Oct 15, 1985 -
NOTICE
THIS DOCUMENT HAS BEEN REPRODUCED FROM MICROFICHE. ALTHOUGH IT IS RECOGNIZED THAT CERTAIN PORTIONS ARE ILLEGIBLE, IT IS BEING RELEASED IN THE INTEREST OF MAKING AVAILABLE AS MUCH INFORMATION AS POSSIBLE
JPL Publication 85 . 80
DOE/MC/20417-1898 Distribution Category UC-90c
0
JAN 188 6 FECEIVED Aft InFAMY
Novel Sorbents for HighTemperature Regenerative H2S Removal
no DST.
Final Report Maria Flytzani-Stephanopoulos George R. Gavalas Satish S. Tamhankar Pramod K. Sharma
(NASA-CB-176449) NOVEL SGR6EATS fOL HIGH TEMPERATURE HEGIVEBATIVE H25 6EMOVAL titiai 8eport (Jet Profulsion Lab.) 110 p HL A06/MP A01 CSCL 07D
N86-1`360 i
U11CIds G3/25
r
05161 s
r,
' +-
October 1985 Prepared for U.S. Department of Energy Morgantown Energy Technology Center Through an Agreement with
National Aeronautics and Space Administration b^
Jet Propulsion Laboratory California Institute of Technology Pasadena. California
t
it
TECHNICAL REPORT STANDARD TITLE PAGE 1. Report No.
85-80
Government Accession W.
__72.
3. Recipient's Catalog No. 5. Report Date
4. Title and Subtitle
October 1985 6. Performing Organization Code
Novel Sorbents for High Temperature Regenerative H 2 S Removal
4
7. Autha(s) M. Stephanopoulos, G. Gavalas, S. Tamhankar,
B. Performing Organization Report No.
9. Performing Organization Name and Address P. Sharma
10. Work Unit No.
JET PROPULSION LABORATORY California Institute of Technology
11. Contract or Grant No. NAS7 -918 13. Type of Report and Period Covered
4800 Oak Grove Drive Pasadena, California 91109
JPL Publication
12. Sponsoring Agency Name and Address NATIONAL AERONAUTICS AND SPACE ADMINISTRATION Washington, D.C. 20546
14. Sponsoring Agency Code
15. Supplementary Notes Sponsored by the U.S. Department of Energy through Interagency Agreement DE-AI21- 83MC20417 with NASA; also identified as DOE /MS/20417-1898 (RTOP or Customer Code 779-00-00). lb. Abstract The overall objective of this program was to develop improved regenerable sorbents for the high-temperature desulfurization of coal-derived gas streams. The aim is to reduce the H2S level, preferably in a single-stage, from 0.51.0 mol% down to 1 ppm. The specific objectives of this project were to synthesize, characterize and test two classes of sorbents. One consists of eutectic mixtures of metal oxides (e.g., ZnO-V 2 05, CUM004 -Mo03) in the form of a melt coating the pore surface of a high surface area solid support. The second class of sorbents includes unsupported mixed oxides, in the form of solid solutions or solid compounds, e.g., Zn-Fe X Oy, Cu-Fe x-Oy, prepared as highly dispersed microcrystalline solids. The principal characteristics of both classes of sorbents are rapid kinetics of absorption, high absorption capacity and good regenerability. The rapid absorption kinetics are realized by eliminating or minimizing the resistance associated with solid state diffusion. With these properties, the sorbents would be suitable for coal gasification/ molten carbonate fuel cell (MCFC) applications, but could also be applied to combined cycle power generation and the purification of synthesis gas.
17. Key Words (Selected by Author(s))
18. Distribution Statement
Chemistry Methods and Equipment Environment Unclassified-Unlimited 19.
Security Classif. (of this report) Unclassified
20. Security Clasif. (of this page) Unclassified
21. No. of Pages
22. Price
109 J ►l 01e4 A 9154
.
