Radio-frequency induction-coupled suspension plasma spraying of dense ceramic layers on metal substrates

1

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

Radio-frequency induction-coupled suspension plasma spraying of dense

ceramic layers on metal substrates

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 29 January 2013

Abstract

Results pertaining to the fabrication of seal-less, metal-supported, electrolytic reverse hydrolysis anode

compartments via radio-frequency induction-coupled suspension plasma spraying (SPS) are presented

herein. The glycine nitrate process (GNP) facilitated the synthesis of contemporary metal oxide powders.

Variable SPS parameters include: GNP-powder tailoring, substrate material, solvent, post-heat-treatment,

substrate cooling arrangement, and substrate holder type/arrangement. The latter most parameter required

a new design in order to eradicate objectionable, plasma-substrate arcing. Depositions were characterized

by energy dispersive x-ray spectroscopy, scanning electron microscopy, and x-ray diffraction. Optimized

anode compartments having La0.7Sr0.3Ga0.7Fe0.2Mg0.1O3-δ as its primary electrolyte, and up to five

additional functional layers, were achieved. The anode compartments may be considered for other

applications, such as intermediate temperature solid oxide fuel cells, and are in the early stages of

electrochemical performance testing.

Keywords: Radio-frequency induction-coupled suspension plasma spraying; Intermediate temperature

solid oxide fuel cell; LSGFM; Thermal spray processes; Coatings

1. Introduction

For over a decade, researchers at CREPE have been investigating the use of thermal spray

processes for the relatively rapid fabrication of metal-supported, intermediate temperature (773±100 ºK)

solid oxide fuel cells (IT-SOFCs). If the anode or cathode layer is deposited onto the porous metal

support prior to the deposition of the solid oxygen anion electrolyte (SOAE) layer, then the integrated

component can be used as anode compartments for the hypothetical electrolytic reverse hydrolysis (ERH)

or electrolytic reverse combustion processes, respectively. These hydrogen storage processes have been

described elsewhere [1-3] and are of principal interest to the author. But the investigations pertaining to

SPS are of interest to the IT-SOFC community, and, more generally, to those interested in thin, dense,

ceramic depositions on metal substrates.

Reviews [4, 5] show that double-doped LaGaO3 perovskites, doped ceria fluorites, and scandia

stabilized zirconia (ScSZ) are the most relevant SOAEs for the processes under investigation. In 1994,

Ishihara et al. [6] reported on the near pure oxygen anion (O2-

) conducting, double-doped, LaGaO3

perovskite family. A review [7] of the first decade of international investigation of this family is

available. Later, the same group reported [8-10] on the extraordinary electronic properties of

La0.7Sr0.3Ga0.7Fe0.2Mg0.1O3-δ (LSGFM). Although hardly investigated since its discovery, LSGFM has O2-

conductivities that are approximately two orders of magnitude higher than the conventional yttria

stabilized zirconia (YSZ) at the intermediate temperature range. Using LSGFM as their primary SOAE

[8], 0.5 mm thick, electrolyte-supported IT-SOFCs were fabricated. An additional La0.9Sr0.1Ga0.8Mg0.2O3-δ

(LSGM), SOAE interlayer was deposited on the LSGFM substrate by pulsed laser deposition (PLD) and

90wt%Ni-Fe electro-catalytic anode and Sm0.5Sr0.5CoO3-δ cathode layers were applied by screen-printing.

2


Despite the thickness of the LSGFM layer, their cells yielded power densities of 0.2 and 0.1 W cm-2

at

873 and 773 °K, respectively. LSGFM has partial electronic conductivity and the additional LSGM layer

(<1µm) blocks electron flow to the anode-SOAE interface which thereby affords nearly pure O2-

conduction through the multi-layer assembly; furthermore, a samarium or gadolinium doped ceria (SDC

or GDC) interlayer (<1µm) was required between the anode and LSGM layers in order to prevent the

formation of insulating, nickel lanthanate phases. Thus far, no group has attempted to fabricate metal-

supported IT-SOFCs having LSGFM as its primary SOAE via TSDPes.

A brief history of radio-frequency induction-coupled suspension plasma spraying (SPS)

deposition technology is warranted. In the patent’s invention summary [11], the physical steps that the

sheared suspension droplets experience upon exiting the atomization probe (into the plasma core) were

described consecutively as: (a) vaporizing the carrier substance; (b) agglomerating the small particles into

at least partially melted drops; (c) accelerating these drops; and (d) projecting the accelerated drops onto

the substrate to form the material deposit. This is true if inert gas or O2 plasmas with aqueous solvent

based suspensions are used.

Consideration of the solvent-plasma interaction(s) and its effect(s) on deposition quality has been

lacking. If hydrocarbons are injected into O2 rich plasmas, reactions tending towards the hydrocarbon’s

complete oxidation will occur. These processes are generally exothermic and should therefore cause

localized heating around the solid particles; hence, if compared to aqueous suspensions, an increase in the

solid’s heating and fusion rates can be expected. No mention of this significant factor was made by

Bonneau [12] who attempted to deposit conventionally sized (5≤dp≤26) electrolyte (e.g. CeO2, ZrO2,

La2O3, etc.) powders as dense SOAE layers by injecting suspensions comprised of glycol-water mixtures

(solvent) into an O2 rich plasma. Due to the difficulties in maintaining the powders in suspension,

negative results were obtained. In Jia’s bibliographic review [13], other than-“low torch power limits the

ability to use high solid content suspensions and often requires suspensions to be alcohol based to lower

the energy required to vaporize the suspending liquid”-no mention of the potential effect(s) of the

suspension’s solvent character on the overall deposition process’s energy balance were made. Schiller et

al. [14] did remark on the possibility of increasing the plasma’s enthalpy by combustion reactions, but this

point has not been sufficiently elaborated on in the literature. A recent review [15] on the preparation of

IT-SOFCs via thermal plasma processing did not discuss the possibility of harnessing exothermic,

solvent-plasma interactions either.

For SPS, suspension comprised of organic solvents being injected into an O2 rich plasma would

more appropriately be characterized as a hybrid between the high velocity oxy-fuel and SPS technologies,

and could be defined as radio-frequency induction-coupled oxygen-fuel vacuum suspension plasma

spraying. Due to the lengthiness of such an acronym, however, it is more practical to use the trademark

acronym, SPS. Rudimentary considerations are possible using the solvents’ available thermodynamic

properties [16]. Assuming the complete oxidation of two solvents, where H2O and CO2 are the reactions’

products, at standard temperature and pressure, the exothermic heat addition per unit time divided by the

power dissipation in the radio-frequency induction-coupled plasma torch, which is typically 45±5 % of

the plate power, is here defined as the maximum power of complete oxidation to power dissipation ratio

( PDCO PP ). In Fig. 1 the ratios have been plotted as a function of the solvent flow-rate for the solvents

methanol (CH3OH), and pentadecane (C15H32). The latter has been used to approximate mineral oil.

Clearly the type of organic solvent and its flow-rate are significant variables in the deposition processes’

overall energy balance. In the studies of Bonneau [12], Schiller [14], and Jia [13], the O2 sheath gas flow-

rates were always well in excess of the stoichiometric amounts required for the complete oxidation of

their respective solvents. More detailed considerations of the effect of the suspension’s wt% solids is

warranted but will not be treated here; rather, the results of a two-year experimental study of SPS of

dense, ceramic layers are presented.

3


flow rate (ml min-1

)

0 2 4 6 8 10 12 14 16 18

PC

O/P

PD

0.0

0.1

0.2

0.3

0.4

0.5

C15H32

CH3OH

Fig. 1 Maximum power of complete oxidation to power of heat dissipation ratios’ dependencies on

MeOH (672.14 kJ mol-1

; 0.792 g ml-1

) and C15H32 (9414.85 kJ mol-1

; 0.769 g ml-1

) flow rates; 45% overall

plasma circuit efficiency operating at 50 kW plate power

1.1 Preparation of integrated, metal-supported SOFCs by TSDPes

The IT-SOFCs that have achieved the highest electrochemical performances were prepared via

the vaporization deposition techniques RF-magnetron sputtering [17, 18] and pulsed laser deposition [19,

20] having deposition rates of 10-4

μm s-1

and 10-3

μm s-1

, respectively. The most exceptional IT-SOFCs

were developed by Ishihara et al. and employed the SOAE La0.9Sr0.1Ga0.8Mg0.2O3-δ. A thin interlayer of

Sm0.2Ce0.8O2-δ was deposited between the anode and LSGM layer in order to prevent the formation of

nickel lanthanates.

Despite the amount of literature related to the preparation of metal-supported SOFCs by TSDPes,

there are few reports detailing the electrochemical performances of complete cells. Recent reviews [21,

22] on integrated, metal-supported SOFC preparation via TSDPes can be compared with the earlier

review by Henne et al. [23].

Takenoiri et al. [24] prepared nickel felt-supported SOFCs via atmospheric plasma spraying

(LaCoO3 cathode and YSZ SOAE) and high velocity oxy-fuel (anode) technologies. Ceramic-coated

metal separators were used in their thirty, seal-less stack that achieved power densities of 0.3 W cm-2

.

Stack degradation was significant while operating at 1243 °K for 2400 hrs. Tsukuda et al. [25]

constructed calcia-stabilized zirconia supported, tubular IT-SOFC having a Ni-YSZ cermet anode, YSZ

SOAE, and LaCoO3 cathode. Their cell performances of ~0.5 W cm-2

were similar to analogous cells

fabricated by co-sintering procedures.

