Sulfur Poisoning on SOFC Ni Anodes: Thermodynamic Analyses within Local Equilibrium Anode Reaction Model

The present construction of the chemical potential diagrams and chemical equilibrium calculations were made with the MALT/CHD/gem software package.^24–26 Some diagrams were constructed with CHD program²⁵ partly modified by one of the authors. This program is based on the polyhedron approach to constructing the generalized chemical potential diagram for multicomponent systems. This makes it possible to construct the generalized Ellingham diagram for multicomponent systems having solution phases.²⁷ The Ellingham diagram for binary systems such as the Ni–O system is convenient for understanding the stabilities of the respective phases.²⁸ This diagram can be regarded as the chemical potential diagram plotted in vs . This type of diagram can also be extended to higher order systems by adopting the same approach as the generalized chemical potential diagrams for the multicomponent system; namely, the utilization of the third axis is powerful for displaying the complicated phase relations, and fixation of the chemical potential of selected species at a given value is another way of handling the multicomponent system.²⁷

In the electrochemical potential diagram, namely, the Pourbaix diagram²⁹ for the multicomponent system, an interesting approach has been adopted to display only the phase relations associated with the selected redox element.³⁰ For example, the Pourbaix diagram for the system³¹ is constructed to display the stabilities of the only Fe-containing species/compound; namely, those species in the subsystem are not explicitly shown. The construction of such diagrams is made by an algorithm, including the selection of redox element.

No such selection of redox element is made in constructing the generalized chemical potential diagrams to be utilized in the present investigation.^{27, 31} However, the same type of diagrams can be constructed using the generalized chemical potential diagrams as follows:

• (1)  
  First, the generalized chemical potential diagram is constructed as the three-dimensional diagram. For the system, can be adopted as the third axis.
• (2)  
  In this three-dimensional diagram, the sulfur-containing species appear in the higher side of the third axis; those stability areas belonging to the sulfur-containing species can be made transparent and essentially eliminated from the diagram.
• (3)  
  When the third axis is ignored, the aimed Pourbaix diagram can be constructed as a two-dimensional chemical potential diagram.

In the present investigation, this technique was adopted frequently to construct the proper diagrams appropriate for the present purpose.

The thermodynamic data for major compounds and gaseous species in the Ni–S–O–H–C system were taken from the main database of the thermodynamic database MALT for Windows.²⁴

In the present investigations, focus was placed on the eutectic formation and the solubility of O, C, H, and S in solid Ni metals. These mixtures were represented in the ideal association solution model; if necessary, hypothetic constituents were introduced, and their thermodynamic properties were determined so as to reproduce the binary phase diagrams for the Ni–X (,^{32, 33} S,^34–37 C,³⁸ H ^39–41) systems. Fitting was made by a trial-and-error method of determining data listed in Table I: enthalpy change for formation, entropy, heat capacity coefficients, and transition information (temperature and enthalpy) for respective states.

Table I. Thermodynamic properties of components of ideal association solutions used in the present investigation: , the enthalpy change for formation; , entropy at 298 K; , , , , and coefficients of heat capacity equation, ; , temperature limit of equation; and , enthalpy change to next equation.

Mixture	State	kJ/mol	J/K mol								Ref.
Ni solid solution
Ni	c1	0.0	0.0	49.41	−80.92	−7.28	100	0	631	0.092	22
	c2			20.42	10.13	16.07	0	0	1728	17.15	22
	1			38.91	0	0	0	0			22
NiS in Ni	ss	−10.4	85.92	58	5	−10	0	0			a
NiO in Ni	ss	−185.7	15.79	50.65	8.73	−3.08	0	0			a
NiH in Ni	ss	17.577	53.87	32	25	−25	0	0			a
NiC in Ni	ss	54	38.87	39	14.5	0	0	0			a
Ni–S–C–O liquid
Ni	c1	0.0	0.0	49.41	−80.92	−7.28	100	0	631	0.092	22
	c2			20.42	10.13	16.07	0	0	1728	17.15	22
	1			38.91	0	0	0	0			22
	L	−70.93	99.124	49.41	−80.92	−7.28	100	0			a
NiS	c	−94.1	52.97	44.69	19.04	−2.89	0	0	652	64.4	22
	1			38.91							a
	S	−133.0	78	64.44	20.75				2000	65.7	22
	L			91							a
NiC		71.93	55	55	0	0	0	0			a
	S	−872.91	92	125.9	41.51				1306	32.8	22
	L			180							a

^aPresent estimates/evaluates.