Eli
• i^ I`
JPL Publication 85.80
DOE/MC/20417-1898 Distribution Category UC-90c
f 4
tf '
Al
Novel Sorbents for HighTemperature Regenerative H2S Removal Final Report Maria Flytzani-Stephanopoulos George R. Gavalas Satish S. Tamhankar Pramod K. Sharma
October 1985 Prepared for
U.S. Department of Energy Morgantown Energy Technology Center Through an Agreement with National Aeronautics and Space Administration by
Jet Propulsion Laboratory California Institute of Technology Pasadena, California
a
M
}
:l 46
Prepared by the Jet Propulsion Laboratory, California Institute of Technology, for the U.S. Department of Energy through an agreement with the National Aeronautics and Space Administration.
This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
0
ABSTRACT
The overall objective of this program was to develop improved regenerable sorbents for the high-temperature desulfurization of coal-derived gas streams. The aim is to reduce the H 2 S level, preferably in a single-stage, from 0.51.0 mol% down to 1 ppm. The specific objectives of this project were to synthesize, characterize and test two classes of sorbents. One consists of eutectic mixtures of metal oxides (e.g., ZnO-V 2 05 , CuMo04-MO03) in the form
of a melt coating the pore surface of a high surface area solid support. The second class of sorbents incl-; ,j , unsupported mixed oxides, in the form of solid solutions or solid ,' ,pounds, e.g., Zn-Fe x -0 , Cu-Fe.-O, prepared as highly dispersed microcrystalline solids. The p A ncipal characteristics of both classes of sorbents are rapid kinetics of absorption, high absorption capacity and good regenerability. The rapid absorption kinetics are realized by eliminating or minimizing the resistance associated with solid state diffusion. With these properties, the sorbents would be suitable for coal gasification/molten carbonate fuel cell (MCFC) applications, but could also be applied to combined cycle power generation and the purification of synthesis gas.
iii
-^ Q
F F t
'6
ACKNOWLEDGMENT
The contribution of Mr. M. Bagajewicz, a graduate student of Chemical Engineering at the California Institute of Technology, is gratefully acknowledged. We also wish to thank Ms. L. Lowry and Mr. R. Ruiz of the Jet Propulsion Laboratory (JPL) for carrying out the X-ray diffraction and Scanning Electron Microscopy analyses, respectively; and Mr. G. Smith, also of JPL, for his technical assistance. Finally, we acknowledge Dr. G. Krishnan of SRI International for the mercury porosimetry work carried out in his laboratory. The senior investigator, Maria Flytzani-Stephanopoulos, is now located at the Massachusetts Institute of Technology. Dr. George R. Gavalas and Satish S. Tamhankar are both with the California Institute of Technology. Pramod K. Sharma is located at JPL. This work was conducted at the Jet Propulsion Laboratory through NASA/JPL Contract NAS7-918 Task Order RE-152, Amendment 396, and was sponsored by the U.S. Department of Energy, Morgantown Energy Technology Center, Morgantown, West Virginia, under Interagency Agreement No. DE-Al21-83MC20417 (S.j. Bossart, Technical Project Officer).
