400 500 600 700 800 900 1000 Reforming equilibrium temperature, (°C)
0%
20%
40%
60%
80%
0%
20%
40%
60%
80%
100%
Methane conversion, (%)
2 MPa 0.1 MPa
S/C = 5.0 S/C = 2.5 S/C = 1.0
S/C = 5.0 S/C = 2.5 S/C = 1.0
FIGURE 4.1 Steam reforming and methane conversion; Reprinted from Rostrup-Nielsen et al. [11], Copyright (2002), with permission from Elsevier.
TABLE 4.1 Reforming Reactions
Steam reforming DHo298; kJ molL1
lnKp[ a D b/T$
a b
1. CH4þ H2O¼ CO þ 3H2 206 30.446 27106
2. COþ H2O¼ CO2þ H2 41 3.798 4160
3. CH4þ CO2¼ 2CO þ 2H2 247 34.244 31266
4. CnHmþ nH2O¼ nCO þ n þ m2ÞH2
Ethane 347 60.42 45256
n-heptane 1107 21.053 141717
5. CH4¼ C þ 2H2 75 12.69% 10779%
6. 2CO¼ C þ CO2 172 21.09 20486
7. COþ H2¼ C þ H2O 131 17.29 16326
$Standard state: 25C, 0.1 MPa, linear regression 500e900C.
%Whisker carbon.
STEAM REFORMING OF NATURAL GAS 51
which catalyst tubes are placed in a fired furnace supplying the required for the desired conversion. The catalyst tubes have a typical diameter of 10 cm and a length of about 10 m.
They are made of high alloy steel to withstand the high tube wall temperatures (typically close to 1000C) and the high thermal stresses caused by the high heat fluxes amounting to average fluxes of 0.1 MW/m2. The industrial break-through for the process 50 years ago was a result of design for operation at high pressure (2e 4 MPa), thereby reducing the energy consump-tion for compression of the large volume of the syngas product. Today, hydrogen plants are built for hydrogen production of about 200,000 Nm3/h with 400e500 reformer tubes. A hydrogen plant reformer is shown inFig. 4.2.
The product gas leaves the reformer tubes close to thermodynamic equilibrium. This
makes it easy to predict the operation and to calculate the need for heat supply from a simple enthalpy balance[1,10].
4.2.3. Nickel-Based Catalysts
Steam reforming catalysts are typically based on nickel. With these catalysts, the gas composi-tion in the reformer tube quickly approaches equilibrium [1,11,12], so that the driving force for the reforming reaction is the slope of the temperature profile as the methane content at equilibrium decreases with increasing tempera-ture. In commercial reactors, there is a large surplus of catalyst activity in a tubular reformer, because the tube dimensions and the number of tubes are determined by the maximum allowable heat flux and other mechanical considerations, leaving the catalyst volume as a dependent variable[1,8]. Still, there is a need for high catalyst activity because high activity means that the heat transfer for a given conver-sion can take place at a lower tube wall temper-ature[10].
The catalyst must ensure equal distribution of flow between the tubes in a tubular reformer and across the catalyst bed in an adiabatic reformer. Maldistribution may cause local over-heating or channeling and hence shorter life of the reformer tubes. The catalyst should be mechanically stable under all process conditions as well as conditions during start-up and shut-down of the reformer. In particular, resistance to conditions during upsets may become crit-ical. Breakdown of catalyst pellets may cause partial or total blockage of some tubes resulting in severe temperature gradients within or between tubes [1,10]. The formation of carbon may result in the same problems.
4.2.3.1. Support
The support must show high mechanical stability because of exposure to high tempera-tures well above 800C and steam partial pres-sures close to 3 MPa [1,12], involving steam FIGURE 4.2 Tubular reformer in hydrogen plant.
