Intermediate temperature water vapor electrolysis

Int. J. Hydrogen Energy, Vol. 16, No. 6, pp. 373--378, 1991. Printed in Great Britain.

INTERMEDIATE

0360-3199/91 $3.00 + 0.00 Pergamon Press plc. International Association for Hydrogen Energy.

TEMPERATURE WATER VAPOR ELECTROLYSIS

M . SCHREIBER, G . LUCIER, J. A. FERRANTE a n d R. A. HUGG1NS

Department of Materials Science and Engineering, Stanford University, Stanford, CA 94305, U.S.A.

(Receivedfor publication 22 February 1991) Abstract--This paper describes a novel electrochemical double-cell concept for water vapor electrolysis at intermediate temperatures (600-700 K) near the thermoneutral voltage, and separation of the hydrogen through a permeable metallic membrane. It involves the use of a combination of liquid electrolytes and inexpensive hydrogen transparent metal membrane materials. Preliminary experiments have verified the overall concept and are very encouraging for further exploration.

1. I N T R O D U C T I O N Although water electrolysis is the most sophisticated and, from an environmental viewpoint, the ideal process to produce hydrogen, its technical application is strongly impeded by economical factors such as the availability and price of electrical energy. The development efforts in electrolysis technology have focussed upon various aspects of the overall process optimization. However, the main goal is the reduction of the cell voltage to bring it as close as possible to the theoretical value. The theoretical energy required for electrolysing water at ambient conditions is 237.141 kJ mol -~, which requires 1.23 V. Part of this energy can be provided thermally. As a result the voltage required decreases if the electrolysis takes place at higher temperatures, Another alternative is to use chemical energy to lower the magnitude of the electrical energy required. An example is the use of a sacrificial iron anode which is converted into iron oxide, with the result that the applied voltage is reduced by an amount equivalent to the chemical free energy change related to the oxidation of the iron [1]. One can divide the different technologies that do not involve extra chemical reactions that are currently being pursued for water electrolysis into two major groups according to their operating temperatures: (1) ambient or near-ambient temperature systems, and (2) very high temperature systems employing solid electrolyte membranes,

One of these employs NAFION, or a similar polymeric electrolyte, such as MEMBREL, that primarily passes protons. They generally operate at about 350-360 K. Several groups are pursuing this approach, and in some cases, catalysts are deployed directly on the membrane [2]. These membranes are quite expensive [3], and cannot be used at higher temperatures. The transferenee number for proton conduction is also somewhat low, leading to a loss of efficiency. Efforts aimed at the use of crystalline solid electrolytes that conduct by the transport of protons or H 30 + ions, although having higher transference numbers and reasonable conductivities, have not been as attractive to date. Attempts to operate these electrolytes at higher temperatures to reduce the voltage and increase kinetics have not been successful, because of loss of structural water from the solid electrolyte, which causes a severe reduction of the conductivity. The third low temperature method employs alkaline electrolytes with inert physical separators. Asbestos is the separator that has been used the most, but there have been some investigations of alternatives, such as nickel oxide-based ceramics, that apparently reduce the cell impedance and voltage, and allow operation at somewhat higher temperatures, 370-390 K [4, 5]. Other experiments [6] on a different alternative, KTiO3 porous TEFLON composites, based upon much earlier work in Japan [7] have also not been very impressive. There has also been some work on improved catalysts and electrode materials to reduce the voltage losses in alkaline environments. Examples are the studies of Raney nickel and titanium-nickel alloys for cathodes [8], and ternary cobalt oxides and related perovskites for anodes [9]. Evidently, the best results show a current density of about 1 Acre-2 of electrode area at a cell voltage of 1.75-1.8 V. The high temperature approaches generally involve operation at 1100-1300 K, and the use of doped zirconia

There are three general types of low-to-moderate temperature systems, and they all generally suffer from electrode kinetic problems, primarily at the oxygen evolution electrode. For thermodynamic reasons, they also require larger voltages than higher temperature approaches, as was mentioned above, 373