Because the Ni–S subsystem has eutectics, this was taken into account by treating the Ni–S liquid solution as the ideal association solution among Ni(l), , NiS(l), and , as given in Table I, although more sophisticated models have been applied to the Ni–S system by several authors. Because the present investigation focuses on a low sulfur content region, liquid sulfur was treated as an individual liquid that was assumed to coexist with the Ni–S solutions. For other binary systems, simple hypothetical constituents were assumed, as shown in Table I. In multicomponent liquids, no further complicated constituents were considered.

Solid solutions of nickel with S, C, O, and H were also treated as the ideal association solution in which NiS, NiO, NiH, and NiC were considered as hypothetical constituents in solid solutions. Their data are also presented in Table I; these thermodynamic data of hypothetical constituents were determined so as to reproduce the solid solubility and solid liquid equilibria reported in the literature.^32–41

Because the present investigation is focused on the sulfur behavior in fuels such as or CO, we utilized the generalized Ellingham diagram for the S–O–H and S–O–C systems, in which gases were included as species having a selected partial pressure.

Figure 1 shows the three-dimensional chemical potential diagram for the S–O–H system. In Fig. 1, the most dominant species in the gaseous phase are plotted in the three-dimensional diagram; the partial pressure of gaseous species belonging to the H–O subsystem (, , and ) is specified at 1 atm, whereas those gaseous species containing sulfur atoms are given at selected values (1 atm in Fig. 1c and in Fig. 1a and 1b). In any plot, the three-dimensional diagram consists of polyhedrons for sulfur-containing species and polygons for , , and . In Fig. 1b and 1c, those polygons for , , and are set as transparent in such a way where only sulfur-containing species are visible. This transparency corresponds to the treatment in the multicomponent Pourbaix diagram, as described in the previous section.

Zoom In Zoom Out Reset image size

In Fig. 2, the predominant areas for the sulfur-containing species/compounds are displayed in two-dimensional diagrams; Fig. 2a corresponds to those given in Fig. 1b and 1c. These predominant areas of the sulfur-containing species are adjacent with those for , , or at 1 atm; therefore, the phase relations given in Fig. 2a are those in equilibrium with (1 atm) in the lower part of the figure or (1 atm) in the upper part; the borderline for the two conditions shown by the dashed line in the figure, , is also given inside the dominant area of in Fig. 2a.

Zoom In Zoom Out Reset image size

In Fig. 2a for , the species of has the dominant region at the oxidative atmosphere, whereas , , and S (condensed phase) appear in a reducing atmosphere, and gas becomes dominant in the hydrogen-rich atmosphere. When the impurity level of sulfur is reduced to an order of , the appearance of species becomes different; that is, species such as , , and S(condensed phase) disappear. At low sulfur concentrations, the borderline between the dominant and dominant areas becomes important. This line is represented by the following chemical reaction

In this reaction, is contributed to the reaction instead of because this borderline is within the dominant region. As a result, the oxygen potential value for this borderline does not depend on a selected partial pressure of SX ( and ) because is always in unity on the borderline.

In Fig. 2b, the corresponding diagrams for the C–S–O system are displayed. Phase relations become complicated because the C–O subsystem has three predominant areas for C, CO, and , which have a corner around and . As sulfur-containing species, , , S(condensed phase), , COS(g), CS(g), and appear as dominant. When compared with the S–O–H system, the wider ranges of S(condensed phase) and are due to the fact that the stability of COS or is not significant compared with . For , the COS predominant area is extended in a similar manner to that for in the S–O–H system. However, a difference can be seen in the location of the borderline between and COS corresponding to the following reaction

This borderline is located in the more reductive side than the borderline in Fig. 2a.

As described above, the sulfur potential is a key property in the present analysis. Because the chemical potential diagram is a projection of phase relations on selected chemical potential values (or intensive variables), it is possible to construct those diagrams that explicitly utilize the sulfur potential as the diagram axis variable. In Fig. 1, the sulfur potential is not used just because the Ellingham-diagram-type variables are already adopted in this investigation. Instead, chemical equilibrium calculations were made to explicitly show the partial pressure of as a function of oxygen potential. Calculations were made on the S–O–H–C system in which a total amount of sulfur is limited to be in 1 mol of fuels (, , and C) at and . Results are shown in Fig. 3, in which mole numbers of only those sulfur-containing species are plotted as a function of oxygen partial pressure, and the mole number of , which is indicated as solid squares, is roughly equal to the logarithmic partial pressure of because the initial mole of the fuel is set at 1 mol.