iv
SUMMARY This report describes work performed by the Jet Propulsion Laboratory (JPL) of the California Institute of Technology during Phase I (October 1983 through July 1984) and Phase II (August 1984 through October 15, 1985) of the project entitled "Novel Sorbents for High Temperature Regenerative H 2 S Removal." This work was carried out for the U.S. Department of Energy, Morgantown Energy Technology Center, under Interagency Agreement No. DE-AI21-83MC20417. The overall objective of this program was to develop improved regenerable sorbents for the high-temperature desulfurization of coal-derived gas streams. The aim is to reduce the H 2 S level, preferably in a single-stage, from 0.51.0 mol% down to 1 ppm. The specific objectives of this project were to synthesize, characterize and test two classes of sorbents. One consists of eutectic mixtures of metal oxides (e.g., ZnO-V 2 0 5 , CuMo0 4 -MO03) in the form of a melt coating the pore surface of a high surface area solid support. The second class of sorbents includes unsupported mixed oxides, in the form of solid solutions or solid compounds, e.g., Zn-Fe x -O Cu-Fex-O•. prepared as highly dispersed microcrystalline solids. The p A ncipal characteristics of both classes of sorbents are rapid kinetics of absorption, high absorption capacity and good regenerability. The rapid absorption kinetics are realized by eliminating or minimizing the resistance associated with solid state diffusion. With these properties, the sorbents would be suitable for coal gasification/molten carbonate fuel cell (MCFC) applications, but could also be applied to combined cycle power generation and the purification of synthesis gas. Work in Phase I was divided into two major tasks. The first task was to synthesize and test sorbents consisting of supported melts of binary oxides, particularly ZnO-V 2 0 5 , on high surface area supports. The second task involved the synthesis, characterization and testing of unsupported mixed oxides such as ZnO-Fe 2 O 3 , and CuO-Fe2O3 in the form of porous solid compounds or solid solutions of relatively high surface area. Tests were carried out in a quartz microreactor with a packed bed of sorbents (-20 +40 mesh particles). The reactor was operated isothermally at 538-700°C, and at a pressure slightly above atmospheric. Phase II of the program included detailed parametric studies of sorbent sulfidation/regeneration with selected sorbent systems. In this phase of the work the most efficient materials from both classes of sorbents were selected and characterized. Extended testing of these materials in consecutive silfidation/regeneration cycles was performed to examine their long-term efficiency and regenerability. Class A Sorbentss: Supported Molten Mixed Oxides. The sorbent systems Zn-V-O an u- o- supporte on porous alumina were examined in this work. These sorbents form complex eutectic melts coating the pores of the support duriny sulfidation. Experiments were performed at 650-700°C and 538-650°C with Zn-V-O and Cu-Mo-O, respectively. Gas mixtures with molar composition H 2 (15-20%), H 2 S (0.2-1.0%), H 2 O (0-27%), CO (0-10%), CO2 (0-12%), balance N 2 were used in sulfidation. Regeneration tests were carried out with air-nitrogen or air-steam-nitrogen mixtures. Boih types of sorbents showed very high H2S removal efficiency (to 75%), and stable over several cycles. Copper aluminate gave similar results. Copper in the +2 or +1 oxidation state was evidenty stabilized in the alumina matrix in tests with copper aluminate. In sulfidation at 650°C, both CF and CA-sorbents gave pre-breakthrough H,,)S levels corresponding to metallic copper sulfidation. At this temperature, rE!duction to metallic Cu is very fast and complete. A stable and high sorbent conversion of ^-80% was obtained even at 650°C. With CA-sorbents, but not with CF, structural changes that occurred at 650°C were reversed when the temperature was lowered to 538°C. In view of the enhanced stabilization of copper oxide in CA-sorbents, a mixed oxide material Cu-Fe-A1-0 (CFA), with a molar composition of 2CuO:Fe203: Al 2 0 was prepared in porous bulk form as a potential sorbent for higher (>600°C) temperature H 2 S removal. Under similar conditions, this sorbent was superior to either copper ferrite or copper aluminate in regard to pre-breakthrough H 2 S levels, which were lower than for the equilibrium of the sulfidation reaction of metallic copper at all temperatures, 538-650°C. Hence, sorbent CFA further stabilized copper oxide towards undesired reduction to metallic copper.