374

M. SCHREIBER et al.

as an oxygen-transporting solid electrolyte. The electrode materials of choice are nickel-zirconia cermets on the anode side, and mixed-conducting oxides, such as strontium-doped lanthanum manganite on the cathode side. The interconnect material is generally a doped lanthanum chromite, There are two types of designs that have been getting attention for some time, the joined-thimble type of construction, and a tubular design that uses the thin film electrolyte that is formed inside the pores of a porous substrate by an electrochemical vapor deposition method. More recently, there has been work on a complex monolithic configuration, and several laboratories are now exploring the possibility of a parallel fiat plate configuration with a very thin solid electrolyte, There are several generic types of problems with these very high temperature approaches, having to do with ceramic fabrication, sealing and thermal expansion matching. These constitute serious challenges to production technology. There is also an inherent high energy cost connected with operation at such high temperatures, and thermal management can be a serious challenge. It is very difficult to assess the real costs of such high temperature systems at the present time, due to the lack of significant information about fabrication costs and operational experience. As has been pointed out in several places, e.g. Ref. [10], operation at intermediate temperatures, high enough to reduce electrode polarization problems, yet not so high as to have to confront the practical problems inherent in the very high temperature zirconia systems, would be optimum, The concept reported here is a novel approach to the electrolysis of water vapor at intermediate temperatures (e.g. 700 K) by the use of a double cell configuration, The hydrogen that is produced is of very high purity and is free of water, for it passes through a metallic membrane between the two cells. This method utilizes cheap metals, that are easy to fabricate and join, as electrodes, intermediate membrane, and construction materials.

Compartment I is the basic electrolysis cell. Water vapor is introduced into a hydroxide molten salt and decomposed at a voltage E~ applied between the oxygen electrode and the membrane. Voltage E l is lower than the decomposition voltage of water to produce hydrogen and oxygen gas at 1 atm pressure at the temperature of operation because the hydrogen enters the membrane at an activity lower than unity. Compartment II acts as a hydrogen concentration cell, and contains a hydrogenconducting molten salt electrolyte. The voltage E2, which is applied between the membrane and the hydrogen electrode, fixes the hydrogen activity at the left side of the membrane at a very low value. This causes hydrogen to move through the membrane by chemical diffusion down its concentration gradient. This voltage also drives the hydrogen across electrolyte II, so that it evolves at the hydrogen electrode. In the absence of losses, E~ + E 2 = Ed. . . . p , and oxygen is evolved at the oxygen electrode on one side of the membrane, and dry, clean hydrogen on its other side. The membrane thus acts not only as a separator between the two electrolytes but also serves as an intermediate hydrogen storage medium at hydrogen pressures below 1 atm. The key role in this approach is played by electrolyte II, a hydrogen-conducting molten salt containing hydride ions, which serves two important purposes. It conducts hydrogen at elevated temperatures and also provides an environment with a very low oxygen activity, so that a number of oxygen-sensitive transition metals can be used as the hydrogen-permeable membrane material. This double-cell arrangement can be described as two separate electrochemical cells, in which the membrane is considered as the negative electrode in cell 1 as well as in cell II. This can be more easily understood by writing down the simplified electrode reactions for each electrode in each of the cells: Cell I: (+)4OH

2. GENERAL CONCEPT

( - ) 4 H + + 4e = 4H_

The double electrochemical cell, which is shown schematically in Fig. 1, consists of two compartments.

Cell II: ( - ) 4H + 4e = 4H (+)4H

---.-t

.<_._.H20_vapor '2 o

®"

~ " _1 • Ol@ e / ~ x ~ lectr°lyte I • | Interface B

=2H20+02+4e

=2H2+4e

where H represents hydrogen dissolved in the membrane. 3. MATERIALS REQUIREMENTS

I

IntertaceA

Membrane

Fig. 1. Schematic illustration of the double cell concept,

Let us consider the important property requirements for the membrane electrode material, both in the bulk and at its two interfaces. In the bulk, (1) it must have high hydrogen permeability, and (2) it must undergo no phase transformations upon heating or cooling.

