Top Catal (2011) 54:977–985 DOI 10.1007/s11244-011-9719-5
ORIGINAL PAPER
Investigation of Mixtures of a Co-Based Catalyst and a Cu-Based Catalyst for the Fischer–Tropsch Synthesis with Bio-Syngas: The Importance of Indigenous Water Matteo Lualdi • Sara Lo¨gdberg • Francesco Regali Magali Boutonnet • Sven Ja¨ra˚s
•
Published online: 2 August 2011 Ó Springer Science+Business Media, LLC 2011
Abstract A series of different mechanical mixtures of a narrow-pore Co/c-Al2O3 catalyst and a Cu-based WGScatalyst has been investigated in the low-temperature Fischer–Tropsch synthesis (483 K, 20 bar) with a model bio-syngas (H2/CO = 1.0) in a fixed-bed reactor. The higher the fraction of WGS-catalyst in the mixture, the lower is the Co-catalyst-time yield to hydrocarbons. This is ascribed to a strong positive kinetic effect of water on the Fischer–Tropsch rate of the Co-catalyst, showing the importance of the indigenously produced water, especially in fixed-bed reactors where the partial pressure of water is zero at the reactor inlet. A preliminary kinetic modeling suggests that the reaction order in PH2 O is 0.3 for the Co/ c-Al2O3 catalyst in the range of the studied reactor-average partial pressures of water (i.e., 0.04–1.2 bar).
low-temperature FT (LTFT) process. The catalyst in this process consists of precipitated and extruded iron promoted with alkali [2, 4]. The water–gas shift (WGS, see reaction 1.2) activity of the iron catalyst makes sure that the H2/CO ratio inside the reactor is high enough to satisfy the stoichiometry of the FT reaction. Since the reaction order of the partial pressure of H2 has been reported to be positive and of similar magnitude (0.5–1) for both iron- and cobalt-based catalysts, there is also a kinetic reason for increasing the H2/CO ratio of a H2-poor syngas [5–7]. The water needed in the WGS reaction (II) is produced in the FT reaction (I).
Keywords Fischer–Tropsch Cobalt Alumina Water effect Copper Water–gas shift
However, the use of iron-catalysts limits the conversion per pass since water has a negative kinetic effect on reaction rate and enhances deactivation strongly [5]. Cobalt-based catalysts, instead, can ensure high conversion per pass but they do not possess WGS activity at such low temperatures as those used in the LTFT process where diesel is the desired end-product [8]. The initial aim of the present study was to investigate the mechanical mixtures of a Co/c-Al2O3 FT-catalyst with a Cu-based low-temperature WGS-catalyst for a model bio-syngas, having a H2/CO ratio of 1.0, in a fixed-bed reactor. The mechanical mixing option has been previously investigated in the literature, and patents can be found, for Co-based catalysts on other supports [9–11]. The findings of the present study together with those of our recent publication [12] and ongoing work on water effects on the FT reaction rate and selectivity, however, suggested the possibility to use these experimental data obtained at very low partial pressures of water as a complement to those
1 Introduction Syngas obtained from gasification of biomass or coal typically has a lower ratio of hydrogen to carbon monoxide compared to syngas produced from natural gas [1–3]. The Fischer-Trospch (FT) reaction (see reaction 1.1), in which syngas is converted into hydrocarbons (HCs), has a H2:CO stoichiometry of 2.1:1. The lowest industrial molar feed H2/CO ratio is 1.7, which is used in Sasol’s coal-based
M. Lualdi (&) S. Lo¨gdberg F. Regali M. Boutonnet S. Ja¨ra˚s Chemical Technology, Royal Institute of Technology (KTH), Stockholm, Sweden e-mail:
[email protected]
FT:CO þ 2H2 ! CH 2 þ H2 O
ðIÞ
WGS:CO þ H2 O $ CO2 þ H2
ðIIÞ
123
978
obtained upon adding water to the feed (external water addition). In fact, the effect of water, which is the main byproduct of FT synthesis, has long attracted significant attention. Traditionally, the rate expressions for Co-based FT-catalysts have only contained the partial pressures of H2 and CO, in which, generally, the reaction order in PH2 is positive while that in PCO is negative [7, 13–15]. Although, water is known to cause deactivation of Co-based catalysts [16], recent literature indicates the possibility of positive kinetic effects of water [17–19] also for Co supported on narrow-pore c-Al2O3 [12] for which the initial waterinduced deactivation is rapid and severe [12, 20].