vii
Additional testing of this CFA sorbent was conducted at 830°C with a reactant mixture containing 30 mol% H 2 , 17 mol% H 2 O, 1 mol% H 2 S, balance N2, simulating the Texaco gasifier-exit gas. Breakthrough of H 2 S took place at complete (100%) sorbent conversion, while the pre-breakthrough H 2 S level was 270 ppmv. The performance was stable over three cycles. This or similar type sorbents, therefore, are very promising with respect to hot gas cleanup for both MCFC and gas turbines applications. A second batch of CFA sorbent was prepared under lightly different conditions in porous form with high surface area (26 m /g) and pore volume ( — 1 cm3 /g). X-ray diffraction (XRD) analysis of this sorbent identified the crystalline phases CuFe 2 04 , CuO and FeAl 2 0 4 ; other compounds, including Al 2 03 were in an amorphous (to XRD) pnase. This sorbent was tested in the microreactor in a series of fourteen cycles of sulfidation/regeneration at 65U°C, and showed remarkahly high H 2 S removal efficiency. Thus, the H2S level was reduced from 10,000 ppmv in the feed gas to 0-15 ppmv at 0.50-0.60 sorbent conversion, while at breakthrough, which took place at about 0.80-0.85 sorbent conversion, the exit H 2 S level was still less than — 40 ppmv. The sulfided CFA sorbent (after cycle 4) was analyzed by XRD. The crystalline phases identified were CuS, Fel_ X S and CuFe22S These results are very imaortant, indicative of stabilization of Cu + in the sorbent matr'x thermodynamic Equilibria (over metallic corper and associated better sulfidation). Different conditions used in regeneration did not affect the performance of CFA sorbent. All regenerations of the sulfided CFA were performed at 650°C. In dry regeneration with air-N 2 (30-70, molar) the sulfur product's consisted of --- 99 M31% SO 2 and —1 mol% elemental sulfur. The effect of steam on the regeneration off-gas composition was studied in regeneration with (a) 30 mol% steam - 30 mol% air - 40 mol% N 2 , (b) 50 mol% steam - 50 mol% air, and (c) 70 mol% steam - 30 mol% air. In cases (a) and (h) the elemental sulfur produced was low, amounting to 1-2 mol% of the total sulfur products (S0 2 , H2 S, and S 2 ). In case (c), a higher amount of elemental sulfur was produced, equal to 10 mol% of the total sulfur products. Other parametric sulfidation studies with the CFA sorbent included the effects of changing (a) the H 2 /H 2 S ratio in the feed gas, b) space velocity, and (c) the fuel gas composition to that of a Lu 'gi -gas simul ant containing 17-18 mol% H ? , 12-13 mol% CO, 10-11 mol% CO 2 , 24-25 mol% H 2 O, 0.5-1.0 mol% H 2 S, balance N 2 . The removal efficiency and breakthrough conversion of the sorbent were not affected by changes (b) or (c). However, some effect was observed in case (a) above, i.e., with higher H 2 /H 2 S ratios in the feed gas, the pre-breakthrough H 2 S levels were higher, apparently because of more extensive reduction of the sorbent under these conditions. The sulfided CFA after cycle 14 was examined by SEM/EDS. The analyses were conducted on particles (as received) as well as their cross-sections (after grinding and polishing). Uniform distribution of Fe and Cu was found on the particle surface, the Fe intensity being stronger than that of Cu. At the same time, very V'Ltle S or Al could be found on the surface. On the particle cross-section, a uniform distribution of S, Fe, Cu, and Al was observed. These tests indicate that a substantial part of sulfur in the sulfided sorbent is retained by the aluminum compounds beneath an iron-rich surface. Further analysis of the sulfided CFA sorbent by XRD supported the SEM/EDS findings. The
viii
4
