MeH cycle)

0360-3199/86 $3.00 + 0.00 Pergamon JournalsLtd. © 1986 International Associationfor Hydrogen Energy.

Int. J. Hydrogen Energy, Vol. 11, No. 7, pp. 459-462, 1986.

Printed in Great Britain.

D E V E L O P M E N T OF A L A B O R A T O R Y CYCLE FOR A T H E R M O C H E M I C A L

WATER-SPLITTING PROCESS (Me/MeH CYCLE) W. WEIRICH, B. BIALLAS, B. KUGLER, M. OERTEL, M. PIETSCH a n d U . WINKELMANN

Lehrstuhl for Reaktortechnik, RWTH-Aachen, F.R.G.

(Received for publication 26 November 1985) Abstraet--Metal-metal hydride (Me/MeH) processes for water splitting using HTR heat are being developed at the Institute for Nuclear Reactor Technology. The research work is concentrated on setting up a laboratory facility and developing metal membranes. It is planned to perform the first experiments in this laboratory facility as from the beginning of 1986. These will be investigations in the transport of Me/MeH suspensions and long-term tests with the metal membranes. TiNi-base alloys and coated materials will be used as membranes. TiNi-alloys did not exhibit any loss of weight due to corrosion in electrolytic experiments lasting more than 500 h. The permeation rates were constant and amounted to approximately 500 A m -2 (s = 50/Jm, PH2 = 1 bar). Pd/Cu-coatings on Ta or Nb, in contrast to pure Pd-coatings are resistant for long duration. Annealing tests at 500°C lasting 4000 h verify this behaviour.

Solid ion conductors are crystalline compounds in which the current is borne by charged atoms (ions). The The aim of thermochemical and hybrid water-splitting ion conduction is made possible by the lattice disorder processes is to make available the non pollutant source [5]. Disorder means that a crystal lattice deviates from of energy, H2, using water and H T R heat [1, 2]. When the ideal structure, i.e. additional particles (interstitial compared to the pure electrolysis of water, these pro- atoms or interstitial ions) can be present between the cesses exhibit thermodynamic advantages. The elec- regular positions of a host lattice. ZrO: doped with trolysis of water necessitates the conversion of a primary approximately 10Mol.% Y203 is a very good oxygen energy resource into electricity via heat in order to ion conductor. The ion conductivity is due to charged subsequently obtain H2. The detrimental conversion oxygen ion vacancies, resulting from the replacement step, heat into electricity, can be partly omitted in of Zr positions by Y. In the solid electrolyte fl-A1203 thermochemical and hybrid processes. In the Me/MeH (Na20.11A1203) the conductivity is due to mobile Na ÷processes, this is performed by introducing HTR heat ions, which are positioned in the lattice planes. at a high temperature level into an endothermic H2 Metal membranes are H+-conductors in which the separation reaction. The basic steps have been devel- hydrogen permeation is determined by the hydrogen oped to such an extent that a laboratory cycle can solubility and the mobility of the hydrogen in the metal be demonstrated at the Institute for Nuclear Reactor lattice. For this reason, the best hydrogen permeation Technology. properties are to be expected in the group IVb and Vb elements. As hydrogen can only diffuse in metals in atomic or proton form, it must be catalytically split MEMBRANES F O R WATER-SPLITTING into 2H" at the metal surface. The elements mentioned PROCESSES above, however, only have a low catalytic activity, and In water-splitting processes the electrolysis cell must be consequently the splitting and hence the permeation is separated by a semipermeable membrane which permits considerably reduced. This problem can be solved by ion transport and prevents back diffusion of the formed alloying. It will be explained later. Pd and Pd-aUoys, products. The following membranes are suitable for which today are the most used for purification of hydrosuch processes: (i) Ion exchange membranes, e.g. gen, exhibit high catalytic activity and good permeation Nation ® and Neosepta ®, (ii) Solid ion conductors, e.g. properties. This material cannot be used in our laborafl - A1203, ZrOe doped with Y203, (iii) Pure metallic tory cycle because it is unstable in alkali metals. H÷-conductors. The mode of operation of the ion exchange membranes is described in detail [3]. Such membrane matMe/MeH PROCESSES erials are based on fluorinated hydrocarbon polymers. The membranes react sensitively to polyvalent cations, Principles e.g. Ca 2÷ and Mg 2÷. Our work has included the develThese processes can be divided into two subsections, opment of concept proposals which use such ion exchange membranes to separate the anode and cathode an exothermic electrolytic (T = 200-400°C) and an endothermic high temperature reaction (T = 700regions [4]. INTRODUCTION

