Oxidation of Metals, Vol. 37, Nos. 3/4, 199"2

The High-Temperature Corrosion of Fe-Nb Alloys in a H2/H20/HzS Gas Mixture W. Kai,* D. L. Douglass,* and F. Gesmundot Receit~ed August 26, 1991; revised October 15, 1991

The corrosion of Fe-Nb alloys containing up to 40 wt. % Nb has been studied over the temperature range 600-980~ in a mixed gas of constant composition having sulfua" and oxygen pressures ranging from 10- 8 to 10- 4 atm. and fi'om 10 -27 to 10-18 atm., respectively. All alloys were two-phase, consisting of an Fe-rich solid solution and an intermetallic compound, Fe2Nb. The scales formed on the Fe-Nb alloys were duplex, consisting of an outer layer of iron sulfide (FeS) and an inner complex layer o f Fe~NbS2(FeNb2S4 or FeNb3S6), FeS and unreacted Fe2Nb. No oxides were detected at any temperature. The addition of Nb reduced the corrosion rate. The corrosion kinetics of Fe-Nb alloys followed the parabolic rate law, regardless of alloy composition and temperature. Platinum markers, attached to the original alloy surfaces, were always located at the interface between the inner and outer scales. KEY WORDS: sulfidation; Fe-Nb alloys; FexNbS2 ; corrosion kinetics. INTRODUCTION Fe-, Ni-, and Co-base alloys are common and versatile engineering materials used in m a n y corrosive, high-temperature environments such as those involved in coal gasification and oil-refining processes. When the alloys are exposed to mixed-gas environments containing high sulfur and low oxygen partial pressures, the predominant corrosion problem is sulfidation. In fact, the highly defective nature of sulfides leads to the growth of less-protective *Department of Materials Scienceand Engineering, School of Engineering and Applied Science, University of California, Los Angeles, California 90024-1595. ]-Istituto di Chimiea, Facolta di Ingegneria, Universita di Genova, 16219 Genova, Italy. 189 0030-770X/92/0400-0189506.50/0 9 1992 Plenum Publishing Corporation

190

Kai, Douglass, and Gesmundo

layers. Furthermore, the low eutectic temperatures in these metal/sulfide systems exacerbates the sulfidation problem. The sulfidation rates of these base metals are generally many orders of magnitude greater than their oxidation rates. Alloys having good oxidation resistance, such as those containing additions of Cr, Si, and A1, generally have poor sulfidation resistance. ~'2 However, some of the refractory metals such as Nb, Mo, and W exhibit excellent sulfidation resistance at elevated temperatures. 34 During recent years, considerable research on high-temperature sulfidation of numerous metals has been developed and summarized by Mrowec and Przybylski 3'4 and by Stratford. 5 Niobium was reported to have the best sulfidation resistance. Recently, the sulfidation behavior of the base metals (Fe, Co, and Ni) with additions of either Mo or Nb in sulfur vapor has been studied in this laboratory 6 ~1 It was shown that a significant reduction in the sulfidation rates of the base metals could be achieved by alloying them with either Mo or Nb, even though the sulfidation rates were still significantly higher than those of pure Mo or pure Nb. It was of interest to extend the previous work to examine the corrosion behavior of the same materials in practical environments containing two oxidants, sulfur, and oxygen. The purpose of the work presented here is to investigate the effect of Nb on the corrosion kinetics and the reaction mechanism of Fe in a gas mixture of H2/H20/H2S. The investigation of the role of Nb on Co j2 and Ni-base 13 alloys in the same mixed gas is currently being studied in this laboratory. EXPERIMENTAL PROCEDURES The starting materials were 99.98% Fe lumps and 99.8% Nb turnings (AESAR Chemical Co.). Pure Fe samples were directly selected from the lumps and fine-polished down to 6 Ftm diamond paste. The alloys and pure Nb were fabricated into 60-g buttons by arc-melting on a water-cooled copper hearth. Ti turnings were melted initially to getter the argon atmosphere. The buttons were flipped over and remelted at least six times to ensure alloy homogeneity. The as-cast samples were cut in half, and then annealed in vacuum at 1000~ for 24 hr as well as at 900~ for another 24 hr. The annealed samples were sliced into about l-mm thick coupons using a lowspeed diamond saw (ISOMET). A suspension hole was drilled through each coupon by a spark-cutter. These coupons were first ground with 600 SiC paper and then polished to a 6 - / t m finish. The specimens were ultrasonically cleaned with either acetone or methanol. The average weight and surface area of the specimens were generally around 0.6 g and 2.5 cm, 2 respectively. Typical optical micrographs of annealed Fe 10Nb, Fe-20Nb, Fe-30Nb, and

Corrosion of Fe-Nb Alloys

191

Fig. 1. Optical micrographs of annealed Fe-Nb samples (a) Fe 10Nb, (b) Fe-20Nb, (c) Fe-30Nb, and (d) Fe-40Nb.