4. STEAM REFORMING FOR FUEL CELLS
52
partial pressures and temperatures without losing strength. Furthermore, it should not contain volatile components. High area supports such asg-alumina, chromia, etc., can be used for catalysts for low-temperature adiabatic reform-ing, but these supports suffer from substantial sintering and weakening at temperatures above 500C. The deterioration is strongly accelerated by the high steam partial pressure and stability tests at atmospheric pressure can therefore be misleading. Because silica is volatile (Si(OH)4) at high temperatures in high pressure steam, it is now excluded from catalysts for steam reform-ing[1,12], unless it is combined with alkali. Silica would be slowly removed from the catalyst (or brickwork) and deposited in boilers, heat exchangers, and catalytic reactors downstream of the reformer.
4.2.3.2. Promoters/Alloys
Alkali metals are used as promoters to elimi-nate carbon formation, but are slowly volatil-ized from the catalyst. The alkali loss is enhanced by high temperature, but may to some extent be controlled by the addition of acidic components (silica). The volatilized alkali may deposit in colder parts of the plant where the resulting hydroxyl ions will strongly promote stress corrosion in stainless steel. More-over, alkali may react with some catalyst support materials such as alumina forming the weak b-alumina, resulting in a decrease of the mechanical strength. While resistant to high temperature, catalysts based on magnesia are sensitive to steaming at low temperatures because of the risk of hydration. The reaction may result in breakdown of the catalyst because it involves an expansion of the molecular volume.
Alloying nickel with groups 1B metals (Cu, Ag) may cause a drastic decrease of the activity [18]although small amounts (1 wt%) of copper may cause the activity to increase.
At steaming conditions at high temperature, formation of the nickel aluminum spinel may
start at temperatures above about 700C[1,10], but a less well-defined interaction between nickel oxide and h- or g-alumina is apparent already at lower temperatures. It is possible to form a “surface spinel” below 600C, which may hardly be identified by X-ray methods alone. A similar trend is observed for the reaction between nickel and magnesium oxide [1,10].
4.2.3.3. Catalyst Particles
The shape of the catalyst pellet should be optimized to achieve maximum activity with minimum pressure drop. The pressure drop depends strongly on the void fraction of the packed bed and decreases with increasing particle size. Similarly, the effective catalyst activity per volume is roughly proportional to the surface to volume ratio of the catalyst pellet due to the low catalyst effectiveness factor at conditions of practical interest (typically below 0.1 for a tubular reformer)[1,10].
In order to fulfill the above-mentioned requirements to the catalyst, the particles are usually made of ceramic material of cylindrical shape with one or more internal holes. Solutions based on the use of ceramic foam as catalyst support have also been considered[13]together with other shapes such as monoliths and even catalyzed hardware [14,15]. The use of such catalyst formulations are often developed together with new reactor concepts taking advantage of the surplus catalyst activity in conventional steam reformers.
4.2.3.4. Activation
The catalyst in industrial plants can be acti-vated by various reducing agents such as hydrogen, ammonia, methanol, and hydrocar-bons added to steam [1]. The reaction with hydrogen is nearly thermoneutral and accord-ingly, the equilibrium constant, Kp ¼ pH2O= pH2varies little with temperature. Metallic nickel will be stable with approximately 0.3 and 0.6 vol% hydrogen in steam at 400C and 800C,
STEAM REFORMING OF NATURAL GAS 53
respectively. In practice,Kpmay be lower, as the free energy of nickel oxide decreases due to interaction with the support material. The reduction of pure nickel oxide by hydrogen starts at temperatures 200e250C depending on the calcination temperature. Supported cata-lysts require higher temperatures to show a reasonable reduction rate. This may be ascribed to interaction with the support. When the formation of nickel aluminum spinel has taken place, temperatures above 800C [10]
may be required for complete reduction.
