C.H. Bartholomew and J.B.
Butt (Editors), Catualyst Deactiuation 1991 515
01991 Elsevier Science Publishers B.V., Amsterdam
DEACTIVATION OF IRON OXIDE CATALYSTS FOR THE SELECTIVE
OXIDATION OF HYDROGEN SULFIDE TO ELEMENTAL SULFUR.
P.J. van den Brink, A. Scholten, A.J. van Dillen, J.W. Geus,
Department of Inorganic C h e m i s v , University of Utrecht, Croesestraat 77A,
3522 AD Utrecht, The Netherlands.
E. Boellaard, A.M. van der Kraan.
Interfacultair Reactor Instituut, Deljt University of Technology, Mekelweg 15,
2629 JB Deljt, The Netherlands.
Summary
A new catalyst for the selective oxidation of hydrogen sulfide in Claus tail gas to elemental sulfur
has been developed. The catalyst consists of highly dispersed iron oxide supported on a silica carrier.
During operation the activity of this catalyst decreases due to transformation of iron(1n) oxide into a
less active component. X-ray diffraction, wet chemical qualitative analysis and Mossbauer
spectroscopy reveal the component comprises iron(I1) sulfate. Although the transformation of
iron(II1) oxide into irongo sulfate causes deactivation, the increase in selectivity results in high sulfur
yields (up to 94%).
Introduction
Oil refineries and natural gas plants often produce large amounts of hydrogen sulfide. The most
applied process for converting this hydrogen sulfide into elemental sulfur is the modified Claus
process. However, 3 to 5% of the H2S feed is not converted into sulfur and has to be treated
alternatively. At the University of Utrecht a new catalytic process has been developed to oxidize
hydrogen sulfide in Claus tail gas selectively to sulfur [1,2]:
In order to obtain high sulfur yields in an excess of oxygen, the following three reactions leading
to SO2 and, hence, affecting the selectivity adversely had to be minimized as much as possible:
- Oxidation of elemental sulfur to sulfur dioxide: '/n Sn + 0 2 + S@ (2)
- Direct oxidation of H2S to sulfur dioxide: H2S + 1l/2 0 2 + SO2 + H20 (3)
- The Claw reaction: 3/" Sn + 2 H2O 2 2 H2S + SO2 (4)
Because Claus tail gas contains large concentrations of water vapour (up to 30%), the equilibrium
results in appreciable concennations of H2S and SO2.
The previous developed catalyst, consisting of iron oxide on a-alumina of a low specific surface
area (6.5 m2/g) and wide pores, exhibited a high selectivity. As reported by Berben et al. [l],
chromium oxide was added to the iron oxide in order to decrease the rate of deactivation of the
catalyst. This catalyst is now used commercially [3].The addition of chromium oxide has, however,
some disadvantages. Firstly, the stability of the catalyst strongly depends on the distribution of the
chromium oxide through the iron oxide, which makes the production of the catalyst difficult.
Secondly, the toxic nature of chromium causes problems with respect to the handling of the spent
catalyst.
516
Without chromia the catalyst deactivates to a low activity, while increasing of the temperature only
results in a gas phase oxidation of sulfur to sulfur dioxide. A higher activity can also be attained by
application of an a-alumina carrier of a higher specific surface area. However, this type of carrier
unavoidably contains some y-alumina or 8-alumina strongly catalyzing the establishment of the Claus
equilibrium (4) [2]. Moreover, under reaction conditions these reactive aluminas react to aluminium
sulfate, causing a loss in mechanical strength. Therefore a carrier had to be found of a high specific
surface area without activity for the Claus reaction and not exhibiting sulfate formation.
A support that meets with these requirements is silica. The use of silica allows one to prepare a
catalyst of a higher specific surface area without the drawbacks of the alumina. Therefore a catalyst
was prepared consisting of highly dispersed iron oxide on silica, This catalyst (catalyst A) was
prepared by incipient wetness impregnation of preshaped silica extrudates (Degussa OX 50,50 m2/g)
with a solution containing ammonium iron citrate as a precursor. A detailed description of the
preparation method has already been reported [4]. Using this method a catalyst containing small
supported iron oxide particles (particle size 2-6 nm) is made. The iron oxide particles cover about
40% of the surface of the silica, and about 80% of the iron oxide is interacting with the carrier. As a
reference a catalyst (catalyst B) with poorly dispersed iron oxide (particle size 5-20 nm) having a low
interaction with the support was investigated. This catalyst was prepared by impregnation with a
solution containing iron nitrate [4].