Fe-40Nb alloys are shown in Figs. l a-d. The dark-gray area is the Fe-rich solid solution, and the light area is the intermetallic compound Fe2Nb. The nature of all phases agreed with the Fe-Nb phase diagram shown in Fig. 2.14 The Fe 40Nb contained two phases although the phase diagram shows that the equilibrium structure is a single-phase intermetallic compound with variable stoichiometry. Obviously, the equilibrium structure was not attained during annealing.

192

Kai, Douglass, and Gesmundo Atomic 0 : 9

l

I0 I

zo i

Percent :~o i

Niobium

40

50

J ......

i .

.

.

.

~o i .

7o

8o

9o

1oo

..... , , 2~c~

2,2.00,

/ / /

~'~O.

/ /

I

1800-

1800-

E

/

/

9 9 ~e

t4OO5~2.

E-,

18.8

(6Fe)

9

38

(~b)-~,

|:~|0"r

IZO0~ 1,5 (-?'re)

9

I000

i $ #

Nl*c7

),7

~:

800 Magnetic Transformation 600

........

Fe

, .........

i0

,

20

, ......

30

, .........

40

Weight

, ............

80

Percent

86 ....

Niobium

~'0 . . . . . . .

8'0 .......

~

......

1oo Nb

Fig. 2. Fe-Nb phase diagram.

The experimental apparatus, schematically shown in Fig. 3, is slightly different from that previously used for sulfidation studies in this laboratory. 6 The major difference involves the use of pre-mixed H2S/H2/ Ar gas, instead of the nature of elemental sulfur vapor in the bottom section. The initial composition of the mixed gas and the calculated equilibrium partial pressures of $2 and 02 are reported in Table I. The mixed-gas inlet pressure was controlled by a regulator set at 2 psig. The mixed gas first passed through a beaker filled with H20 at room temperature, and then passed through a gas washer containing H20 within an ice container. This was done to obtain saturated H20 vapor at 0~ (4.579 mmHg)J 5 The flow-rate of the mixed gas was kept constant at 150 cc/min during the experimental runs. The kinetics measurements were performed by means of a quartz-spring thermobalance. The spring elongation, measured by a cathometer, was converted into weight gain by multiplication by the spring constant. Then, the weight-gain data were plotted against the square root of time (sec 1/2) and fit by a linear-least-square function. The parabolic rate constants were calculated from the slope by a linear regression. The corroded samples were analyzed by x-ray diffraction using monochromatic C u - K c t radiation at 35 KV applied voltage and 15 mA current.

Corrosion of Fe-Nb Alloys

193

(~

gas exhgmte d

__ ~ z t i a r Tape

9~

-

~na~ple [o~.ded

]l~,rm~,~ple

Furitlce

Sample

,~------------ m i x e d

Fig. 3. Schematicdiagram of the experimentalapparatus. A sequential grinding technique was adopted to study the distribution of the different phases within the corroded scales, x-ray diffraction data were compared with the JCPDS standard files to identify the various phases present.

194

Kai, Douglass, and Gesmundo Table I. Partial Pressures of $2 and 02 in Mixed Gas a

Temp.(~ 600 700 800 900 980

P,S2(atm.)

P,O2(atm.)

4.452 x l0 -8 5.594x 10 7 4.496x 10 6 2.517x10 5 8.114x10 5

2.719 X 10 _27 2.881 x 10 _24 8.878x 10 ..22 1.020x10 19 2.634x10 18

"The composition of the mixed gas is H2S (1.92 vol.%), H2 (13.40 vol.%) and Ar (84.68 vol.%). The mixed gas passes through H20 at 0~ to obtain the saturated vapor pressure (0.6025%). Also, the partial pressure of H2 is about 0.158 atm. in the temperature range of 600-980~

Corroded samples were prepared for metallography by cold mounting in an epoxy resin. The samples were then ground and polished through 0.05/zln Al203-powder, cleaned in acetone and finally etched by a picric acid/methanol solution. The general microstructure of the corroded samples was examined by an optical microscope and a scanning electron microscope (SEM). An electron probe microanalyzer (EPMA) with a wavelength dispersive x-ray spectrometer (WDS) was used for compositional studies and xray mapping/line-scanning analyses. Platinum wire (25/.tin diameter) was used as a marker by spot-welding on the surface of the samples. The marker samples were corroded and then prepared for microstructure examination by the same procedure described above. RESULTS Kinetics

The reaction kinetics of Fe-Nb alloys containing up to 40 wt.% Nb are shown as parabolic plots in Fig. 4. In general, the corrosion rates followed the parabolic rate law, but a two-stage parabolic behavior was noted, consisting of an initial slow transient stage followed by a faster steady-state stage. The length of the transient stage decreased with increasing temperature. The parabolic rate constants for pure Fe, pure Nb, and Fe-Nb alloys in the mixed-gas environment are summarized in Table IIa. These data show that alloys containing up to 30 wt.% Nb corroded slightly slower than pure iron (2-6 times); however, when the Nb content was increased from 30 to 40 w/o, a substantial decrease in the corrosion rate occurred at 900~ by a factor of 37. However, the corrosion rates of the Fe-40Nb are still one to one-and-a-half orders of magnitude higher than those of pure Nb. The decrease in the parabolic rate constants with increasing Nb content is shown

