On the Compatibility of Electronic Insulators and Ionic Conductors With the Na2BH5-Na4B2O5-H2 Quasi-ternary System

1

*Corresponding author Tel.: +1 819 821 8000ext63238; fax: +1 819 821 7095 E-mail:[email protected]

On the compatibility of electronic insulators and ionic conductors with the Na2BH5-

Na4B2O5-H2 quasi-ternary system

Daniel L. Calabretta

Centre de recherche en énergie, plasma et électrochimie,

Département de génie chimique et de biotechnologique, Université de Sherbrooke, Sherbrooke, Qc., Canada, J1K 2R1

Available online 12 February 2013

Abstract

As it pertains to simple electrolytic reverse hydrolysis (ERH) test cells, discussions and experimental

results regarding the thermodynamic stability of electronic insulators and ionic conductors with the

liquid, anhydrous Na+,B

3+,H

+/H

-,O

2- system (823±10 °K, 1.00±0.01 MPa (H2)) are given. Inductively

coupled plasma optical emission spectrometry, energy dispersive x-ray spectroscopy, scanning electron

microscopy, and x-ray diffraction facilitated the post-equilibration analyses of the hydridic electrolytes

behaviour with the investigated materials. The reduction of hematite, by sodium hydride, to metallic

iron negates the possibility of having the most desirable solid oxygen anion electrolyte,

La0.7Sr0.3Ga0.7Fe0.2Mg0.1O3-δ, in direct contact with these hydridic electrolytes. The other metal oxides

of the relevant double-doped, LaGaO3 perovskite family are significantly soluble in the melts. The

thermodynamic analyses identifies scandia stabilized zirconia as the most auspicious solid oxygen

anion electrolyte; hence, understanding the nature of the liquid, anhydrous Sc3+

,Zr4+

,B3+

,Na+,H

+/H

-,O

2-

system (823±50 °K; ≤1.0 MPa) is paramount to further proof-of-concept efforts. Keywords: Sodium borohydride; Sodium hydride; Hydrogen storage; Oxide ion membranes;

Compatibility testing; Electrolytic reverse complete oxidation

1. Introduction

The relevant Na

+,B

3+/H

-,O

2- liquidus topology with its relationship to a simple but hypothetical

electrolytic reverse complete oxidation (ERCO) process is available [1]. The electro-synthesis of 50

mol%NaBH4-NaH (Na2BH5) from sodium pyroborate (Na4B2O5) is a convenient example of an ERCO

process that is suitable for automotive applications as the theoretical, onboard energy capacities of its

complete oxidation is akin to the status quo. A proceedings [2] article on a so-called thermodynamic

amphoteric formalism, which employs a contemporary concentration scale, allows for a simple

description of the hypothetical electrolytic reverse hydrolysis (ERH), electrolytic reverse combustion

(ERC), and ERCO processes. While some experimental attempts have been made to prove the ERH

concept [3], the metal-supported anode compartments, which were furnished by radio-frequency

induction-coupled suspension plasma spraying, were unsuitable for a fundamental study [4].

Furthermore, in regards to the compatibility of the liquid H2-Na4B2O5-Na2BH5 quasi-ternary system

with both solid ionic electrolytes and electronic insulating materials, mostly negative results were

obtained. Here, following a discussion of the relevant literature, the materials compatibility results are

given.

1.1 Electrolytic molten salt, solid oxygen anion electrolyte processes

Electrolytic molten salt-SOAE (MS-SOAE) processes, which have been discussed since 1971

[5], pertain to the production of metals and O2, or other gaseous oxide species, from mixtures of metal

oxide containing molten salt mixtures. In their investigation of metal and O2 production for lunar

2


applications, Sammels and Semkow [6] investigated electrolytic cells having an yttria-stabilized

zirconia (YSZ) solid oxygen anion electrolyte (SOAE), La0.89Sr0.10MnO3 (LSM) anode, FeSi2 or Pt

cathode, and catholyte mixtures comprised of Li2O(24 mol%)-Na2O, Li2O(66 mol%)-B2O3, or Li2O

(23.9 mol %)-LiF (23.6 mol %)-LiCl. For the electrolytic cell (-) FeSi2| Li2O-LiF–LiCl | YSZ | LSM

(+), operating at 973 °K with an overpotential of ~1 V, they reportedly obtained a current density of 60

mA cm-2

. Their observed voltage plateaus were said to be due to the formation of ternary metal alloys.

No post-experimental information was given for the YSZ tubes.

The group from the University of Boston and their associates have published the most

extensively on MS-SOAE processes [7-13]. Their bench-scale studies have produced magnesium [7-

9], tantalum [10], and titanium [12] metals. In their initial investigation [7] on magnesium production

(1023 °K), which operated with H2 flowing through the anode compartment, pathetically low current

densities were obtained. In the follow-up investigation, by applying convection, large over-potentials

(~ 5 V), an extended anode compartment comprised of a carbon rod immersed in liquid copper, and

MgF2 or MgF2-CaF2 flux saturated with MgO (1423-1575 ºK), Krishnan et al. [8] achieved high

current densities (> 1 A cm-2

). Magnesium vapors were condensed in a third compartment and the

observed dissociation potential for the overall cell reaction was 0.59-0.63 V.

Martin et al. [14, 15] investigated the production of calcium metal from liquid CaO-CaCl2

mixtures at ~1123 °K by both galvanostatic and potentiostatic electrolysis. Their cells were equipped

with an Ag/AgCl reference electrode and YSZ tubes having 0.1 cm wall thicknesses. O2 and calcium

metal were produced at a Pt (inert) anode and a stainless steel cathode, respectively. Using the former

electrolysis technique, excessive over-potentials were applied, and the limiting current density was

reportedly due to the rate of diffusion of complex CaxOCl2xO2-

ions. A noticeable color change

occurred at the YSZ tubes’ exterior surfaces that were in contact with the molten salt, and the tubes

broke after operating for extended periods of time. No color change occurred under potentiostatic

conditions. However, current densities decreased significantly with time until the YSZ tubes broke.

Mechanisms for YSZ’s degradation were discussed in terms of ‘hot corrosion’ processes [16-19].

Depending on the pO2-

(= -log [O2-

]) of the melt, which is analogous to the pH scale used in aqueous

media, the SOAE can undergo degradation due to oxoacid (low pO2-

) or oxobasic (high pO2-

)

processes, which are given for Y2O3 by (Eq. (1)) and (Eq. (2)), respectively. 23

32 32)( OYYSZOY (1)

2

2

32 2)( YOOYSZOY (2)

Others have investigated the stability of YSZ in the presence of soda lime glass [20, 21], LiF-

NaF-KF [22, 23], Na2SO4 [24] and Na2SO4-NaVO3 [25] melts at temperatures substantially higher than

those being considered here. Electrochemical impedance spectroscopy [20], laser Raman spectroscopy

[22-24], and x-ray diffraction [24, 25] have been used for the investigations of corrosion mechanisms

under both polarized [24] and non-polarized [22, 23, 25] conditions. Deterioration in YSZ

performance has generally been attributed to the leaching of Y2O3 from the stabilized structure which

gives rise to the more fragile ZrO2 monoclinic phase.