2 Experimental 2.1 Catalyst Preparation A Co-based FT-catalyst composed of 12 wt% cobalt supported on c-Al2O3 was prepared by the incipient wetness technique. The c-Al2O3 (Puralox SCCa-5/200 from Sasol) was first calcined in air flow for 10 h at 773 K (ramp 1 K/ min) and then impregnated with an aqueous solution of Co(NO3)26H2O (ACS reagent, C99.0%, Fluka). The catalyst was subsequently dried at 393 K for 3 h and calcined in air at 573 K for 16 h (ramp = 1 K/min). A Cu-based WGS-catalyst was also prepared. In order to avoid residual alkali-metals, which have been reported to be detrimental to the FT activity of Co-catalysts [21, 22], a co-precipitated 30 wt% Cu/30 wt% ZnO/Al2O3 catalyst was prepared according to Ref. [23]. In brief, aqueous ammonia was added to an aqueous solution of nitrates of Cu, Zn and Al to form a precipitate at a pH-value around 7. The resultant gel was dried in air at 373 K for 12 h and then calcined at 773 K for 3 h. As Cu-based WGS-catalysts are usually activated at significantly lower temperatures than Co-based FTcatalysts [23, 24], the Co-catalyst was first ex situ reduced, and then activated in situ together with the WGS-catalyst. The Co-catalyst reduction was performed in pure hydrogen flow (*200 Ncm3/gcat, min) at atmospheric pressure for 16 h at 623 K (heating rate: 1 K/min) and then the catalyst was passivated. The passivation was carried out at roomtemperature flowing N2 (*250 Ncm3/gcat, min) through the sample overnight and then flowing 0.5% O2 in N2 (*250 Ncm3/gcat, min) for 1 h, monitoring the sample temperature.
Top Catal (2011) 54:977–985
overnight prior to analysis. The BET area was estimated by N2 adsorption at liquid nitrogen temperature at relative pressures between 0.05 and 0.2. In order to estimate the cobalt dispersion (D, %) and the cobalt crystallite size (d, nm) of the Co-catalyst, hydrogen static chemisorption was performed on the reduced catalyst. The chemisorption measurement was conducted on a Micromeritics ASAP 2020C unit at 308 K, after reducing about 0.15 g of the calcined catalyst in flowing hydrogen at 623 K (heating rate: 1 K/min) for 16 h according to the procedure described in refs. [20, 25]. The average particle size of Co0, assuming spherical shape, was estimated according to [26–28]: 96 DOR d Co0 H ¼ D
ð1Þ
where DOR is the degree of reduction. Hydrogen temperature-programmed-reduction (H2-TPR) was used to study the reducibility of the Co-catalyst. TPRs of the calcined Co-catalyst and of the reduced and passivated Co-catalyst (0.1–0.2 g) were conducted on a Micromeritics Autochem 2910 flowing 5% H2 in Ar increasing the temperature from ambient to 1200 K (10 K/ min) while monitoring the H2 consumption with a thermal conductivity detector (TCD). The degree of reduction (DOR) was estimated from H2-TPR for the ex situ reduced and passivated Co-catalyst after activation and also for an in situ reduced Co-catalyst. The activation procedure of the ex situ reduced and passivated catalyst comprised reduction at 493 K (heating rate: 5 K/min) for 2 h in flowing H2. After the reduction/activation, the samples were flushed with inert gas for 30 min, the flowing gas was changed to 5% H2 in Ar and a standard TPR was performed. The TCD signal was calibrated with Ag2O as standard and the reported TPR profiles are normalized per gram catalyst. X-ray diffraction (XRD) measurements on the calcined catalysts were performed on a Siemens D5000 instrument with Cu–Ka radiation (2h = 10–908, step size = 0.028) equipped with a Ni filter. Average crystallite diameter of Co3O4 in the Co-catalyst was estimated by using the Scherrer formula and assuming spherical crystallites [29]. The Co crystallite size was estimated from that of Co3O4 using the following formula [30]: ð2Þ d Co0 ¼ 0:75 dðCo3 O4 Þ according to the relative molar volumes of metallic cobalt and Co3O4.