4


Shiller et al. [26] and Lang et al. [27] reported on SOFCs prepared by vacuum plasma spraying

(using Laval-type nozzles) the functional layers on nickel-alloy, porous, metal filters and nickel-felt

substrates. Power densities were measured to be 0.3-0.4 W cm-2

between 1023-1073 °K. Compared to

atmospheric plasma spraying, vacuum plasma spraying can decrease the SOAE’s permeability by about

an order of magnitude. The high melting temperature YSZ and scandia stabilized zirconia (ScSZ) SOAEs

were sprayed directly into the arc while electrode materials were injected downstream. Electrochemical

characterization was later reported [28] for larger SOFCs using the same fabrication procedure. Schiller

et al. [29] assembled anode-supported SOFCs and achieved similar electrochemical performances to those

reported on by Lang et al. [27]. By obtaining a maximum power density of 0.5 W cm-2

at 1073 °K, Stöver

et al. [30] set a new benchmark for mostly TSDP-prepared SOFCs that were comprised of NiO-YSZ

anodes and YSZ SOAEs.

In regards to TSDP for the fabrication of SOFCs having double-doped, LaGaO3 electrolytes, Ma

et al. [31] reported on near theoretical open circuit voltages for their Ni-YSZ/LSM/LSM button cells.

SOAE layer thicknesses of 65±15μm were deposited by atmospheric plasma spraying, which rendered

power densities of 0.08-0.15 W cm-2

between 773-1073 °K; however, in order to obtain an acceptable

level of perovskite phase, a post-deposition, heat-treatment step was required. Hwang et al. [32] later

reported on Ni-YSZ/LSGM/LSCF SOFCs prepared by atmospheric plasma spraying technology on nickel

substrates. Compared to the cells prepared by Ma et al., they obtained superior power densities.

To the author’s knowledge, at the start of this project, a fully-functional, metal-supported IT-

SOFC had not yet been assembled and tested at CREPE. Efforts towards the development of small IT-

SOFCs have always been in regards to the individual cathode [33, 34] or electrolyte [11, 12, 35, 36]

layers.

2.0 Experimental

2.1 The glycine nitrate process

Several metal oxide powders were synthesized by the glycine nitrate process (GNP). These

materials’ coefficients of thermal expansion (CTEs), along with those of the principal metal substrate

materials, are listed in Table 1. Reagent grade starting materials having the following purities were

obtained from Alfa Aesar: La(NO3)3*6H2O (99.9 wt%), Fe(NO3)3*9H2O (98.1 wt%), Sr(NO3)2 (99.9

wt%), Mg(NO3)2*6H2O (99.9 wt%), Sm(NO3)3*6H2O (99.9 wt%), Ce(NO3)3*6H2O (99.5 wt%),

Ni(NO3)2*6H2O (99.5 wt%), Ga (99.999 wt%) and C2H5NO2 (98.5 wt%). Gd0.1Ce0.9O2-δ SOAE (Sigma

Aldrich; 734667) having particle diameters (dp)≤0.5µm was also used without further tailoring.

For the synthesis of powders by the glycine nitrate process (GNP), the vessel used at CREPE for

the synthesis of nano-powders [37] by radio-frequency induction-coupled solution plasma spraying was

slightly modified. Fig. 2 shows the block-schematic for the set-up. Briefly, metal nitrate salts of the

corresponding metal oxides are dissolved in water, or the metals, i.e. gallium, is reacted and dissolved in

[HNO3]. Glycine (C2H5NO2) is then dissolved into the solution which is subsequently heated and agitated

on a hot-stir-plate. After a few minutes, the majority of the solution’s water is evaporated and the

homogeneous, viscous fluid reacts exothermically rendering powders with sub-micron grain sizes.

During the synthesis of GNP powders, best results were obtained when the double-walled vessel

was placed in a horizontal position and maintained at 343±10 °K with a thermostatic water bath. This

avoided excess condensation on the reactor walls thus making it more agreeable to collect the powders.

The system’s pressure was maintained at 54±14 103 Pa and the reaction vessel’s temperature was

monitored with a hand-held recorder and a 3.175 mm o.d., SS-sheathed, K-type, grounded thermocouple

(Omega™) that was fitted into place with a Swagelok™ component. As well, a 1.7±0.1 C2H5NO2 to

metal mole ratio was used and 75±25 ml volumes of GNP solution were reacted each time. The reaction

vessel’s temperature would stagnate after several minutes of heating; shortly thereafter, simultaneous

increases in the system’s pressure and reaction vessel’s temperature indicated the initiation of the powder

formation process.

5


Typically, in order to eliminate phase impurities, following their synthesis, ~100 g samples of

GNP powders were placed in MgO crucibles (Ozark Technical Ceramics, MO, USA) and heated to ≤1780

°K at 1.5±0.5 °K min-1

in a RHF 17/3M (Carbolite Ltd., London, UK) resistance furnace. Then, the

powders were pulverized in an 8000M Mixer/Mill (Spex SamplePrep LLC, NJ, USA), and subsequently,

with a Sonic Sifter (ATM Corp.; WI, USA), sieved to dp≤53 μm.

Table 1 mean CTEs for the ERC and ERH anode compartment materials

*mass basis

Fig. 2 Schematic of the arrangement used for the synthesis of GNP powders

2.2 SPS for the preparation of ERH anode compartments

The SPS apparatus’s cross-view schematic is shown in Fig. 3. Details on the apparatus’s

suspension atomizer, torch, and substrate holder, along with other relevant points on the preparation of

substrates have been given elsewhere [38]. Constant SPS parameters are listed in Table 2.

Material (mole basis) CTE (10-6

m m-1

°K-1

) Reference

Sm0.2Ce0.8O2-δ 12.7 45

La0.9Sr0.1Ga0.8Mg0.2O3-δ 11.6 7

La0.8Sr0.2Ga0.8Mg0.2O3-δ 12.1 46

La0.7Sr0.3Ga0.7Fe0.2Mg0.1O3-δ 13.6 10

Gd0.2Ce0.8O2-δ 11.6 5

Inconel™ 600 14.5 -

Hastelloy™ X 15.6 -

*Ni0.9Fe0.1 13.2 ca.

reaction

vessel

hot-plate

stirrer

double-walled

powder

collection-vessel

filtration-

vessel

thermostat

to vacuum

pump

6


Table 2 Constant SPS parameter values

Parameter Value

icentral plasma gas(Ar) flow rate (slpm) 22±1

isheathe plasma gas (O2) flow-rate (slpm) 61±1

atomization gas (Ar) flow-rate (slpm) 11±0.5

deposition chamber pressure (kPa) 13.3±6.7

plasma plate power (kW) 50±1

suspension injection flow rate (ml min-1

) 5.5±1.5

spraying distance (cm) 7±3

atomization gas pressure (kPa) 345±10

Fig. 3 Cross-sectional schematic of the apparatus used for SPS-(1) electrically-grounded water-cooled

mask; (2) copper plaque; (3) SS304 water-cooled substrate holder; (4) double-walled, aluminum, water-

cooled jacket; (5) flexible SS316 hose; (6) dielectric fitting; (7) Teflon nipple; (8) metallic adapter

Initial investigations of the character of SPS depositions were done on commercial, porous or

solid, metal substrates, which were neither pre- nor post-heated by the plasma jet. While ideally the

water-cooled

automated arm

high-pressure

water inlet

vacuum

high-pressure

water outlet

high-pressure

water outlet

Faraday cage

peristaltic pump suspension

stir-plate

torch

deposition

chamber

1

2

3

5 4

6

7

8

7


deposition rates would refer to the coatings’ thicknesses per unit area and unit time, and would be

reported in units of µm cm-2

s-1

, where the coatings’ thicknesses-determined from their cross-sectional

micrographs-, and the area and time factors- determined from the cooling mask’s inner diameter (Fig. 3)

and the substrate-plasma jet exposure-time during each loop, which are known to be 3.834 ± 0.001 cm

and 2.5 ± 0.1 s, respectively-, the accuracy of the time variable renders meaningless deposition rates;

hence, the deposition rate are given in µm loop-1

2.3 Design and preparation of ERH anode compartments

Both ‘pipe’ (Fig. 4(a)) and ‘NPT’ (Fig. 4(b)) anode compartment designs were investigated; after

several negative results, the former was discarded for the latter. It was imperative for the substrates’

interior and exterior walls to be intensively cooled amid SPS; this necessitated the NPT design.

A 1.91 cm o.d., Inconel™ bar was machined, in-house, to the GKN™ flow-restrictor

specifications (1.27±0.01 cm i.d.), sand-blasted, cleaned, and shipped to the manufacturer where SIKA-

HX-5 powders were pressed and sintered into the part. There appeared to be excellent porous-solid joint

integrity, and further machining of the integrated part was done as required throughout the SPS

optimization. For the final preparation of the integrated parts, its non-porous region were sand-blasted

prior to washing it in alternating baths of 1 N HCl and 20 vol% H2O2 -as was deemed necessary-and

placing it in an EtOH, ultra sound bath for several hours.

Fig. 4 Cross sectional schematics of the metal-supported, ERH anode compartments substrates – (a) pipe

design; (b) NPT design

For the preparation of ERH anode compartments, protocols were taken from the studies of Shiller

et al. [26], Lang et al. [27] and Ishihara et al. [19, 39, 40]; the deposition of a dense layer of NiO-Fe2O3 on

nickel-alloy, GKN™ porous metal filters, followed by its H2 reduction, prior to or after the depositions of

the electrolyte layers, is essentially a hybrid of the research groups’ designs. The selection of SPS

feedstock materials coincided with those used by Ishihara et al; however, LSGFM typically took the place

of LSGM as the principal electrolyte.