INTERMEDIATE TEMPERATURE WATER VAPOR ELECTROLYSIS The membrane interface that is exposed to the water vapor side, indicated as interface A in Fig. 1, has to exhibit three major materials properties: (1) it must have high interracial hydrogen permeability, (2) it must be stable under the chemical and electrochemical conditions in electrolyte I, and (3) it must not be sensitive to interfacial oxygen poisoning as a result of the diffusion of molecular oxygen from the oxygen evolution electrode. The requirements for the membrane material at interface B are: (1) high interfacial hydrogen permeability, and (2) stability in the chemical and electrochemical environment imposed by electrolyte II. All of these have one common feature, the requirement for high hydrogen permeability. The transport of a species in a bulk metallic material under these conditions is driven by the internal chemical concentration gradient, and the resulting permeability is the product of the chemical diffusion coefficient and the imposed concentration gradient of the moving species. Thus several materials that might be considered because of their high chemical diffusion coefficients, but which have low solubilities, such as iron, cannot be employed for this purpose in practical cells, Many metals react with hydrogen to form solid solutions, and at higher hydrogen concentrations form metal hydride phases. Such second phase precipitation generally is accompanied by changes in specific volume, which can cause large local stresses and dislocation generation and motion. This often results in changes in shape, the formation of microcracks, or even the mechanical disintegration of the metal. The presence of dissolved hydrogen can also cause deleterious changes in the ductility. These effects are often included under the general term "hydrogen embrittlement". We can prevent this potentially serious problem if we use metals that do not form hydride phases at the imposed hydrogen activities, or restrict the experimental parameters, such as temperature so that metal hydride phases will not be formed. This can be accomplished with some otherwise attractive metals that form hydride phases at low temperatures if we maintain the electrochemical cell temperature in a range such that it is always within the solid solution region of the metal hydride system, Some of the transition metals are known to possess very high permeabilities for hydrogen if their surfaces can be kept free of blocking oxide films. Their bcc and fcc structures are very attractive host lattices for interstitials such as oxygen, nitrogen, carbon, and hydrogen, sometimes in restricted ranges of temperature. Since their affinity for oxygen at room temperature is generally extremely high, most of them are typically covered with oxide layers. These oxide layers act as barriers for the entry of hydrogen into the lattice, and thus cause a high impedance for hydrogen permeation,

375

Several of these metal-hydrogen systems have been shown to exhibit fast transport kinetics if their surfaces are sufficiently clean. Examples are V, Ta, Nb and Ti. However, Pd and Pd-Ag alloys are the most commonly employed hydrogenmixed-conductorstodate, especially when gases containing oxygen or water species are involved. This is because they remain noble in these environments. However, they are also quite expensive, and do not have as high values of the permeability as several of the transition metals. Several reviews are available concerning experimental measurements of hydrogen transport in metals [11-17], evaluated by the use of a variety of different techniques. A thorough and detailed discussion of the data collected from a number of studies can be found in Refs [12, 13]. Birnbaum and Wert [13] showed that the published results of fcc metals (e.g. Ni, Cu and Pd) were generally consistent. However, for bcc metals (e.g. Fe, Nb, Ta, Mo, W and V), widely scattered results have been reported, probably due primarily to the influence of the presence of other interstitial species within the bulk, or to blocking species on the surface. Thus it is clear that a very important factor that has contributed to the inconsistent results reported in the literature is the extreme sensitivity of transition metals to surface effects, or to the inadequacy of surface treatments. Because of their thermodynamic properties, bcc metals usually suffer more from surface oxide problems, which retard hydrogen reactions, then fcc metals. In addition, the greater solubility and diffusivity of other interstitials, such as oxygen, nitrogen, and carbon, in bcc metals can interfere with the determination of their hydrogen diffusivity. Therefore, extremely careful materials preparation is necessary for consistent measurements. One of the successful surface cleaning techniques involves annealing under ultra-high vacuum at a relatively high temperature, followed by coating the surface with a protective thin Pd film that allows the passage of hydrogen. This was described in Refs [15-17]. The other way to avoid oxide formation at the metal surface, is to provide such a low oxygen activity environment that the metal cannot form its oxide. The oxygen activity above which a particular metal will tend to form an oxide can be calculated from thermodynamic data. Table ! shows such data for several potential membrane metals at four different temperatures. As can be seen from this table, the oxygen activities, expressed in atmospheres oxygen partial pressure, that must not be exceeded are often so low that they never can be achieved by the use of any commercially available vacuum system. For example, a good working ultra-high vacuum system can provide pressures in the range of 10 -11 to 10 ~3 torr. This is equivalent to an oxygen activity of 10-14-10 ~6atm relative to pure oxygen, or 10-20_10 22 atm for an inert gas system in which oxygen is 1 ppm of the total gas present. Table 1 shows that in the case of titanium, oxygen activities, equivalent to partial pressures lower than a b o u t l 0 -69 atm have to be achieved in order to