2.2 Catalyst Characterization Methods
2.3 Fischer–Tropsch Synthesis
Brunauer-Emmet-Teller (BET) surface area and porosity measurements were performed in a Micromeritics ASAP 2000 unit. The samples were evacuated and dried at 523 K
The FT synthesis was carried out in a down-flow stainlesssteel fixed-bed reactor (i.d. 9 mm) with a loading of about 1 g Co-catalyst (pellet size: 53–90 lm) mixed in different
123
Top Catal (2011) 54:977–985
979
proportions with the WGS-catalyst (pellet size: 53–90 lm) and diluted with 5 g SiC (average pellet size: 88 lm), resulting in a bed length of *5–7 cm. The small pellet size used for the Co-catalyst assures the absence of any mass transfer limitations on the observed Co-catalyst-time yields [31]. The different mechanical mixtures investigated were 0WGS-100FT, 15WGS-85FT, 33WGS-67FT, and 50WGS50FT, where the figures indicate wt% of catalyst in the mixture. The process conditions used were 483 K, 20 bar and an inlet H2/CO ratio of 1.0. The reactor tube was heated by means of an oven regulated by cascade temperature control with one sliding thermocouple in the catalyst bed and another one placed in the oven. This system, together with an aluminum jacket placed outside the reactor, allowed for an even temperature profile in the catalyst bed (483 ± 0.8 K). Prior to reaction, the catalyst (or catalyst mixture) was activated in situ in pure H2 (200 Ncm3/gCo-cat, min) at atmospheric pressure at 493 K for 2 h (heating rate: 5 K/ min). After activation, the reactor was cooled to 443 K and then flushed with He before increasing the pressure to the desired level. Then, the feed was switched to syngas, containing 3% N2 as internal standard. Subsequently, the temperature was slowly increased to 483 K (rate: 0.15 K/ min). The initial gas hourly space velocity (GHSV) was about 8200 Ncm3/gCo-cat, h (period A) and held for approximately 24 h in order to reach a (pseudo) steady state condition. Then, the GHSV was lowered to *1700– 3300 Ncm3/gCo-cat, h (period B) in order to increase the CO conversion and held for another 24 h. The heavy HCs and most of the water were condensed in two consecutive traps kept at 393 K and room-temperature, respectively. The product gases leaving the traps were depressurized and analyzed on-line by means of a gas chromatograph (GC) Agilent 6890 equipped with a TCD and a flame ionization detector (FID). H2, N2, CO, CH4, and CO2 were separated by a Carbosieve II packed column and analyzed on the TCD. C1–C6 products were separated by an alumina-plot column and quantified on the FID allowing for determination of the C5? selectivity (SC5?). The CO2-free SC5? (i.e., SC5? if excluding CO2 from the C-atom balance) is defined as follows: SC5 þðCO2 freeÞ ¼ 100 ðSC1 + SC2 + SC3 + SC4 ÞðCO2 freeÞ ð3Þ The reported HC selectivities are all C-atom based and CO2-free. Aqueous products were analyzed off-line with a GC Agilent 6890 equipped with a FID and a capillary column (Hewlett-Packard HP-5). An internal standard (CH3CN) was added to the aqueous phase prior to the injection.
The H2/CO usage ratio (u.r.) is used as a measure of the relative WGS activity and defined as follows: u:r: ¼
3 SC1 þ 2:1 SC2 C4 þ 2 SC5þ SCO2 100
ð4Þ
The lower the usage ratio, the higher is the relative WGS activity. If all water produced in the FT reaction is used in WGS, the usage ratio is 0.5. For a process to be able to reach 100% syngas conversion (theoretically), the inlet H2/CO ratio must equal the H2/CO usage ratio. 2.4 Preliminary Modeling As the experimental data set of the present study is limited, we have confined ourselves to perform broadbrush/preliminary fittings of three reaction rate expressions to the experimental data. This was done by interpreting the Co-catalyst-time yields (i.e., mol CO converted to HCs (mol/gCo-cat, s)) obtained in the integral reactor for the different WGS/FT mixtures as the FT reaction rates (rFT) at the ‘‘inlet–outlet’’ average partial pressures in the reactor, meaning that linear concentration profiles along the reactor were assumed. In order to estimate the partial pressures at the reactor outlet it was assumed that the C5? fraction of the HCs is present as a liquid. The presence of the WGScatalyst was assumed not to affect the kinetics of the Co-catalyst (which is in line with the findings of Chanenchuk et al. [9]), mainly since the increase in selectivity to alcohols (mainly methanol) with the addition of the WGScatalyst was found to be very low (see 3.2 Activity results) and as the WGS-catalyst used in the present study was free from traces of potential Co-catalyst poisons such as alkalimetals (see 2.1 Catalyst preparation). Furthermore, we assume that the cobalt surface is not contaminated with copper as both reduction and reaction temperatures are far below the Tamman temperature of Cu (679 K), making the migration of copper to cobalt improbable. The following three reaction rate expressions (models) with possible positive water effects were fitted to the experimental data points. Model 1 [17]: 3=2