Corrosion of Fe-Nb Alloys

195

/

. g=o-

r ~ ,

__

"

,,1///,oo..J

1

r

/

o

~~

b

:, )oo> ,~EE ~o~

~Io

iiI 20,"

Nb

oo'c

Fe 0. N 9o0 c

~ ' J ~ 700~ Time 1/2 (sec ~ )

d Fig. 4 Reaction kinetics of Fe-Nb alloys (a) Fe-10Nb, {~o) Fe-20Nb, {c) Fe-3ONb, and (d) Fe 40Nb.

in Fig. 5. For comparison, the previous results 7:~ concerning the corrosion of Fe-Nb alloys and pure Fe in sulfur vapor (Psi=0.01 arm.) are also reported in Table IIb. The parabolic rate constants in the mixed-gas environment are lower than those in sulfur vapor, under the same testing temperatures and alloy composition, by a factor of ranging from 19 to 15. An Arrhenius plot of the rate constants is shown in Fig. 6. The activation energies of Fe-Nb alloys and pure Fe are reported in Table III. There are apparently two distinct activation energies of approximately 22 27 Kcal/ mole for pure Fe, Fe-10Nb, Fe-20Nb, Fe-3ONb and pure Nb, and of 17

196

Kai, Douglass, and Gesmundo Table

l l a . P a r a b o l i c R a t e C o n s t a n t (grnZ/cm4/sec) o f F e - N b Alloys in M i x e d G a s a

Temp.(~

600

700

800

900

980

PureFe

2.07 • 10 -8

9.92•

Fe-10%Nb Fe-20%Nb Fe-30%Nb

7.52 6.84 6.05 1.29

1.69 x 10 -8 1.66 • 10 -8 1.64• 8

1.23 x 10 -7 9.74 • 10 -8 9.02• 8

2.89 x 10 -7 3.47 • 10 -7 1.57•

6.64 x 10 7 6.39 • l0 7 4.63•

6.60 x 10 -9 1 . 5 8 x 1 0 io

8.08 x 10 -9 2.55• io

1.47 • 10 -8 9 . 7 1 x 1 0 l0

3.82 • 10 8 1.76• 9

Fe-40%Nb PureNb

• x • •

10 9 10 -9 10 9 10 -9

--

-8

2.63x10

7

5.42x10

7

1.12•

"Kp values at low temperatures are for steady-stateconditions. Table Ilb. Parabolic Rate Constant (grn2/cm4/sec) of Fe Nb Alloys in Sulfur Vapor (Ps2=0.01 atm.) Temp.(~

600

700

800

Pure F e

3.70 • 10 -7

9.50 x 10 -7

4.60 x 10 -6

Fe-10%Nb Fe-20%Nb Fe-30%Nb

3.30 x 10 _7 1.30 x 10 -7 6.10• -~

8.00 • I0 7 4.70 x 10 -7 1.20x10 7

1.50 • 10 6 1.30 • 10 -6 3.60• 7

900 6.00 X [ 0 - 6 2.30 x I0 6(850~ 2.70 • 10 6 5.10• 7

Kcal/mole for Fe-40Nb. These differences in the Q values do not have a special significance because the gas used had a constant composition, so that the actual equilibrium partial pressures of both oxygen and sulfur were both increasing with increasing temperature. Scale Morphology and Constitution The single-layered scale formed on pure Fe was iron sulfide, containing large columnar grains as shown in Fig. 7a. On the contrary, only a thin layer of NbO2 formed on pure Nb as shown in Fig. 7b. All the alloys formed instead duplex scales, consisting of an outer layer of iron sulfide and a Complex inner layer of FeNb2S4 (and/or FeNb3S6), FeS and unreacted Fe2Nb. These scales are similar to those formed by reaction of these alloys with elemental sulfur vapor. 7 A continuous protective NbS2 layer did not form, and internal sulfidation was also absent. Based on the x-ray diffraction results, there was no evidence of oxidation for either Fe or Nb in any of the alloys. In addition, the outer FeS layer spalled severely at high temperatures ( > 800~ on Fe-10, 20, and 30Nb, whereas only mild spalling was observed below 800~ The spalled FeS always had a strong texture as evidenced by high-intensity (300) peaks. By contrast, the scales formed on Fe-40Nb exhibited nearly no spalling even at very high temperatures. The nature of the scales is exemplified in Fig. 8, which shows a BSE micrograph and the x-ray maps of Fe, S, and Nb of a sample of Fe 10Nb exposed for 29 hours at 600~ The two-phase nature of the substrate and

197

Corrosion of Fe-Nb Alloys

10-61~..q) 03 10 -7. 9I~0~

E o O4

9~30~

E)~l0 -8.

8bo~

700~

13_

5s

1 0 -9 I 0

t 10

I 20

wt.