During their investigations of MS-SOAE aluminum metal production process at ~1200 ºK,

which succeeded the one described by Marinek [5], Rapp and Zhang [26] reported on the excessive

dissolution of Y2O3 in cryolite compared to alumina (Al2O3). As a minor variation, reforming gases

flowed through the anode compartment to react with O2-

. Their second report [27] on aluminum

production investigated the use of sodium sulfate (Na2SO4) solvent. Although YSZ’s components had

lower solubility in the solvent, the formation of sodium aluminum sulfide compounds occurs.

1.2 SOAE for the ERH and ERC processes

The aforementioned MS-SOAE studies (1.1) operated at high temperatures (1233-1473 °K)

with SOAE-supported anode compartments. Due to NaH decomposition (~923 °K; 1 MPa), the ERH

3


and ERC processes [1-3] of interest here are limited to a significantly lower operating temperatures.

For practical cells, under desirable operating conditions, YSZ’s low oxygen anion conductivity ( 2O )

will require SOAE layers of <20 μm. If suitable, thin-filmed (~20 µm), SOAE anode compartments

are achievable, then double-doped LaGaO3 perovskite, ceria-based fluorite, and scandia stabilized

zirconia (ScSZ) SOAE have 2O that can accommodate the aforementioned operating temperature

constraint. The criteria put forth by the thin-filmed, intermediate temperature SOFC (IT-SOFC)

community [28] can be extended to the ERH and ERC metal-supported type cells: the SOAE must

have excellent chemical stability with all other connecting materials (i.e. molten salt, anode material

and the anode’s chemical environment); cells must have long operational life-spans (4000-5000 hrs);

the cells’ functional layers must have similar coefficients of thermal expansion (CTE); the cells must

be tolerant to thermal shock and have the good mechanical properties; at operating temperature the

SOAE must have an 2O ≥ 0.01 S cm-1

; and so on.

The ceria doped (Gd, Sm, etc.) fluorite family of SOAEs are not stable in reducing

atmospheres [29]. Macro-fissuring occurs due to the change in the metal’s oxidation state (Ce4+

to

Ce3+

) and cannot be in direct contact with either the ERC or ERH cathode or ERH anode

compartments.

In 1994, Ishihara et al. [30] reported on the near pure O2-

conducting doped LaGaO3 perovskite

family. A review [31] of the first decade of international investigation of these materials is available.

The extraordinary electronic properties of La0.7Sr0.3Ga0.7Fe0.2Mg0.1O3-δ (LSGFM) at low temperatures

were later reported on by the same group [32-34].

1.3 Thermodynamic assessment of materials’ compatibilities

Experimentally, other than a preliminary pressure differential thermal analysis investigation

[35] of YSZ and La0.8Sr0.2Ga0.8Mg0.2O3-δ sintered pellets with molten Na2BH5 (923±5 °K; 1.00±0.01

MPa), nothing has been reported on materials’ compatibilities/solubilities with/in the anhydrous, liquid

Na+,B

3+,H

+/ H

-,O

2- system [1]. There was an obvious white to black color change in the sintered YSZ

after exposing it to liquid Na2BH5 whereas La0.8Sr0.2Ga0.8Mg0.2O3-δ sintered discs appeared to be stable;

however, the analytical technique lacked the resolution required to determine the solubility of the

components in the hydridic melt. The thermodynamic analyses given in this sub-section will be

followed by crude experimental investigations of the compatibilities and solubilities of ionic

conducting and electronically insulating materials with and in the hydridic melts (823±10 °K,

1.00±0.01 MPa (H2)). For commercial ERH and ERC cells under both non-polarizing and polarizing

conditions, the materials must be non-reactive and very sparingly soluble with and in the hydridic

electrolytes.

Potential detrimental processes under non-polarizing conditions will now be considered.

Firstly, for the rare-earth metal, hydrogen binary systems, only the binary MH2 hydrides should be

considered as compounds [36]; therefore, reactions between the hydridic electrolytes and the SOAE’s

rare-earth oxides would result in H2 and rare-earth dihydride formation; hence, as given by (Eq. 3) and

(Eq. 4), the rare-earth metal would have its oxidation state diminished. Secondly, reactions expressed

by (Eq. 5) parallel those that were encountered by Hayward and Rosseinsky [37] where NaH reduces

one of a SOAE’s metal oxide components to a metal oxide of a lower oxidation state; furthermore,

NaH may have the capacity to further reduce a metal oxide in its lowest state to the metallic phase (Eq.

6). The combination of (Eq. 5) and (Eq. 6) results in the process given by (Eq. 7). In an analogous

fashion, H2 may also have the capacity to reduce a metal oxide to either a lower oxidation (Eq. 8) or

metallic (Eq. 9) state. Finally, the metathesis reactions given by (Eq. 11) - (Eq. 13) are illustrated in

the isothermal, isobaric, quaternary, composition triangular-prism for the Na+,B

3+,M

q+/O

2-,H

- system

[3] shown in Fig. 1. Given the decomposition temperatures of the corresponding rare and alkaline

earth borohydrides (Table 1) processes corresponding to (Eq. 10) and (Eq. 14) should be unfavourable.

4


However, due to the lack of available thermodynamic data for these borohydrides, the presumption

cannot be confirmed by thermodynamic calculations.

HMHM

HONaMHNaHMO

121141

2

1

22

1

2

1

2222

2222 (3)

HMHM

HONaMHNaHOM

1213

2

1

2232

1

22

22332 (4)

HMHM

HONaMONaHOM

1213

2

1

2

1

32

1

)2()2(

)2()2()2( (5)

HMHM

HONaMNaHMO

121121

2

1

2

111

)2()4()2()4(

)2()2()2()2( (6)

2232

1 )2/3()2/3(3)2( HONaMNaHOM (7)

HMHM

OHMOHOM

1213

2

1

2

1

32

1

)2()2(

)2()2()2( (8)

HMHM

OHMHMO

121121

2

11

2

11

)2()4()2()4(

)2()2()2()2( (9)

ONaBHMNaBHOM 2

1

34

1

432

1 )2()3()6( (10)

524

1

2

1

52

11 )10()2()5()2( OBNaMHBHNaMO (11)

ONaMHNaHMO 2

1

2

11 )2()2()2( (12)

2

1

2

1

4

11 )4()2()4()2( NaBOMHNaBHMO (13)

ONaBHMNaBHMO 2

1

24

1

4

1 )2()()2()2( (14)

Table 2 lists the Gibbs energy of reaction for (Eq. 3)-(Eq. 12) obtained from the available

thermodynamic data [38]. The values in italics implies that the referred to metal oxide species are in

the vapour phase; thus, only if (Eq. 5) and (Eq. 8) are spontaneous should (Eq. 6) and (Eq. 9) be

considered. Aluminum has been included in the analysis to adjudge the stability of the solid sodium

cation electrolyte, β”Al2O3, for mixing studies, for hypothetical complex ERH processes, or for its

possible use as an electrically insulating material.