rFT ¼
e PH2 PCO =PH2 O ð1 þ f PH2 PCO =PH2 O Þ2
ð5Þ
Model 2 [18, 19, 32]: rFT ¼
k P0:25 P0:5 CO H2 ð1 þ m PH2 O =PH2 Þ
ð6Þ
Model 3: 0:25 rFT ¼ a P0:5 PbH2 O H2 PCO
ð7Þ
Model 1 is a conventional Langmuir–HinshelwoodHougen-Watson (LHHW) expression derived from
123
980
mechanistic assumptions, e.g. that the hydrogenation of surface carbon is the rate limiting step [17]. It predicts that the kinetic effect of water on FT rate for Co-catalysts is positive at low partial water pressures, while turning to negative at higher partial pressures. This was suggested to be due to small amounts of water reducing the inhibition of the FT reaction by surface carbon, while larger amounts lead to a very low concentration of surface carbon and the rate decreases. However, it was found that Model 1 deviated significantly from the experimental data at high water partial pressures, predicting a too low FT rate for a Co/MgO/ThO2/SiO2 catalyst [17]. In Model 2, we have fixed the reaction orders in PCO and PH2 at -0.25 and 0.5, respectively, as our small data set does not contain enough information about the effect of CO and H2 on the FT rate. The values chosen are typically found for Co-based catalysts for CO and H2 partial pressures in our range, see for instance refs. [18, 19]. The original Model 2 [32] was a semi-empirical model derived for Pt-promoted Co/Al2O3 catalyst showing a negative ‘‘apparent’’ kinetic effect of water where m was defined as a ‘‘water effect parameter’’. The expression was based on a theory of competitive adsorption of water and CO and/or H2. In later publications [18, 19], the authors also fitted Model 2 to kinetic data for Co/SiO2 catalysts reacting positively to added water (m \ 0), which implies that Model 2 was used as a purely empirical expression. Model 3 is a simple power-law expression in which we also have fixed the reaction orders in PCO and PH2 . The constants e, k, and a in the models are measures of the intrinsic activity of the catalyst (i.e., *rate constants) and will depend on the number of active Co sites present in the catalyst and, hence, should decrease with timeon-stream (e.g., from period A to B) due to deactivation. The three models were fitted to the experimental data points (see Table 3) in period A, B and A ? B using the regression method of least squares by applying the function fminsearch in MATLABÒ V7.10 (R2010a). Some catalyst mixtures were run twice and therefore two and one additional data point(s) were used in the model fittings for period A and B, respectively.
Top Catal (2011) 54:977–985 Table 1 Textural properties of support and catalysts (calcined) as determined by N2 adsorption Sample
BET surface area (m2/g)
Pore volume (cm3/g)
Pore diametera (nm)
c-Al2O3
193
0.50
10.4
12Co/c-Al2O3
173
0.40
9.2
75
0.12
6.4
30Cu/30ZnO/Al2O3 a
The average pore diameter (dp) estimated from surface area (A) and pore volume (V) according to dp = 4 V/A
Table 2 shows the physicochemical properties of the Co-catalyst. The DOR after activation is similar for the ex situ reduced and passivated sample as for the in situ reduced sample, and hence the Co-particle size estimation for the ex situ reduced sample is similar to that for the in situ reduced sample. The TPR profiles of the Co-catalyst before and after the reduction and passivation treatments are shown in Fig. 1. The calcined catalyst (Fig. 1a) shows three peaks: peak I at 540 K has been attributed to removal of residual nitrates from the catalyst preparation [33]. Increasing the temperature, a two-step reduction of cobalt oxides [34] can be observed: peak II at 625 K is ascribed to reduction of Co3O4 to CoO, while the broad peak III at around 895 K corresponds to the reduction of differently sized CoO crystallites into Co. The TPR profile of the reduced and passivated sample (Fig. 1b) shows a sharp peak at around 539 K which could probably be ascribed to the reduction of the protective outer oxide layer (CoO) formed during passivation, and a much smaller peak III, which most probably is attributed to hard-to-reduce cobalt-alumina species that were not reduced under the preceding reduction at 623 K. The XRD pattern for this catalyst (see Fig. 2) shows that the only detectable phases (in addition to c-Al2O3) are CoO and Co, which is in agreement with the findings for a reduced and passivated Co/SiO2 catalyst [35]. For the calcined Co-catalyst, Co3O4 is the only detectable cobalt-containing phase. The phases identified in the co-precipitated WGS-catalyst are mainly copper and zinc spinels (CuAl2O4, ZnAl2O4), in agreement with the findings of Tanaka et al. [23]. Also copper oxide (CuO) is detected in the WGS-catalyst.
