfurization Processes for Coal Gas. Desulfurization of Hot Coal Gas; Springer: Berlin, Heidelberg, 1998; pp 117–136, 10.1007/978-3-642-58977-5. [DOI] [Google Scholar]
- Yumura M.; Furimsky E. Comparison of Calcium Oxide, Zinc Oxide, and Iron(III) Oxide Hydrogen Sulfide Adsorbents at High Temperatures. Ind. Eng. Chem. Process Des. Dev. 1985, 24 (4), 1165–1168. 10.1021/i200031a045. [DOI] [Google Scholar]
- Ayala R.; Venkataramani V.; Chuck T.; Gupta R.. Sorbent Development for Low-Temperature Moving-Bed Desulfurization; Advanced Coal-Fired Power Systems ‘96 Review Meeting; Morgantown Energy Technology Center (METC): Morgantown, WV, USA, 1996.
- Yang H.; Sothen R.; Cahela D. R.; Tatarchuk B. J. Breakthrough Characteristics of Reformate Desulfurization Using Zno Sorbents for Logistic Fuel Cell Power Systems. Ind. Eng. Chem. Res. 2008, 47 (24), 10064–10070. 10.1021/ie8008617. [DOI] [Google Scholar]
- Samokhvalov A.; Tatarchuk B. J. Characterization of Active Sites, Determination of Mechanisms of H2S, COS and CS2 Sorption and Regeneration of ZnO Low-Temperature Sorbents: Past, Current and Perspectives. Phys. Chem. Chem. Phys. 2011, 13 (8), 3197–3209. 10.1039/c0cp01227k. [DOI] [PubMed] [Google Scholar]
- Bakker W. J. W.; Kapteijn F.; Moulijn J. A. A High Capacity Manganese-Based Sorbent for Regenerative High Temperature Desulfurization with Direct Sulfur Production. Chem. Eng. J. 2003, 96 (1–3), 223–235. 10.1016/j.cej.2003.08.022. [DOI] [Google Scholar]
- Wakker J. P.; Gerritsen A. W.; Moulijn J. a. High Temperature H2S and COS Removal with MnO and FeO on Y-Al203 Acceptors. Ind. Eng. Chem. Res. 1993, 32, 139–149. 10.1021/ie00013a019. [DOI] [Google Scholar]
- Dutta B.Adsorption. Principles of mass transfer and separation processes; Prentice Hall, 2010; pp 609–677. [Google Scholar]
- Yaşyerli S.; Ar I.; Doǧu G.; Doǧu T. Removal of Hydrogen Sulfide by Clinoptilolite in a Fixed Bed Adsorber. Chem. Eng. Process. 2002, 41 (9), 785–792. 10.1016/S0255-2701(02)00009-0. [DOI] [Google Scholar]
- Suyadal Y.; Erol M.; Oǧuz H. Deactivation Model for the Adsorption of Trichloroethylene Vapor on an Activated Carbon Bed. Ind. Eng. Chem. Res. 2000, 39 (3), 724–730. 10.1021/ie990612w. [DOI] [Google Scholar]
- Garces H. F.; Galindo H. M.; Garces L. J.; Hunt J.; Morey A.; Suib S. L. Low Temperature H2S Dry-Desulfurization with Zinc Oxide. Microporous Mesoporous Mater. 2010, 127 (3), 190–197. 10.1016/j.micromeso.2009.07.022. [DOI] [Google Scholar]
- Sasaoka E.; Taniguchi K.; Uddin A.; Hirano S.; Kasaoka S.; Sakata Y. Characterization of Reaction between ZnO and COS. Ind. Eng. Chem. Res. 1996, 35 (7), 2389–2394. 10.1021/ie950370+. [DOI] [Google Scholar]
- Sasaoka E.; Taniguchi K.; Hirano S.; Uddin M. A.; Kasaoka S.; Sakata Y. Catalytic Activity of ZnS Formed from Desulfurization Sorbent ZnO for Conversion of COS to H2S. Ind. Eng. Chem. Res. 1995, 34 (4), 1102–1106. 10.1021/ie00043a011. [DOI] [Google Scholar]
- Zhao S.; Yi H.; Tang X.; Jiang S.; Gao F.; Zhang B.; Zuo Y.; Wang Z. The Hydrolysis of Carbonyl Sulfide at Low Temperature: A Review. Sci. World J. 2013, 2013 (8), 739501. 10.1155/2013/739501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Guo L. F.; Pan K. L.; Lee H. M.; Chang M. B. High-Temperature Gaseous H2S Removal by Zn-Mn-Based Sorbent. Ind. Eng. Chem. Res. 2015, 54 (44), 11040–11047. 10.1021/acs.iecr.5b02078. [DOI] [Google Scholar]
- Bu X.; Ying Y.; Zhang C.; Peng W. Research Improvement in Zn-Based Sorbent for Hot Gas Desulfurization. Powder Technol. 2008, 180 (1–2), 253–258. 10.1016/j.powtec.2007.03.039. [DOI] [Google Scholar]
- Kim K.; Jeon S. K.; Vo C.; Park C. S.; Norbeck J. M. Removal of Hydrogen Sulfide from a Steam-Hydrogasifier Product Gas by Zinc Oxide Sorbent: Effect of Non-Steam Gas Components. Ind. Eng. Chem. Res. 2007, 46, 5848–5854. 10.1021/ie0614531. [DOI] [Google Scholar]
- Geankoplis C. J.Transport and Separation Process Principles (Includes Unit Operations), 4th ed.; Pearson Education, 2003. [Google Scholar]
- Novochinskii I. I.; Song C.; Ma X.; Liu X.; Shore L.; Lampert J.; Farrauto R. J. Low-Temperature H2S Removal from Steam-Containing Gas Mixtures with ZnO for Fuel Cell Application. 2. Wash-Coated Monolith. Energy Fuels 2004, 18 (2), 584–589. 10.1021/ef030138d. [DOI] [Google Scholar]
- Lew S.; Sarofim A. F.; Flytzani-Stephanopoulos M. Modeling of the Sulfidation of Zinc-Titanium Oxide Sorbents with Hydrogen Sulfide. AIChE J. 1992, 38 (8), 1161–1169. 10.1002/aic.690380804. [DOI] [Google Scholar]
- Konttinen J. T.; Zevenhoven C. A. P.; Hupa M. M. Hot Gas Desulfurization with Zinc Titanate Sorbents in a Fluidized Bed. 1. Determination of Sorbent Particle Conversion Rate Model Parameters. Ind. Eng. Chem. Res. 1997, 36 (6), 2332–2339. 10.1021/ie960686x. [DOI] [Google Scholar]
- Sun J.; Modi S.; Liu K.; Lesieur R.; Buglass J. Kinetics of Zinc Oxide Sulfidation for Packed-Bed Desulfurizer Modeling. Energy Fuels 2007, 21 (4), 1863–1871. 10.1021/ef060521t. [DOI] [Google Scholar]