CC BY
52
УДК 536.4
P. Pfeifer, K. Haas-Santo, O. Gorke, J. Thormann, K. Schubert Германия
SMALL EFFICIENT MICROCHANNEL SYSTEMS FOR MOBILE AND DECENTRALISED HYDROGEN PRODUCTION
INTRODUCTION
Fuel cells are of great interest for mobile and decentralised generation of electricity to fulfil exhaust gas regulations and to save resources. Higher efficiencies and lower noise emission can be realised and pollution can be reduced. Hydrogen supply and storage are difficult due to low energy density of hydrogen leading to heavy tank systems. The production of hydrogen on site is an alternative to the hydrogen transport. However, a high efficiency and highly dynamic behaviour for system components like chemical converters and heat exchangers is necessary. Microstructure technology can be a solution since high heat and mass transport coefficients allow fabrication of smaller, light weight and very efficient devices.
As different sources are available and different types of processes - i.e. (steam) reforming, autothermal reforming and partial oxidation of hydrocarbons -are useful for different applications, the Institute for Micro Process Engineering at Forschungszentrum Karlsruhe is investigating a variety of processes for hydrogen generation:
- H2 by reforming of methane - H2 by reforming of bio-ethanol
- H2 by partial oxidation or - H2 by reforming of gasoline
autothermal reforming of - H2 by reforming of diesel
propane / butane - H2 clean up by catalytic processes
- H2 by reforming of methanol - mass production of devices for H2-
- Heat production from fuel cell off-gas generation
EXAMPLES FOR HYDROGEN PRODUCTION
Oxidative reforming of methane [1]
As oxidative reforming was known to possess fast kinetics a microstructured honeycomb out of pure active metal was used for the oxidative steam reforming of methane. Rhodium was chosen as the substrate material, having high activity for reforming and high heat conductivity, so that heat from total oxidation at the reactor inlet can be distributed to regions of CO-shift at the reactor end. A method based on micro wire EDM (electro discharge machining) was developed to fabricate the microchannels with a minimum channel width of 60 ^m and depth of
'j
up to 140 ^m into the Rhodium. A microstructured cross section of 5.5 x 5.6 mm with up to 1152 channels and a channel length of up to 20 mm was formed by a stack of 200 ^m thick foils bonded by electron beam welding (EB) or diffusion bonding (Fig. 1). As supposed, even at operating pressures of up to 1.2 MPa with oxygen and up to 2.0 MPa with air, the catalyst was not found to be adversely affected by the reaction heat caused by the total oxidation taking place at the
catalyst inlet. The high thermal conductivity of rhodium contributed to a good heat distribution in the flow direction (DT measured with a pyrometer at reactor inlet versus outlet was 50 - 120 K), which was reflected by significantly increased conversions (CO and H2 selectivities were compared to those obtained by using a stack of commercial Pt/Rh gauzes). At 1100°C a hydrogen selectivity of up to 88 % was reached at a GHSV of 0.195*106 h-1 STP and a fuel to oxygen ratio of CH4/O2 = 1.75 (Fig. 1). The weight of the microstructured body was 11.5 g and the pressure drop was in the range of only 0.025 MPa.
Fig. 1: Microstructured Rhodium honeycombs fabricated by EDM (left) and obtained conversion and selectivity versus temperature (right)
Steam reforming of methanol
The steam reforming of methanol was done together with Mannesmann AG to build up a complete hydrogen system suitable for dynamic hydrogen generation for automotive fuel cells. Main parts of the system, reformer, catalytic burner and selective oxidation reactor were developed at IMVT. The intermediate state of the reformer demonstrator has been published recently [2]. The PdZn alloy catalyst used was developed on the basis of a washcoating procedure with nanoparticles (Fig. 2) described in more detail elsewhere [3]. This coating applicable to already bonded microreactors was used for two different reformer foil materials, an aluminum alloy (ALMg3) and copper. To a total number of e.g. 48 microstructured copper foils with channels of 100 x 150 ^m , the empty channels yielding a
-5
reaction volume of 19.3 cm , 13.2 g of catalyst was applied (20 ^m thick layer). The necessary heat for reaction was delivered by heat cartridges within the reformer body. The demonstrators (Fig. 2) were tested under steady state and dynamic operation. At 310°C a maximum conversion of methanol of 90% was reached when feeding 70 g/h methanol with a molar water/methanol ratio of 1:2. This is enough for a 200 W fuel cell system assuming a total efficiency of 42%. Scale up effects, which reduce body to foil weight ratio to approximately 20%, result in a value of 0.7 kg/kW for the reformer. Dynamic experiments with the demonstrators highlighted that the hydrogen demand during load changes and temperature changes was met as fast as in other already well optimised systems although the weight of was not optimised in the microsystems.