459

460

w. WEIRICH et al. Hydrogen separation

900°C). The electrolytic reaction is H 2 0 + (2 alkali Me) ~ (2 alkali MeH) + ½02.

The oxygen is produced in gaseous form, the hydrogen chemically bound as metal hydride. In order to perform this reaction, the alkali metal and the alkali metal hydride must be separated from the hydrous electrolyte by means of a H+-conducting metal membrane. The high affinity of the alkali metal for the hydrogen causes a drop in the electrolysis voltage. Figure 1 presents the theoretically possible decrease in the electrolysis voltage---with Li as hydrogen aceptor--as compared to the electrolysis of water. The difference between the two straight lines gives the maximum theoretically possible voltage saving as a function of the electrolysis temperature. Furthermore, Fig. 1 shows that the largest depolarization gain is to be expected at low electrolysis temperatures. The temperature level of the electrolysis cell must, however, be higher than approx. 200°C in order to attain technical current densities, because the permeation of hydrogen through metal depends greatly on the temperature. Beyond that, a high electrolysis temperature reduces the overvoltage losses. In the high temperature step the hydrogen is separated from the metal hydride using heat as follows: 2 alkali MeH ~ 2 alkali Me + H2.

(2)

The heat required for this purpose is to be supplied by a high temperature reactor. Figure 2 presents the flow sheet. Membrane deoelopment

The work is concentrated on improving metal membranes which meet the following requirements: high hydrogen permeation rates corresponding to approximately 4000A m -2, corrosion resistance to the electrolyte melt and in the hydrogen acceptor and long service life. These requirements place considerable restrictions on the possible membrane materials. The highly permeable materials V, Nb, Ta, Ti and Zr cannot be used in a pure form either because they are partly corroded by the electrolyte---hydrous NaOH/LiOH

H 2 0 ( g ) ' - " H 2 + 1/20z

I0

:3

. / e

05

~ i 300

HzOtg ) + 12 Li)-,=,-(2 L i H ) I 400

J 500

I 600

H.

(1)

~

+l/zO z I 700

T[K]

Fig. 1. Equilibrium voltage for the water-splitting reaction.

I/202

i

High temperature reactor

I

t

r. . . . . . . . . . .

;. . . . . .

Steam power process = I ~ .......... HzO-~J

[-]

"i E

~ = = i

L-;

r ELectroLysis

_.LT J call

High temperature heat. ...... HzO , H z , 02 ..... Me -- - - ~ MeH

Fig. 2. Me/MeH block diagram.