i s crystalline phases CuFe204, CuAl2p 4 , some a-A1 0 3 , Cu, and CugFegS1 were identified by Xk9. Moreover, there was evidence o an amorphous phase containing most of the s-ilfided compounds). These analyses have provided a better understanding of the sulfidation characteristics of the very promising CFA sorbent, and have shown the almost complete pore accessibility of this material. A new type of sorbent was considered and developed in this project. This involves mired Cu-Mo-A1-0 materials (CM-sorbents) which were prepared in highly porous form. The evolution of bulk CM-sorbents followed that of aluminasupported Cu-Mo-O sorbents mentioned above. All CM-sorbents studied exhibited very high H 2 S removal efficiency and breakthrough sorbent conversion. Based on visual observations of the sulfided sorbent and XRD analyses, complex sulfide-containing eutectics formed during sulfidation appear to be responsible for they high H 2 S capacity of CM-sorbents. The effect of composition (Cu/Mo ratio) on the stability and performance of the molybdenum/CM-sorbents was studied in detail. Lower molybdenum contents enhanced both the stability and the overall sorbent conversion. The most promising results were obtained with sorbent CM-6 containing a molar ratio of CuO:Mo03:Al203 = 1:1/3:1. This sorbent was prepared with high surface area of 32.3 m 2 /g, and consisted of crystalline CuO, CuM004 and an amorphous phase containing all the aluminum compounds. In sulfidation at 650°C with a feed gas containing 15 mol% H 2 , 0.5 mol% H 2 S, 25 mol% H 2 O, balance N2, the breakthrough conversion of this sorbent was higher than 0.90 (based on Cu2S) with pre-breakthrough H 2 S levels lower than 10 ppmv (up to 0.70 sorbent conversion). At 538°C, the exit H 2 S level was reproducibly , l
N
N
1
1^
1
I
=
1
LO ^
N
1
1 1
1 1
1 1
1
I
1
1 1
M
.+
O
OD
Obi
1c•1
Q^i
t
A
O
N
1 1
^ ^
1
N
1 1
1 1
000 OOO
ilf0 1-0 -40
00 tli0 -4 0
0 n 0
0 Op 0
d l
o o
cv)
O 1 1
1^
1 1
1
C^ N
(30
kn
N S
v r►
00
0000
00
N r W •: E
OO
OOOM
0
C, C) 0
d
io O 0
z
M (n
e
x
O
V
d ill N .r
H
i0 1 1
1
1
0
d
I
M N ip
l l
a+
MI
O
d
E 0 u
rW cc
yl ry.+ L NWv
1
C
O 1
O
N
N
in O iO N NM
0 0M 00 M d /, dn f- .r NM
O LO d d
ill .-t
NN
0 0 ilt O N O Md
NO
COOCiO
MO
NN
OD CD
N00
N
v
Q
O• 1 1 / •.•
Li'I
C
m
I
ill O
N
ill
C-4
.+ to ip ... .r r+ C%j •'+ r-tN
d Oi
N
i
O I
N
Z
C9 1L
0 mco ^
O
.+
IC
H
S
I.M
pl
O
N
t
IN
dK
d
N
H
d p d 0 1 I^
i0
ill O to O i0 0 .^ N
ili O ill O0 iD ^N
N
dOr+
dLn
r^ON
r, P,
itLli P`P ► P,N
.r
.4
N
-4
N
1 1 01 ,
N
N
O 00
N
1
to IN
ill O
N I
O 0
N
1
O _pl
H H W
N
N
0000 ^M
1
N 4) 41
^o
C 6
1 1 d
O 01
O ° 1
O
N
N
C G. QJ CA N
10
10
to
i }
O
.Y
1 1 1
1
tp
ilOn
Oi i0 N O
O
d
N1
Oi
.^ 1 t 1
0
0
d
u c
8
J
H
lu
.+
N
Q
a.i
d .+ 1
1
0
0
r
O
N
Ni^ N r+
00
N N 1 1
00
N N1
00
N N1
N.N
0
0
0
^L^
iQ
O O .+
A C
Nttn NN ^
N I LD NO
i,Q
\ i0 C ^-
u y ^ rnu
O) 7 t-
t^
W
V
W O
.0 goo allo
J m Q H
O Hd
^WZU LL_ J W
^Z Li W
N OC
N d OC OC
N O:
d
LA
^Z
LL. W J0 W
JCC 00 »WW
L 0
• -10 W
LL J >
N
^Z Li. W
-j0 >W
t/1 CG
=
Li W -10 W t / f OC
""'Z LL W
.J0 MW N1 ad
W V O
(%j
N
Z
V
_
^ O
m
1 Z
O Q o .C C o QI t
M
c v .+
O *J D AOC
i
do .r N
NV `U-'
N
y
N
41
# }to%% ?a
17
a
Tests with CAT-NZ
This sorbent was prepared with 2.1 wt % ZnO on the UCI-alumina. Figure 2 shows breakthrough curves with CAT-NZ. Under dry conditions (cycles 1, 4), considerable loss of zinc by evaporation was observed in the form of deposits on the cooler parts of the quartz reactor. These deposits were later confirmed to contain zinc by AA analysis. Consequently, the H 2 S absorption capacity of this supported ZnO sorbent dropped rapidly in consecutive cycles. Another reason for lack of regenerability may be loss of specific surface area (sintering) through growth of zinc oxide crystallites at the high temperature of operation (70O°C).