I 30

X

630~

I

40

of Nb

Fig. 5. The composition of dependence of the reaction kinetics of Fe Nb alloys.

the complex nature of the inner layer are clearly evident. A thin, outer layer of FeS formed on top of an inner layer, with good adherence between the two layers. From the Nb map (Fig. 8d), it is evident that Nb was not present in the outer layer. Another example (Fig. 9a), shows a sample of Fe 10Nb corroded at 980~ for 1.7 hours. Obviously, the outer FeS layer spalled. The detailed inner-layer structure is shown in Fig. 9b. The light phase (pt. 1) is Fe2Nb. The black spots (pt. 2) inside Fe2Nb are double sulfides having an average composition of FeNb].gSzv. Compositional variations across the scale in Fig. 9a are shown in the EPMA traverses in Fig. 9c. The heterogeneous nature of attack can be seen in Fig. 10 which shows a large particle of Fe2Nb extending across the inner reaction zone into the

198

Kai, Douglass, and Gesmundo

1 0 -6.

d r./) 10

-~"

E

X

C)

!!!

r

E:)'~1 0

b b 4b

-e.

0..

b

1 0 -9

I

7

8

I

1

9

10

I

I

11

12

1/T , 1 0 0 0 0 (1/t<)

Fig. 6. Temperature dependence of corrosion rate constants.

Table III. Activation Energy of Fe-Nb alloys Activation energy Alloys

Mixed-gas corrosion (Kcal/mole)

Sulfidation only (Kcal/mole)

Pure Fe Fe- 10%Nb Fe 20%Nb Fe-30%Nb Fe-40%Nb Pure Nb

22.2 26.7 27.4 24.7 17.2 21.8

20.6 15.0 20.7 15.3

13

Corrosion of Fe-Nb Alloys

199

Fig. 7. Cross-sections of corroded samples (a) pure Fe at 800~ for 4 hr, (b) pure Nb at

900~ for 4 hr, (c) Fe-10Nb at 600~ for 29 hr, and (d) Fe 20Nb at 800~ for 7 hr.

substrate in F e - 2 0 N b exposed 8 hours at 700~ The high N b content o f this phase imparts greater sulfidation resistance than that of the matrix, so that the particle is virtually unreacted even at the exterior of the inner layer. Point 1 in Fig. 10b was found by E P M A to be Fe2Nb, while point 2 is a partially-sulfidized tip of the intermetallic particle. The variation in composition across the various regions, as determined by the microprobe analyzer, is shown in Fig. 10c.

200

Kai, Douglass, and GesmundQ

Fig. 8. (a-d) EPMA BSE image and X-ray maps of Fe-10Nb exposed 29 hr at 600~

Corrosion of Fe-Nb Alloys

201

Substrate-----~[

In ner-la yer------~-.,~O u ter-layer o

r

~

4

%-

2

Nb 0

0

20

L---I

40

i

13

l

i

8o

60 Fe

100

120

x (urn) +

Nb

14o ,0

160

180

200

S

Fig. 9. (a) BSE micrograph of scale formed on Fe-10Nb exposed 1.7 hr at 980~ (b) high magnification, and (c) electron microprobe traverse of (a).

The scale formed on a sample of Fe-30Nb corroded for 7 hr at 800~ is shown in Fig. 11. The outer layer of iron sulfide (Fig. 1 la) shows two different morphologies: a thick compact columnar as region similar to that of iron sulfide formed on pure Fe and an irregular find grained region grown on top of the inner layer. Some unreacted Fe3Mo2 particles (bright phase) are found in the inner layer even though they are located at the inner/outerlayer boundary, as shown in Fig. 1 lb. A typical cross-section of a scale formed on Fe-40Nb after corrosion for 8 hours at 700~ is shown in Fig. t2a and b. The E P M A traverses of Fe, S, and N b across the scale are shown in Fig. 12c. The structure of the

202

Kai, Douglass, and Gesmundn

t,,-~-.,(- - Substrate

O u t e r - l a y e r - - D ~ I n net-layer [(

Fe2Nb----

~

6

4

ii-

9

a~ v

a

1

o

. . . .

[1

a 40

Fe

i

. .i. . ~

i . . . . !. . . 80

~

! *" '" ; ? "r ; LgO

8

Fig. 10. (a) BSE micrograph of scale formed on Fe 20Nb exposed 8 hr at 700~ (b) high magnification, and (c) electron microprobe traverse of (a).

inner layer of the scales formed on these samples differed from that found for Fe-10Nb, Fe-20Nb, and Fe-30Nb, while the outer layer was still composed of pure iron sulfide. In fact, due to the two-phase nature of this alloy, the formation of sulfides is limited mainly to the Fe-rich solid solution (dark regions), while the large islands of the Nb-rich phase Fe2Nb are only slightly sulfidized. At higher magnifications (Fig. 12b), the micrograph of the inner layer shows at least three different phases. EPMA analysis shows that the white area (pt. 1) is unreacted Fe2Nb, the gray area (pt. 2) is a double sulfide (FexNb2S4), and the dark area (pt. 3) is FeS. Moreover, some small islands (pt. 4) within the FeS zone also contained Nb. Quantitative analysis gives

Corrosion of Fe-Nb Alloys

Fig. 11. (a) BSE micrograph of scale formed on Fe 30Nb exposed 8 hr at 700~ and (b) high magnification.