5


Fig. 1 Isothermal, isobaric, quaternary, composition triangular-prism for the anhydrous

Na+,B

3+,M

q+/O

2-,H

- system

Table 1 Decomposition temperatures of relevant hydrides and borohydrides (0.1MPa)

Mz+

H-z+ Td (ºK) Ref. MBH4 Td (ºK) Ref.

NaH 693 39 NaBH4 838 39

SrH2 948 39 Sr(BH4)2 683-693 45

MgH2 600 39 Mg(BH4)2 580 45, 46

FeH2 <RT 40 Fe(BH4)2 <RT 47

YH3 >1200 41 Y(BH4)3 595 48

LaH2 800-900 41 La(BH4)n ~RT 48

GaH3 <RT 42 Ga2(BH4)6 <RT 45

ScH2 1123 43 Sc(BH4)n <500 48, 50

ZrH2 363 44 Zr(BH4)4 ~RT 45,48

NaH

(2)-1

Na2O

(2qM)-1

M2O

(5)-1

Na2BH5

(4)-1

NaBH4

(qM)-1

MHqM

(10)-1

Na4B2O5

(4)-1

NaBO2

6


Table 2 Temperature dependent Gibbs energy of reaction (kJ) for (Eq. 3)-(Eq. 12) for the metals Al,

Fe, Ga, La, Mg, Sc, Sr, Y, and Zr [40]

Theoretically, under ERH or ERC polarizing/operating conditions, if a large enough potential

is applied across the cell, a sparingly soluble oxide will dissociate; consequently, this would cause the

Eq. T(ºK) Al Fe Ga La Mg Sc Sr Y Zr

3 600 - - - - - - - - 128.6

700 - - - - - - - - 124.3

800 - - - - - - - - 120.4

900 - - - - - - - - 116.8

4 600 - - - - - - - 312.6 -

700 - - - - - - - 309.6 -

800 - - - - - - - 307.4 -

900 - - - - - - - 305.7 -

5 600 599.3 -53.0 521.8 481.5 - 595.4 - 614.8 -

700 574.3 -60.8 497.7 459.0 - 571.6 - 591.3 -

800 549.5 -68.5 473.8 436.9 - 548.1 - 568.1 -

900 525.0 -75.9 450.3 415.2 - 524.8 - 545.4 -

6 600 -167.6 -47.4 -276.7 -75.7 108.5 -106.4 105.7 -115.0 -163.8

700 -164.8 -51.8 -274.9 -73.5 101.8 -103.4 99.5 -112.3 -161.1

800 -162.1 -56.2 -273.2 -71.3 95.3 -100.5 93.4 -109.8 -158.4

900 -159.4 -60.5 -271.4 -69.1 88.7 -97.8 87.3 -107.3 -155.8

7 600 264.1 -147.8 -31.5 330.0 - 382.5 - 384.9 -

700 244.7 -164.5 -52.1 312.1 - 364.9 - 366.6 -

800 225.4 -180.8 -72.5 294.4 - 347.1 - 348.6 -

900 206.3 -196.9 -92.6 277.0 - 329.2 - 330.9 -

8 600 652.1 -0.2 574.6 534.3 - 648.2 - 667.7 -

700 631.0 -4.1 554.4 515.7 - 628.3 - 648.0 -

800 610.1 -7.9 534.4 497.5 - 608.7 - 628.7 -

900 589.5 -11.4 514.8 479.7 - 589.3 - 609.9 -

9 600 -114.7 5.4 -223.8 -22.9 161.3 -53.6 158.5 -62.1 -

700 -108.1 4.9 -218.2 -16.8 158.5 -46.7 156.2 -55.6 -

800 -101.5 4.4 -212.6 -10.7 155.8 -39.9 154.0 -49.2 -

900 -94.9 4.0 -207.0 -4.6 153.2 -33.3 151.8 -42.8 -

10 600 238.0 - - - - - - - -

700 235.7 - - - - - - - -

800 234.1 - - - - - - - -

900 233.3 - - - - - - - -

11 600 - - - - 68.4 - 15.1 - -

700 - - - - 69.1 - 16.6 - -

800 - - - - 69.8 - 18.2 - -

900 - - - - 70.4 - 20.0 - -

12 600 - - - -128.9 110.8 - 57.5 -151.1 -206.6

700 - - - -119.4 111.1 - 58.7 -140.8 -196.2

800 - - - -109.9 111.5 - 60.0 -130.4 -185.7

900 - - - -100.4 112.0 - 61.5 -119.9 -175.0

7


oxide’s further dissolution. If an ERC cell is considered, the standard potential of combustion (ECombustion ; Eq. 15) can be compared with a soluble metal oxide’s standard formation potential, (

EMO ; e.g. Eq. 16). Similarly, the standard potential of hydrolysis, (EHydrolysis ; Eq. 17), can be

compared with a soluble oxide’s standard potential of metal combustion, (EMetCom ; Eq. 18), which is

simply its EMO minus the standard potential of water formation (

EHO ; Eq. 19). Table 3 lists the

calculated [38], linear, temperature-dependent functions for the considered oxides’EMO s at 0.1 MPa.

As well, their EMO and

EMetCom values at 823 °K have been tabulated so that they can be

compared with ECombustion and

EHydrolysis . The latter potentials for Na2BH5 have values of 2.10 and

1.05 V, respectively.

Table 3 Temperature dependence of standard formation and metal combustion potentials of relevant

oxides (0.1 MPa)

Formation Reaction

∆E° (T) (773-823 °K)

∆E° (823 °K)

2

3

3

2

12

3

3

2

1

32

1

2

11

)6()6(

)6()4()3(

OScOSc

OScOSc

T*0.0005-3.2696

2.86

1.81

2

3

3

2

12

3

3

2

1

32

1

2

11

66

)6()4()3(

OYOY

OYOY

3.284 - 0.0005*T

2.87

1.82

2

3

3

2

12

3

3

2

1

32

1

2

11

66

)6()4()3(

OLaOLa

OLaOLa

T*0.0005-3.0861

2.68

1.63

2

2

2

2

12

2

2

2

1

1

2

11

44

242

OSrOSr

SrOOSr

T*0.0005-3.0562

2.64

1.59

2

2

2

2

12

2

2

2

1

1

2

11

44

242

OMgOMg

MgOOMg

T*0.0006-3.1163

2.62

1.57

2

3

3

2

12

3

3

2

1

32

1

2

11

)6()6(

)6()4()3(

OAlOAl

OAlOAl

T*0.0005-2.8909

2.45

1.40

2

4

4

2

12

4

4

2

1

2

1

2

11

88

444

OZrOZr

ZrOOZr

T*0.0005-2.8339

2.42

1.37

2

3

3

2

12

3

3

2

1

32

1

2

11

66

642

OGaOGa

OGaOGa

T*0.0006-1.893

1.40

0.35

2

2

2

2

12

2

2

2

1

1

2

11

44

242

OFeOFe

FeOOFe

T*0.00005-1.2228

0.02

2

3

3

2

12

3

3

2

1

32

1

2

11

66

643

OFeOFe

OFeOFe

T*0.0002-1.2706

-0.05

2

1

524

1

2

1

52

121045 HOBNaOBHNa

(15)

8


2

5

5

2

12

5

5

2

1

524

1

2

1

2

1

1010

1045

OXOX

OBNaOBNa (16)

2524

1

2

1

52

11025 HOBNaOHBHNa

(17)

2

1

524

1

2

1

2

121052 HOBNaBNaOH

(18)

2

2

12

2

1

2

1

2

1

2

1

)2()2(

)2()2()4(

OHOH

OHHO (19)

The preeminent findings from the temperature-dependent, thermodynamic analyses given in Tables

1-3 can be enumerated as follows:

1. EMO for Fe2O3 and Ga2O3 are less than Na2BH5’s

ECombustion and EHydrolysis

2. Fe2O3 and Ga2O3 should react with NaH (Eq. 7);

3. H2 should reduce Fe2O3 to FeO (Eq. 8);

4. ScSZ has thermodynamic stability under both polarized and non-polarized ERH and ERC,

operating conditions;

5. ScSZ is the most promising SOAE;

6. Al2O3, La2O3, MgO, Sc2O3, SrO, Y2O3, and ZrO2 can be considered as potential electronically

insulating materials.