Fig.2: Demonstrators for methanol reforming (left) and PdZn catalyst coating
(right)
Steam reforming of methane
For the steam reforming of methane a demonstrator (Fig. 3) for a 500 W fuel cell system was fabricated and tested together with the project partner Technical University of Munich (TUM) in steady state and dynamic operation. For heat supply methane was burned catalytically in a second passage in counter current flow. The reactor weighs 1.2 kg, has a complete length of 95 mm, a width of 22 mm and a height of 60 mm. The reformer passage was coated with a NiAl2O4 catalyst, the burner passage was impregnated with Pt-solution. The use of the microstructured reformer enables the production of up to 2 l/min hydrogen (STP). The reactor has been operated autonomously without external heating and achieved quite high methane conversions in excess of 70% at high loads. Methane conversions of 90 % were achieved at lower loads. The temperature could be well controlled combusting a mixture of methane and hydrogen in co-flow with the reforming reaction. Start-up tests showed that a constant product composition is achieved at less than 20 s (Fig. 3) after admittance of methane [4].
Fig. 3: Demonstrator and catalyst coating for methane reforming (left) and product
composition during load changes (right)
Oxidative and autothermal reforming of propane/butane
The oxidative as well as autothermal reforming of propane / butane, i.e. energy supply by oxidation in equilibrium with heat loss and heat demand of
reforming, was started together with SINTEF and the Norwegian University for Science and Technology (NTNU) in Norway by applying the rhodium honeycombs to propane experiments. Due to progress in fabrication of high temperature alloys and catalyst coating technology it was possible to switch to rhodium impregnated microstructured honeycombs out of FeCrAlloy© (Fig. 2). The substrate material having high aluminum content segregates aluminum oxide at elevated temperatures (annealing) under air conditions. From the reaction experiments it is clear that this oxide layer provides enough surface area for the desired reaction. Still using the same channel geometries as in rhodium honeycombs it was possible to increase conversion and selectivity considerably by applying the high temperature material. Rhodium and nickel impregnation were compared and rhodium was found to be superior in conversion [7]. Temperature measurements were performed within the honeycomb and compared with foam systems at SINTEF (Fig. 4). Results showed that the reduced temperature gradients within the honeycombs were the reason for superior long term stability of the catalyst [8]. The weight of the honeycombs can be reduced by use of the high temperature alloy by 36%. The hydrogen selectivity for oxidative reforming is already as high as 60% at 1000°C with a C/O ratio of 0.8 and a residence time of 12.6 ms (STP conditions).
Fig. 4: Microstructured FeCrAlloy honeycomb (left) and comparison of conversion and selectivities (right) during autothermal reforming on honeycomb (0 - 0.04 s) and foam (0.02 - 0.12 s); dotted lines represent an inlet flow of 1000
ml/min STP.
Steam reforming of gasoline
Most of the work done on gasoline reforming systems has been done in the BMBF project called “Micromotive” together with DaimlerChrysler and is described in more detail elsewhere [9]. The main objective of this work is to evaluate the mass fabrication possibilities of microstructured components for a hydrogen generation system with a fuel cell power of 5 kW. Different microstructured heat exchangers for evaporation, heat recovery and condensation as well as a reactor for the system (Fig. 5) have been built. Mass production is
feasible with today’s knowledge, integrating microembossing (Fig. 5), microetching and soldering.