Zoom In Zoom Out Reset image size

For hydrogen fuel, the dominant species changes from in the reducing side to in the oxidative side. This is essentially the same as Fig. 1b and 2a, although is assumed in those diagrams. However, the amount of has a maximum in the same oxygen potential region as that appearing as the -dominant area in Fig. 1c and 2a. The borderline between and is shown as line 1 (L1) at in Fig. 3a, whereas the borderline between and is given as line 2 (L2) at .

For carbon fuel shown in Fig. 3c, the sulfur-containing dominant species changes from via COS(g) to . Again, the species does not appear as the dominant species in Fig. 3c. In Fig. 2b, the oxygen potential width for the area at is larger in the S–O–C system than in the S–O–H system. Correspondingly, a maximum amount of on the borderline in Fig. 3c is higher than that in the S–O–H system in Fig. 3a. This borderline shown as line 3 (L3) at is about 1 order of magnitude lower than that for line 1 (L1) in Fig. 3a. In the COS dominant area, the carbon dominant chemical form changes from to CO and C(graphite). Their borderlines are given as line 4 (L4) at and line 5 (L5) at . About the amount, there appears one interesting feature in the S–O–C system; that is, the amount of again increases with decreasing oxygen potential down to the region where becomes dominant. Because the -dominant area in Fig. 2b is within the C-dominant region in the C–O binary system, the partial pressure becomes constant through the following reaction

where has the fixed partial pressure and carbon activity is also fixed at unity. Because the results of Fig. 3c were obtained at a total pressure under a selected oxygen potential, becomes dominant around line 6 (L6) at and corresponds to a gaseous mixture consisting of 2/3 of COS and 1/3 of .

Figure 3b shows the sulfur-containing species for the case where is used as fuel. When compared with Fig. 3a and 3c, the interesting features can be seen as follows:

• (1)  
  shows higher stability compared with COS(g) and and therefore becomes dominant even in hydrocarbon fuels. The percentage of COS(g) is about 2% in maximum. The concentration of is extremely small.
• (2)  
  Because is dominant, the borderline oxygen potential from to is about the same as that for hydrogen. Furthermore, the peculiar behavior in the carbon stable region of the sulfur potential in the S–O–C system shown in Fig. 3c [at lower than L6] disappears in the hydrocarbon system.
• (3)  
  In the SOFC anode chamber, therefore, the partial pressure of is a good measure to make further considerations on thermodynamic equilibria associated with sulfur poisoning. Even so, when the sulfur potential becomes crucial, care must be exercised because the sulfur potential changes rather sensitively to the behavior of other substances, although the major form of sulfur is always . This is particularly true for the Ni–S interaction, as seen below.

The chemical potential diagrams were constructed for the Ni–S system and are displayed in Fig. 4. The same phase relations are plotted as the Ellingham diagram in Fig. 4a and also as the Arrhenius-type diagram in Fig. 4b.

Zoom In Zoom Out Reset image size

The most important feature in the Ni–S system is that the lowering melting temperature due to the eutectic formation is quite significant in the Ni–NiS subsystem.^{34, 35} Another important feature associated with such eutectics is that with changing temperature under a constant sulfur partial pressure in Fig. 4b, the liquid eutectics appear in between solid metal and solid metal sulfide. This feature plays an important role in considering the nickel sulfur interaction in the SOFC anode environment.