4.2.3.5. Nickel Dispersion
The activity of a nickel catalyst is related to the nickel surface area. It is important to refer reaction rates to unit surface area in terms of the specific activity or turnover frequency (TOF) in order to compare catalysts and to analyze catalytic phenomena. Sintering of the nickel crystals results in no significant change in TOF. However, for small nickel particles (dNi
< 10 nm), there is a significant increase of TOF with increased dispersion[19].
The most widely used method for measure-ments of the nickel surface area is chemisorp-tion (of hydrogen [16] or hydrogen sulfide [10]). The average nickel particle size (dNi; in nm) and the nickel dispersion (Disp; in %) can be estimated from the nickel surface area:
dNi ¼ 6.8 (Ni content (wt%))/(Ni area (m2/g))
Disp¼ 101/dNi
The TOF (molecules/site/s) may be calcu-lated from:
TOF ¼ 10.9 (rate (mol/g/h))/(Ni area (m2/g))
The nickel surface area is generally increased with higher nickel contents in the catalyst[10], but the dispersion or utilization of the nickel tends to decrease with increasing nickel content.
Accordingly, many commercial catalysts are optimized at nickel contents around 15 wt%.
Through special preparation methods, it is possible to obtain high dispersions even at
high nickel contents, but because of sintering effects at high nickel loadings practice often shows an optimum in nickel area depending on the nickel content.
4.2.3.6. Sintering
Most tubular reformers operate above the Tammann temperature of nickel (TTammann¼
1/2Tmelt ¼ 581C), at which nickel is prone to sintering, which is the growth of the nickel particles and a decrease of nickel surface area and of activity. However, the sintering phenom-enon is more complex and may take place below the Tammann temperature[17]as well.
Sintering is influenced by many parameters, including temperature, chemical environment, catalyst composition and structure, and support morphology. The most important parameters are the temperature and the atmosphere in contact with the catalyst. The presence of water greatly accelerates sintering.
One mechanism considers particle migration and coalescence, where particles move over the support and collide, forming larger crystals with a loss of overall metal surface area. This mechanism dominates at low temperatures. At high temperatures, particle coarsening proceeds via transport of metal atoms or small agglomer-ates over the carrier and between metal particles (atom migration and Ostwald ripening). The time dependencies of crystal growth are t1/7 and t1/3for particle migration and coalescence and atom migration, respectively. The transition temperature between particle migration and Ostwald ripening depends on process condi-tions (H2O/H2, etc.) and the catalyst composi-tion[1,17].
4.2.4. Non-Nickel Catalysts
The group VIII metals are active for the steam reforming reaction[20,21]. The TOF is typically 2e5 s1 at 500C for steam reforming of methane on nickel (H2O/CH4¼ 4. H2O/H2 ¼ 10, 0.1 MPa [10]). Most of the group VIII noble
4. STEAM REFORMING FOR FUEL CELLS
54
metals have higher Tammann temperatures than nickel and should be more resistant to sin-tering. All group VIII metals show a significant decrease in TOF when the catalyst contains alkali[1,22]. It means that even traces of alkali metal may blur TOF measurements if not recog-nized. The deactivation depends strongly on the type of support, the effect being less on acidic supports that bind alkali more strongly. This all hints that alkali works on nickel via the gas phase. All group VIII metal catalysts are sensi-tive to sulfur poisoning (refer Section 4.2.7).
Cobalt shows a lower activity than nickel [10]e probably attributable to the process condi-tions, with the H2O/H2 value being close to levels that cause oxidation of the metal. Iron is active for steam reforming, but only at strongly reducing conditions. Rhodium and ruthenium are the most active and nickel the most preferred because of its cost. Rhodium and ruthenium show TOF values about ten times higher than nickel, platinum, or palladium. Most studies [10,19,20]find a sequence of activity (TOF); Ru, Rh > Ni, Ir > Pt, Pd. This sequence has also been confirmed by DFT calculations[19].
4.2.5. Non-Metal Catalysts
Difficulties in desulfurizing heavy feedstocks have led to attempts to use non-metallic catalyst for steam reforming. However, these non-metal catalysts have far less activity.