Experimental
Catalvtic performance, The conversion and selectivity were measured in a continuous integral
microflow apparatus at atmospherical pressure. The definitions used are:
H2S converted
Conversion = H2S supplied
Sn produced - H2S converted - SO2 produced
= H2S converted - HzS converted (2)
Yield = Conversion * Selectivity (3)
400 mg (- 1 ml) catalyst with a sieve fraction between 0.40 and 0.63 mm was placed into a quartz
reactor (010 mm). Special care was taken to ensure plug flow conditions in the catalyst bed. The
feed consisted of 1.0 vol% HzS, 5.0 vol% 0 2 , and 30 vol% H20 in He. The total flow rate was
200 "'I(sK')/min. The H2S and SOzcontent of the effluent was analysed with either a gas
chromatograph (Car10 Erba 6000) or a UV-spectrophotometer (Perkin Elmer Lambda 2). Two
different experimental procedures were applied with the fresh catalysts, viz., an isothermal, and a
temperature-programmed procedure. With the isothermal procedure the temperature was kept constant
to determine the conversion, selectivity, and yield as a function of time. With the temperature-
programmed procedure the temperature of the reactor was varied stepwise (10 K) from 453 K
(180OC) to 593 K (32OOC) and back down to 453 K. This cycle was repeated three times. After each
temperature step the effluent gas was analyzed twice at an interval of 9.0 minutes causing a total cycle
time of 9 hours. Subsequently the catalysts were either cooled down i n helium and removed from the
reactor for characterization or kept at 573 K (300OC) for 5 hours in a flow of 1.0 vol % S02, 5.0
vol% @, and 30 vol% H20 to mimic typical industrialconditions that can occur during operation and
can cause deep sulfation of the active material. The effect of the treatment was checked in two
additional subsequent temperature-programmed cycles.
517
The catalysts taken out after three cycles were characterized with:
rav Diffraction: Catalysts A and B, before and after reaction conditions, were ground and pressed
.o a sample holder and examined using a Philips diffractometer (FeKa, h=1.93735 A).
M" . The spectra were recorded with a constant acceleration spectrometer
ing a 57Co in Rh source (50 mCi). Isomer shifts are reported with respect to the NBS standard
iium nitropxusside. Magnetic hyperfine fields were calibrated with the 515 kOe field of a-Fe2O3 at
3m temperature. The spectra were not corrected for the varying distance between the source and the
tector and, hence, the curved background in the measured spectra is of instrumental origin. The
ossbauer parameters were determined by fitting the collected spectra with reference subspectra
misting of Lorentzian-shapedlines using a non-linear iterative minimisation routine. In the case of
adrupole doublets the line width as well as the absorption areas were constrained to be equal. All
ectra were ex-situ measured in a vacuum of 2x10-5 torr at 77 K. Before the measurements the
nples had been stored for about three months in capped bottles.
et chemical analvsis; In order to obtain additional information about the valence state of the iron in
used catalyst a sample of 50 mg was ground and dispersed into a small amount (ca. 1 ml) of
idified (HCl, pH 3) water. The thus obtained solutions were analysed with solutions consisting of
W N S , K4[Fe(CN)6I7and K3[Fe(CNkl PI.
Results
Italytic perfmancg First the catalytic behavior of catalyst A was measured as a function of time at
3 K (200°C). The results measured with the UV spectrophotometer are shown i n figure 1. As
Jnd with the alumina-supported catalyst [l] the initial conversion decreases during the first hours of
eration. Steady state conversion was reached after 50 hr of operation. The selectivity to sulfur,
wever, increases to a high level of 96%.
0 86
084
01 0 82
0.0 08
10 20 30 40 50
Time (hours)
figure 1 Isothermal experiment at 473 K (200°C)with catalyst A.