203

204

Kai, Douglass, and Gesmundo

~-~In ner-layer,,--I~ Outer-la yer

Substrate

d

5

4

c

.

1

20

40 D

(SO Fe

80 X+(um)$

100 <,

120

140

Nb

Fig. 12. (a) BSE micrograph of scale formed o]1 Fe 40Nb exposed 8 hr at 700~ (b) high magnification, and (c) electron microprobe traverse of (a). the composition as FeL~3NbS. The larger amount of FezNb in the innerscale region for this alloy resulted in a significant reduction in the corrosion rate, as shown in Table II. Short-Term Corrosion

To help in understanding of the reaction kinetics and to determine the initial corrosion behavior of the two-phase alloys, short-term corrosion tests

Corrosion of Fe-Nb Alloys

205

were carried out at 800~ for 3 min. A transient stage was not observed for either pure Fe or Fe-10Nb specimens due to the formation of a continuous layer of iron sulfide on the surfaces but was present for the higher Nbcontent alloys. A typical initial-stage scale formed on Fe-20Nb is shown in Fig. 13. The plan view of the surface scale (Fig. 13a) revealed that small white nodules of iron sulfide formed on the surface. The BSE micrograph (Fig. 13b) clearly shows that the intermetallic compound Fe2Nb (bright image) had not been corroded at all. A cross-section showing the preferential growth of the iron sulfide is shown in Fig. 13c. It is evident that the iron sulfide formed only in the region of the Fe-rich solid solution (dark phase), while no sulfide formed on the interrnetallic compound (bright phase).

Fig. 13. (a) SEM plan view of the surface of Fe-20Nb after corrosion at 800~ for 3 rain, (b) BSE image of (a), and (c) BSE micrograph of the cross-section of (a) (unetched).

206

Kai, Douglass, and Gesmundo

Marker Studies

Three marker studies were performed on different alloys and temperatures to determine the corrosion mechanism of the overall reaction. According to the BSE images of Pt-wire markers shown in Fig. 14, all markers were located at the interface between the inner layer and the outer layer. DISCUSSION The behavior of the Fe Nb alloys in a mixed-oxidant gas containing both sulfur and oxygen was, surprisingly, virtually the same as that observed in pure sulfur vapor. 7 No evidence of any oxides was found by XRD, EPMA, etc. The main features observed were: (1) the parabolic rate was obeyed, although in some cases, a two-stage parabolic behavior was noted, (2) bilayered scales formed, (3) no Nb was detected in the outer layer of the scale, and (4) unreacted particles of the intermetallic compound Fe2Nb always existed in the inner-scale layer for all the alloys. The parabolic rate law was generally followed, although two-stage kinetics for some alloys were noted at certain temperatures. The transient stage seems to be due to initial nucleation and growth of scales on the surface. According to the short-term corrosion experiments the initial corrosion product is only iron sulfide, in agreement with the fact that its growth kinetics is faster than that of any other sulfides. Also, at the outset, its formation will essentially result from the corrosion of the Fe-rich solid solution (a-Fe) particles. After a short period, the kinetics behavior reaches a steady-state stage and both a-Fe and the intermetallic compound (Fe2Nb) react simultaneously. This later stage corresponds to the true reaction kinetics. The corrosion rate of pure-Fe in the mixed gas is about 1.5 orders of magnitude lower than in sulfur vapor (Ps2=0.01 atm.), as shown in Table II. This is in agreement with the dependence of Kp on Ps2, as reported by the sulfidation of pure iron. j6,j7 Thus, in principle, the Kp at lower Ps2 should be also lower. However, there may also be a kinetics control due to the surface reaction of Fe with H2S/H2 ]7,~8 Furthermore, although the corrosion products of the F e - N b alloys in the mixed gas are almost identical to those found in sulfur vapor, the corrosion rates in the mixed gas are also markedly lower. More precisely, up to 800~ the corrosion rates of F e - N b alloys (10-30 w/o) in the mixed gas are about 1.5 orders of magnitude lower than in sulfur vapor, whereas at 900~ the reduction of the corrosion rates in the mixed gas is of only 0.6-1.0 orders of magnitude. It is reasonable to expect that the corrosion becomes even more severe above 900~ because the sulfur partial pressure is higher and the diffusional processes both in the alloys and in the scales are faster, thus enhancing the reaction rate.

Corrosion of Fe-Nb Alloys

207

Fig. 14. BSE micrographs illustrating the position of Pt markers on Fe Nb alloys after reaction (a) Fe 10Nb at 980~ 1.7 hr, (b) Fe 20Nb at 700~ 8 hr, and Fe 300Nb at 800~ 7 hr.