4.2.5.1. Ceria
Ceria has an activity almost two orders of magnitude less than that of nickel [23,24].
However, the low activity might be sufficient for the reforming process [25] in a high-temperature fuel cell operating at 800C, where the balance between the rate of reform-ing and the rate of the electrochemical reaction is critical [22]. With an activation energy of 105 kJ/mol, the ratio of rates at 800C and 500C is about 96.
4.2.5.2. Carbides
Molybdenum carbide and tungsten carbide are catalysts for steam and CO2 reforming and catalytic partial oxidation (CPOX) [26,27].
However, in synthesis gas, the carbides are stable only at elevated pressures (approximately 0.9 MPa) and they are transformed into the oxides at ambient pressure. Molybdenum car-bide will hardly be stable in a plug flow reactor [26], because the carbide will be oxidized at the inlet.
4.2.5.3. Non-Catalytic Reforming
Non-catalytic steam reforming requires high temperatures. Methane cracks above 1000C into radicals leading to the formation of ethylene, acetylene, and coke [28]. These radicals may react with steam radicals, but temperatures above 1500C are necessary for significant conversion[29]. One approach to improve rates is the use of plasma technology (see Chapter 8 for more details) [8], the key issue being the power consumption. As thermal plasma reform-ing is not sensitive to sulfur poisonreform-ing, it may represent a solution for steam reforming of logistic fuels (jet fuel and diesel)[30].
The thermal cracking of higher alkanes becomes significant above 650C [31,32] with the formation of alkenes, aromatics, and coke.
This is applied in steam crackers in ethylene plants where steam is added as a diluent and for minimizing coke formation. There have been attempts to improve the steam cracking process by installing a catalyst (K/ZrO2) in the cracking tubes [1,32]. This resulted in co-production of syngas and light alkenes from heavy gas oil and naphtha.
4.2.6. Mechanism and Kinetics
The mechanism of steam reforming of methane is well described by recent funda-mental studies[19,33]. DFT calculations[34,35], adsorption studies [11,34], and observations by in situ high-resolution transmission electron
STEAM REFORMING OF NATURAL GAS 55
microscopy (HRTEM)[36]have shown that step sites play a key role in methane activation. This is illustrated by the energy diagram for the indi-vidual reaction steps[34] shown inFig. 4.3 on the dense Ni(111) surface and the stepped Ni (211) surface.
Methane may be activated on any single surface atom[19,35], but stabilizes on high coor-dination sites at steps having a lower barrier for the further dissociation of methane. The activa-tion of water does not depend strongly on the coordination number of the surface metal atom [35].
For both surfaces, the highest barrier is the surface reaction to carbon monoxide which may indicate a two-step mechanism involving both the CH4 activation as well as the CO desorption[10,11,19]:
CH4D/CHxD1=2ð4LxÞH2
CHxDOH/COD1=2ðxD1ÞH2D2 (4.2) H2OD[ OHD1=2H2
This was supported by an expanded analysis [19]by using the free energy (and not the total energy (enthalpy) as used in Fig. 4.3). It was shown for nickel that the barrier for formation decreases with temperature indicating that
a one-step mechanism may dominate at high temperatures. This trend is reflected by kinetic studies.
In general, the kinetics of the steam reforming reaction is found to be first order with respect to methane partial pressure [1,11,37e39] with an activation energy in the range of 100e120 kJ/mol.
However, at lower temperatures (<550oC), a denominator term may result in an overall zero dependency of total pressure[11]. One example [38] is a complex LangmuireHinshelwood expression, using a classical approach for the steam reforming reaction:
Rint;1 ¼ kint;1 p2:5H2$Z2
$ pCH4$pH2Op3H
2$pCO
Keq;1
! (4.3)
in which,
Z ¼ 1 þ Ka;CO$pCOþ Ka;H2$pH2þ Ka;CH4$pCH4
þ Ka;H2O$ pH2O
pH2
It has been shown that the denominator term depends on the composition of the catalyst[10].