518
During the deactivation the color of the catalyst changes from pale rearown to white indicating the
formation of a sulfate. The sulfur yield as calculated from this experiment does not represent the
maximum obtainable yield. The optimal yield has been determined by the temperature programmed
experiments.
- a
01
0
180
, ,
200
, ,
220
, ,
240
, ,
260
, ,
280
, ,J44
300 320
153 473 693 513 533 553 573 5s3 633 453 473 493 51, 533 553 573 5Sl 615
Temperalure (“C, K) Temperature (“C. K)
W 473 d’J9 513 533 553 573 593 613 153 d7P 4993 513 533 553 573 133 613
Temperature (“C. K) Temperature (“C. K)
1
160
,
zw
173
,
220
$93
, ,
240
573
, ,
260
-3
,
280
153
, ,
300
573
,wo
320
593
40
613
,80
453
zoo
413
220
493
240
513
260
533
280
553
300
173
320
593
340
6(3
Temperature (‘C, K) Temperature ( C,K)
figure 2 a-f. Catalytic performance versus temperature of catalyst A.
a) fresh catalyst, fust cycle: b) second cycle; c) third cycle; d) after sulfation, first cycle; e) after sulfation, second cycle;
f ) third cycle, (see 2 c); -0- conversion at rising T; 4- conversion at decreasing T;
-EL selectivity at rising T; --C selectivity at decreasing T;
-A- yield at decreasing T
519
Figure 2 shows the performance of the highly dispersed catalyst A as a function of the temperature
and the number of applied temperature cycles. Analogously to the isothermal experiment, the
conversion of the fresh sample (2a) below 493 K (220°C) decreases, whereas the selectivity increases
during the interval of 9 minutes at which the activity was measured at constant temperatures. The
most considerable loss of activity is exhibited after operation at high temperatures. The rate of the
formation of sulfate as indicated by the color change is also higher at high temperatures. The
formation of sulfate is also apparent from a comparison of the selectivity at rising and at decreasing
temperatures. Due to the consumption of sulfur dioxide caused by the formation of sulfate, the
apparent selectivity (2) temporary increases. In the second cycle deactivation after operation at high
temperatures is also evident. Because the color of the catalyst is already white and no uptake of sulfur
dioxide is seen, the deactivation during the second cycle must be attributed to any structural change,
such as sintering or removal of defects. After three cycles the performance at rising and decreasing
temperatures is about the same (figure 2c), showing a behavior typical of a catalyst that controls the
extent to which the different reactions (1-4) proceed [6]. At low temperatures the low activity of the
catalyst limits the sulfur yield. At high temperatures and high conversions (>95%) the yield falls due
to the sequential oxidation (2) of sulfur to sulfur dioxide. At low conversions the selectivity
asymptotically approaches a level of about 96%. The parallel oxidation of H2S (3) now brings about
formation of small quantities of S 0 2 .
The maximum obtainable sulfur yield is high (94%) in spite of the high water content of the feed
and the high oxygen to hydrogen sulfide ratio. Because the reaction is kinetically controlled, the
sulfur yield can be adjusted by simply controlling the temperature of the reactor.
After the suIfation procedure with S@ and 0 2 , the activity has strongly decreased (figure 2d). The
selectivity however still remains high. At rising temperatures the activity strongly increases. This
increase is not only caused by the normal relation between temperature and activity, but also the result
of some other additional effect is seen. This is clearly demonstrated by the higher conversion at
decreasing temperatures. In the second cycle after the sulfation procedure (figure 2e) the same
catalytic performance is displayed as in the third cycle before sulfation (figure 2c). Apparently
sulfation by SO2 and 0 2 causes formation of some phase of a lower activity, which reacts again into
the active phase present before sulfation under reaction conditions.
m:In figure 3 the XRD patterns of catalyst A and B before and after use are shown. It is clear that
the high dispersion of the active component of catalyst A impedes identification by XRD. However
the low dispersed catalyst B shows a more distinct XRD pattern. It can be seen that during operation
the hematite (a-Fe203) phase is transformed into orthorhombic iron(I1)sulfate (FeS04) [ 7 ] . As
indicated in the figure the large hematite particles of catalyst B do not completely react to FeS04.