208

Kai, Douglass, and Gesmundo

An outer layer of thick columnar-shape of iron sulfide is always present in the scales formed on the F e - N b alloys, which tends to spall for most specimens. The tendency of a metal to form a protective scale is indicated by the Pilling-Bedworth (P.B.) ratio, whose optimal value is between 1 and 2. j9 Since the P.B. ratio for F e S / F e is about 2.56, large compressive stresses are likely to exist in the sulfide layer, thus leading to extensive spalling. Ideally, the corrosion resistance of Fe Nb alloys in a mixed gas is expected to increase markedly if either a continuous layer of NbS2 or niobium oxides form. In fact, these layers would serve as a protective barrier to reduce the outward diffusion of iron or to block the inward diffusion of sulfur. However, a continuous layer was never observed, so that an outer layer of iron sulfide always formed, regardless of the alloy composition and temperature. Stability diagrams are useful to predict the nature of the phases to be formed in the mixed gas (Table I). The standard Gibbs free energies of formation of some corrosion products were calculated and are tabulated in Table IV, based on thermodynamic data from the J A N A F Tables. 2~ The stability diagrams of the F e - S - O and N b - S - O systems at different temperatures were calculated and are shown in Figs. 15 and 16, respectively, assuming unit activities for the condensed phases, where the gas composition is shown by point A. The phases in equilibrium with the mixed gas are FeS and Nb2Os, regardless of temperature. As noted above, short-term corrosion tests show that iron sulfide is always the first phase to form during the transient stage. This reaction removes sulfur from the gas and may produce a local increase of the oxygen activity at the metal/sulfide interface, thereby favoring the formation of niobium oxides, if the gas flow is too slow. In the case of the corrosion of pure Nb, one would expect to find a three-layered scale, containing NbO in contact with the metal followed by NbO2 and finally by Nb2Os, according to the N b - S - O stability diagram Table IV. The Standard Gibbs Free Energiesof Forma-

tion (AG}) of Corrosion Products at 600 and 800~ a Compound FeS FeO Fe304

FezO3 NbS2 NbO NbO2 Nb205

"AG}unit: Kcal/mole.

600~

800~

-24.76 -51.32 -198.47 - 141.76 -73.79 -81.20 -151.73 -362.12

-22.31 -48.36 -184.13 - 129.95 -66.18 -76.09 -143.49 -342.05

Corrosion of Fe--NbAlloys O-

209

600"~

S(~

F2.~_.~

;2(S04)~ A r

-w 13 - 1 0 -

13_ F'e(,)

Un 0 -20" -30

Fe20s

04

T

-20

-10

log Po, (atm.)

/

8000C F'eS

Aj

FeS04

/

J

""-' 13 - 1 0 -

V

or 12_

Fe(,)

FeO

Fe~04

cr~ 0 -20 -30

-20

log Po, (atm.)

- 10

Fig. 15. (a-b) Stability diagrams of Fe-S-O at different temperatures. (Fig. 16). However, according to the X R D results, only NbO2 was observed on pure N b at all temperatures. It is very likely that a layer of Nb205 existed at the scale surface, but it may have been too thin to be detected. In fact, the scale should be in a condition of quasi-equilibrium with the gas in view

210

Kai, Douglass, and Gesmundo

0

-

600o C

-

NbS2 .~..- 10-

/

E ..l..a

a

~Nb=0.1

Nb20~

U~--20- ~a~= 1 13_ E~ 0

NbO2 tlNb(

Nb(4

I I

-30 -50

-40

log Po,

-30

-20

--10

O-

NbS2 /---,-104..a

~=0.1 b

~Nb = 1

El

tNbOz

~-20-

13_

N b(.)

Nb20~

IINtO

C3n 0 -30-50

-4O

log Po,

-30

-20

- 10

Fig. 16. (a-b) Stability diagrams of ND-S-O at different temperatures. of the very low reaction rate. Also the inner layer of NbO may be either too thin for detection or even be absent for kinetics reasons. For the alloys, there was no evidence by both X R D and EPMA that either NbS2 or niobium oxides formed, even for alloys containing 40 wt.% Nb. On the contrary, only iron sulfides and double sulfides (F.exNbS2) were