The presence of alkali results in a large adsorp-tion term for steam [10]. Alkali also results in a significant decrease of the TOF, probably because alkali atoms block the step sites[11].
There is a general agreement that CO2
reforming on nickel can be described by the same kinetics as steam reforming [36,39e43].
The change in mechanism with carbon dioxide instead of steam would have little practical impact on reforming, because steam will be present not far from the inlet, but also in the center of the catalyst particle as a consequence of the low effectiveness factors of catalysts in industrial reformers. CO2 reforming results in lower atomic ratio of H/C, which means a higher risk of carbon formation (refer Section 4.2.8). CO2 reforming is of less interest for the manufacture of hydrogen, but it may be applied when converting biogas (CO2 and CH4) for fuel cells.
FIGURE 4.3 Energy diagram for steam reforming of methane. DFT calculations; Reprinted from Bengaard et al.
[34], Copyright (2002), with permission from Elsevier.
4. STEAM REFORMING FOR FUEL CELLS
56
4.2.7. Sulfur Poisoning
The group VIII metals are subject to sulfur poisoning with nickel being most sensitive[2].
The H2S/Ni system is well described in terms of a two-dimensional sulfide [2]. The sulfur coverage can be estimated from the expression [10,44]:
qs ¼ 1:45 9:53$105$T þ 4:17$105$T ln
pH2S
pH2
!
(4.4)
As an example, a sulfur coverage of 0.5 at 500C corresponds to H2S/H2 ¼ 1.6 1012. In practice, this means that all sulfur is retained at this temperature. At 850 oC, a ratio H2S/
H2 ¼ 1 106 corresponds to a coverage of 0.7. The sulfur coverage is independent of the H2O/H2 ratio[10]. The chemisorption process is reversible, but normally the driving force for desorption is low[10]. This means that the cata-lyst life is dependent on the efficiency of the desulfurization of the feedstock.
Catalytic partial oxidation (CPOX) over noble metals is less sensitive to sulfur poisoning[45].
In the presence of oxygen, sulfur is oxidized to SO2which is not adsorbed on the catalyst. Nickel will be oxidized in the presence of oxygen, but rhodium, a typical catalyst for CPOX, will not.
This means that rhodium stays active as long as oxygen is present. After depletion of oxygen, SO2will be reduced to H2S, which will be chem-isorbed on the catalyst as well as on downstream catalysts and anodes.
Removal of hydrogen sulfide and lower mercaptans is easily accomplished over zinc oxide, whereas more heavy sulfur components require hydrodesulfurization (HDS) over CoMo catalysts. The most difficult sulfur components are dibenzylthiophenes [46] as illustrated in Fig. 4.4.
It is evident that even after deep desulfuriza-tion, the dibenzylthiophene remains almost untouched. The amount of this component
increases strongly with the final boiling point of the feedstock, which corresponds to a signifi-cant difference between difficulty of desulfur-ization of kerosene and various diesel fuels.
Special CoMo catalysts and high hydrogen partial pressure are required to obtain desulfur-ization below 1 ppm sulfur[46].
4.2.8. Carbon Formation
Carbon formation may take place by different routes[1,10]:
e whisker carbon e gum formation e pyrolytic coke
Whisker carbon is formed on nickel as fibers from carbon monoxide, methane, and higher hydrocarbons. They grow from a nickel crystal with a diameter close to that of the nickel crystal [1,36]as illustrated inFig. 4.5.
FIGURE 4.4 HDS of diesel. Gas chromatogram of desulfurized product; Reprinted from Cooper and Knudsen [46], Copyright (2006), with permission from Springer.