The thermodynamic analyses should eliminate the possibility of having the double-doped,

LaGaO3 perovskites in direct contact with the hydridic electrolytes under consideration. However, it is

worth verifying the supposed instability of the most promising members of this family. For, in solid

solution, the metal oxides may have unpredictable stability.

2.0 Experimental

2.1 Materials’ compatibilities and solubilities testing

The following protocols allowed for the investigations of the compatibilities/solubilities of

commercial electronic insulators, relevant SOAE materials, and ß"-Al2O3 with/in the pre-synthesized,

hydridic electrolytes.

While solubility measurements would preferably be done via electrochemical or quenching

techniques [51], these methods were not considered. The former was ill-advised because along with

the number of materials that needed to be investigated and the high cost of β”-Al2O3 tubes, prior to this

investigation, nothing was known about the compatibility/solubility of β”-Al2O3 with the hydridic

melts. The latter technique was not used because of the potential hazards that could arise during the

quenching process. Although the methodology given below is simple, it allows for the rudimentary

investigation of numerous materials at relatively low cost. For example, incompatibility (i.e. reactions)

between the hydridic electrolytes and the investigated materials could be detected simply by observing

the overall texture and colour of the mixture and substrate after the equilibration period and comparing

it to its corresponding blank hydridic electrolyte.

Described elsewhere are the details regarding the modifications made to the as-found, custom

glove-box at CREPE [3], the design and construction of the custom pressure vessel [3], and the

synthesis of SOAE powders by the glycine nitrate process (GNP) along with their subsequent tailoring

9


and analyses [4]. The custom glove-box and pressure vessel was used for the synthesis of the hydridic

electrolytes, for the subsequent material compatibility tests, and for the very limited ERH experiments

[3]. The pressure vessel’s thermocouple temperatures and pressures that are cited in this text are

estimated to carry errors of ±10 °K and ±0.01 MPa, respectively.

2.4±0.1 cm o.d., 2.5±0.5 mm thick, green discs of SOAEs were prepared with a Model 3850

axial hydraulic press (Carver Co., IN, USA) that was operated at 30±5 MPa for 1.5±0.5 minutes. In

order to obtain stable green discs, when required, polyvinyl butyral (Sigma-Aldrich, B0154) binder was

added to the tailored powders (~5 wt%). They were subsequently air-sintered at 1773±10 °K for 9±3

hrs.

Commercial, electronically insulating paints included Resbond™ 919 and 903 HP (Cotronics

Corp.; NY, USA), Pyro-paints™ 634-BN, 634-SIC, 634-AL (Aremco Products Inc.; NY, USA), AlN

(Alfa-Aesar – 43959) and 96wt% ZrO2-CaO (Alfa Aesar - 40480); the commercial powders included

99.9 wt% La2O3 (Sigma-Aldrich; L4000), 99.9 wt% SrO (Sigma-Aldrich; 41513), 99.0 wt% MgO

(Sigma-Aldrich; 342793), 99.99 wt% Y2O3 (Sigma-Aldrich; 205168), 99.9 wt% Al2O3 (Sigma-Aldrich;

318767), and 99.9 wt% Al2O3 (Alfa-Aesar; 318767; the sintered ceramic materials included ß"-Al2O3

(Ionotec Inc. – B2-220-LN; Cheshire, UK), BN (Carborundum Co. – WT118098024; NY, USA) and

Si3N4 (McMaster Carr – 9576K24; IL, USA).

To determine their general adherence, the commercial paints were painted onto sand-blasted,

ultra-sounded, 1.27 cm o.d., SS316 discs and then heated according to the suppliers’ specifications.

This was followed by a secondary heating to 873±10 °K under 99.9 vol% H2. Materials that adhered

well were used for the investigation.

Rings of BN and strengthened ß"-Al2O3 (7wt% ZrO2, 86wt% ß"-Al2O3, 6.5wt% ß-Al2O3,

0.5wt% NaAlO2) were cut from their respective tubes with a diamond saw before being oven dried

under air at 373±5 °K; they were subsequently held under vacuum in the commercial glove-box’s

entrance port for several days.

NaH mineral oil suspensions (Alfa Aesar - 13431), NaBO2*4H2O (Alfa Aesar - 36736), and

99.2 wt% silica free NaBH4, were the starting reagents for the hydridic electrolytes; the latter was

donated by the Rohm and Haas Co. (PA, USA). NaH was stripped of its mineral oil by vacuum

filtration in the commercial glove-box. Copious amounts of 99.5 vol% anhydrous toluene (ACP

Chemicals Inc.; Qc, Canada) solvent was used and vacuum was applied for prolonged periods of time.

Excess toluene was then evaporated from the stripped NaH by placing it in a non-hermetic, dried, glass

bottle in a commercial glove-box’s entrance port for about a week. NaH’s purity was measured using a

custom hydrolysis apparatus described elsewhere [1].

Soon after the custom glove-box had been made operational, in order to avoid its future

wetting, 0.50±0.05 kg samples of NaBO2*4H2O were dehydrated. The starting reagent was placed in

custom 8 cm o.d., 20 cm tall, custom copper crucibles that were capped at their bottoms with copper

discs and silver solder (Braze 450, Lucas Milhaupt Ltd., WI, USA), which has a liquidus eutectic

temperature of 1016 °K. The sample was heated at 2 °K min-1

and soaked under vacuum at 423±10 °K

for several hours before it was further heated to 673±10 °K and soaked for several more hours at 50±10

°K intervals. Then the sample was cooled to ambient and its hydrous portion, which had agglomerated

in the vessel’s headspace, was crushed and compacted with a SS rod and subsequently mixed with the

anhydrous material contained at the bottom of the crucible. Lastly, the sample was reheated under

vacuum to 923±10 °K and held for a day, or until the XRD pattern indicated a pure product.

Compounds contained within the sub-systems of the H2O-Na2O-B2O3, ternary system were

undetectable [3].