2.4 Analyses

XRD patterns of were acquired using an X’Pert Pro™, multi-purpose, x-ray diffractometer

(Philips, Eindhoven, Netherlands); it utilized Cu Kα radiation (45 V; 45 mA).

A model S-4700 (Hitachi, Tokyo, Japan) field-emission scanning electron microscope (SEM),

equipped with an energy dispersive x-ray spectroscopy (EDS) apparatus, facilitated the micrographic and

elemental analyses of the GNP synthesized powders.

influent water

effluent water

exterior water-cooling

interior water-cooling

copper plaque

integrated NPT ERH anode

compartment (b)

0.48 cm OD SS316 tube

0.16 cm OD SS316 tube

TIG weld

1.27 cm OD

GKN™-HX-5

1.91 cm OD machined SS316 bar

1.91 cm OD machined Inconel bar

(a)

8


Prior to acquiring an SPS coatings’ cross-sectional, SEM micrographs– with either the S-4700, a

JEOL J840A (JEOL Ltd., Tokyo, Japan) or a variable pressure N-3700 (Hitachi, Tokyo, Japan)

apparatus-a homogeneous mixture of epoxy resin and hardener (Precision Surfaces Int., TX USA) was

poured into the sample containing molds and left to cure at ambient conditions for two days. The

hardened, sample-containing epoxy was cut with an ISOMET 2000 (Buehler Co., IL, USA), diamond

saw; the corresponding face was manually polished with an Ecomet 3-variable speed grinder (Buehler

Co.) with 180-600 grit silica-papers, and the final polishing employed 6 µm diamond suspensions

(Precision Surfaces Int.). Lastly, a thin layer of 80wt%Au-Pd was applied with a Hummer IV Sputtering

System (Anatech; CA, USA), which was operated at 10 mA and 90±10 Pa for ~1 min.

On one occasion the adherence and porosity of SPS depositions on the ERH anode compartment

was investigated by measuring its open circuit voltage (OCV) using the set-up depicted in Fig. 5.

Following the depositions, 9912-G silver conductor paste (ESL Electroscience; PA, USA) was applied as

a cathode (SOFC) layer. Then a layer of Aremco 634-AL Pyro-paint™ was coated on the ERH anode

compartments’ exterior walls with a paint brush. A nickel wire, which was inserted in an Al2O3 tube and

coiled at its end, was used to make the electrical contact with the silver cathode. Electrical contact was

made with the anode via a Swagelok™, SS316, male, NPT fitting. The OCV measurements were taken

with a hand held multi-meter. Electrical contacts were made at the exterior of the furnace at the 1.56 mm

o.d., SS316, influent H2 tube and the nickel wire. 99.9 vol% H2 and 99.9 vol% O2 were passed through

their respective compartments at 200 ± 50 ml min-1

. Teflon tubing and Swagelok™ components were

used to electrically insulate the influent H2 line from its reservoir, and a grounded, k-type, Omega™

thermocouple was inserted in an alumina tube such that its end was in physical contact with the ERH

anode compartment’s electrically insulating layer. The arrangement was heated at 2 K min-1

to 923±10

°K in a Lindberg 55035 mini mite, horizontal, tubular furnace (Lindberg/MPH; MI, USA).

Fig. 5 Experimental arrangement used for OCV measurements

3. Results and discussion

Results and discussions regarding the synthesis and characterization of the SPS feedstock

powders will be given first. This will be followed by the results and commentary on the initial

investigations pertaining to SPS and succeeded by the presentation of its relative, follow-up research,

particularly as it relate itself to the eradication of plasma-substrate arcing. Consecutively, inquests

towards the fabrication of ERH anode compartments are given.

O2

Al2O3 tube

influent H2 effluent H2

quartz tube

Al2O3 tube Ni wire

634-AL pyro-paint™

tube furnace resistance coils

1.56 mm o.d.

SS316 tube

4.76 mm o.d.

SS316 tube

NPT-type ERH anode

Swagelok™-NPT fitting

sheathed k-type

thermocouple

9


3.1 Characterization of GNP powders

During the GNP synthesis of the supposed nickel/iron oxide mixture (NFO), there was a

remarkable increase in the reaction vessel’s temperature (673 ± 10 °K), and, as shown by its XRD pattern

in Fig. 6, the as-synthesized powder was identified as Awaruite (Ni3Fe). However, subsequent air

sintering at 1773 ± 10 °K for 8 hrs rendered the desired composition, which was ferro-magnetic and

tended to agglomerate which therefore made it arduous to sieve to the preferred particle sizes (dp < 32

μm).

Position (°2 theta)

20 30 40 50 60 70 80

Rela

tive I

nte

nsit

y

0

2000

4000

6000

8000

10000

Fe3O4

NiO

FeNi3

Fig. 6 XRD patterns for GNP synthesized NFO powder before (bottom) and after air sintering at 1773 ±

10 °K for 8 hrs

The La0.8Sr0.2Ga0.8Mg0.2O3-δ, La0.7Sr0.3Ga0.7Fe0.2Mg0.1O3-δ, and Sm0.2Ce0.8O2-δ GNP synthesized

powders that were obtained from the GNP collection vessel were ethereal. During the reaction, the

reaction vessel’s temperature did not exceed 473±10 °K, over-pressurization of the collection vessel did

not occur while applying the specified vacuum conditions, and about half of the materials could be

collected directly from the reaction vessel (Fig. 2).

Following the GNP synthesis of La0.8Sr0.2Ga0.8Mg0.2O3-δ, the resultant phases coincided to with

what was already known [44]. As shown in Fig. 7(a), the insulating, SrLaGa3O7 compound, which was

initially significant, diminished with the air-sintering temperature, and became negligible after soaking it

at 1748±10 °K for 6 hrs. Fig. 7(b) is the EDS spectrum of La0.8Sr0.2Ga0.8Mg0.2O3-δ following its air-

sintering at 1748±10 °K. EDS does not allow for the detection of magnesium; however, the subsequent

two-electrode conductivity tests affirmed its composition [41].

10



20 30 40 50 60 70

Rela

tiv

e In

ten

sit

y

0

2000

4000

6000

8000

10000

12000

14000

16000

? ? ? ? ? ? ?

La0.8Sr0.2Ga0.8Mg0.2O3-

SrLaGa3O7

SrLaGaO4

uknown?

1523 ºK

1748 ºK

1623 ºK

(a)

keV

2 4 6 8 10

co

un

ts

0

200

400

600

800

1000

O

LaLa

La

La

La

La

La

La

Ga

GaSr

(b)

Fig. 7 Analyses for GNP synthesized La0.8Sr0.2Ga0.8Mg0.2O3-δ powder-(a) XRD patterns before (bottom)

and after air sintering for 6 hrs at 1523, 1623, and 1748±10 °K; (b) EDS spectrum for

La0.8Sr0.2Ga0.8Mg0.2O3-δ after air sintering at 1748±10 °K

For La0.7Sr0.3Ga0.7Fe0.2Mg0.1O3-δ (LSGFM), if the same sintering sequence as the one used for

LSGM is employed, it was not possible to completely eliminate the SrLaGa3O7 phase. This is

exemplified by LSGFM’s XRD patterns that are shown Fig. 8(a).

11


Fig. 8 Analyses for GNP synthesized La0.7Sr0.3Ga0.7Fe0.2Mg0.1O3-δ powder − (a) XRD patterns after GNP

synthesis (bottom) and after air sintering for 8 hrs at1373, 1623 1673, 1743, and 1773±10 °K; (b) EDS

spectrum after air sintering at 1773±10 °K for 8 hrs; (c) SEM micrograph after GNP; (d) SEM

micrograph after air sintering at 1623±10 °K for 8 hrs; (e) SEM micrograph after air sintering at 1748±10

°K for 8 hrs, pulverizing, and sieving to dp≤32µm


20 30 40 50 60 70 80

Re

lati

ve

In

ten

sit

y

0

2000

4000

6000

8000

10000

12000

14000

? ? ? ? ?

La0.7Sr0.3Ga0.7Mg0.2Fe0.1O3-

SrLaGa3O7

La4Ga2O9

unknown?

1373 ºK

1623 ºK

1673 ºK

1743 ºK

1773 ºK

(a)

keV

2 4 6 8 10

co

un

ts

0

100

200

300

400

500

O

LaLa

La

La

La

La

LaLa

Ga

Ga

Sr

Fe

Fe

(b)

(c) (d) (e)

50µm 50µm 5µm

12


Iron’s presence can be found in LSGFM’s EDS spectrum (Fig. 8(b)). The SEM micrographs for

the as-synthesized powders (Fig. 8(c)) shows that the dp is greater than several hundred microns and that

the particles are comprised of interlocking grains having sizes in the order of 0.20±0.05 µm. Although dp

had remained essentially the same after sintering at 1623±10 °K (8 hrs), the grain sizes augmented to

0.75±0.25 µm. It was discovered that, for SPS, it was essential to pulverize and sieve the powders such

that their dp≤32µm. The SEM micrograph shown in Fig. 8(e) exemplifies the conventional sized powders

that were used for the majority of the SPS depositions.

Following its GNP synthesis, Sm0.2Ce0.8O2-δ (SDC) was sintered at 1773±10 °K for 8 hrs. Its

XRD patterns, given in Fig. 9, shows that a purely fluorite phase was obtained.