Fig. 5: Microstructured reactor for a 5 kW fuel cell system (left) and heat exchanger design for microembossing and a SEM picture of the embossed
structure
Steam reforming of diesel fuel
Steam reforming on gasoline is performed in PhD work at IMVT in collaboration with the University of Clausthal. Due to the complexity of diesel fuel, its conversion by steam reforming causes problems of e.g. coke formation and dependence on fuel origin. Therefore reforming of different diesel components and conversion intermediates is investigated in a lab micro reactor (Fig. 6) systematically. Experiments so far were conducted over a rhodium catalyst on a support layer of Al2O3. The reaction conditions in the test reactor varied between 400-700°C, steam to carbon ratios from 3 to 6 and residence times from 40-160 ms. The catalyst regeneration to remove carbonaceous residues was performed when necessary at 750°C in air and subsequently in 2 vol% hydrogen. For the evaporation of higher boiling hydrocarbons such as hexadecane a new injection nozzle has been fabricated to create a fine hydrocarbon spray which evaporates in water vapour. Furthermore a complex gas chromatographic method to analyse hydrocarbons up to C16 and permanent gases (hydrogen, nitrogen, oxygen, carbon monoxide and dioxide) in one analysis run has been developed.
Turnover frequency of the fuel molecules decreases linearly with increasing number of carbon atoms in the feed for linear hydrocarbons. Branched molecules as e.g. isooctane seem to be converted easier and therefore show higher turnover frequencies (Fig. 6). Calculations show that the observed conversions and product gas compositions are close to the thermodynamic equilibrium.
0,0 -I------------------.---------------■-----------------1----------------.----------------T
0 0,2 0,4 0,6 0,8 1
1 minilrei of C-atoms [-]
Fig. 6: Micro structured test reactor (right) at operation temperature 700°C containing 14 replaceable foils with 100 micro channels each (Channel length 80 mm, width 200^m, depth 200^m) and TOF for steam reforming of different
hydrocarbons
CO removal by catalytic processes
Necessary components for gas cleaning and heat generation have been developed and are still objective of further work. The gas cleaning includes selective methanation, selective oxidation and water gas shift.
Methanation in microreactors is a good possibility to reduce the CO content in a reforming gas mixture in presence of CO2 [10]. The reactor used (Fig. 7) at IMVT consists of 27 foils (length = 78 mm, width = 20 mm, thickness = 230 ^m) with 17 channels (width = 600 ^m, height =150 ^m). The experiments were performed with a gas mixture of 1 vol.% CO, 1 vol.% 02, 25 vol.% H2 (N2: balance). At a temperature of 250 °C a CH4 space time yield of 9 mol l^-h"1 was reached. At temperatures up to 250 °C the CH4 space time yield by CO methanation is higher than by CO2 methanation. If CO (in a mixture of CO and CO2) has to be converted over a Ru/SiO2-catalyst by methanation a sufficient amount of O2 has to be added and temperature has to be controlled precisely. Because of the dimension and the heat transfer coefficients of the microchannel reactor the latter demand can be met very easily. Temperature ranges can be controlled exactly, which is important to maximize the ratio of CO to CO2 methanation rate. A microchannel reactor is an excellent tool for studying the reaction network of methanation of CO in presence of oxygen, CO2 and hydrogen without heat transfer limitation.
Fig. 7: Microstructured reactor for lab experiments on selective methanation
Also the water gas shift reaction (WGS) and the selective oxidation of CO
(Selox) are good possibilities to reduce the CO content in reforming gas mixtures [11]. At small average residence times of less than 30 ms in a relatively small microreactor (structured body: length 20 mm, width = 20 mm, height = 2.5 mm), the water gas shift reaction at 250°C - 300°C using a Ru/ZrO2 catalyst allowed to reduce the CO content by more than 95%. So it is possible to reach a very high H2 space time yield of 720 moM^-h"1.
By means of the selox reaction, which was performed in a microreactor with same geometrical dimensions like the WGS reaction, a residual CO content of 1 vol.% could be oxidized to CO2 with a conversion of more than 99% when using a CuO/CeO2 catalyst at less than 150°C and an average residence time of 14 ms (Fig. 8). However, CO-selectivity amounted to 20% for the selox reaction only. That corresponds to a C02 space time yield of 75 moM'^h"1. The selox catalysts prepared by the IMVT, allow the temperature range of selective CO oxidation to be extended considerably. The usable temperature range is increased from 60°C -120°C (conventional reactors) to 60°C - 180°C. Because of the small dimension and the high heat transfer coefficients of a microreactor system, temperatures can be reached very fast. Additional temperature ranges can be controlled exactly, which is important to maximize the CO-selectivity.