Table 4 lists the hydridic electrolytes that were prepared in both mol% Na4B2O5 (Na4B2O5-

Na2BH5 quasi-binary) and elemental equivalent ionic fractions [2, 3] ( 2OY , 4.0Na

Y ). The starting

reagents, which were well mixed, included NaBH4, NaH, and anhydrous NaBO2, and the resulting

phases are a consequence of the quasi-reciprocal process [1]. The mixtures were subsequently used for

the investigations of materials’ compatibilities and solubilities. To allow for the quasi-reciprocal

10


process to go to completion, metal oxide containing mixtures were allowed to equilibrate for ~2 days at

848±25 °K. MSi-iii and MSiv belong to the Na+/B2O5H2

6-, BH4

-, H

- and Na

+/B2O5H2

6-, B2O5

4-,BH4

-

additive ternary systems, respectively. In some cases the behaviour of relevant materials were studied

in both systems.

Copper crucibles of all dimensions were washed with 1 N HNO3 followed by distilled water

before being dried and heated in the custom pressure-vessel to 773±10 °K under a mild purge of H2 for

about a day.

Table 4 Synthesized hydridic electrolytes belonging to the Na4B2O5-Na2BH5 quasi-binary system

Mixture Na4B2O5 (mol%) 2O

Y , 4.0NaY

MSi 0.0 0.0

MSii 15 0.26

MSiii 20 0.33

MSiv 30 0.46

Fig. 2 shows the compatibility testing arrangements that were used. For the first arrangement

(A), material masses (1.5±0.5 g) were positioned in the bottom of 3.2 cm o.d. copper crucibles (tube

caps) before hydridic electrolytes (7.5±0.5 g) were added over top of them. For the second

arrangement (B), commercial and non-commercial powders (1.35±0.65 g) were placed in the bottom of

small copper crucibles (1.3 cm o.d.) that had small holes drilled slightly above the midpoints of their

heights; they were put in larger 3.2 cm o.d. copper crucibles before the addition of the hydridic

electrolyte (7-10 g); the objective being to attain a saturated hydridic electrolyte free of solid powder

contamination so that the ICP elemental analyses (2.2) represented the powder’s solubility.

Arrangement C was used to saturate large quantities of hydridic electrolytes that were used for the

subsequent ERH experimentation. Following the saturation period, samples were taken from the large

crucibles’ tops, middles and bottoms for ICP analysis. This was done to determine the variability in

the materials’ solubilities as a function of the crucibles’ heights. Furthermore, inert-XRD analysis was

done above and below the interface that separated the SOAE and hydridic electrolyte. For

arrangements A and B, 3-4 samples were tested simultaneously.

Fig. 2 Schematics for materials’ compatibilities and solubilities testing arrangements A, B, and C

2.2 Analyses

The inert-XRD patterns of hydridic materials were recorded using the X’Pert Pro™, multi-

purpose, x-ray diffractometer at that utilized Co K α radiation (45 V; 45 mA). In preparing the

hydridic materials, previously molten samples were firstly ground to fine powder with an agate mortar

and pestle in a commercial glove-box and transferred into 1 mm o.d. soft glass capillaries (Charles

Supper Co., MA, USA). The capillaries were sealed with a propane torch moments after their exit

from the glove-box. For comparative reasons, the hydridic materials’ XRD patterns were converted to

A B C

11


Cu Kα radiation with Jade™ 7 XRD Picture Processing, Identification and Quantification software.

XRD patterns were measured between 20-80 °2θ.

Hydridic materials’ hydride concentrations were measured using a familiar gas evolution

technique [1]. The concentrations of materials that had been immersed in the hydridic electrolytes

(2.1) were measured by inductively coupled plasma optical emission spectrometry (ICP) using a

Optima™ 2100 DV ICP-OES (PerkinElmer Inc., MA, USA) apparatus. Following equilibration

periods, ~1 g of the hydridic electrolytes were removed with pliers and a spatula from the regions of

the crucible that were not in physical contact with either the sintered ceramic or the powder containing

crucible; thereafter, they were placed in PTFE bottles and diluted with 20.00±0.01 ml of D.I. water.

The ICP apparatus was calibrated with standard solutions prior to measuring the metal concentrations

(wt% ppm) that are herein reported with respect to the composition of the molten salt. Concentrations

were calculated under the assumption that only NaH, and not NaBH4, was hydrolyzed during the

samples’ preparations.

3 Results and Discussion

3.1 Hydridic electrolyte synthesis

Fig. 3 shows the post-synthesis photos of 0.25±0.05 kg quantities of Na2BH5 mixtures. The

measured temperature differences between the furnace thermocouple, which was in contact with the

bottom of the glove-box’s custom pot, and the copper-sheathed, interior thermocouple that was situated

3.5±0. 5 cm above the custom copper crucible’s bottom was in the order of 135±15 °K for the first

several hours. It took ~24 hrs for their difference to equilibrate at 50±5 °K. The temperature of the

pressure vessel’s lid was 373±10 °K during the first several hours and slowly decreased to 323±10 °K.

The pressure vessel’s cooling coils caused salt vapours to condense in the pressure-vessel’s headspace;

ostensibly, the condensations impeded heat transfer between the furnace’s hot zone and the pressure

vessel’s lid; the latter’s temperature stagnated at 373±10 °K for the first several hours before it slowly

decreased to 323±10 °K. The photos given in Figs. 3(a)-(b) demonstrate the nature of the

condensations. Caution must be taken in their handling as they react violently with water and are

believed to be rich in NaH. For example, MSi’s (Table 4) condensation yielded only 53±1 % of the

theoretical volume of H2 as it corresponds to complete Na2BH5 hydrolysis (Eq. 17).

At ambient temperature, metal oxide-hydride mixtures have a distinct yellow colour. In order

to obtain reasonably homogeneous mixtures on this scale, it was necessary to soak them at 823±10 °K

for 2-3 days. Fig. 4 shows the photos of MSiii that was used for subsequent investigations of

materials’ compatibilities and solubilities. Again, there was a minor secondary phase at the bottom of

the crucible. As the interior thermocouple was not in direct contact with crucibles’ bottoms, the phase

may stem from some slight decomposition of NaH; consequently, small amounts of liquid sodium

metal may have diminished the eutectic temperature of the copper crucibles’ zinc-containing, silver

solder. Compared to the primary phase’s mass, the secondary phase’s mass was minor and had

crystalline character. As was found to be the case while saturating electrolytes with SOAE materials

for ERH experimentation [3], longer equilibration times may result in more homogenous mixtures.

12


Fig. 3 Photos of MSi (Table 4) after stability test f (Table 6) (a)-(b) condensations in the pressure

vessel’s headspace; (c)-( d) mixture following the manipulation from its crucible

Fig. 4 Photos of MSiii (Table 4) following its synthesis and before further use in material’s

compatibilities and solubilities experiments (Tables 5 and 6)

(c)

(a) (b)

(a) (b)

(d)

13


3.2 On materials’ compatibilities and solubilities

Listed in Table 5 and Table 6 are the experimental conditions, general post-equilibration

stability remarks, and analytical techniques(s) used in the investigations of electronic insulator and

ionic conductor compatibilities/solubilities with/in the hydridic electrolyte(s). Materials that were

deemed as unstable (NS) with the hydridic electrolytes are those that had clearly undergone a reaction.