Chemical potential diagram at selected temperatures

Phase equilibria in the Ni–S–O system have been investigated well in terms of chemical potential diagrams such as a vs plot^{8, 23} because this system is very important in hot corrosion science since the 1970s.^42–44 In addition, the above consideration on the concentration of sulfur impurities in fuels makes it clear that in the anode environment, the partial pressure of becomes a key, as indicated in Fig. 3. This suggests that the variable of should be included in the chemical potential diagrams for the Ni–S–H–O or Ni–C–S–O–H systems; there can be several ways of constructing two-dimensional diagrams for such multicomponent systems. For example, Sasaki et al. adopted the condition of .⁸ Another method is to draw the isoactivity (isopartial pressure) lines corresponding to the selected partial pressures of and .⁴³ Here, such diagrams are displayed in Fig. 5. In the Ni–C–S–O–H system, two chemical potentials have to be fixed at selected values to draw a two-dimensional diagram. Here, and are adopted; these conditions correspond to the anode atmosphere at a rather high fuel utilization. As indicated in Fig. 5, nickel exhibits a complicated liquid formation; at 1073 K, the Ni–S eutectics appear between Ni and and another eutectic appears between NiS and , whereas at 1273 K, the Ni–S eutectics extend to more wider composition regions and combined with others. The isopartial pressure lines for (dotted lines in Fig. 5) are straight and increase with increasing oxygen potential, whereas the isopartial pressure lines for (dashed lines in Fig. 5) decreases with increasing oxygen potential. As a result, those redox points, where the partial pressures of and become equal to each other, have the maximum sulfur potential. During the fuel cell operation with fuel at a constant concentration, the sulfur potential increases with increasing fuel utilization. This indicates that the partial pressure of is determined using given partial pressures of and as follows

Because the sulfur content in the anode environment is characterized in terms of a constant partial pressure, Fig. 5 strongly suggests that the sulfurization of Ni can take place in the more oxidative side. The minimum partial pressure of corresponding to the sulfurization of Ni can be seen around at 1073 K. In Fig. 5c, corresponding isopartial pressure lines for COS in the Ni–S–O–C system at are given; in this case, the minimum partial pressure of COS corresponding to the sulfurization of Ni can be seen around at 1073 K.

Zoom In Zoom Out Reset image size

Temperature dependence of sulfurization equilibria

Temperature dependence of the above phase relations was obtained as follows: In the Ni–S–O–C–H system, the partial pressures of and are fixed at 0.50 and 0.25 atm, respectively, in the same manner. Then, the chemical potential diagram of a vs type was constructed with a parameter of the partial pressure of SX ( or ). Figure 6 shows the results for (a) and (b) . This diagram was constructed as follows: A three-dimensional diagram was constructed by including S-containing gaseous species at the selected partial pressure ( or ). In a similar manner to Fig. 1, the dominant areas of and were made transparent; as a result, the chemical potential diagram, without explicit indication of and , is shown in Fig. 6.

Zoom In Zoom Out Reset image size

Figure 6a is based on the predominant species for sulfur compounds, which are given in Fig. 2; therefore, roughly speaking, this diagram indicates the most stable chemical form of Ni within the dominant area or the dominant area. When compared with Fig. 5a and 5b, phase relations given in Fig. 6 correspond to those on the isopartial pressure lines for and in Fig. 5. This clearly shows why the sulfur potential is high around the borderline between the dominant and dominant areas. Under these conditions, the sulfurization of Ni or NiO starts from the lower temperature side along the borderline. At , the sulfurization takes place below 700 K (see Fig. 6a). The eutectic liquid starts to be formed as a result of sulfurization above in the intermediate temperature region of 900–1100 K (see Fig. 5a). With increasing partial pressure of , the Ni–S eutectic stability area (formation area) is spread up to the higher temperature side. Sulfurization is most significant on the borderline between the dominant and dominant areas. Also, the stability area for eutectics is near the NiO/Ni borderline, and that is also in contact with the NiO stable area as well as the Ni stable area.

Solid solubility of sulfur and oxygen atoms in nickel as functions of temperature and oxygen partial pressure

The present thermodynamic considerations take into account the solid solubility of carbon, oxygen, hydrogen, and sulfur within the ideal association model. This makes it possible to calculate sulfur solubility as a function of temperature, oxygen partial pressure, or fuel utilization. In Fig. 7, calculation results are given as a function of fuel utilization or oxygen potential with a parameter of the partial pressure of , namely, (1 ppm) and (100 ppm).

Zoom In Zoom Out Reset image size

With increasing fuel utilization or oxygen potential, the sulfur solubility increases. This is by the same mechanism shown in Fig. 5 in which increases with increasing under constant . For comparison, other elements such as hydrogen, carbon, and oxygen also increased, although these solubilities do not depend on the partial pressure of . Hydrogen concentration is rather high and insensitive to the fuel utilization, whereas carbon solubility decreases rapidly with fuel utilization. Oxygen solubility shows similar behaviors as sulfur.