STEAM REFORMING OF NATURAL GAS 57
The fibers are strong and may result in a breakdown of the catalyst pellet. This may have a serious impact on the operation of the tubular reactor as the carbon and the broken catalyst pellets may result in maldistribution of feed and overheating of the reformer tubes [1,10]. As shown in Fig. 4.3, adsorbed carbon atoms are most stable on the Ni(211) surface [39,47]. This was confirmed by HRTEM studies [36]showing that the nucleation of carbon takes place at step sites. The nickel particle size has an impact on the nucleation of carbon. The smaller the crystals, the more difficult is the initiation of carbon formation. The carbon formation depends on the kinetic balance between the surface reaction of the adsorbed hydrocarbon with oxygen species and the further dissociation of the hydrocarbon into adsorbed carbon atoms, which can nucleate into solid carbon.
The rate of carbon formation is far less on noble metals than on nickel[42]. The whisker growth mechanism is also blocked by sulfur poisoning of the nickel surface[48].
For the reversible decomposition of carbon monoxide and methane, the potential for carbon can be assessed by the “principle of equilibrated
gas” [1,10] as a simple guideline using simple thermodynamic calculations.
Principle of equilibrated gas: Carbon formation is to be expected on a nickel catalyst if the gas composition corresponds to one with a thermo-dynamic driving force for carbon after the estab-lishment of the methane reforming and the shift equilibrium. This implies that the composition used to calculate the equilibrium corresponds to the composition after accounting for the reforming and shift equilibrium.
The thermodynamic properties used to calcu-late the driving force for carbon should be for the whisker carbon and not graphite. This implies smaller equilibrium constants for the two decomposition reactions (Reactions 5 and 6 inTable 4.1).
When using the decomposition of methane (Reaction 5 inTable 4.1) this means:
DG ¼ RT$In ðKp$
pCH4=pH22
eq
(4.5)
in which Kp is the equilibrium constant for methane decomposition and (pCH4=pH22)eq are the partial pressures after equilibration of the reforming and shift reactions.
Carbon is predicted for DG > 0. As the surface energy of the whisker carbon increases with smaller whisker diameter and because this is related to the diameter of the nickel crystal, the equilibrium constants for the carbon forming reactions depend on the nickel crystal size.
The carbon limit is often expressed as temper-atures above which or below which there is a ther-modynamic potential for carbon formation. The carbon limits are a function of the atomic ratios O/C, H/C, and inert/C, and of total pressure.
The equilibrium calculations result in no simple upper or lower carbon limits. There may be upper and lower carbon limits for certain conditions[1].
In general, however, there is a tendency for higher carbon limits from steam reforming and lower carbon limits for CO2reforming.
One way of representing areas for potential for carbon formation is demonstrated in FIGURE 4.5 Whisker carbon on nickel catalyst.
4. STEAM REFORMING FOR FUEL CELLS
58
Fig. 4.6 [1,49]. The diagram illustrates that CO2
reforming is more critical than steam reforming.
For steam to carbon ratios applied in hydrogen plants (H2O/CH4 > 1.5), there is no risk of carbon formation. The carbon limit curve can be moved to the left when using noble metal catalysts, because the noble metals show signif-icantly smaller equilibrium constants for decomposition of methane and carbon monoxide.
The principle of equilibrated gas is not a “law of nature,” but an empirical guideline. It is possible to exceed what may appear to be the thermodynamic limit. This can be done by sulfur-passivated catalyst as practiced in the SPARG process [49]. The principle of equili-brated gas predicts conditions where carbon
formation is expected (except for noble metals and SPARG). It does not guarantee that carbon is not formed if the principle predicts no poten-tial in the equilibrated gas. Methane may decompose to carbon instead of reacting with steam to form the required syngas even when there is no potential for carbon in the equili-brated gas. This is of course not possible in a closed system, but in an open system carbon may be stable at steady state and the accumula-tion of carbon may continue [1,10]. This is the situation for steam reforming of higher hydrocarbons.