Evidence for this was given by a noticeable colour change in the melt combined with an obvious

change in the investigated material’s volume. If the two aforementioned conditions were not met,

materials were deemed to be non-reactive/stable (S) but soluble. Very stable (VS) materials refer to

those that were easily recovered from their crucibles and maintained their integrity during and after the

hydrolysis of the excess hydridic electrolytes.

The hydridic electrolytes are reactive towards most of the commercial ceramic paints; or,

towards their proprietary binders, such as perhaps phosphates. Only aluminum nitride (AlN), Pyro-

paint™ 634-AL, and ZrO2-CaO were deemed to be somewhat stable. Depending on its form,

aluminum’s solubility varies significantly. (One might expect that Al2O3 would have rendered higher

solubilities because it shares a common ion (O2-

) with the melts.) While hydrolyzing the residual

hydrides that covered the 634-AL coated SS316 disc, hydrides were present through the ceramic layer

up to the metal-ceramic interface, and the coating dissolved into the aqueous solution. Residual water

in the ceramic layers, significant open porosity, or complex chemistry may have contributed to the

penetration. The melts, however, did not penetrate either the AlN or ZrO2-CaO layers; for AlN,

hydrolysis testing (2.2) of some recovered salt, which was not in physical contact with the substrate,

rendered 65±1 vol% of H2; ZrO2-CaO, remained adhered to the metal following the hydrolysis

procedure but its colour had changed from white to black during the equilibration period, and an

infinite electrical resistance was measured between neighbouring points at ambient temperature. The

sintered boron nitride (BN) sample was not compatible with the hydridic electrolyte and post-

hydrolysis of the sample resulted in foaming and sludge formation. Only the sintered Si3N4 bearing

appeared to be completely unaffected by the hydridic electrolyte.

SDC and Y2O3 seemingly reacted with the select molten salts. Because of the reduction of

Ce4+

to Ce3+

, it is improbable that doped ceria SOAEs are stable in these highly reducing media.

Although YSZ does not have the ionic required for either ERH or ERC (1.3), ScSZ does. Y2O3 is not

a material of great interest; its reactivity, however, may stem from the reaction given by (Eq. 4).

Notably, YH2’s thermodynamic properties are dubious [38]. The Zr concentration that was obtained in

compatibility test m (356 ppm) was close to the one obtained while investigating MgO*ZrO2 powder

(Table 5). The solubility/compatibility of Sc2O3 was not investigated.

LSGM and LSGFM were investigated by varying the experimental set-up, the hydridic

electrolytes’ composition, the equilibration times, and the materials’ history. The solubilities of the

LSGM’s individual metal oxides were also measured (h-l). Fig. 6 and Fig. 7 show the measured

solubilities. To determine the measurements’ variability as a function of the samples’ depths, after

tests f and g, samples were taken from the top, middle, and bottom of each of the large crucibles (Fig.

5).

i

14


Table 5 Electronically insulating materials’ compatibilities and solubilities testing summary (823±10 °K, 1.00±0.01 MPa (H2))

Material Source SU Analysis Stability Base MS ICP (ppm) ET (hrs) Form

903 HP* Cotronics A N/A NS Al2O3 iii - 24 P

919 Cotronics A N/A NS MgO-ZrO2 iii - 24 P

634-BN** Aremco B N/A NS BN iiv - 48 P

634-SIC** Aremco B N/A NS SiC iiv - 48 P

634-AL** Aremco B ICP S Al2O3 iiv 2670 48 P

AlN Alfa Aesar B ICP, H S AlN iii 147 (Al) 72 P

BN Carborundum Co. B N/A NS N/A iiv - 72 SC

MgO-ZrO2 Sultzer-Metco B ICP S N/A iiv 338 (Zr) 72 Po

ZrO2-CaO Alfa Aesar B RM S N/A iiv - 72 A

Al2O3 Alfa Aesar B ICP S N/A iiv 915 72 Po

Cu - A ICP S N/A iiv 321 48 C

Si3N4 McMaster Carr B ICP VS N/A iiv < 72 SC

SU - set-up (3.2); MS - molten salt (Table 4); ET - equilibration time; NS - not stable; AS - appeared stable; VS - very stable; H - hydrolysis; RM - resistance measurement; P - paste;

SC - sintered ceramic; Po - powder; A - Aerosol; C - crucible; *Resbond™, **Pyro-paint™

Table 6 Ionic conducting materials’ compatibilities and solubilities testing summary (823±10 °K, 1.00±0.01 MPa (H2))

Test Material Source SU Analyses Stability MS ET (hrs) form

Aa LSGFM GNP A SEM, EDS, XRD, ICP NS iii 24 SC

Bb LSGFM GNP A SEM, ICP NS iv 24 SC

Cc LSGFM GNP B ICP NS iii 72 Po

Dd LSGM GNP A XRD, ICP S iv 72 SC

Ee LSGM Praxair B ICP S iii 72 Po

Ff LSGFM GNP C ICP, inert-XRD NS ii 168 Po

Gg LSGFM GNP C ICP, inert-XRD NS iv 168 Po

Hh La2O3 Alfa B ICP S iv 72 Po

Ii La2O3 Alfa B ICP S iii 72 Po

Jj SrO Alfa B ICP S iv 72 Po

Kk Ga2O3 GNP B ICP S iv 72 Po

Ll MgO Alfa B ICP S iv 72 Po

Mm ZrO2 Alfa A ICP (356 ppm) S iv 72 Po

Nn SDC GNP B - NS iv 48 SC

Oo Y2O3 Alfa B - NS iiv 72 Po

Pp ß”-Al2O3 Ionotec B XRD, SEM S iiv 96 SC

SU - set-up (2.1); MS -molten salt (Table 4); ET - equilibration time; S - stable; NS - not stable; S - stable; H - hydrolysis; SC - sintered ceramic; Po - powder

15


Fig. 5 Photos of MSiii (Table 4) after test g (Table 6)

Contrary to the thermodynamic analyses (1.3), NaH seemingly did not reduced Ga2O3 (k) to its

metallic phase (Eq. 4). Minus the anomalous instances of the relatively high metal concentrations

measured at the bottoms of the large crucibles (f and g)-particularly the levels of strontium and

gallium-the data suggests that the solubility of La > Sr ≈ Mg > Ga > Fe with [La] ≤ 3000 ppm. In the

case of f, there was a noticeable color difference between the top and bottom of the large copper crucible

after the equilibration period. Hydrolysis testing (2.2) yielded 110% (top) and 68 vol% (lower) of the

theoretical volume of H2 (Eq. 17), respectively. The colour differential is less apparent in the case of

stability test g, where the metal concentrations are considerably less compared to stability test f, but high

compared to stability tests a-e. The most probable reason for these findings is that, upon cooling the

system, the soluble oxides crystallized out of solution; hence, a protocol that invokes sample quenching

will be required to more accurately determine the metal oxides’ solubilities. To be clear, a large error

should be included in the ICP values as the samples were not quenched, the equilibration times may have

been insufficient, accidental cross contamination during preparation of the samples may have occurred,

the metal oxides of interest may have been insufficiently soluble in the highly caustic solutions used for

the ICP analyses, metal oxides that are reduced to their metallic phase may have been insoluble in the

hydridic melts, and so on.