376

M. SCHREIBER et al. Table 1. Oxygen activities (atm) and electrochemical potentials E(V) vs 02 (at I atm), required for oxide formation of some metals at different temperatures Oxide NbO

Po2 (atm) E(V) vs 02 P02 (atm) E(V) vs 02 Po2 (atm) E(V) vs 02

Ta205

TiO VO

Po2 ( a t m )

ZrO2

E(V) vs 02 Po2 (atm) E(V) vs O:

300 K

500 K

700 K

900 K

8 x 10 134 - 1.98 9 x 10 134 --1.98 4 × 10 -173 -2.57 4 X 10 -141 -2.09 1 x 10 t8J -2.69

6 x 10 vv - 1.89 8 x 10 77 -l.89 9 x 10 101 -2.48 3 × 10-81 -2.00 3 x 10-105 -2.59

2 x 10 52 - 1.80 2 x 10 52 -1.80 9 x I0 70 -2.40 2 x I0 43 -1.91 1 x 10-72 -2.50

6 x 10 39 - 1.67 6 x 10 39 -1.71 1 x 10-52 -2.32 2 x 10 41 -1.82 2 x l0 54 -2.40

clean the surface a n d prevent local oxide f o r m a t i o n at 700 K. Earlier work in o u r l a b o r a t o r y showed that a m e t h o d t h a t can be used to provide such low oxygen activities is to employ a chloride m o l t e n salt c o n t a i n i n g a hydride, e.g. LiH as the reducing species [18-20]. To calculate the oxygen activities that can be produced by the use of this system, let us consider the schematic simplified isothermal ternary phase stability diagram for the lithium, hydrogen, a n d oxygen system at 700 K s h o w n in Fig. 2. The oxygen activities in triangles A and B can be calculated from the following equations: F o r all c o m p o s i t i o n s in triangle A: 4LiH + 02 = 2 L i 2 0 + 2H2 Po2 = 9

x 10 -66

a t m or 2.26 V vs O2 at 1 atm.

F o r all compositions in triangle B: 4Li + O2 = 2 L i 2 0 Po2 = 1 x 10 -76 a t m or 2.64 V vs 0 2 at 1 atm. F o r the metals considered in Table 1 this means that in case of Nb, T a a n d V we can provide a low e n o u g h oxygen activity by p r o d u c i n g conditions equivalent to o p e r a t i o n in triangle A to prevent the f o r m a t i o n of surface oxide films. However, Ti and Z r require even lower oxygen activities in order to prevent oxide formation. This can be achieved by operating u n d e r conditions equivalent to triangle B. Similar calculations can be done for d e t e r m i n i n g the critical H 2 0 pressure above which surface reactions will O

Li O

I_i

H O

H kill Fig. 2. Phase stability diagram of the ternary system Li H O.

take place. The values given below were calculated for 700 K. Triangle A: 2LiH + H 2 0 = L i 2 0 + 2H 2 Pn2o = 2 x 10-~Tatm. Triangle B: 4Li + H 2 0 = L i 2 0 + 2LiH P~2o = 2 × 10 28 atm. These simple calculations show that while surface oxide f o r m a t i o n o n such materials can not be avoided in m o d e r n ultra-high v a c u u m systems (10-13 torr), this can indeed be done by the use of a p p r o p r i a t e molten salt environments. Because of this, the t h e r m o d y n a m i c a n d kinetic properties of a n u m b e r o f oxygen-sensitive materials can be studied using electrochemical m e t h o d s with molten salts containing hydride ions t h a t provide these very low oxygen activities [18-20]. A n u m b e r of o t h e r potential practical applications of such salts have also been proposed, including their use in a solid/liquid/solid (S/L/S) system as a h y d r o g e n - t r a n s p o r t i n g composite "solid" electrolyte that can easily be produced in a variety of shapes [20-22] for h y d r o g e n - t r a n s p o r t i n g fuel cells, e n h a n c e d catalysis, hydrogen-cycle thermoelectric devices [20] a n d sensors [23, 24].