melt---or form a permeation-hindering oxide layer. Two concepts for developing membranes are therefore being pursued: (a) homogeneously alloyed membranes and (b) coated membranes. In the case (b), a base material which is highly permeable for hydrogen is coated with a thin, corrosion-resistant layer on the electrolyte side (A-region). This coating also exhibits good permeation properties. Palladium is suitable as coating material, but it diffuses into the base material Ta, Nb or V at higher temperatures, thus eliminating the protective effect. This fact is confirmed by examinations using the scanning electron microscope. Figure 3 shows the element analysis of a Ta foil which is vapour-plated with Pd (Pd layer thickness = 5000/~) before and after a 2-h annealing process in vacuum at 570°C. In the unannealed state, it is possible to distinguish a definite Pdpeak beside the Ta-peak, which is visible because of the penetration depth of the used electron beam. There is hardly any Pd on the surface after only a 2-h annealing treatment because it has diffused into the base material. We have developed an alloy coating which prevents the diffusion of the Pd into the base material [6]: The alloying element should form a thermodynamically stable compound with the Pd, but, at the same time, should only be slightly soluble in the base material. Cu and Ag are suitable alloying elements. Figure 4 presents the element analysis of a Ta specimen coated with Pd/ Cu after an annealing period of 4000 h (T = 400-500°C). The definite Pd-peak is clearly visible. The Pd remains on the surface and does not diffuse into the base material as a result of the alloying effect. The ratio Pd : Cu should be about 60: 40 wt%. In this concentration range Pd/Cu forms a r-phase which is characterized by high hydrogen permeation rates [7]. Their practicability still has to be verified under stringent electrolysis conditions. Homogeneously alloyed membranes are developed on the basis of Zr and Ti. Both elements have greater affinity for hydrogen than the planned hydrogen acceptors Li and Na. Consequently Zr and Ti would be hydrogenated and destroyed during electrolytic operation. This imposes a further requirement on such a membrane, i.e. no formation of a brittle hydride phase

THERMOCHEMICAL WATER SPLITTING

UnanneaLed Ta Pd

I

I

2

4

I

I

6

8

.c_

t

Ta

AnneaLing time : 2h T ~" 570"C

To

461

reduced form at a cell voltage of U >t 0.8 V. In Ti alloys it is possible to reduce the hydride stability, in contrast to Zr alloys. As an example, Fig. 5 shows the hydrogen solubility in TiNi and TiAg in comparison to pure Ti at 400°C. In the Ti-H2 system, a hydride phase is formed at a hydrogen pressure of 3.10 -2 mbar. The corresponding pressure in the TiAg-H2 system is 5.10 -1 mbar. In contrast, the alloy TiNi does not exhibit any hydride phase. The decomposition pressure of sodium and lithium hydride is also indicated in Fig. 5. Permeation tests in an electrolytic cell at T = 350°C show that current densities of approximately 500 A m -2 are achieved with TiNi (s = 50/~m; PH: = 1 bar). During the test (t = 500 h) the permeation rate remains constant. No corrosion is detected on the membrane by gravimetry. As a result TiNi can be used as a membrane and will be the material of our choice [8]. Further permeation improvements with TiNi seem possible, e.g. by grain refinement.

The laboratory cycle

I

1

2

4

I

I

6

8

Energy (keY)

Fig. 3. Element analysis at the surface (Pd at Ta).

The flow scheme of the laboratory cycle is shown in Fig. 6, where thick lines represent the liquid metalbearing components and thin lines the gas-bearing ones. The main cycle components are: electrolysis cell, hydrogen separator, liquid metal pump, surge tank for regulating the pressure compensating any changes in volume, gettering tank filled with Ti-sponge for precleaning the liquid metal, cold trap for fine purification

in the membrane. This requirement may be met by alloying with Ni, Cu, or Ag. These alloying elements must simultaneously keep part of the membrane surface flee of oxides and so permit good absorption of the electrolytically offered hydrogen. They are present in

103

,No,,o, / 102

T=,oooc I

/// TiNi 7

Annealing time==4000 h T= 400 - 500"C Pd

io t Cu/To

'-2 g E

i0 o

ix i0-1 c

Ta To

I

//T,-H,

10-2 LI/LI~

/

[

10-3 10-4

Energy Pd.'Cu at To : element analysis

Fig. 4. Element analysis at the surface (Pd/Cu at Ta).

10-3

10-2

I0-~

n [H/Me] Ti - H - systems

Fig. 5. Hydrogen solubilities in Ti-alloys.

I0 0

462

W. WEIRICH et al. t._~q

ELectroLyte

-

[

+H2o÷o2fi +H2o

Ar

Vac. ~Lo. i /.1>¢ .........

ELectroLysis H2 - separator

1

ceLL

EM

ELectroLyte

meter ~ - ~

~I~-~

I i

¢1:

._

H2

Na+NaH~

~

""

[ ~ [ ~

"

J CoLd

Vac.