The LSGFM sintered ceramic discs from stability tests a and b withstood the hydrolysis of the

residual hydrides and were subsequently prepared so that their cross-sectional, SEM micrographs could

be recorded (Fig. 8). As shown in the bottom panels of the figure, the EDS analysis indicates that NaH

reduced Fe2O3 to metallic phases (Eq. 5 and Eq. 6); the other metal oxides were either washed away

during their preparation or had dissolved into the hydridic electrolyte during the equilibration periods.

top

middle

bottom

Interior crucible

LSGFM

(a) (b)

(d) (e)

16


The top-sectional, XRD pattern (Fig. 9(a)) further confirms that the reduction took place; hence, the

findings coincide with the thermodynamic analyses (1.3). By the thicknesses of the iron phases-which for

test a and b are 65±5 and 120±10 µm, respectively, the solubilities of LSGFM’s components increase

with the hydridic electrolyte’s [O2-

]. These analyses explain why iron’s measured solubilities in the

hydridic electrolytes were so low.

element

so

lub

ilit

y (

pp

m)

0

200

400

600

800

1000

1200

La Sr Ga Fe Mg

(a

solubility tests a-e

element

so

lub

ilit

y (

pp

m)

0

500

1000

1500

2000

2500

3000

3500

h

i

j

k

l

La Sr Ga Mg

(b

solubility tests h-l

Fig. 6 ICP atomic solubility measurements (ppm) of SOAE materials corresponding to the test conditions

given in Table 6-(a) circles - a, diamonds - b, squares – c, open hexagons – d, half circles - e; (b) tests h-l

are noted in the figure

element

so

lub

ilit

y (

pp

m)

0

500

1000

1500

2000

2500

3000

La Sr Ga Fe Mg

(b

solubility test g

element

so

lub

ilit

y (

pp

m)

0

2000

4000

6000

8000

10000

12000

La Sr Ga Fe Mg

(a

solubility test f

Fig. 7 ICP atomic solubility measurements (ppm) of LSGFM’s components (Table 6) as a function of the

large copper crucibles’ depths diamonds (tops), squares (middles), circles (bottoms) (a) test f; (b) test g

17


Fig. 8 Cross-sectional, SEM micrographs of a LSGFM sintered disc after exposure to the hydridic

electrolytes (Table 6)-(a) test a; (b) test b; their corresponding EDS, elemental line-scans are given in

their respective, bottom panels

Position (°2 theta)

20 30 40 50 60 70 80

Re

lati

ve

In

ten

sit

y

100

200

300

400

500

Fe

Fe

(a)


20 30 40 50 60 70 80

Re

lati

ve

In

ten

sit

y

0

200

400

600

800

LSGM

(b)

Fig. 9 (a) Top-sectional, XRD pattern of the LSGFM sintered disc following solubility test a (Table 6);

(b) XRD pattern of LSGM powder from its ground disc following solubility test d (Table 6)

Fe(Ka1) Ga(La1 2) Ga(La1 2) Fe(Ka1)

La (La1) Sr (La1) La (La1) Sr (La1)

90 µm 100 µm

(a) (b)

18


During the manipulation from its crucible, the LSGM sintered disc from stability test d could not

be kept intact; furthermore, upon hydrolyzing the residual hydridic electrolytes, large pieces broke up into

smaller ones. Prior to taking its XRD pattern (Fig. 9(b)), the sample was further dried for several days in

an oven at 373±5 °K and then crushed to a fine powder. Significant levels of water insoluble phases,

other than the desired perovskite phase, were not detected. From the inert-XRD analyses on the

materials contained in the interior crucibles (e.g. Fig. 5) after tests f (Fig. 10(b)) and g (Fig. 10(a)),

perovskite or other oxide phases associated with LSGFM were not detected, which is perhaps due to the

insufficient depths at which the samples were taken. Clearly, other than hematite (Fe2O3), which was

reduced, LSGFM’s components are significantly soluble in the melts. The inert-XRD pattern, shown in

Fig. 10(a), contains several unidentified peaks which may correspond to other complex phases stemming

from hydridic electrolyte-SOAE interactions. Because of the formation of metallic iron in the more

hydridic electrolyte, unequivocally, the most promising SOAE for the ERH or ERC processes cannot be

in direct contact with these media; rather, a thin layer of LSGM or ScSZ could perhaps be used to protect

the LSGFM layer. Peaks that correspond to the previously identified [1] compound, Na6B2O5H2, are

listed in Table 7.


25 30 35 40 45 50

Rela

tiv

e In

ten

sit

y

500

1000

1500

2000

?? ?

? ?

?

NaBH4

Na6B2O5H2

(a


25 30 35 40 45 50

Rela

tiv

e In

ten

sit

y

0

500

1000

1500

2000

2500

3000

NaBH4

NaH

Fe

(b

Fig. 10 (a) Inert XRD patterns for samples collected above (top), at (middle), and below (bottom) the

catholyte-LSGFM interface after test g (Table 6); (b) Inert XRD patterns of the samples collected above

(top), at (middle), and below (bottom) the catholyte-LSGFM interface after test f (Table 6)

Table 7 XRD data for the compound Na6B2O5H2 (Cu Kα)

Position (º2θ) Relative Intensity (%)

34.28 55

34.73 68

39.39 100

40.22 52

44.47 80

47.38 64

56.71 25

57.61 31

59.33 26

63.19 27

19


Among other matters, the determination of metal oxide solubility limits could be accomplished with

concentrations cells that utilize ß”-Al2O3. By comparing the relative intensities of the before and after

XRD patterns, it is found that ß”-Al2O3’s presence diminishes while ZrO2’s presence remains relatively

constant. As well, the top-sectional, SEM micrographs shows material loss. It is inconclusive to what

extent the subsequent hydrolysis affected the material loss. ß”-Al2O3’s is known to be hygroscopic

material; post-equilibration, neither the melt nor tube appeared affected by their interaction.


10 20 30 40 50 60 70

Rela

tiv

e In

ten

sit

y

0

200

400

600

800

? ?? ???

?

?

????

? ?

ß"-Al2O3

ZrO2

Uknown?


10 20 30 40 50 60 70

Rela

tiv

e In

ten

sit

y

0

200

400

600

800

??? ?

??

?

?? ?

??

? ?

ß"-Al2O3

ZrO2

Uknown?

Fig. 11 Analyses of ß”-Al2O3 before (left panels) and after (right panels) test p (Table 6) – (a)-(b) SEM

micrographs (top section); (c)-(d) EDS spectra (top sections); (e)-(f) XRD patterns (top sections)

4 Conclusions

The rudimentary, thermodynamic analyses indicate that the double-doped, LaGaO3 perovskite family

is unstable with the Na2BH5-Na4B2O5-H2 quasi-ternary system as NaH reduces both Ga2O3 and Fe2O3 to

their metallic states. The former could not be proven but the latter was unambiguously validated.

However, ScSZ, may be thermodynamically stable under ERH and ERC polarizing/operating conditions.