Corrosion of Fe-Nb Alloys

211

detected at all temperatures. Obviously, the assumption of unit activity of niobium does not apply for the alloys. The effect of a reduction of the Nb activity on the stability of the Nb compounds was considered and is shown in Fig. 16. A reduction of the Nb activity to 0.1 enlarges the stability range of Nb(s~ and shrinks the stability range of NbS2 and NbO (dashed lines). The lines on the stability diagram would shift even more, if the activity of Nb was lowered further. It may be noted that the correct type of phase diagram to be used is that for the F e - N b - O - S system, which becomes threedimensional at constant temperature} I In any case, the thermodynamic data available are not sufficient for their calculation. Actually, the nature of the products formed during corrosion in a mixed gas does not depend only on the thermodynamic stabilities of the phases but also greatly on the kinetics of their nucleation and growth. Thus, the kinetics boundary in a M O-S stability diagram which divides the region of gas composition where oxides are kinetically stable from that where sulfides or mixtures of oxides and sulfides form may not be identical to the corresponding thermodynamic boundary. In particular, for pure metals such as chromium 22 or for alloys of the M - C r type, 21'23"24the kinetics boundary lies in the field of stability of chromium oxide. This is essentially the result of the larger rate of growth of the chromium sulfides with respect to chromia. The location of the kinetics boundary for the present F e - N b alloys concerns only the niobium compounds, because iron can form only sulfide for the thermodynamic reasons mentioned above. The location of the kinetics boundary for pure niobium is not yet known and could not be studied here because of the fixed initial composition of the gas mixture. However, the experimental results concerning its corrosion in the present ternary gas mixtures show that when the composition of the gas phase is in the oxide stability field, as it applies to the present gas mixture, only niobium oxides form. This result is reasonable because in the case of niobium the sulfide is the slowestgrowing phase, even though NbO2 also grows quite slowly. Thus, one may expect that the kinetics boundary for pure niobium may lie in the sulfide rather than in the oxide stability field but in any case it should be rather close to the thermodynamic equilibrium line. The situation for the F e - N b alloys could be similar, and the gas composition is probably located within the region where the niobium oxides are expected to form, i.e. to the right of the kinetics boundary. In spite of the previous predictions, the formation of niobium oxides has never been observed for these alloys, at variance with what has been found for similar alloys based on cobalt ~2 or nickel. 13 Thus, there must be special kinetics reasons. In fact, niobium does not dissolve significantly into the outer layer of iron sulfide, so that it remains out of direct contact with

212

Kai, Douglass, and Gesmundo

the gas phase, except possibly during the initial stages of the reaction. Thus, the possibility of formation of niobium oxides depends on the ability of the oxygen to penetrate in some way through the outer layer of pure iron sulfide and to reach, at the locations where niobium is present, partial pressures sufficiently high to make the niobium oxides more stable than the corresponding sulfides. This may occur either by dissolution of oxygen and solidstate diffusion inwards or by means of penetration of H20 in the gas phase through cracks or pores in the scale. These processes take place to a different extent in the case of the Co-Nb ~2and Ni-Nb j3 alloys but not for the present system. Thus, it is concluded that the outer FeS layer formed on the present alloys is so compact as to prevent the penetration of the gas phase and that oxygen is neither sufficiently soluble nor does it diffuse sufficiently rapidly through FeS to provide the oxygen activity which is thermodynamically required to form the niobium oxides within the inner layer of the scale. In other words, the partial pressures of oxygen and sulfur prevailing at the location within the scale where niobium is present correspond to the stability of the sulfide, even though the gas-phase composition is in the oxide-stability field. Because only sulfides formed on the scales of the alloys at all temperatures, neither the Fe S-O nor the Nb-S O stability diagram are really needed to depict the true reactions for Fe-Nb alloys. The appropriate phase diagram corresponds instead to the Fe-Nb-S ternary system. However, since the thermodynamic data regarding the intermetallic compounds and the double sulfides needed to calculate their dissociation partial pressures are not available, only approximate stability diagrams may be constructed. Based on the phase constitution of the inner layer as identified by XRD and EPMA, a hypothetical stability diagram was drawn as shown in Fig. 17. The double sulfides are noted by the formula FexNbS2, where x could be smaller than 1. However, only two double sulfides (FeNb2S4 and FeNb3S6) have been reported in JCPDS files and are shown in the diagram. Besides, from Table IV, one can expect that NbS2 is more stable than FeS because of its more negative value of the free energy of formation. This stability diagram shows that both the a-Fe rich solid solution and the intermetallic compounds will react with sulfur to form NbS2 at the partial pressure P1. Upon further reaction, the alloy composition will shift towards pure Fe and double sulfides will form at higher sulfur pressures. As described before, the transport of iron through these scales is very rapid, which causes FeS to form always in the outer scale during the initial corrosion stage. Further, the formation of FeS is continuously favored due to rapid iron diffusion through the nonprotective layer structure of double sulfides, as discussed by early investigators. 7 It is possible that the absence of NbS2 is due to either the

Corrosion of Fe-Nb Alloys

213

F e S + FeNb2S4

z

z

O~ + FeNb2S 4

0-,

(J. + FeNb3S6 C[.+ NbS 2

P~

E

Z

\

u

Fe2Nb

At.O/oNb

Nb

Fig. 17. Hypothetical stability diagram for Fe-Nb-S system.

presence of amounts below the x-ray detection limits ( < 5%) or to its consumption by the "multiphase effect" if its growth kinetics are slow enough. 25 The location of Pt markers (Fig. 13) agrees with previous observations during sulfidation and suggests that the outer layer formed by outward diffusion of iron cations, and the inner layer formed by means of a partial centribution from inward sulfur diffusion. Finally, it should be pointed out that the mixed gas also contained hydrogen. As seen in Table I, the hydrogen partial pressure is much higher than that of sulfur or oxygen. As reviewed by Mrowec and Janowski, 16 hydrogen may dissolve in various forms in sulfides and change the concentration of defects and affect the corrosion rate. This effect is probably negligible in the case of common-metal corrosion (such as pure Fe, Ni, and Co), because of the high concentration of defects in the metal sulfides. However, the effect of hydrogen ions on the sulfidation of a refractory metal such as Mo is significant because of the low defect concentration of MoSs.26 The situation of Nb is different because both the niobium sulfides, 2s- and 3sNb~ +xS2, have high concentration of defects in the cation sublattice. 27 Thus, the rate of sulfidation of pure Nb in H2/H2S mixtures is nearly the same as in sulfur vapor. 2s Therefore, the hydrogen effect on the corrosion kinetics of the Fe-Nb alloys is not expected to be really significant.