3 Results and Discussion 3.2 Activity Results 3.1 Characterization of the Supports and Catalysts Table 1 shows the physical properties of the pure support of the Co-catalyst, a narrow-pore c-Al2O3 (10 nm), and of the Co- and WGS-catalysts. The typical decrease in surface area, pore volume and pore diameter of the pure support upon impregnation is seen.
123
It should first be mentioned that the Co-catalyst-time yield and HC selectivities were similar for the ex situ reduced Co-catalyst as for an in situ reduced Co-catalyst, although the results of the latter is not reported here. The catalytic performances in period A (higher GSHV) and B (lower GHSV) are summarized in Table 3. As expected, by
Top Catal (2011) 54:977–985
981
Table 2 Physicochemical properties of the Co-catalyst Sample
DOR (%)
Metal dispersion H2-chemisorption
XRD d(Co3O4) (nm)
d(Co0)c (nm)
Dd (%)
Decorr (%)
d(Co0)fH (nm)
5.5
8.6
11.2
–
–
12Co/c-Al2O3
64a
16.8
13
12Co/c-Al2O3 red ? pass
62b
–
–
a
–
Estimated by H2-TPR after reduction at 623 K for 16 h in H2
b
Estimated by H2-TPR after activation at 493 K for 2 h in H2
c
Cobalt particle size according to Eq. (2)
d
Metal dispersion, after reduction at 623 K for 16 h in H2
e
Dispersion corrected for DOR
f
Cobalt particle size according to Eq. (1)
H2-uptake per gram catalyst (a. u.)
539
•
∗
898
(b)
12Co/γ-Al2O3
•
red+pass
•
∗
Intensity (a. u.)
625 895
540
(a)
♦
12Co/γ-Al2O3
♦
♦
♦
♦
calcined
o
200
400
600
800
1000
1200
Isothermal
+
Temperature (K)
30Cu/30ZnO/Al2O3 o
Fig. 1 H2-TPR profile for calcined Co-catalyst (a) and reduced and passivated Co-catalyst (b)
increasing the fraction of WGS-catalyst in the mixture an increase in CO2-selectivity and a decrease in usage ratio is generally seen for both period A and B. The usage ratios show only a weak dependence of GHSV, being around 2 for the pure Co-catalyst and about 0.65–0.75 for the WGScatalyst-rich mixtures. Interestingly, the addition of WGS activity, and thus an increase in average partial pressure of H2 and a decrease in average partial pressure of CO, does not correspond to an increase in the Co-catalyst-time yield (mol CO converted to HCs/gCo-cat, s) as predicted by traditional Co-catalyst rate expressions, but instead to a decrease. This finding is in contrast to what has been reported by Chanenchuk et al. [9] for a slurry reactor. In Fig. 3, a correlation between the average PH2 O inside the reactor and the Co-catalyst-time yield to HCs is shown, indicating that at the present process conditions water has a larger positive effect on the FT rate than has hydrogen.
calcined
+
o o
o
0
20
40
60
80
100
2θ Fig. 2 XRD diffractograms for calcined 30Cu/30ZnO/Al2O3, calcined 12Co/c-Al2O3 and reduced and passivated 12Co/c-Al2O3. Phases: (open circle) CuAl2O4/ZnAl2O4; (?) CuO; (open square) c-Al2O3; (closed diamond) Co3O4; (closed circle) CoO; (*) Co0
From Table 3 it can be seen that there is no correlation between the fraction of WGS-catalyst in the mixture and the HC selectivites. One would expect the SC5? to decrease with increased WGS activity as the H2/CO ratio in the reactor is higher [27] and the water partial pressure is lower [20]. These findings are, however, in agreement with those of Chanenchuk et al. [9] who reported similar selectivities for a pure Co-catalyst as for a 67WGS-33FT mixture. There is, however, a clear change in product distribution in terms of olefin to paraffin ratio (o/p). The o/p for C3 drastically decreases with increased WGS activity (see
123
0.1
0.2 0.65
0.71
5.0 45.6
88.1
5.0 43.9
89.5
2.05 1.49 5.3 6.2 0.5 18.8
88.5 86.7
1.72 Selectivities are CO2-free
2.22 1.22
1.07 17.3
43.5 1657
3309 0.3
0.6 0.74
0.76 89.3
88.8 5.2
4.2 41.0
43.2 2.06
1.81 1.02
1.02
8318 50WGS-50FT
6.5
8287 33WGS-67FT
8.0
3.45 2.33 0.75 0.91 19.3 15.2 3245 3280 4.3 2.8 2.04 1.58 87.3 90.8 6.0 4.1 0.8 14.8 2.95 2.25 0.93 0.96 8153 8155 0WGS-100FT 15WGS-85FT
6.5 5.8
GHSV (Ncm3/ gCo-cat, h)
a
u.r. (%) SaC5? (%) SaC1 (%) SCO2 (%) Mol CO conv. to HCs 9 106 (mol/gCo-cat, s) Outlet H2/CO XCO (%) GHSV (Ncm3/ gCo-cat, h) (%) (%)
Table 3). This has been reported to be due to that olefins undergo secondary hydrogenation over Cu-based catalysts [9], which seems as a reasonable explanation since the hydrogenation activity of the Co-catalyst, apparently, is invariant in the range of the studied process conditions. Since the used WGS-catalyst is also a methanol synthesis catalyst, it is worth to mention that the C-atom selectivities to alcohols (in particular to methanol) were below 1% for all mixtures at the studied process conditions.