I trap

I

I

)

ELeCtroLysis ceLL

Fig. 8. Electrolysis cell.

Dump tank

Gettering tank

Fig. 6. Scheme of the Me/MeH bench scale unit. of the liquid metal and dump tank which works as emergency discharge tank in an accident and which holds the liquid metal during facility shutdown. The behaviour of the facility is to be tested in the first experiments using liquid Na as hydrogen acceptor. The Na-bearing pipes have a diameter of 16 mm and are made of austenitic stainless steel, DIN 1.4571. The system pressure o f p = 1 bar is obtained using extremely pure Ar, which simultaneously acts as protective gas. In the hydrogen separator (Fig. 7) the hydrogen is drawn off into the vacuum via a membrane (material Nb). This type of separation was selected in the interests of safety. Figure 8 presents the electrolysis cell. When viewed from a safety standpoint the membrane is the most critical component. This is why the hydrogen is introduced into the Na via the membrane in gaseous form in the initial experiments. The design of the cell allows such an operation. It is planned to connect the electrolysis unit later on. If the membrane ruptures, all the Na in the cycle is passed into the dump tank within a few seconds. The hydrogen concentration in Na remains low ( < 1 A t o m %) during steady-state operation because on the one hand the quantity which will be introduced is limited by the membrane surface and on the other hand the Na inventory is relatively high ( ~ 2000 cm3). However, it cannot be reduced by means of smaller pipe cross-sections as it is to be expected that solid particles will block the pipes.

1

No/NoH

I

tO0

Niobium mm I

StainLess steel 1.4571

Fig. 7. Hydrogen separator.

Hydrogen production rates of up to 500 cm 3 h -I are expected with TiNi membranes and effective membrane areas up to 50 cm 2. It is planned to use Li as hydrogen acceptor at a later date, as this element is more suitable for thermodynamic reasons. CONCLUSIONS The M e / M e H process is set up as a laboratory cycle. A significant detail of the cycle is a TiNi membrane separating the hydrogen acceptor from the water containing electrolyte melt ( L i O H / N a O H ) in the electrolytic step. Referring to the mentioned conditions, tests at 350°C exhibit a constant hydrogen permeation rate and no surface corrosion of the membrane is detected. Furthermore, developed coated membranes Pd/Cu on Ta or Nb promise high hydrogen permeation rates and long time stability, suitable up to 500°C. Acknowledgements--The work described in this paper is sponsored by the Deutsche Forschungsgemeinschaft (German Research Society) as part of the SFB 163 at the RWTHAachen.

REFERENCES 1. C. E. Bamberger and D. M. Richardson, Hydrogen production from water by thermochemical cycles, Cryogenics 16, 197-208 (1976). 2. C. E. Bamberger, Hydrogen production from water by thermochemical cycles; a 1977 update, Cryogenics 18, 170183 (1978). 3. S. Yea and A. Eisenberg, Physical properties and supermolecular structure of perfluorinated ion-containing (Nation) polymers, J. appl. Polym. Sci. 21,875-888 (1977). 4. B. Biallas, F. Behr, K. Huns~inger, B. Kiigler and W. Weirich, The methane-methanol hybrid cycle, Proc. 4th WHEC, Pasadena, California 2, 579-590 (1982). 5. H. Rickert, Feste Ionenleiter-Grundlagen und Anwendungen, Angew. Chem. 90, 24-48 (1978). 6. F. Behr, R. Schulten and W. Weirich, Wasserstoffdiffusionswand. Patentschrift DE 3149084 C2 (1984). 7. A. G. Knapton, Palladium alloys for hydrogen diffusion membranes, Platin. Metals Rev. 21, 44-50 (1977). 8. F. Behr, B. Kiigler, M. Pietsch and W. Weirich, Nicht por6se Wasserstoffdiffusionsmembran. Patentschrift DE 32 11 193 C1, (1983).