4. D E S I G N A N D C O N S T R U C T I O N O F A PROTOTYPE DOUBLE CELL Initial experiments to test this general double-cell concept were performed using a simple configuration with a P d - A g alloy sheet m e m b r a n e . A K O H / N a O H eutectic salt was used o n b o t h sides of the m e m b r a n e in a T E F L O N cell. The system was externally heated to only 450 K, which is within the practical t e m p e r a t u r e range of the T E F L O N . The P d - A g m e m b r a n e was 0.127 m m thick a n d h a d a surface area of a b o u t 1.5 cm 2. A current density of a b o u t 350 m A cm 2 o f electrode surface was o b t a i n e d at a total voltage (E 1 + E2) of 1.3V. These values were very encouraging, a n d confirmed the double cell concept.

INTERMEDIATE TEMPERATURE WATER VAPOR ELECTROLYSIS ,

Inlet for

~i

Water vapour 02

- ~ H~

iiiii!il]

iiiiiiiii

.............

377

that could be employed were limited, however, by the relatively low permeability of the iron membrane, due to the small value of the maximum solubility of hydrogen in iron at these temperatures and experimental conditions.

!iliiii Anode I Membrane

Cathode

.... J Fig. 3. Schematic illustration of the concentric tubular double cell design, A second experiment was undertaken to test the concept of the use of the hydride-ion containing salt as electrolyte II, so that less expensive metals could be considered for the membrane. This also involved the use of a concentric tube design, to provide a large membrane surface area, and operation at about 700 K. This design is shown schematically in Fig. 3. Cell I, containing the hydroxide salt as electrolyte I, was placed inside the membrane, which had the shape of a closed end tube. Iron was chosen as the membrane material in this case because of its thermodynamic stability in the KOH/NaOH electrolyte of cell I under the imposed experimental conditions. The hydridecontaining salt in cell II also provided a sufficiently low oxygen activity that no oxide formed on that side of the membrane, A nickel electrode was placed in the electrolyte (electrolyte I) inside the tubular membrane to act as the positive electrode of cell I. Water vapor was supplied to this cell by passing an inert gas stream through a heated water bubbler, and then through a smaller tube into the electrolyte. Both the oxygen that was evolved at the nickel electrode and any excess water vapor that did not dissolve into the salt were allowed to escape through a gas outlet in a lid of insulating material that was placed over the salt within the tubular membrane, The electrolyte of the outer cell II was contained in a second concentric tube made of aluminium. A wire of molybdenum was used as an electrode to evolve hydrogen from this outer region, which acted as cell II. A few experiments have been done with this configuration. They again showed the validity of this double cell concept, and that hydrogen can be transported across the membrane and discharged in cell II by the use of the hydride-ion containing molten salt. The currents

5. SUMMARY AND DISCUSSION A novel concept for the electrolysis of water vapor at intermediate temperatures (600-900 K) to produce high purity and very dry hydrogen has been described. It involves the use of an electrochemical double-cell with an intermediate hydrogen permeable metal membrane. In addition to having high hydrogen permeability through the bulk, it is important that blocking layers not be present on the membrane/electrolyte interface. The necessary thermodynamic conditions have been presented, and it has been shown that the use of a molten chloride salt containing lithium hydride provides an environment with a sufficiently low oxygen activity to allow the use of several transition metals with high hydrogen permeabilities as the separation membrane. This approach allows operation at desirable intermediate temperatures around 700 K. It also provides a clean separation of the hydrogen from both oxygen and water vapor, for they are produced on different sides of an oxygen blocking, yet hydrogen-permeable, metallic membrane. The present choice for the electrolyte in cell I is a molten hydroxide, and such electrolytes have been investigated for use in this intermediate temperature range in Germany [25]. However, the solubility of water vapor in this electrolyte is somewhat low at 700 K [26], which can also limit the useful current density. Initial experiments have demonstrated the validity of this double-cell concept, as well as showing that the hydride-containing molten salt can be usefully employed in the second cell, where hydrogen gas is generated. An investigation of alternative electrolytes to use in cell I instead of the hydroxide electrolyte is underway, and further experiments will be performed using membrane materials known to have higher hydrogen permeabilities. The use of metallic construction materials and the simple concentric design provide an opportunity for the use of novel configurations with high surface/ volume ratios, such as clusters of fine tubes, and relatively simple construction methods. This also avoids potential problems with expansion matching and metalceramic seals.