3.3 Preliminary Modeling
Sample
XCO (%)
Outlet H2/CO
Mol CO conv. to HCs 9 106 (mol/gCo.cat, s)
SCO2 (%)
SaC5? SaC1
u.r. (%)
o/p C3
Period B Period A
Catalytic performances
Table 3 Activity and selectivity data for periods A (end of) and B (after stabilized conversion). Experimental conditions: 483 K, 20 bar, inlet H2/CO = 1.0
123
4.1 2.4
Top Catal (2011) 54:977–985 o/p C3
982
The results of the model fittings are summarized in Table 4, and in Fig. 4 the resulting rate expressions obtained from experimental data in period A and B, respectively, are plotted versus PH2 O , fixing PH2 and PCO at 9.5 bar each. Also the experimental data points are included in Fig. 4, but it should be mentioned that PH2 and PCO for these deviate somewhat from 9.5 bar. Model 1 provides the poorest fittings as seen from the high residuals in Table 4. The rate constant e increases from period A to B, which makes no practical sense as we would rather expect a deactivation of the catalyst. Also parameter f, being a constant in a LHHW expression, which should be more or less constant for a wide range of process conditions, increases significantly. This suggests that Model 1 does not describe our data satisfactorily. The best fit is obtained with Model 2 when using data from period A and B separately (see Table 4). When combining the data from period A and B the residual of this model fitting increases significantly, suggesting that the low residuals obtained when treating period A and B separately are merely a coincidence. This is further strengthened by noticing that the water effect parameter m changes value drastically from period A to B, implying a much stronger water effect for the rather fresh catalyst (i.e., in period A) while, at the same time, the rate constant k is unchanged, which does not make any practical sense. An unchanged k should indicate that no deactivation has taken place from period A to B, which in turn should imply that the catalyst shows the same reactivity or sensitivity to water. The fact that PH2O in period B ranges to higher values than in period A could, of course, affect the obtained water effect, but such an effect should reasonably be moderate. Our recently published results on a Re-promoted Co/c-Al2O3 catalyst [12] suggests that the positive kinetic water effect for these narrow-pore c-Al2O3-supported catalysts is leveling off at higher partial pressures and does not increase as predicted by Model 2. The fittings of Model 3 show intermediate residuals (see Table 4). The rate constant a decreases between period A and B, while the water effect (i.e., b) is rather unchanged,
983
3.0
0.18 2.5
0.16 0.14
2.0
0.12 1.5
0.10 0.08
1.0
0.06 0.04
PH2O, avg (bar)
0.5
0.02 0.0
0.00 33
50
Period B 3.0
0.20 0.18
2.5
0.16 0.14
2.0
0.12 1.5
0.10 0.08
1.0
0.06 0.04
PH2O, avg (bar)
Model 1 4 3 2 Exp. A Exp. B Mod. A Mod. B
1 0
5
Model 2 4 3 2 Exp. A Exp. B Mod. A Mod. B
1 0
0.5
0.02
6
Co-catalyst-time yield of HCs (g/gCo-cat, h)
5
6
15
Mol CO converted to HCs ∗ 10 (mol/gCo-cat, s)
0
Weight percent of WGS-catalyst in the mixture
0.0
0.00 0
15
33
50
Weight percent of WGS-catalyst in the mixture Fig. 3 Co-catalyst-time yields and reactor-average PH2O versus weight percent of WGS-catalyst in the mixtures in period A (upper) and B (lower). Experimental conditions: 483 K, 20 bar, H2/CO = 1.0
which is as would be expected for a rate expression that describes our data fairly well. Figure 5 shows the parity plots for Model 3 fitted to experimental data from period A and B separately. The rather high positive reaction order in PH2 O (*0.3) is interesting and explains why in a fixed-bed reactor the removal of indigenous water by addition of a WGS-catalyst may result in a lower Co-catalyst-time yield to HCs, while in a well-mixed slurry reactor (in which the
Mol CO converted to HCs ∗ 10 (mol/gCo-cat, s)