The solubility of oxygen and sulfur is also shown in Fig. 8 as a function of temperature and oxygen potential, which should be compared with Fig. 6b for . Oxygen solubility was experimentally determined along the Ni–NiO two-phase line;^{32, 33} although there is some discrepancy in temperature dependence between the chemical³² and electrochemical³³ determination, the electrochemically determined values³³ are adopted to be reproduced in the present thermodynamic modeling. Along this Ni–NiO two-phase line (bold line in Fig. 8), oxygen solubility decreases with decreasing temperature. However, in the vs plot, the isosolubility lines are rather flat, as shown in Fig. 8. Compared with oxygen, sulfur solubility shows stronger temperature dependence under a constant oxygen potential, , or a constant electrical potential referred to as , [referred to as ] .

Zoom In Zoom Out Reset image size

Sulfurization in the Ni–S–O–C system

The chemical equilibrium calculations given in Fig. 3c suggest that the sulfur potential as a function of oxygen potential may exhibit quite a different behavior between the H–O–S system and the C–O–S system. Here, the appearance of the Ni–S eutectics in the Ni–O–C–S system is considered. As shown in Fig. 2b, any point in the Ellingham diagram can be characterized in terms of the S-containing dominant species as well as the dominant regions for the C–O subsystem and, as a result, a stability diagram consists of many polygons and makes it complicated to draw the Ni–S interaction. Here, Fig. 9 is constructed as a combination of two three-dimensional diagrams; one is obtained by fixing at with making the C-, CO-, and -dominant areas transparent, the other being ; the former corresponds to the lower part in Fig. 9, the latter being the upper part; a comparison between Fig. 8 and 9 indicates the following interesting features:

• (1)  
  Because Fig. 8 is constructed under the condition of , the sulfur content can be regarded as 1 order of magnitude lower in Fig. 9. Even so, the eutectic formation region appears in a similar way. This indicates that the sulfur potential becomes 1 order of magnitude higher in the Ni–S–C–O system.
• (2)  
  Another important difference in the appearance of eutectics is the oxygen potential values. For carbon fuels, about 1 order of magnitude in the oxygen potential is shifted to the more reductive side because the dominant borderline in Fig. 9 lies at oxygen potentials 1 order of magnitude lower than that at the dominant borderline in Fig. 8, as suggested by the difference between L3 and L1 in Fig. 3.
• (3)  
  The eutectics appear in the C-dominant region as well as in the -dominant region. This is mainly because there is a symmetrical change in the sulfur potential between the dominant region and the C(graphite) dominant region; the sulfur potential in both regions can be determined from the following reactions

  

  

These relations explain that the relation of the sulfur potential to the oxygen potential is completely inverse.

Zoom In Zoom Out Reset image size

Sulfur adsorption on nickel surface

In the catalyst field, the adsorption of sulfur on nickel has been intensively investigated for a long time. Bartholomew et al.⁴⁵ summarized the accumulated data in the vs plot. More recently, Hansen¹⁴ attempted to represent the sulfur coverage on nickel with the Temkin-like isotherm. Here, in Fig. 10, his proposed data for and 0.6 are plotted and compared with the present phase equilibria, including the solid solubility of sulfur. Clearly, the coverage of sulfur on the nickel surface is large even with low , and its temperature dependence is quite steep; the coverage of sulfur becomes about 60% even though the concentration is only 1 ppb at 1000 K, and it increases quite steeply by decreasing temperature. In the same plot, the sulfur solubility³⁶ exhibits inverse temperature dependence to that of the adsorption.

Zoom In Zoom Out Reset image size

In a normal operation, the fuel utilization is limited at 70–90%. To examine the sulfurization equilibria at the same fuel utilization, the phase equilibria given in Fig. 8 are replotted as a function of ; when this axis variable is equal to unity, this corresponds roughly to 90% fuel utilization. Figure 11 indicates that below 90% fuel utilization, sulfurization does not take place in the equilibrated state even for . Even so, in the electrochemical situation for the SOFC anode, there are several important nonequilibrium effects, such as overpotential, nonequilibrium between two and fuel systems, and temperature distribution due to reforming reaction.

Zoom In Zoom Out Reset image size

Before sulfur poisoning on the nickel anode is examined, the nickel anode reaction mechanism is described here in terms of the local equilibrium approximation for the elemental distributions in nickel and oxide [yttria-stabilized zirconia (YSZ), ScSZ, or doped ceria] in cermet anodes.