20 30 40 50 60 70 80

Rela

tive

In

ten

sit

y

0

1000

2000

3000

4000

CeO2

Fig. 9 XRD patterns for GNP synthesized Sm0.2Ce0.8O2-δ powder before (bottom) and after air sintering at

1773±10 °K for 8 hrs (top)

3.2 Initial investigations of SPS

Initial attempts at SPS depositions of double-doped, LaGaO3 SOAE employed parameters that

were very similar to those used by Jia and Gitzhofer [42]. Early on, however, it was found that the back-

pressure in the atomization probe’s suspension inlet-caused by the pressure exerted by the injection gas

line (Fig. 3)-often prevented the EtOH-based suspension from flowing altogether. Furthermore, EtOH,

GNP powder suspensions were not uniform; thus, at the time of the depositions, EtOH flowed faster than

the solids while the latter obviously settled in the peristaltic tube; therefore, it was impossible to know,

even approximately, what the suspensions’ wt% solids were at the atomization probe’s exit. This

13


occurred if solid GNP powders were added to the solvent either directly following their synthesis or

following their sintering and subsequent placement in an ultrasound bath. The problem was particularly

severe using the former arrangement. The effect of the addition of dispersant(s) on EtOH suspensions’

stabilities and fluidities was not investigated as white mineral oil was discovered to be a far superior

solvent. For the flow-rates used in this study (2.5-8.0 ml min-1

), mineral oil suspensions were uniform;

this stems from mineral oil’s kinematic viscosity (34.5 cSt), which is high compared to EtOH’s (1.5 cSt).

Furthermore, as shown in Fig. 1, mineral oil has a higher volumetric power density. All of the SPS

results given here are in regards to mineral oil suspensions.

A summary of the SPS (i-xi) parameters are given in Table 3. The corresponding samples’

analyses were done using the given protocols (2.3). For this preliminary study, 7 wt% solid suspensions

were used. As well, the substrate type and the GNP powders’ post-synthesis manipulations varied for this

portion of the optimization. Either a mortar and pestle (MP) or a miniature ball-mill (MBM) was used to

pulverize (~ 1 g) the sintered, GNP powders.

Table 3 Summary of the varied parameters’ values during the initial SPS investigations

Deposition ST FR (ml min-1) SOAE SD NL Analyses PT

Ii Ti 4 LSGFM 10 55 SEM, EDS MBM

iii Ti 4 LSGM 10 20 SEM, EDS MBM

iii Ti 4 LSGMF 10.5 20 SEM MBM

iiv Ti 4 LSGM 10 23 SEM, EDS MBM

iv Ti 4 LSGMF 14 15 SEM, EDS MBM

ivi Ti 8 LSGM 20 4 SEM, EDS MBM

ivii SIKA-SS-10 4 LSGMF 10 20 SEM MP

iviii SIKA-HX-10 4.5 LSGMF 10 10 SEM, EDS MP

iix SIKA-HX-1 4 LSGFM 11 7 SEM 2.1

ix SIKA-HX-1 4 LSGFM 11 60 SEM 2.1

ixi SIKA-HX-1 4 LSGFM 11 55 SEM 2.1

ST – substrate type; FR – flow rate; SD – spraying distance; NL – number of deposition loops; PT

– powder tailoring.

Figs 10(a)-(c) shows that the coating suffers from a high level of open porosity; despite the

number of passes, its EDS, elemental spectrum indicated the presence of titanium. The distribution of

elements, however, was similar to the feedstock’s (Fig. 8(b)).

14


Fig. 10 Analyses for deposition i (Table 3) − (a)-(c) SEM micrographs (top section)

Deposition ii’s cross sectional, SEM micrographs are shown in Fig.s 11 (a)-(b). The LSGFM

layer is well adhered to the Ti substrate, but there is significant variation in its thickness. Furthermore,

the coating is not homogeneous. Regions with significant open porosity abruptly become dense. Copper,

which assuredly came from the fusion of the copper, cooling insert that Jia [13] used for his SPS

depositions, was present in the EDS, elemental spectrum; hence, Jia’s claim that the copper cooling mask

rendered superior coatings could not be substantiated.

10 µm 10 µm

(a) (b)

I(c)

1 µm

LSGFM LSGFM

LSGFM

15


Fig. 11 Analyses for deposition ii (Table 3) − (a)-(b) SEM micrographs (cross section); (c) SEM

micrograph (top section)

The global, SEM micrograph of deposition iii’s top-section is given in Fig. 12(a).

Although the coating has a high surface density of caverns, their characters do not correspond

well with those stemming from the discovered, plasma-substrate arcing. In addition, the

coating’s microstructure is highly porous and it appears as though the powders were heated to a

quasi-vapour state amid SPS. The effect may be a result of the amount of time the post-sintered,

GNP powders were subjected to ball-milling-which, for this portion of the optimization, varied

between several minutes to several hours. A substantial portion of the powders were lost during

the ball-milling. Powders having an excessively small particle size distribution would be more

susceptible to vaporization; hence, careful control over the powder’s post-tailoring process is

critical. The SEM micrograph of the deposition’s cross section (Fig. 12(c)) shows that the rate of

deposition was in the order of 1.0 ± 0.1μm loop-1

. However, the coating did not adhere to the

titanium substrate and suffers from a significant level of open porosity.

10 μm

Ti

LSGFM

(a)

LSGFM

(b)

10 μm Ti

29.4 μm

23. μm

1 μm

LSGFM (c)

16


Fig. 12 Analyses for deposition iii (Table 3) (a)-(b) SEM micrographs (top section); (c) SEM

micrograph (cross section)

As shown in Figs. 13(a)-(c), deposition iv’s top global and micro characteristics are

similar to those of deposition iii’s. The layer’s thickness, which is approximately half of the

previous two depositions, indicates that either its density is significantly higher or that a

significant portion of the in-flight particles were rebounded upon contacting the substrate.

Again, the coating has poor adherence to the titanium substrate and its non-homogeneity is

presumably due to the powder’s post-tailoring sequence. As well, the top-section’s EDS,

elemental spectrum was similar to the feedstock’s (Fig. 8(b)).

100 μm

LSGM LSGM

10 μm

1 μm

18 μm

LSGM

(a) (b)

(c) epoxy

17


Fig. 13 Analyses for deposition iv (Table 3) − (a)-(b) SEM micrographs (top section); (d) SEM


In comparison to the previous depositions, the only remarkable parameter change for deposition v

was the greater stand-off spraying distance. Two different types of microstructures can be observed. On

the one hand, as shown in Fig. 14(a), there are regions in which splats having diameters ranging from a

tenth to several tens of microns are deposited upon seemingly larger particle deposits that had reached a

quasi-liquid state. A vertical crack is also present in the top left hand side of the micrograph. On the

other hand, as shown in Fig. 14(b), there are regions that correspond to the previous depositions where the

material attained a quasi-vaporized state prior to being deposited. There is substantial variation in the

deposition’s thickness and the cross-section micrographs (Figs. 14(c)-(d)) shows that, along with open

porosity, the deposition suffers from both vertical and horizontal cracking. In addition, either during or

after the deposition, portions of the coating having lengths and widths in the order of several microns

broke away from the rest of the deposition. The initial layer of the deposit, however, is well adhered to

the titanium substrate. Finally, other than the presence of titanium, the deposition’s top-section, EDS

spectrum coincided with the feedstock’s.

100 μm

LSGFM

10 μm

LSGFM (a) (b)

1 μm

LSGFM (c) (d)

1 μm

epoxy

Ti

9.7 μm

18


Fig. 14 Analyses for deposition v (Table 3) (a)-(b) SEM micrographs (top section); (c)-(d) SEM


During deposition vi, the suspension’s flow-rate, the SPS stand-off distance, and the number of

deposition loops differed. High-resolution, SEM micrographs of the coating’s cross-view are shown in

Fig.s 15(a)-(c); it can be seen that the well-adhered deposit suffers from a significant level of open

porosity. Additionally, the substrate was not well centered amid the deposition; hence, the coating’s

thickness varies greatly (7.9 ± 6.9 μm). Other than the presence of titanium, the deposition’s cross-

sectional, EDS, elemental spectrum corresponded to its feedstock (Fig. 7(b)).

--

LSGFM

1 μm

(a) LSGFM (b)

(c) (d) Ti

LSGFM

1 μm

1 μm

Ti

LSGFM

8.2 μm

1 μm

19


Fig. 15 Analyses for deposition vi (Table 3) − (a-c) SEM micrographs (cross section);

Due to the substantial powder loss that was experienced while pulverizing the post-sintered, GNP

powders in the miniature ball-mill, attempts were made to reduce the powders’ particle sizes with a

mortar and pestle. As well, commercial GKN™ porous metal filters made from both SS316 (SIKA-SS-

10) and Hastelloy X (SIKA-HX-10 and SIKA-HX-1) were used throughout the remainder of the initial

optimization procedure.

Deposition vii’s global top-view, SEM micrograph, given in Fig. 16(a), shows that the powder

sizes are excessively large; hence, they are insufficiently melted in the plasma core, which gives rise to

discontinuous coatings. The high-resolution, top-sectional, SEM micrograph (Fig. 16(b)) suggests that

particle splat formation did not occur and that the particles were slightly liquefied in the plasma’s core.

The global, cross section, SEM micrograph (Fig. 16(c)) infers that the SS316 substrate’s pore size is

excessive and that the coating is poorly adhered to the substrate. As well, its thickness (Fig. 16(d)) is less

than what was achieved while spraying the ball-milled powders under the same operating conditions. This

indicates that solid particles were rebounded off of the substrate.