Acknowledgements--This work was supported by the BrookhavenNational Laboratory under Contract No. BNL 274318-S REFERENCES 1. P. H. Koske, presented at the 8th World Hydrogen Energy Conference,Honolulu, Hawaii (1990). 2. R. Oberlin and M. Fischer, in Hydrogen Energy Progress v, p. 333. Pergamon Press, New York (1984).

378

M. SCHREIBER et al.

3. G. Scherer, Poster P 139, presented at the 6th World Hydrogen Energy Conference, Vienna (1986). 4. J. Divisek et aL, in Hydrogen Energy Progress V, p. 655. Pergamon Press, New York (1984). 5. J. Divisek et al., in Hydrogen Energy Progress VI, p. 258. Pergamon Press, New York (1986). 6. Annual Progress Report, International Energy Agency Program of Cooperative Research and Development, Task IV, p. 6 (January 1988). 7. E. Yorikai et al., German patent 2,717,512 (I0 November 1977). 8. C. Fabjan et al., Poster P 253, presented at the 6th World Hydrogen Energy Conference, Vienna (1986). 9. W. Schnurnberger and R. Henne, Poster P 237, presented at the 6th World Hydrogen Energy Conference, Vienna (1986). 10. F. J. Salzano, G. Skaperdas and A. Mezzina, in Hydrogen Energy Progress V, p. 787. Pergamon Press, New York (1984). 11. J. V61kl and G. Alefeld, in A. S. Nowick and J. J. Burton (eds), Diffusion in Solids, Recent Developments. Academic Press, New York (1974). 12. J. V61kl and G. Alefeld, in G. Alefeld and J. V61kl (eds), Hydrogen in Metals I, Basie Properties, Springer, Berlin (1978). 13. H. K. Birnbaum and C. A. Wert, Ber. Bunsenges. Physik. Chem. 76, 806 (1972).

14. J, V61kl, in G. Bambakidis (ed.), Metal Hydrides, p. 105. Plenum Press, N.Y. (1981). 15. N. Boes and H. Zfichner, Z. Naturforsch. 31a, 754 (1976). 16. N. Boes and H. Z/ichner, J. Less-Common Metals 49, 223 (1976). 17. G. Alefeld and J. V61kl (eds), Hydrogen in Metals lI, Application-Oriented Properties. Springer, Berlin (1978). 18. G. Deublein and R. A. Huggins, Solid State Ionics 18-19, 1110 (1986). 19. G. Deublein and R. A. Huggins, J. Electrochem. Soe. 136, 2234 (1989). 20. G. Deublein, B. Y. Liaw and R. A. Huggins, Solid State lonics 28-30, 1078 (1988). 21. G. Deublein, B. Y. Liaw and R. A. Huggins, Solid State lonies 28-30, 1084 (1988). 22. B. Y. Liaw and R. A. Huggins, Z. Phys. Chem. N.F. 164, 1533 (1989). 23. G. Deublein, B. Y. Liaw and R. A. Huggins, Solid State lonies 28-30, 1660 (1988). 24. B. Y. Liaw, G. Deublein and R. A. Huggins, in D. R. Turner (ed.), Proc. Syrup. Chemical Sensors, p. 91. Electrochemical Soc. (1988). 25. W. Weirich et al., in Hydrogen Energy progress V, p. 467. Pergamon Press, New York (1984). 26. V. Egger, Dissertation, Ludwig-Maximilian Universitfit, M/inchen (1964).