Wet chemical Qualitativeanalvsis: Catalyst samples analyzed directly after the reactor experiment only
showed the presence of ferrous ions. The catalysts samples after exposure for a week to air revealed
also the presence of ferric ions. This indicates that the iron sulfate formed in the reactor is indeed an
iron(II) sulfate, which can easily be converted into basic iron(II1) sulfate salts in wet air.
Mossbauer sDectroscoy: In order to obtain more information about the change in chemical
composition of the catalyst during operation, Mossbauer spectroscopy was performed on catalyst A
and B, before and after use (figure 4). The spectra of the fresh catalysts clearly illustrate the
difference in panicle size of the catalysts. At 77 K catalyst B gives rise to a number of superimposed
520
sextuplets representing a broad size distribution of rather large hematite particles. The spectrum of the
fresh catalyst A at 77 K is dominated by a doublet due to superparamagnetichematite indicating a
particle size of about 4 nm. This was also confirmed by other techniques [4].
-10 -8 -6 -4 -2 0
Catalyst A before use
t I , , , , ,
-
, , / . , . , , , ,
90" 60" 30" 15"
I I I
-10 -8 -6 -4 -2 0 2
I I 2
75" 45" 4 6 8 10
Velocity (-1s)
20
figure 3 XRD patterns of tigure 4 MCIssbauer spectra of
catalyst A and B, before and after use catalyst A and B, before and after use
After operation the iron oxide phase has transformed into other phases. Both catalysts show about
the same spectrum, containing at least two doublets, viz., one due to a divalent iron phase as
indicated by the large quadrupole splitting, and one of a trivalent iron phase as indicated by the small
quadrupole splitting. The central part of the spectra is fitted with sets of two doublets, one
representing the bulk, and the other representing the surface. This detailed analysis is based on high-
resolution spectra measured in the Doppler-velocity range of interest, which are not shown.
The results of the fitting procedure (table 1) reveal that the divalent iron phase is Szomolnokite
([Link]). The mvalent iron phase is probably a hydroxylated basic iron sulfate like Butlerite
(FeS04OH.2H20) [8].The data also reveal that the catalysts after operation contain well crystallized
Szomolnokite, as follows from the observed small line width of its spectral contribution. The
presence of the sextuplet in the spectrum of the used catalyst B shows that still some iron(III) oxide is
present as also found by XRD.
At first sight the presence of trivalent iron species in both catalyst is in contrast with the data found
with XRD and wet chemical analysis. However, well dispersed iron(I1) sulfate can react with water
521
vapour and oxygen to basic iron sulfates, like Butlerite even at room temperature. The time between
the catalytic measurements and the Mossbauer experiments was significantly longer than the time
before the XRD experiment was performed. The transformation was also indicated by the observed
change in color of catalyst A, viz., from white to palish brown.
TABLE 1 Mossbauer parameters of the catalysts obtained by fitting with the experimental data
'ample Hyperhe spectral Compound
7
contribution (%)
* 0.03 +5
htaMA
kesh catalyst 0.83 44 Hematite
I .31 43
0.00 494 13
Wter reactor exp. 0.51 25 Butlerite
0.93 42
2.66 6 Szomolnokite
hlvst B
+esh catalyst Hematite
ifter reactor exp. 0.81 0.63 20 Butlerite
0.81 1.06 18
1.66 2.71 15 Szomolnokite
1.65 3.09 28
0.76 0.00 534 12 Hematite
0.75 -0.16 543 7
Resuming it may be concluded that after the catalytic oxidation of hydrogen sulfide the catalyst is
containing iron(I1) sulfate. However, Mossbauer spectra also indicate the presence of basic iron(II1)
sulfate Butlerite, which contribution could not be confirmed by XRD and wet chemical analysis.
Reaction of Szomolnokite to Fe(II1) compounds during the storage period before the Mossbauer
experiments were performed, cannot be excluded. Since the surface-to-volume ratio of the iron
containing particles in catalyst B is much smaller than that of the highly dispersed iron containing
particles in catalyst A, the extent of reaction to Fe(III) can be expected to be much lower with catalyst
B. As can be seen in table 1, the fraction of Fe(II1) within catalyst A is indeed much smaller than
within catalyst €3 containing much larger particles.