LSGM Ti

(a) (b)

LSGM

Ti

10 μm 1 μm

14.7 μm

LSGM

(c)

10 μm

Ti

20


Fig. 16 Analyses for deposition vii (Table 3) − (a)-(b) SEM micrographs (top section); (c)-(d)

SEM micrographs (cross section)

Fig. 17 contains the analyses for deposition viii. While operating with the specified parameters, it

can be seen that a large percentage of the solid particles, especially those having diameters in excess of 30

µm, are not sufficiently fused in-flight, and therefore the coating is not dense. That is to say that only the

particles’ exterior surfaces attain the liquid state as the heat transfer rate was insufficient to liquefy their

cores’. However, as indicated by Fig. 17(d), the larger particles fuse together during the quenching

process. Clearly, the suspension’s large particle size distribution results in irregular depositions hence

further tailoring of the synthesized powders is required.

100 μm

(a) LSGM (c)

1 μm

LSGM

10 μm

SIKA-SS-10

LSGM

SIKA-SS-10

1 μm

7 μm

6.4 μm

21


Fig. 17 Analyses for deposition viii (Table 3) − (a)-(b) SEM micrographs (top section); (c)-(d) SEM

micrographs (cross section)

Other than the severe plasma-substrate arcing, which results in “micro-lunar” surfaces, SPS of

post-sintered, GNP-synthesized, LSGFM powders, tailored to dp≤32μm, on SIKA-HX-1 substrates

renders depositions of uniform thicknesses (Fig. 18). The author is unable to specify the reason(s) for the

seemingly absence of the arcing in the prior depositions. In the plasma core, the outer volumes of the

feedstock powders with 32 μm < dp < 10 μm attain a majority liquid state in their outer volumes, while

smaller particles are completely liquefied. The quenched mixture results in depositions having minor

amounts of open porosity (e.g. Fig. 18(e)).

LSGFM LSGFM

300 μm 20 μm

50 μm 50 μm

SIKA-HX-10

LSGFM

LSGFM

SIKA-HX-10

(a) (b)

(c) (d)

22


Fig. 18 Analyses for deposition ix (Table 3) − (a)-(f) SEM micrographs (top sections)

The SEM micrograph of deposition ix’s global top-view (Fig. 18(a)) indicates that the arc rate

density is 1.5 ±0.25 arcs loop-1

mm-2

. Microstructures for three of the deposition’s craters are displayed in

Figs. 18(b)-(d); the arcs penetrate the depth of the ceramic layer leaving craters with diameters of 45 ±15

μm. Two types of craters are noted. One type, which is exemplified by Figs. 18(b)-(c), is where only the

removal of material occurs during the arcing. The other type is where the arc results in both the removal

of material in combination with its vaporization, which, upon quenching, results in ceramic bubble

formation in the crater’s exterior ring (Fig. 18 (d)).

500 μm

LSGFM (a)

50 μm

(b) LSGFM

50 μm 100 μm

(c) (d)

1 μm

(e)

23


Deposition x’s global, top section SEM micrograph (Fig. 19(a)) shows the augmentation in crater

density, compared to deposition ix’s, because of the greater number of deposition passes. The severe

environment also renders a high density of vertical cracks (Fig. 19(b)).

Fig. 19 Analyses for deposition x (Table 3) − (a)-(d) SEM micrographs (top section)

Cross-view, SEM micrographs for a 55 loop, LSGFM deposition are given in Fig. 20. The

operating conditions allow for a deposition rate of 0.8±0.1 μm loop-1

and the use of SIKA-HX-1

substrates largely eliminates the coating discontinuities and depressions that occur on SIKA-HX-10

substrates. Although the level of open porosity is very minor and the LSGFM coating adheres well to

Hastelloy X, the presence of vertical cracks-caused by either the arcs or the mismatch of the materials’

CTEs, or a combination of the two effects-are highly objectionable.

500 μm

LSGFM

200 μm

LSGFM

100 μm

LSGFM LSGFM

50 μm

(a) (b)

(c) (d)

24


Fig. 20 Analyses for deposition xi (Table 3) − (a)-(c) SEM micrographs (cross section)

In summary, as it pertains to obtaining dense ceramic coatings by SPS, despite what was reported

[42], major parameter adjustments were required. Firstly, GNP powder-EtOH suspensions are not

homogeneous. Rather, viscous solvents that can be pumped at greater than 4 ml min-1

, such as mineral

oil, are required. Secondly, to obtain coatings of uniform thicknesses, post-air-sintered, GNP synthesized

powders must be pulverized and sieved to dp≤32 μm, and commercial substrates having low pore sizes,

such as SIKA-HX-1, are required. Thirdly, the plasma-substrate arcing, which produces craters at a rate

1.5±0.5 arcs loop-1

mm-2

, is highly objectionable as it is responsible for significant amounts of coating

loss and cracking–both horizontal and vertical.

The results from the secondary investigations of SPS of dense ceramic layers on commercial,

porous, metal substrates, particularly as it relates to the eradication of the plasma-substrate arcing, is

presented in the following sub-section.

3.3 SPS on GKN™ porous, metal filter substrates

Although several problems associated with SPS of dense ceramic layers on metal substrates were

exposed and resolved in the previous sub-section, several issues still need to be addressed before the

fabrication of metal-supported, ERH anode compartments can be considered. In this sub-section, findings

from the optimization of dense ceramic coatings on GKN™ porous metal filters by SPS are elaborated.

This particularly relates itself to material compatibility issues, the eradications of the discovered plasma-

substrate arcing, and multi-layer depositions.

LSGFM

100 μm

LSGFM

SIKA-HX-1

(a) (b)

SIKA-HX-1 100 μm

20 μm

(c)

LSGFM

SIKA-HX-1

25


Table 4 Summary of the varied parameters’ values for the SPS of dense, ceramic layers on GKN™

porous metal filters

Deposition Powder Wt% solids iNo. loops Analyses

Ii NFO 10.6 50

SDC 10 3

LSGM 10 3

LSGFM 10.1 60 SEM, EDS

Iii NFO 10.6 50

SDC 10 3

LSGM 10 3

LSGFM 10.1 60 SEM

Iiii NFO 10.6 50

LSGM 10 3

LSGFM 10.1 80 SEM, EDS

Iiv NFO 9.7 60

SDC 10 2

LSGM 10 2

LSGFM 10.1 60 SEM

Iv NFO 6.9 25

LSGFM 8 25 SEM

Vi NFO 8 20

GDC 7.4 2

LSGM 7.4 2

LSGFM 8 70 SEM, EDS, XRD

The remainder SPS results presented in this text involves the use of sintered, GNP synthesized

powders that were tailored to dp<53 μm (NFO) and dp<32 μm (LSGM, SDC, LSGFM) using the protocol

given in 2.1. Depositions i-iv and v-vi were done using the electrically-non-insulated (old) and

electrically-insulated (Fig. 3) substrate holders, respectively. During every deposition, the cooling mask

was electrically grounded and neither the pre-heating of the metal substrates nor the post heating of the

coatings was done. As well, the spraying distance (nozzle tip to substrate) was 5.5±1.5 cm, and the

suspension flow-rate was 4.0±0.1 ml min-1

. A summary of the varied parameters’ values for the SPS of

dense, ceramic layers on GKN™ porous metal filters are listed in Table 4.

SEM micrograph of deposition i’s cross-sectional view, shown in Fig. 21(a), does not focus on

the craters that were still present to a minor degree. For the NFO, SDC, and LSGFM deposits, respective

deposition rates of 0.72±0.02, 1.0±0.02, and 1.45±0.02 μm loop-1

were obtained. Although the NFO layer

does not adhere to the SIKA-SS-5 substrate, there is excellent adherence at the three ceramic interfaces.

Other than the LSGM/LSGFM interface, it is possible to visually distinguish the distinct layers by the

abrupt colour variations. While the level of open porosity is low, the vertical crack that extends itself

across the SOAE layer is objectionable.

26


Fig. 21 Analyses for deposition i (Table 4) − (a) SEM micrographs (cross section)

Deposition ii was done on a SIKA-HX-1 substrate. The SEM micrographs of its top section are

shown in Fig. 22. A crater formation rate of 0.06±0.01 craters loop-1

mm-2

is low compared to the ones

obtained while performing depositions with an ungrounded, cooling mask. Note that the calculation was

made using the number of LSGFM passes as it is believed that craters in the previous layers were

undetectable. The high resolution micrographs (Figs. 22(b)-(c)) indicate that the vast majority of the

feedstock is liquefied in the plasma’s core and that the coating is essentially free of vertical cracks.

Fig. 22 Analyses for deposition ii (Table 4) − (a)-(c) SEM micrographs (top-section)

1000μm

LSGFM (a) (b)

50μm

50μm

LSGFM

LSGFM

NFO

50μm epoxy

SIKA-SS-5 (a)

SDC

LSGM

(c) LSGFM

50μm

27


Deposition iii was done on a SIKA-HX-1 substrate. From its cross-sectional, SEM micrograph,

shown in Fig. 23(b), deposition rates of LSGM/LSGFM and NFO are 1.34±0.14 and 0.92±0.10 μm loop-1

,

respectively, have been determined. The contrasting layers are firmly attached to one another and there

are no signs of discontinuities along their interfaces. Additionally, the EDS, elemental mapping of the

individual elements, which are shown in the bottom inserts, affirms that excellent phase separation is

achieved. Furthermore, the EDS spectrum (Fig. 23(a)) of the LSGFM layer coincided with its feedstock.

However, unwanted vertical cracks, which are only present in the SOAE layer(s), exist at intervals of one

for every 100±20 μm and extend themselves through the layer’s thickness, appear to have initiated

themselves at the top surface of the coating.