At the electrochemical reaction sites, the local equilibrium is established among the electrochemically active gaseous species and the electrode/electrolyte materials. Within the condensed materials including surfaces, we assume that the elemental chemical potentials are uniquely defined, and their change within the condensed materials is continuous. This is the same as the electrochemical treatment of high temperature corrosion proposed first by Wagner. Thus, the local equilibrium approximation makes it possible to treat the mass transfer through nickel metal or through the oxide component of cermet, although many other models of the anode reaction mechanisms^46–48 consider only surface species. The present focus on mass transfer even inside solid materials is due to the following reasons:

• (1)  
  In cathode reaction mechanisms, a similar reaction model based on the local equilibrium approximation can provide a good basis of interpreting the electrode/electrolyte chemical interaction^{49, 50} and also chromium poisoning.⁵¹
• (2)  
  In our other experiments with the secondary-ion mass spectrometry technique,^52–56 we observed a rather high concentration of dissolved oxygen atoms within a thin layer just below the nickel surface in addition to dissolved hydrogen or carbon atoms in the hydrocarbon fuel environment. The amount of such oxygen in nickel is enhanced under the anodic polarization⁵⁵ so that we have a tentative working idea that such oxygen can contribute to the electrochemical reaction in such a way that the dissolved (or adsorbed) oxygen catalytically assists the formation of the electrochemically active hydrogen in/on nickel from hydrogen molecules.⁴⁷ Such dissociated hydrogen atoms are electrochemically oxidized and emitted as water vapors at the reaction sites; such water vapors, in turn, provide oxygen atoms to be adsorped/dissolved at places other than the reaction sites.
• (3)  
  In our electrochemical impedance analysis using hydrogen or higher hydrocarbon fuels with pelletlike disk cells with Ni–ScSZ cermet anodes, we recognized that the two observed semicircles both depend on water vapor pressure regardless of fuels (hydrogen or higher hydrocarbons).¹⁵ From this observation, the above anode reaction model can probably be applied to both hydrogen and hydrocarbon. Reforming processes^{57, 58} may proceed by reacting with such adsorped oxygens on nickel in a similar manner as the hydrogen dissociation process.

From the above considerations, we adopt the idea that the mass flow associated with the electrochemical reaction should be accompanied with the distribution of the chemical (or electrochemical) potentials of corresponding chemical species and therefore the elemental chemical potentials of involved elements.

Distribution of oxygen potential and associated flows of oxygen and hydrogen

Because the hydrogen formation process and the charge-transfer process both proceed in the vicinity of TPBs, a steep oxygen potential gradient corresponding to the overpotential is developed in the vicinity of TPBs. When the flow of oxide ions/oxygen atom is considered, the TPBs become the major reaction sites where almost all oxygen atoms pass through. This is schematically illustrated in Fig. 12a, where the oxide ion's flow in the electrolyte is given in terms of the electrochemical potential gradient. Correspondingly, hydrogen molecules are chemisorped on the nickel surface and dissociated hydrogen atoms go to TPBs to meet with the oxide ions and to evolve water vapors. The hydrogen potential in and on nickel should become a minimum, and the oxygen partial pressure in the gaseous phase should be a maximum in the TPB area. In Fig. 12a, the oxygen potential distribution is also given inside nickel; in Fig. 13, the distribution of chemical potential, coverage on nickel, and concentration of dissolved atoms inside nickel are shown schematically. A rather large gradient is due to the slow diffusion of oxygen atom in nickel. A more moderate distribution can be expected for the hydrogen potential in nickel because of the higher diffusion coefficient in bulk nickel; at 1273 K, , whereas in Ni. Although the main hydrogen flux passes through a thin surface layer toward the TPBs, there can be some flux even inside Ni.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In the lower part of Fig. 12a, the oxygen potential, hydrogen potential, oxide ion electrochemical potential, proton electrochemical potential, and chemical potential of water vapor are also given as a function of location on (I) the YSZ surface, (II) the nickel and YSZ interface from the TPB A to B, and (III) the nickel surface from the TPB of points B to C; it is assumed that there is no potential gap across the nickel and YSZ interfaces. The hydrogen potential and the oxygen potential both have higher values at the interfaces (region II) than those at TPBs (points A and B). As a result, there is some possibility that some parts of nickel can be oxidized to be NiO even when the anode potential is lower than that for the Ni/NiO equilibrium. Also, the hypothetical vapor pressure of water has a maximum along this nickel–electrolyte interface.