1.0
0.8
A
4 *
Cu0/Ce02/stainless steel
o
Au/Ce09/stainless steel
u u uo
O 0-4
O
O
0.2
0DA A □
a o D A
□A O
□ A o
A
A
A A
□
□
Au/Fe203/stainless steel
100
250 100
150 200
Reaction Temperature [°C]
i = 14 ms; [CO]in = 8vol.%; [H2]in = 40 vol.%; [02]in = 6 vol.%; [N2]in = 46 vol.%
150 200
Reaction Temperature [°C]
r 0.8
0.7
0.6
0.5 &
.fe
0.4 u 0)
«
0 1 W
O
o
0 2
- 0.1
0
250
Fig. 8: Conversion and selectivities for selective oxidation experiments on different catalysts still exhibiting high selectivities at temperatures up to 200°C due to increased heat transfer in the microstructured reactor
Heat generation for endothermal processes in fuel cell technology
First results in collaboration with the Max Planck Institut fur Kohlenforschung Muhlheim about heat generation from hydrogen combustion on platinum impregnated aluminum oxide layers prepared by a CVD technique in cross flow devices have been published early in 2000 [12]. A different approach concerning the catalyst preparation based on platinum impregnated aluminum oxide obtained by anodic oxidation of aluminum was presented on the 4th IMRET [13]. At the 5th IMRET further experiments have been presented for hydrogen combustion on sol-gel supported platinum catalysts highlighting the modularity of IMVT’s microstructured heat exchangers, mixers and cross flow reactors. The reaction product water was condensed in a subsequent heat exchanger and prior to the reactor a mixer was connected for security reason [14] (for space applications). In all three cases about 1 kW of heat was generated in approximately 1 cm3 of structured body.
To reduce the thermal stress in the microstructured cross flow devices (Fig. 9) further development was done and a new design (Fig. 9) was found which is capable to prevent hot spots [15]. The hydrogen mixing with air/oxygen is distributed along the reaction channels by small foil interconnecting holes so that reaction rate is lowered (Fig. 10). The overall maximum heat transfer is reduced in the new design of the so called catalytic burner by a factor of about 10 but temperature distribution is excellent and cold start properties (Fig. 10) remained nearly constant.
Fig. 9: Hot spots demonstrated by IR-photography in a cross flow device (left) and new design for e.g. evaporation of water by oxidation of hydrogen without hot-spot
(right)
Air
Fig. 10: Schematic of the flow distribution in the new design of a catalytic burner (left) and gas/body temperatures in the burner during cold start (T1 - T6 located along the reaction channels in equal distances)
CONCLUSION
The results obtained for hydrogen generation by the use of micro technology are promising in terms of temperature control, catalytic conversion and mass fabrication. However, challenges like quality control of catalytic coatings and improved dynamic operation and strategies therefore have to be faced when a new fuel has to be evaluated for hydrogen generation.
Literature
1 Mayer et al., Ind. Eng. Chem. Res. 2001, 40, 3475-3483
2 Pfeifer et al., Chem. Eng. Res. Des. 2003, 81, A7, 709-720
3 Pfeifer et al., Appl. Cat. 2004, 270, 165-175
4 Cremers et al., Proc 7th IMRET 2003, 202
5 Reuse et al., Chem. Eng. J. 2004, 101, 133-141
6 Cremers et al., Proc 7th IMRET 2003, 167
7 Aartun et al., Chem. Eng. J. 2004, 101, 93-99
8 Aartun et al., 2005, IMRET 8, Proc. on CD-ROM
9 Pfeifer et al., 2005, IMRET 8, Proc. on CD-ROM
10 Gorke et al., 2005, IMRET 8, Proc. on CD-ROM
11 Gorke et al., Appl. Cat. 2004, 263, 11-18
12 Janicke et al., J. Cat. 2000, 191, 282-293
13 Wunsch et al., Proc. 4th IMRET 2000, 481-487
14 Haas-Santo et al., Proc. 5th IMRET 2001, 313-321
15 Pfeifer et al., Chemie Ingenieur Technik 2004, 76, 607-613
16 Pfeifer et al., AIChE J. 2004, 50, 418-425
© P. Pfeifer, K. Haas-Santo, O. Gorke, J. Thormann, K. Schubert, 2006