The formation of an iron(II) phase in an excess of oxygen is quite remarkable. Thermodynamic
calculations indicate that under reaction conditions anhydrous iron(II1)sulfate Fez(SO& is the only
stable compound. Kinetic experiments reveal that in the catalytic redox cycle the reoxidation of the
active sites is the rate-determining step: The order with respect to oxygen is almost unity, whereas the
order with respect to H2S is close to zero [9]. Under steady state conditions therefore the
concentration of the iron(III) species and, hence, the concentration of reactive oxygen species will be
low. The high steady state surface concentration of Fe(I1) species brings about the formation ofthe
thermodynamically less stable compound, viz., Fe(II) sulfate. The low surface concentration of
522
reactive oxygen explains on one hand the relatively low activity, and on the other hand the high
selectivity to elemental sulfur.
When sulfation is carried out in a mixture of SO2 and 0 2 at higher temperatures, most probably the
thermodynamically predicted iron(1II) sulfate phase is formed. This phase is less active, because no
Fe2+/Fe3+ redox cycle is possible, unless decomposition of the sulfate takes place. However, when
subsequently the H2S / mixture is passed through the catalyst bed, the iron(III) sulfate reacts again
completely or partially to the more active iron(1I) sulfate. Consequently, the activity and selectivity
return to their original level. To verify this reaction mechanism in-situ Mossbauer measurements are
in progress [9]. These experiments will also elucidate the observed discrepancy between the ex-situ
Mossbauer measurements and XRD and wet chemical analysis.
There are indications from the relative small h e width in both XRD patterns and Mossbauer
spectra that the iron(I1)sulfate easily sinters, even at the relative low temperatures of the catalytic
reaction. This is also indicated by the catalytic performance test of the catalyst B. Catalyst B
containing iron oxide not interacting with the support shows a large deactivation. The time required
for stabilization is much longer, and the remaining activity and selectivity is low. The effect of the
dispersion, and the interaction with the carrier stresses the great importance of the preparation
procedure [4].
Conclusions
Well dispersed iron oxide on silica of a specific surface area of 50 m2/g has proven to be a very
suitable catalyst for the selective oxidation of hydrogen sulfide to elemental sulfur. Under reaction
conditions the iron oxide is transformed into iron(I1) sulfate as revealed by X-ray diffraction,
Mossbauer spectroscopy, and wet chemical analysis. The presence of an iron(II1) component as
observed by ex situ Mossbauer spectroscopy can not be excluded. Although the transformation of
iron(1II) oxide into iron(II) sulfate causes initial deactivation, the increase in selectivity (96%) results
in high sulfur yields (up to 94%) .
REFERENCES
1. [Link], AScholten, [Link], [Link], W.J.J. van der Wal and [Link], in
Stud. Surf. Sci. Catal. (Catalyst Deactivation), 34,303-316, (1987).
2 [Link] and [Link], in Proceedings of the 9th international congress on catalysis,
[Link] and [Link] (Us.), the chemical intitute of Canada, Ottawa, 284-291, (1988).
3 [Link], [Link], [Link], "SUPERCLAUS - The answer to Claus plant
limitations", 38th Canadian chemical engineering conference, Edmonton, Canada, (1988).
4 P.J. van den Brink, [Link], [Link] Wageningen, [Link], A.J. van Dillen and
[Link] in Preparation of Catalysts V, B. Delmon, P. Grange, P.A. Jacobs, and G. Poncelet
(Eds.),Elsevier, Amsterdam, preprints 533-542, (1990).
5 A.I.Voge1, A textbook of qualitative chemical analysis. Longmans,Green and co, London.
6 C.N. Satterfield, Mass transfer in heterogeneous catalysis, M.1.T Press, Cambridge, (1970).
7 Powder Diffraction File Search Manual, Alphabetical Listing and Search Section of Frequently
Encountered Phases, Inorganic Compounds. [Link] [Link].(Eds.),Swarthmore,USA, (1976)
8 Mossbauerhandbook, Mineral, [Link] [Link].(Eds.), Mossbauer Effect Data Centre, North
Carolina, (1983).
9 To be. published by the same authors