On the basis of the above anode reaction model and associated oxygen potential distribution in the local equilibrium approximation, accumulated experimental results on sulfur poisoning can be considered in several points as follows.

Features of sulfur poisoning: Rapid poisoning and reversible but slow recovering

Features of sulfur poisoning: Irreversible processes

These degradation behaviors should be compared with the equilibrium properties described in the present investigation with the aid of the anode reaction mechanism model based on the local equilibrium approximation; particularly, the partial pressures of , , and CO(g) or oxygen potential, hydrogen chemical potential, and sulfur chemical potential in the TPB vicinity should be crucial.

Before detailed analyses are made, the adsorption/desorption processes are first examined from the electrochemical point of view. Because the adsorption of sulfur from the gaseous molecule on nickel is a rather fast process, this process can be described as the following reaction

From the equilibrium thermodynamic point of view, this provides the appropriateness to treat the adsorption as a function of . Even so, this chemisorption is similar to the hydrogen gas dissociation (hydrogen atom formation) process associated with the electrochemical oxidation of hydrogen gas. This can be written as

Because H(on/in Ni) should be consumed in the charge-transfer process, the combined reaction can be written as

This is the electrochemical oxidation of to form and S adsorption. This is quite similar to the electrochemical reduction process of in the Cr poisoning of the lanthanum strontium manganite cathode.⁵¹

In a similar manner, the desorption process can be discussed by two processes. The normal chemical process is the reverse reaction of Eq. 7. The electrochemical process available can be the following

although this reaction is not so favored from the thermodynamic equilibrium point of view. No reverse reaction of Eq. 9 is expected because such a process is electrochemical oxidation and will not proceed under the fuel cell operation mode. Here, we come to realize an important fact that the side electrochemical reaction such as Eq. 10 is irreversible, whereas a similar process of sulfur deposition such as Eq. 7 proceeds as a reversible process because no oxide ion in the electrolyte nor electron in nickel participated in the process.

For the "reversible" sulfur poisoning stage and the subsequent slow recovering stage, the followings features should be noted:

• (1)  
  Blocking by chemisorped sulfur: As suggested by Hansen,¹⁴ the coverage of sulfur on nickel is one of the crucial measures. Because the poisoning speed of the initial performance drop is rather fast, the chemisorped sulfur blocks the critical processes of the electrochemical oxidation of hydrogen molecules or hydrocarbons. This strongly suggests that chemisorptions take place preferentially as the electrochemical oxidation of , as given in Eq. 9.
• (2)  
  In some cases, an induction period of time is observed before a stepwise decrease in voltage. This can be explained in terms of time for sulfur to reach all available TPBs; when only a part of the TPBs are covered with sulfur, a redistribution of current density takes place so as to maintain a high voltage. Furthermore, the trapping ability of nickels may be different from anode to anode.
• (3)  
  Slow recovery process is discussed here in terms of the reversible adsorption/desorption process or the electrochemical volatilization reaction. Desorption is a reaction of the chemisorped sulfur to be removed with hydrogen to form so that the equilibrium pressures of and and a flow rate of hydrogen are factors of determining the recovering speed. As shown in Fig. 10, an extremely low concentration of is needed to reduce the surface coverage of sulfur on nickel. In addition, the hydrogen partial pressure is small in the TPB vicinity where sulfur tends to be adsorped. However, the electrochemical oxidation reaction (Eq. 10) of chemisorped sulfur to form can also be considered as a candidate for the desorption process. As described above, this reaction is favored only in the high oxygen potential region and, therefore, the rapid removal cannot be expected. Even so, an important feature is that the removal process takes place first from sulfur near the TPBs. In view of these two different possible mechanisms, it is interesting to identify which gaseous species is dominant during the recovery process. Anyway, from the equilibrium property for the adsorption of sulfur, the slow recovering process is understandable.

To interpret the irreversible damage of nickel anodes, it must be necessary to make further considerations on other factors influencing sulfur behavior.

Diffusion

Effects of Ni sintering

Flooding

Enhancement effects

Effects of hydrocarbon fuels