Zr’s solubility is relatively low but nothing is known about the compatibility of Sc2O3 with such liquid

20


systems. Future solubility testing may be accomplished with concentration cells that invoke ß”-Al2O3, or

via the quenching method. If ScSZ’s components are stable and only slightly soluble with the

compositions near the liquid eutectic of the liquid, anhydrous Na+,B

3+,H

+/H

-,O

2- system, then the

conditions would be such that fundamental ERH or ERC investigation that utilize a LSGFM-supported

design.

Acknowledgements

The financial support received from Kingston Process Metallurgy Inc., AUTO21, and NSERC made this

project possible. The construction and design of the custom pressure-vessel and glove-box, along with

SEM, XRD and EDS analyses, were aided by F. Barrette, S. Gutierrez, K. Béré, B. Couture, and I Kelsey-

Léveque. As well, K. Gholshani (KPM Inc.) and A. Grant (Queen’s University) were instrumental in the

ICP and inert-XRD analyses.

References

[1] Calabretta DL, Davis BR (2007) J. Power Sources 164(2): 782-791

[2] Calabretta DL (2009) Proceedings of the 2009 Hydrogen and Fuel Cells Conference

(Vancouver, Canada)

[3] Calabretta DL (2012) Université de Sherbrooke (PhD thesis)

[4] Calabretta DL (2013) retrieved February 6, 2013, from

http://www.scribd.com/doc/122853366/Radio-frequency-induction-coupled-

suspension-plasma-spraying-of-dense-ceramic-layers-on-metal-substrates

[5] Marincek B (1971) U.S. Patent 3562135

[6] Sammels AF, Semkow KW (1988) J. Power Sources 22: 285-291

[7] Pal UB, Woolley DE, Kenney GB (2001) J. Metals 32: 32-35

[8] Woolley DE, Pal UB, Kenney GB (2001) High Temp. Mater. Processes 20(3-4):

209-218

[9] Krishnan A, Lu XG, Pal UB (2005) Metall. Mater. Trans. B 36B: 463-473

[10] Krishnan A, Lu XG, Pal UB (2005) Scand. J. Metall. 34(5): 293-298

[11] Pal UB, Powell IV AC (2007) J Metals May: 44-49

[12] Suput M, DeLucas R, Pati S, Ye1 G, Pal UB, Powell IV AC (2008) Trans. Inst. Min.

Metall., Sect. C 117(2): 118-122

[13] Pal UB (2008) J Metals February: 43-47

[14] Martin A, Lambertin D, Poignet JC, Allibert M, Bourgès G, Pescayre L, Fouletier

(2003) J. Metals October: 52-54

[15] Martin A, Poignet JC, Fouletier J, Allibert M, Lambertin D, Bourgès G (2010) J.

Appl. Electrochem. 40: 533-542

[16] Combes R, Vedel J, Trémillon B (1975) Electrochim. Acta 20: 191–200

[17] Combes R, Trémillon B (1977) J. Electroanal. Chem. 83: 297-308

[18] Combes R, De Andrade F, De Barros A, Ferreira H (1980) Electrochim. Acta 25:

371-374

[19] Rapp RA (2000) Metall. Mater. Trans. A 31A: 2105-2118

[20] Rodrigues CMS, Labrincha JA, Marques FMB (1997) J. Electrochem. Soc. (Solid

State Science and Technology) 144(12): 4303-4309

[21] Rodrigues CMS, Labrincha JA, Marques FMB (2000) Solid State Ionics 136-137:

671-675

[22] Stournaras CJ, Tsetsekou A, Zambetakis T, Kontoyannis CG (1995) J. Mater. Sci. 30

4375-4379

[23] Kontoyannis CG (1996) J. Mater. Sci. Lett. 5: 222-224

[24] Windisch CF, Bates JL, Boget DI (1987) J. Am. Ceram. Soc. 70(9): C220-C221

http://www.scribd.com/doc/122853366/Radio-frequency-induction-coupled-suspension-plasma-spraying-of-dense-ceramic-layers-on-metal-substrates

http://www.scribd.com/doc/122853366/Radio-frequency-induction-coupled-suspension-plasma-spraying-of-dense-ceramic-layers-on-metal-substrates

21


[25] Nagelberg AS (1985) J. Electrochem. Soc. 132(10): 2502-2507

[26] Rapp RA, Zhang Y (2002) Light Metals 463-468

[27] Rapp RA, Zhang Y (2002) Light Metals 469-474

[28] Gong W, Gopalan S, Pal U (2006) J Power Sources 160: 305-315

[29] Ivers-Tiffée E (2009) Encyclopedia of Electrochemical Power Sources 181-187

[30] Ishihara T, Matsuda H, Takita Y (1994) J Am Chem Soc 116: 3801-3803

[31] Ishihara T (2006) B Chem Soc Jpn 79(8): 1155-1165

[32] Ishihara T, Enoki M, Yan J, Matsumoto H (2005) Solid Oxide Fuel Cell (SOFC),

Electrochemical Society Proceedings: 1117-1126

[33] Ishihara T, Ando M, Enoki M, Takita Y (2006) J Alloy Compd 408-412: 507-511

[34] Enoki M, Yan J, Matsumoto H, Ishihara T (2006) Solid State Ionics 177: 2053-2057

[35] Calabretta DL (2006) unpublished result

[36] Mulford RNR, Holley CE (1955) J. Phys. Chem. 59: 1222-1226

[37] Hayward MA, Rosseinsky MJ (2000) Chem. Mater. 12: 2182-2195

[38] FactSage™ 6.2 ThermFact and GTT Technologies

[39] Lyci G, Surendra KS (2010) Int. J. Hydrogen Energy 355: 5454-5470

[40] Tkacz M (2002) J. Alloys Compd. 330–332: 25-28

[41] Mulford RNR, Holley CE (1955) J. Phys. Chem. 59: 1222-1226

[42] Klaveness A, Swang O, Kjekshus A, Fjellvåg H (2006) Inorg. Chem. 45: 10698-

10701

[43] Purewal J, Hwang SJ, Bowman RC, Rönnebro E, Fultz B, Ahn C (2008) Journal of

Physics and Chemistry C 112: 8481-8485

[44] Wojciech G, Edwards P (2004) Chem. Rev. 104: 1283-1315

[45] Schrauzer GN (1955) Die Naturwissenschaften 42: 438

[46] Nakamori Y, Li HM, Matsuo M, Miwa K, Towata S, Orimo S (2008) J. Phys. Chem.

Solids 69: 2292-2296

[47] Schaeffer GW, Roscoe JS, Stewart AC (1956) J. Am. Chem. Soc. 78: 729-733.

[48] Łodziana Z (2010) Phys. Rev. B: Condens. Matter 81: 144108

[49] Marks TJ, Kolb JR (1977) Chem. Rev.77(2): 263-293

[50] Nakamori Y, Li HW, Kikuchi K, Aoki M, Miwa K, Towata S, Orimo S (2007) J.

Alloys Compd. 446-447: 296-300

[51] M. Blander; Molten Salt Chemistry, John Wiley and Sons Inc., New York, ©1964

Documents

On the Compatibility of Electronic Insulators and Ionic Conductors With the Na2BH5-Na4B2O5-H2 Quasi-ternary System