Fig. 23 Analyses for deposition iii (Table 4) − (a) EDS, elemental spectrum of the LSGFM layer (cross

section); (b) SEM micrograph (cross section); EDS, elemental mapping of the elements Fe (Ka1), Ga

(La1,2), Ni (Ka1), O (La1), and Sr (La1)

Deposition iv’s top section, SEM micrographs are shown in Figs 24(a)-(f). Again, while

operating with the grounded, cooling mask, and assuming 60 deposition loops (LSGFM), the crater

formation rate is 0.06±0.01 craters loop-1

mm-2

. The high resolution micrographs shows that not all of the

feedstock powders attain a purely fused state in the plasma core, particularly those having dp>20 μm.

100μm

LSGFM (b)

NFO

SIKA-HX-1

(a)

Ni (Ka1) La (La1) Ga (La1_2)

O (Ka1) Sr (La1)

Fe (Ka1)

28


Fig. 24 Analyses for deposition iv (Table 4) − (a)-(f) SEM micrographs (top section)

For deposition v, to investigate the possibility of liquefying a greater portion of the feedstock’s

particles, the suspension’s solid wt% was lowered. Furthermore, this was the first trial that employed the

electrically insulated sample holder (Fig. 3). Numerous depressions are shown by the global SEM

micrographs of their top sections (Fig.s 25(a)-(c)). The cavernous character differs from the craters that

result from the plasma-substrate arcing and it can be concluded that they are a corollary of the larger pore

size of the SIKA-SS-10 substrates. Ostensibly, the plasma-substrate arcing can be eliminated by

operating with the new sample holder in combination with the grounded, cooling mask. As well, as

anticipated and exemplified by the high-resolution micrographs given in Figs 24(c)-(d), the percentage of

200μm

LSGFM LSGFM (a) (b)

1000μm

LSGFM (c)

50μm 20μm

LSGFM (d)

LSGFM (e)

10μm

(f)

20μm

LSGFM

29


the feedstock that is liquefied in the plasma core is augmented when operating with suspensions

comprised of lower wt% solids.

Fig. 25 Analyses for deposition v (Table 4) − (a)-(d) SEM micrographs (top section)

Deposition vi’s top-sectional, XRD pattern coincides with the feedstock’s (Fig. 8(a)). Figs.

26(a)-(f) are its top- and cross- sectional, SEM micrographs of varying resolutions. Because a SIKA-HX-

1 substrate was used, the deposition is devoid of the caverns previously found (i.e. deposition v). Judging

by the quantity of splat formations (Figs. 26(c)-(d)), presumably, the vast majority of the feedstock

powders were liquefied in the plasma core. The deposition rates for the NFO, GDC, and LSGFM/LSGM

layers were 0.99±0.15, 0.65±0.30, 1.00±0.50 μm loop-1

, respectively. While there is some slight open

porosity at the GDC-LSGM interface, the deposition is dense. Only one vertical crack-which again

appears to have commenced at the top of the coating-can be found along the depositions cross-section;

but, to a lesser extent, horizontal cracks are also present. The EDS, elemental line-scans, which are

shown in the bottom panels of Fig. 27, demonstrate the excellent separation between the material

deposits. Using these optimized SPS parameters, similar results were achieved (i.e. reproducible);

multilayer, anodic, IT-SOFC half-cells of varying LSGFM thicknesses were prepared for further

investigations [37]. Notably, for these coatings, the coating thicknesses were an order of magnitude less

than those presented here and they did not suffer from vertical or horizontal cracks.

500μm

LSGFM

500μm

LSGFM

(b)

50μm

LSGFM

(a)

10μm

LSGFM (c) (d)

30


Fig. 26 Analyses for deposition vi (Table 4) − (a)-(d) SEM micrographs (top sections); (e)-(f) SEM

micrographs (cross sections)

100μm

LSGFM (b)

50μm

LSGFM (c)

5μm

LSGFM (d)

100μm

LSGFM

(e)

50μm

LSGFM

(f)

500μm

LSGFM (a)

31



20 30 40 50 60 70 80

Re

lati

ve

In

ten

sit

y

0

100

200

300

400

500

600

LSGFM

SrLaGa3O7

(b)

Fig. 27 Analyses for deposition vi (Table 4) − (a) SEM micrographs (cross section); (b) XRD pattern (top

section); EDS, elemental line-scans for the composition line shown in (a) are given in the bottom panels

The experimental findings from this sub-section can be summarized as follows:

grounding the cooling mask drastically diminishes the rate of crater formation;

grounding the cooling mask in combination with the electrically-insulated sample-holder

eliminates the plasma-substrate arcing;

depositions on SIKA-HX-10 substrates results in cavernous coatings;

depositions on SIKA-HX-1 substrates renders coatings having uniform depths;

feedstock liquefaction in the plasma core is largely dependent on the suspension’s solid content;

compared to the relevant TSDP literature pertaining to dense, ceramic coatings, the optimized,

multi-layer depositions (e.g vii) are of very high quality.

At this juncture, there are several recent claims that have been made by CREPE researchers

that need to be addressed. Jia and Gitzhofer [42] stated that the lowest viscosity suspensions

were ideal for SPS, which contradicts this investigation’s findings. In regards to their high

resolution, 79 µm2 micrographs of the SPS deposition of GDC, they claimed that negligible gas

permeability along with low coating porosity (1.4%) was attainable while using both induction

coupled plasma and GNP synthesized powders, which indicates that they did not observe the

(a)

NFO

SIKA-HX-1

GDC

Ce (La1) Gd (La1)

Ga (La1_2) Sr (La1) La (La1) O (Ka1)

Ni (Ka1) Fe (La1)

20μm

LSGFM

32


plasma-substrate arcing. However, this does not correspond to the author’s understanding of

matters. In their description of the test rig they used for the measurement of gas permeability, no

mention of how their coated substrates were fixed into the test rig’s enclosure was given. The

author deemed that open circuit voltage readings would be more accurately allow for the

determination of gas permeability and coating porosity hence their test rig was not used.

Moreover, despite claiming that powder synthesis was achieved by two different methods, the

high-resolution SPS micrograph that were presented in the separate articles [42, 43] were the

same, and no low-resolution micrographs were presented. Lastly, the SPS apparatus set-up for

the coatings that were presented in the Gitzhofer et al.’s follow-up article [37] did not correspond

to reality; rather, in order to eradicate plasma-substrate arcing, the set-up shown in Fig. 3 was

used.

Attempts at extending these promising results to ERH anode compartments will now be

elaborated.

3.4 SPS for the fabrication of ERH anode compartments

Table 5 lists the parameters/post-deposition analyses used in the investigations of, ERH anode

compartment (2.3) prepared by SPS.

For these investigations, the SPS stand-off distance was 4.5 ± 0.5 cm and, unless stated otherwise,

the suspension flow-rate was 4.0 ± 0.1 ml min-1

. Furthermore, in order to achieve a denser GDC inter-

layer, its suspension’s solid content was diminished from 8 to 3 wt%. GDC has a significantly higher

fusion temperature (~3000 °K) compared to those of NFO and the double-doped LaGaO3 (~1800 °K).

All other suspensions were comprised of 8 wt% solids. Finally, as will be discussed in more depth below,

the sequencing of the post-deposition, H2 reduction (H2R in Table 4) plays an important role in the

successful fabrication of the ERH anode compartments. Components were placed under a flowing (100 ±

25 ml min-1

) H2 (99.9 vol%) atmosphere at 923 ± 10 °K for 6 hrs. In order to determine if the reduction

of the anode layer(s) had indeed taken place in the allotted time-frame, electrical resistance measurements

were taken before and after their depositions. Similar tests proved that the incorporated Fe2O3, in

LSGFM is not reduced by H2 to its metallic state under the aforementioned conditions.

At the time of this writing, Ishihara et al.[19] had obtained the highest power densities in the IT-

SOFC literature. This was achieved by the in-situ reduction of the NiO-Fe2O3 substrate after the

deposition of the SOAE layers were deposited by pulsed laser deposition. Being that the ERH process to

be tested was to be done with a metal-supported, anode compartment fabricated by TSDP, which are

susceptible to vertical cracks, relatively thick (≥ 100 μm) SOAE layers were deposited so to lessen the

possibility of catholyte cross-over.

The SEM micrographs of deposition i’s top-section are shown in Fig. 28. Judging by the

coating’s surface, ostensibly, while in the plasma core, the feedstock had reached a quasi-vaporized state.

The post-deposition measurement of the suspension’s flow-rate rendered an unexpected value of 2.5 ± 0.1

ml min-1

due to a worn out tube.

33


Table 5 Summary of the varied parameters’ values during the SPS on ERH anode compartments having

the NPT design

**3NFO7SDC-30wt%NFO-SDC; *7NFO3SDC-70wt%NFO-S

Deposition Powder no. loops H2R

Ii NFO 15

LSGFM 20

Iii NFO 35

LSGFM 99 R

Iiii NFO 35 R

GDC 4

LSGM 3

LSGFM 110

Iiv NFO 35 R

GDC 4

LSGM 3

LSGFM 99 R

Iv NFO 35

GDC 4

LSGM 3

LSGFM 99 R

Ivi NFO 35 R

*7NFO3SDC 3

**3NFO7SDC 3 R

GDC 4

LSGM 3

LSGFM 150 R

Ivii NFO 30

*7NFO3SDC 3

**3NFO7SDC 3 R

GDC 3

LSGM 3

LSGFM 99 R

Iviii NFO 30

*7NFO3SDC 3

**3NFO7SDC 3 R

GDC 3

LSGM 3

LSGFM 70

34


Fig. 28 Analyses for deposition i (Table 5) − (a)-(d) SEM micrographs (top section)

To test the functionality of the ERH anode compartment that resulted from deposition ii, OCV

measurements were taken using the arrangement depicted in Fig. 5. As the system’s temperature was

augmented between 873±10 and 923±10 °K, the OCV increased from 0.385±0.003 and 0.645±0.003 V.