Co-catalyst-time yield of HCs (g/gCo-cat, h)
Period A 0.20
Mol CO converted to HCs ∗ 10 (mol/gCo-cat, s)
6
Top Catal (2011) 54:977–985
5
Model 3 4 3 2 Exp. A Exp. B Mod. A Mod. B
1 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
PH2O (bar) Fig. 4 Model predictions (and experimental data points) of the Co-catalyst-time yield of HCs (mol CO/gCo-cat, s) versus PH2 O for constant PH2 and PCO (9.5 bar each). Model 1 (top), Model 2 (center), Model 3 (bottom)
Table 4 Obtained constants and residuals upon fitting of three reaction rate models to the experimental data Model 1
Model 2 b
Constants
6
Model 3 b
RES 9 10 (mol/gCo-cat, s)
Constants k 9 106 (mol/(gCo-cat, s, bar0.25))
m (-)
Period(s)a
e 9 106 (mol/(gCo-cat, s, bar1.5))
f (bar-1)
A
0.012
0.0029
0.19
1.01
-11.4
B A?B
0.066 0.026
0.0150 0.0060
0.22 0.22
1.03 1.20
-3.7 -3.2
a
6
RES 9 10 (mol/gCo-cat, s)
Constants
RESb 9 106 (mol/gCo-cat, s)
a 9 106 (mol/(gCo-cat, s, bar0.25?b))
b (-)
0.032
2.41
0.28
0.10
0.045 0.140
1.89 1.95
0.32 0.22
0.09 0.11
Experimental data points from indicated period(s) were used in the regression
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi P 2 ðExpðnÞCalcðnÞÞ
b
Indicates the average residual of the n experimental data points according to RES ¼
n
n
123
984
Model 3
6
Calculated mol CO converted to HCs ∗ 10 (mol/gCo-cat, s)
Fig. 5 Parity plots for Model 3 in period A and B, respectively
Top Catal (2011) 54:977–985
4.0
Period A
Period B
3.5
3.0
2.5
2.0
1.5
1.0 1.0
1.5
2.0
2.5
3.0
3.5
4.0
1.0
1.5
2.0
2.5
3.0
3.5
4.0
6
Experimental mol CO converted to HCs ∗ 10 (mol/gCo-cat, s)
average partial pressure of water is relatively high) addition of a WGS-catalyst may result in an increased productivity of HCs due to the higher reaction order in PH2 and also due to the negative reaction order in PCO. Finally, it should be commented that the FT reaction rate at PH2 O = 0 in reality is higher than zero (as given by Model 3 (and Model 1)), since the reaction would otherwise never occur without external water addition. This means that the fitted Model 3 (i.e., with the constants given in Table 4) probably would underestimate the Co-catalysttime yields at partial pressures of water lower than the average PH2 O values obtained in the present study. A better fit of Model 3 for low water partial pressures would, however, be achieved if a reactor model (e.g., a plug-flow reactor (PFR) with inlet PH2 O * 10-6 bar) was included in the regression instead of using the ‘‘inlet–outlet’’ average partial pressures of H2, CO and water.
4 Conclusions A narrow-pore Co/c-Al2O3 catalyst was mixed in different proportions with a WGS-catalyst in a fixed-bed reactor with the initial intention to increase the production rate of HCs by increasing the partial pressure of hydrogen and decreasing that of CO inside the reactor. The usage H2/CO ratio could be adjusted to match the inlet H2/CO ratio of 1.0 by choosing the correct amounts of the two catalysts. However, the Co-catalyst-time yields to HCs of the WGScatalyst-containing mixtures were lower than for the pure Co-catalyst. This is ascribed to a strong positive kinetic effect of water on the FT rate of the Co-catalyst, showing the importance of the indigenously produced water, especially in fixed-bed reactors where the partial pressure of water is zero at the reactor inlet.
123
Work in progress comprises inclusion of a PFR model in the reaction rate expression regressions, as well as expanding the experimental data set with initial rate data obtained for broader partial pressure (of H2, CO, and H2O) ranges. Acknowledgment The authors acknowledge the financial support provided by the Swedish Energy Agency and by SGC (Svenskt Gastekniskt Center). The authors also want to express their gratitude to Truls Liliedahl, Chemical Technology, for fruitful discussions.