214

Kai, Douglass, and Gesmundo

CONCLUSIONS 1. T h e corrosion of F e N b alloys in H R / H 2 0 / H 2 S resulted only in sulfidation. The kinetics followed the p a r a b o l i c rate law. The addition of N b reduced the corrosion rate. 2. The scales formed on the F e - N b alloys were bilayered, consisting of a p o r o u s FeS outer-layer a n d a complex inner-layer of the double sulfides (FexNbS~), FeS, a n d unreacted FeRNb. 3. The rate-controlling step of the reaction for F e - N b alloys involves the o u t w a r d t r a n s p o r t o f iron through the outer layer of iron sulfide as well as t h r o u g h the i n n e r layer as well as the inward sulfur diffusion t h r o u g h the i n n e r mixed-sulfide layer.

ACKNOWLEDGMENTS This work was supported by a g r a n t from the Electric Power Research Institute, Palo Alto, California. The interest a n d s u p p o r t of Dr. John Stringer is gratefully appreciated. Travel funds for c o l l a b o r a t i o n of Profs. G e s m u n d o a n d Douglass were furnished by N A T O , G r a n t No. 850607.

REFERENCES

1. 2. 3. 4. 5.

F. Gesmundo and F. J. Viani, J. Electrochem. Soc. 128, 470 (198l). G. C. Wood and F. H. Stott, Mater. Sci. Technol. 3, 519 (1987). S. Mrowec and K. Przybylski, High Temp. Mater. Proc. 6(1) (1984). S. Mrowec and K. Przybylski, Oxid. Met. 23, 107 (1985). K. N. Stratford, The Degradation of Metallic Materials by Sulphidizing Gases, and the Potential o f the Refractory Metals for High-Temperature Service in Complex Sulphur-Bearing Atmospheres of Low Oxygen Potential (Institute of Metallurgists Conference on

6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

Environmental Degradation of High Temperature Materials, Isle of Man, U.K., 1980J. B. Gleeson, D. L. Douglass and F. Gesmundo, Oxid. Met. 31,209 (1989). G. Wang, R. Carter and D. L. Douglass, Oxid. Met. 32, 273 (1989). M. F. Chen and D. L. Douglass, Oxid. Met. 32, 185 (1989). B. Gleeson, D. L. Douglass and F. Gesmundo, Oxid. Met. 33, 425 (1989). R. V. Carter, D. L. Douglass and F. Gesmundo, Oxid. Met. 31, 341 (1989). M. F. Chert, D. L. Douglass and F. Gesmundo, Oxid. Met. 31,237 (1989). C. Shing and D. L. Douglass (to be published). Y. He and D. L. Douglass (to be published). T. B. Massalski, J. L. Murray, L. H. Benett, and H. Baker, Bina~3' Alloy Phase Diagrams, Vol 1, (ASM, 1986). R. C. Weast, Handbook of Clwmistl3' and Physics, 58th Ed. (CRC Press, 1977-78). S. Mrowec and J. Janowski, in Advances in Ceramics Nonstoichiometric Compounds, C. R. A. Catlow and M. C. Markrodt, eds., Am. Ceram. Soc. 23, 585 (1986). D. J. Young, Rev. High Temp. Mater. 14, 299 (1980). J. P. Orchard and D. J. Young, J. Electrochem. Soc. 136, 545 (1986). N. B. Pilling and R. Z. Bedworth, J. h~st. Met. 29, 529 (1923). JANAF TherhTochemical Tables, 3rd Ed. (American Chemical Society and American Institute of Physics for National Bureau of Standards, 1986). A. Rahmel, M. Schorr, A. Velasco-Tellez, and A. Pelton, Oxid. Met. 27, 199 (1987).

Corrosion of Fe-Nb Alloys

215

22. G. J. Yrek and M. H. LaBranche, in Proceedings of the Conference on Corrosion-ErosionWear of Materials in Emerging Fossil Energy System, ed. (A. V. Levy, NACE, Houston (1982), p. 993. 23. R. A. Perkins, High Temp. Corrosion, ed. (R. A. Rapp, NACE, Houston, 1983), p. 345. 24. K. Natesan, Corrosion 41,646 (1985). 25. G. Wang, B. Gleeson and D. L. Douglass, Oxid Met. 31,415 (1989). 26. B. S. Lee and R.A. Rapp, J. Eleetrochem. Soe. 131, 2998 (1984). 27. F. Kadijk and F. Jellinek, J. Less-Common Met. 19, 421 (1960). 28. F. Gesmundo (unpublished results).