Thereafter, it diminished rapidly (~3 mV min-1

) until a zero voltage reading was obtained. The measured

value at 923±10 °K is in reasonable accordance with the one obtained by Ishihara et al. [19]. This was

the only ERH anode compartment that was characterized in this way. The subsequent SEM analysis of

the sample’s cross-section (Figs. 29(a)-(e)) shows that, during the examination period, other than the far

left portion of the coating (Fig. 29(a)), the LSGFM layer had completely delaminated itself from the

quasi-dense, 30±5 μm thick, 90wt%Ni-Fe, anode layer. The latter material remained flush and fully

adhered to the well adjoined SIKA-HX-1 and Inconel materials that constitute the integral, flow-restrictor

substrate. Five minor vertical cracks of variable spacing between each (50-400 μm) can be observed

along the 100±10 μm thick, LSGFM layer (Fig. 29(c)), which was therefore deposited at a rate of

1.01±0.10 μm loop-1

. From these analyses it can be concluded that the custom cooling arrangement can

achieve the goal of substrate temperature management but a protocol that allows for an adherent, multi-

layer deposition following the H2 reduction of NFO is still required.

LSGFM (b)

50μm

LSGFM (c)

20μm

LSGFM (d)

5μm

LSGFM (a)

200μm

35


Fig. 29 Analyses for deposition ii (Table 5) following the OCV testing − (a)-(d) SEM micrographs (cross

sections)

Deposition iii’s H2 reduction sequence allows for the investigation of the cohesion of the SOAE

layer(s) on the 90wt%Ni-Fe anode layer. So to conform, as much as possible, to the desired final product,

GDC and LSGM inter-layers were included. The SEM micrographs of its cross sections are displayed in

Figs. 30(a)-(f).

The apparent LSGFM/LSGM and GDC deposition rates are 1.19±0.18 and 0.19±0.06 μm loop-1

,

respectively. As the deposition of the NFO layer was succeeded by its H2 reduction, a corresponding

value cannot be assigned; hence, only a “superficial” (i.e. 90wt%Ni-Fe) deposition rate of 0.74±0.12 μm

loop-1

can be given. The anode’s sub-micron porosity is particularly notable (Fig. 30(f)). Although there

is excellent coherence between the various layers, it can be seen that, in comparison with the previous

LSGFM

(a)

500μm

SIKA-HX-1

Inconel

Ag

90wt%Ni-Fe LSGFM

(b)

500μm

SIKA-HX-1

Inconel

Ag

90wt%Ni-Fe

LSGFM

(c)

500μm

SIKA-HX-1

Ag

90wt%Ni-Fe

LSGFM

(e)

200μm

SIKA-HX-1

Inconel

Ag

90wt%Ni-Fe

LSGFM (f)

100μm

Inconel

90wt%Ni-Fe SIKA-HX-1

LSGFM

(d)

500μm

SIKA-HX-1

Inconel

Ag

90wt%Ni-Fe

36


depositions of simple substrates, the number of vertical and horizontal cracks is extensive; a significant

percentage of the former type, which occur at a frequency of 5-6 cracks mm-1

, penetrate the multi-layer

coating all the way up to the substrate; the latter type, which often but not always span themselves

between two neighbouring vertical cracks, appear to coincide with an individual layer resulting from a

single, deposition pass. Speculative reasons for the formation of these macro-fissures include: (a)

insufficient substrate cooling at the time of the deposition which results in the coating’s subsequent, rapid,

post-deposition cooling; (b) abrasive, post-deposition handling of the component during its removal from

the cooling apparatus; (c) CTE mismatch between the substrate and deposited layers; and (d) the inherent

vibrations that the robotic arm perpetuates during the deposition.

A Swagelok™, SS316, bottom ferrule was used to make the thermal contact between the

underside of the substrate and the interior cooling apparatus. The ferrule’s thermal conductivity (16.3 W

m-1

K-1

), in combination with its irregular shape, lends credence to (a). Moreover, the obnoxious macro-

fissures are more probably a consequence of reasons (a) and (b) together. Because of copper’s excellent

ductility and high thermal conductivity (386.0 W m-1

K-1

), it is a preferable thermal contact material;

hence, copper foil was used for depositions iv, v, vii, and viii. The remaining aforementioned reasons are

regarded with less confidence but are worthy of mention for future researchers.

Prior to depositing its three SOAE layers, deposition iv’s NFO layer was H2 reduced to its

metallic state. The completed deposition was then exposed to a H2 environment; thereafter, the SEM

micrographs for both the coating’s top (Figs 31(a)-(d)) and cross (Figs 31(e)-(h)) section were taken. Due

to the cross-sectional micrographs’ poor colour resolution it is not possible to ascertain the GDC layer’s

thickness. However, the LSGFM/LSGM and superficial anode deposition rates are 1.04±0.07 and

0.79±0.07 μm loop-1

, respectively. The anode layers sub-micron porosity is visible in Figs 31(e)-(f).

37


Fig 30 Analyses for deposition iii (Table 5) − (a)-(f) SEM micrographs (cross sections)

LSGFM

(b)

300μm

90wt%Ni-Fe

SIKA-HX-1

GDC

(c)

LSGFM

300μm

90wt%Ni-Fe

SIKA-HX-1

GDC

LSGFM

300μm

90wt%Ni-Fe

SIKA-HX-1

GDC

LSGFM

300μm

90wt%Ni-Fe

SIKA-HX-1

GDC

LSGFM (f)

50μm

Inconel

90wt%Ni-Fe

SIKA-HX-1

GDC

(e)

(d)

LSGFM

(a)

500μm

Inconel

90wt%Ni-Fe

SIKA-HX-1

GDC

LSGM

LSGM

LSGM

LSGM

LSGM

LSGM

38


o

Fig 31 Analyses for deposition iv (Table 5) − (a)-(d) SEM micrographs (top section); (e)-(h) SEM

micrographs (cross section)

LSGFM (a)

1000μm

Inconel

LSGFM (b)

200μm

LSGFM LSGFM

50μm

200μm

SIKA-HX-1

90wt%Ni-Fe

GDC

(d) (c)

LSGFM

(e)

500μm

500μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(g)

LSGM

LSGM

(f)

500μm

LSGFM

SIKA-HX-1

90wt%Ni-Fe GDC

Inconel

LSGM

90wt%Ni-Fe GDC

(h)

500μm

LSGFM

SIKA-HX-1

LSGM

39


Deposition v’s cross-sectional micrographs are shown in Fig. 32. The SOAE layers completely

delaminate themselves from the anode, which remained adhered to the integral metal component. Despite

the addition of the LSGM and GDC SOAE interlayers, the overall result is similar to what occurred in the

case of deposition ii. Tentatively, one may conclude that it is essential to reduce the anode layers prior to

the deposition of the SOAE layers as the volumetric change that the anode layer endures during the

reduction step renders the undesirable result.

Fig. 32 Analyses for deposition v (Table 5) − (a)-(f) SEM micrographs (cross sections)

Deposition vi’s cross-sectional SEM micrographs are shown in Fig 33. As a Swagelok™, SS316,

bottom ferrule was used to make thermal contact between the substrate and the interior cooling apparatus

GDC

500μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(b)

Inconel

LSGM

500μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(a)

Inconel

LSGM

500μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(d)

LSGM

200μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

(e)

Inconel

500μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(c)

LSGM

LSGM LSGFM

200μm

SIKA-HX-1

90wt%Ni-Fe

GDC

(f)

Inconel

LSGM

40


(Fig. 4), the deposition’s character resembles deposition iii’s. This is illustrated by the prevalence of

horizontal (6-7 cracks mm-1

) and vertical fissures that are of the particularity previously described.

However, there is good coherence between the anode, cermet, and electrolyte layers; therefore, the cermet

layers are effective in diminishing the interlayer tension that results from materials’ differing CTEs.

Finally, the loss of the electrolyte layers shown in Fig. 33(b) may have occurred while removing the

component from cooling apparatus following the final depositions.

Fig. 33 Analyses for deposition vi (Table 5) − (a)-(f) SEM micrographs (cross section)

With the exception of the use of copper foil as the cooling apparatus-substrate thermal connection

and the number of LSGFM deposition passes, the protocol used for deposition vi and vii were the same;

(a)

500μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

GDC

500μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(c)

500μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(d)

200μm

SIKA-HX-1 90wt%Ni-Fe

LSGFM

GDC

(e)

Inconel

100μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(f)

Inconel

(b)

500μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

LSGM

3NFO7SDC

7NFO3SDC

LSGM

3NFO7SDC

7NFO3SDC

Inconel

LSGM

3NFO7SDC

7NFO3SDC

LSGM

3NFO7SDC

7NFO3SDC

LSGM

3NFO7SDC

7NFO3SDC

LSGM

3NFO7SDC

7NFO3SDC

41


the latter’s cross-sectional, SEM micrographs, along with its top-sectional XRD pattern, are shown in Fig.

34.