References 1. Berglin N, Berntsson T (1998) Appl Therm Eng 18:947–961 2. Dry ME (2002) Catal Today 71:227–241 3. Johansson M (1998) Catalytic combustion of gasified biomass for gas turbine applications. KTH-Royal Institute of Technology, Stockholm 4. Wender I (1996) Fuel Process Technol 48:189–297 5. Dry ME (1996) Appl Catal A 138:319–344 6. Botes FG, Breman BB (2006) Ind Eng Chem Res 45:7415–7426 7. Yates IC, Satterfield CN (1991) Energy Fuels 5:168–173 8. Newsome DS (1980) Catal Rev Sci Eng 21(2):275–318 9. Chanenchuk CA, Yates IC, Satterfield CN (1991) Energy Fuels 5:847–855 10. Post MFM, Sie ST (1985) E.P. Application 11. Post MFM, Sie ST (1985b) E.P. Application 12. Lo¨gdberg S, Boutonnet M, Walmsley JC, Ja¨ra˚s S, Holmen A, Blekkan EA (2011) Appl Catal A 393:109–121 13. Outi A, Rautavuoma I, van der Baan HS (1981) Appl Catal 1:247–272 14. Sarup B, Wojciehowski BW (1989) Can J Chem Eng 67:620–627 15. Laan GPVd, Beenackers AACM (1999) Catal Rev Sci Eng 41:255–318 16. Tsakoumis NE, Rønning M, Borg Ø, Rytter E, Holmen A (2010) Catal Today 154:162–182 17. Van Steen E, Schulz H (1999) Appl Catal A 186:309–320 18. Das TK, Conner WA, Li J, Jacobs G, Dry ME, Davis BH (2005) Energy Fuels 19:1430–1439
Top Catal (2011) 54:977–985 19. Ma W, Jacobs G, Sparks DE, Gnanamani MK, Pendyala VRR, Yen CH, Klettlinger JLS, Tomsik TM, Davis BH (2011) Fuel 90:756–765 20. Storsæter S, Borg Ø, Blekkan EA, Holmen A (2005) J Catal 231:405–419 21. Balonek CM, Lillebø AH, Rane S, Rytter E, Schmidt LD, Holmen A (2010) Catal Lett 138:8–13 22. Duvenhage DJ, Coville NJ (2005) Catal Lett 104:129–133 23. Tanaka Y, Utaka T, Kikuchi R, Sasaki K, Eguchi K (2002) Appl Catal A 238:11–18 24. Ayastuy JL, Gutie`rrez-Ortiz MA, Gonza´lez-Marcos JA, Aranzabal A, Gonza´lez-Velasco JR (2005) Ind Eng Chem Res 44:41–50 25. Lo¨gdberg S, Lualdi M, Ja¨ra˚s S, Walmsley JC, Blekkan EA, Rytter E, Holmen A (2010) J Catal 274:84–94 26. Borg Ø, Dietzel PDC, Spjelkavik AI, Tveten EZ, Walmsley JC, Diplas S, Eri S, Holmen A, Rytter E (2008) J Catal 259:161–164 27. Tristantini D, Lo¨gdberg S, Gevert B, Borg Ø, Holmen A (2007) Fuel Process Technol 88:643–649
985 28. Jones RD, Bartholomew CH (1988) Appl Catal 39:77–88 29. Lemaitre JL, Menon PG, Delannay F (1984) In: Delannay F (ed) Characterization of heterogeneous catalysts. Marcel Dekker, Inc, New York 30. Schanke D, Vada S, Blekkan EA, Hilmen AM, Hoff A, Holmen A (1995) J Catal 156:85–95 31. Borg Ø, Eri S, Blekkan EA, Storsæter S, Wigum H, Rytter E, Holmen A (2007) J Catal 248:89–100 32. Li J, Zhan X, Zhang Y, Jacobs G, Das T, Davis BH (2002) Appl Catal A 228:203–212 33. Lo¨gdberg S, Tristantini D, Borg Ø, Ilver L, Gevert B, Ja¨ra˚s S, Blekkan EA, Holmen A (2009) Appl Catal B 89:167–182 34. Borg Ø, Rønning M, Storæster S, van Beek W, Holmen A (2007) Stud Surf Sci Catal 163:255–272 35. Huber F, Yu Z, Lo¨gdberg S, Rønning M, Chen D, Venvik H, Holmen A (2006) Catal Lett 110:211–220
123