Although present, the incidence and character of the vertical and horizontal cracks was less than

what was found in the previous deposition. This supports the hypothesis that: amid deposition, maximum

substrate cooling increases the uniformity of the coating. The reason for the phenomenon may be because

the transformation of energy that is endured during the relaxation of residual stresses in a “macro-

quench”-i.e. deposited material resulting from several deposition passes-is in excess of what would

otherwise be experienced if the liquefied particles from an single pass are quenched and relieved of the

their residual stresses prior to the deposition of the succeeding layer.

The electrolyte layers are coherent to the anode layer. If compared with the previous depositions,

the addition of the cermet layers is responsible for the positive result. Due to substrate misalignment,

there is considerable variation in the electrolyte layer’s thickness ranges between 33±3 to 90 ± 5 μm,

going from left to right.

Other notable observations include: (I) the sub-micron porosity of the anode layer (Figs.

34(a),(c)-(d)); (II) the LSGFM, perovskite phase was present following the deposition (Fig. 34(h)); (III)

the poor coherence between the porous SIKA-HX-1 and solid Inconel housing at the integrated

substrate’s left side (Figs. 34(a)-(c)); and (IV) the detachment of the SOAE layers at the outer edge (Figs.

34(a),(d)). The latter two observations may be a direct result of the handling of the post-deposition

component. Further efforts are required to resolve minor issues surrounding the custom cooling

apparatus.

42



20 30 40 50 60 70 80

Re

lati

ve

In

ten

sit

y

0

5

10

15

20

LSGFM

SrLaGa3O7

(h)

Fig. 34 Analyses for deposition vii (Table 5) − (a-g) SEM micrographs (cross section); (h) XRD pattern

(top section)

LSGM

3NFO7SDC

7NFO3SDC

100μm


LSGFM

GDC

(c)

Inconel

LSGM

100μm

90wt%Ni-Fe

LSGFM

GDC

(d)

Inconel

100μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(a)

Inconel

LSGM

300μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(b)

Inconel

LSGM

LSGM

3NFO7SDC

7NFO3SDC

LSGM

3NFO7SDC

7NFO3SDC

LSGM

3NFO7SDC

7NFO3SDC

50μm

SIKA-HX-1

90wt%Ni-Fe

LSGFM

GDC

(e)

LSGM

3NFO7SDC

7NFO3SDC

50μm


LSGFM

GDC

(f)

LSGM

3NFO7SDC

7NFO3SDC


GDC

(g)

50μm

LSGFM LSGM

3NFO7SDC

7NFO3SDC

(h)

43


Points worth reiterating from the investigations pertaining to the fabrication of NPT type, ERH anode

compartment via SPS include:

NFO experiences a volumetric change during its H2 reduction to its corresponding sub-micron,

porous metallic form which causes the SOAE layer(s) to delaminate; hence, the deposition of the

electrolyte layers must succeed the reduction of the NFO layer;

degrees to which substrate cooling occurs reflects upon the homogeneity of coatings, particularly

as it pertains to the incidence of both horizontal and vertical cracking -where maximum cooling

amid deposition renders the best result;

diminishing the interlayer CTE mismatch via additional cermet layers mitigates the SOAE

layer(s) from delaminating from the anode;

unlike the optimal depositions in sub-section , crack-free coatings were not achieved during the

course of this optimization;

modifications to the SPS’s cooling apparatus (Fig. 3.4) that will allow for more intensive

substrate cooling during SPS and less abrasive addition and removal of the substrates are

required.

A TSDP coating study that unequivocally demonstrates the outright eradication of the fissuring

phenomena is unknown to the author. This, of course, does not bode well for the experiments that were

supposed to be at the heart of this multidisciplinary study. Only a superficial investigation of the

hypothetical process was possible in the allotted time-frame.

4. Conclusions

Significant gains were made in the field of SPS, multi-layered, ceramic-to-metal deposition

technology. Necessarily, many of the previously stated parameters were significantly modified. This

includes: (a) the suspension’s solvent; (b) the post-tailoring of the GNP-synthesized powders; (c) the

anode’s post-deposition, H2 reduction sequence; (d) the substrate type; (e) the substrate’s cooling amid

deposition; and (f) the substrate holder. Particularly, the undisclosed, deleterious, plasma-substrate arcing

was completely eradicated via the development of an electrically-insulated substrate-holder.

In regards to the TSDP literature pertaining to the deposition of dense, ceramic layers on metal

substrates, the optimized depositions given in 3.3 are of the highest quality known to the author.

Although the results were not fully transposed to the ERH anode compartments that were desired for a

fundamental investigation of a novel electrochemical process, they could instead be considered for

intermediate temperature water electrolysis, IT-SOFC, or other applications. For economic reasons,

Hastelloy X could be substituted by cheaper iron substrates, and air, as opposed to O2, could be

considered as the principal plasma gas.

More generally, because of the increase in the volumetric power density that results when

operating SPS with hydrocarbon solvents and O2 rich plasmas, this sub-family of TSDPes is very

versatile.

Acknowledgements

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

project possible. Along with F. Gitzhofer, CREPE technicians F. Barrette, S. Gutierrez, K. Béré, B.

Couture, I Kelsey-Léveque, and N. Simard were instrumental in the manifestation of this work.

References

[1] Calabretta DL (2006) Queen’s University (masters thesis)

44


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

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

(Vancouver, Canada)

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

[5] Badwal SPS, Ciacci FT (2000) Ionics 6: 1-20

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

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

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

Electrochemical Society Proceedings: 1117-1126

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

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

[11] Gitzhofer F, Bouyer E, Boulos MI (1994) US patent 5609921

[12] Bonneau M-E (2001) Université de Sherbrooke (masters thesis)

[13] L. Jia (2010) Université de Sherbrooke (PhD thesis)

[14] Schiller G, Müller M, Gitzhofer F (1999) J Therm Spray Technol 8(3): 389-392

[15] Hui R, Wang Z, Kesler O, Rose L, Jankovic J, Yick S, Maric R, Ghosh D (2007) J

Power Sources 170: 308-323

[16] FactSage™ 6.2 ThermFact and GTT Technologies

[17] Sasaki K., Muranaka M, Suzuki A, Terai T (2008) Solid State Ionics 179: 1268-1272

[18] Yoo Y (2006) J Power Sources 160: 202-206

[19] Ishihara T, Yan J, Shinagawa M, Matsumoto H (2006) Electrochim Acta 52: 1645-

1650

[20] Hui R, Yang D, Wang Z, Yick S, Deces-petit C, Qu W, Tuck A, Maric R, Ghosh D

(2007) J Power Sources 167: 336-339

[21] Hui R, Wang Z, Kesler O, Rose L, Jankovic J, Yick S, Maric R, Ghosh D (2007) J

Power Sources 170: 308-323

[22] Henne R (2007) J Therm Spray Technol 16(3): 381-403

[23] Henne R, Muller M, Prob E, Schiller G, Gitzhofer F, Boulos M (1999) J Therm

Spray Technol 8(1): 110-116

[24] Takenoiri S, Kadokawa N, Koseki K (2000) J Therm Spray Technol 9(3): 360-363

[25] Tsukuda H., Notomi A, Hisatome N (2000) J Therm Spray Technol 9(3): 364-368

[26] Schiller G, Henne RH, Lang M, Ruckdaschel R, Schaper S (2000) Fuel Cells

Bulletin 3(21): 7-12

[27] Lang M, Henne R, Schaper S, Schiller G (2001) J Therm Spray Technol 10(4): 618-

625

[28] Lang M, Franco T, Schiller G, Wagner N (2002) J Appl Electrochem 32(8): 871-874

[29] Schiller G, Henne R, Lang M, Muller M (2004) Fuel Cells 4(1-2): 56-61

[30] Stöver D, Hathiramani D, Vaßen R, Damani R (2006) Surf Coat Technol 201: 2002-

2005

[31] Ma XQ, Hui S, Zhang H, Dai J, Roth J, Xiao TD, Reisner DE (2003) Thermal Spray:

Advancing the Science and Applying the Science, ASM International: 163-168

[32] Hwang C, Tsai C, Lo C, Sun C (2008) J Power Sources 180: 132-142

[33] Bouchard D (2006) Université de Sherbrooke (masters thesis)

[34] Gheriani M (2008) Université de Sherbrooke (masters thesis)

[35] Mailhot K (1998) Université de Sherbrooke (masters thesis)

[36] Vallet-Renouard G (2004) Université de Sherbrooke (PhD thesis)

[37] Shan Y, Almeida VAB, Gitzhofer F (2010) J Therm Spray Technol 19(3): 566-574

[38] Calabretta D; (2012) Université de Sherbrooke (PhD thesis)

[39] Ishihara T, Yan J, Enoki M, Okada S, Matsumoto H (2008) J Fuel Cell and Technol

5: 031205

[40] Yan J, Matsumoto H, Enoki M, Ishihara T (2005) Electrochem Solid State Letters

45


8(8): A389-A391

[41] Bilodeau J (2012) Université de Sherbrooke,(masters thesis)

[42] Jia L, Gitzhofer F (2010) J Therm Spray Technol 19(3): 566-574

[43] Gitzhofer F, Jia L (2008) Int J Appl Ceramic Technol 5(6): 537-547

[44] Cong L, He T, Ji Y, Guan P, Huang Y, Su W (2003) J Alloy Compd 348: 325-331

[45] Wang Z, Berghaus JO, Yick S, Deces-Petit C, Qu W, Hui R, Maric R, Ghosh D

(2008) J Power Sources 176: 90-95

[46] Lee D, Han J, Chun Y, Song R, Shin D (2007) J Power Sources 166: 35-40

Documents

Radio-frequency induction-coupled suspension plasma spraying of dense ceramic layers on metal substrates