The H2S breakthrough curve correspondi-ig to cycle 3 in Figure 2 was obtained with 6.5 mol % H2O in the sulfidation feed gas, which also contained 2530 ppm H 2 S 0120/112S = 26/1). Although the effect of steam cannot be properly assessed in this case due to the observed loss of zinc during the preceding dry operation, the drastic drop in breakthrough conversion is mainly due to the large 1120/H2S ratio.
Similar results were obtained with the sorbent CAT-
NZ2, which had a higher loading of ZnO (see Table 3).
Tests with CAT-NV
This sorbent contained 4 wt % V205 impregnated on the UCI-alumina support. Figure 3 shows H 2 S breakthrough curves, with CAT-NV (the abscissa in this figure is the real time of reaction). Ir the presence of hydrogen, vanadium (+5) is reduced to a lower oxidation stzte. The resulting solid oxides (e.g., V 203, V02, etc.) are probably highly dispersed since their crystallites precipitate from the molten V 20 5 phase.
In dry sulfidation tests (cycles 1, 2), CAT-NV consistently retained 0.18-0.20 mol H2S/mol V205,
Most of the H 2 S retained was subsequently recovered quanti-
tively by desorption in nitrogen at the same temperature (700%).
A small
fraction of H2S decomposed into elemental sulfur and hydrogen, and was
18
MAL
4
O
c1 W,:
a^
E U^ r.4
O o 'C
Z
n
R
Q
1
U
Lm
U
N
Z
•
g
00 •
•
U
1
U
C v
n n NO> CA
^
N •
=
r-
4-)
Z
r
w
n
r
•
O •r
r N N
~^=
>
ro
U
t.
7 II +•1
U
t/1
E
r^
• 4- N
Z
Q)
•o _O X
a
1
U
P
1)
^ 1= N U ua-R
V)3
x
O
1
N co
x
O W
N
x
=N
I N
d
J
N N OJ
L C U^ Ol ^S0
Of N O i t
N Z 1
3
Yr0 dU
x
d
C -4
•
O cz
M O 1
Uv
Z
O
O O O O O 00 O 1` C 1.
i 4-
co O
N
O O O O O O O O L^ lqr %0 M imAd `
J° O
O O N
O O
(\i 0)
L
7
rn LL-
O
r-+
SZH III ino
o 19
accounted for by titration of the elemental sulfur produced. The nitrogen
N2 -air mixture. Neither S0 2 nor sulfur
purge was followed by exposure to a
were detected in this step indicating the absence of any vanadium sulfides. These results are important in view of conflicting reports
(13,14) on
equilibrium data and formation of vanadium sulfides. The present study has
650-700% no bulk vanadium sulfides are
shown that in a fuel gas atmosphere at formed; instead,
H2S is chemisorbed on vanadium oxides.
In the presence of steam (cycle equal to to
3, Figure 3) with a molar ratio of H20/ H2S
34.2/1, the H2S retention at breakthrough was substantially reduced
0.04 mol H 2 S/mol V205. The inhibiting effect of H2O appears to be
due to competitive chemisorpt;on of
H2O and H2S on the available vanadium
oxide sites.
When
V205 was exposed to H 2 S-N 2 mixtures, with no hydrogen present, it
was reduced according to the following overall reaction:
3V205 + 2H 2 S = 3V 2 U3 + 2S02 + 2H2O The uptake of
H2S was much higher in this case as shown by the breakthrough
curve of cycle 6 in Figure
3. However, the S02 produced by this reaction is
clearly undesirable. During sulfidation in the presence of hydrogen, as with coal-derived fuel gas, the above side reaction is largely suppressed. However,
j
a small amount of S02 is still released in the beginning of each cycle. This constitutes a disadvantage of vanadium-based sorbents. Another disadvantage is their low sulfur capacity.
Since vanadium oxides were found to chemisorb substantial amounts of H2S under reaction conditions, it was decided to investigate their adsorptiondesorption characteristics in some detail. CAT-NV2 with a slightly lower
V205 loading was tested for this purpose. The results depicted in Figure 4 show an overall behavior identical to that of CAT-NV. Cycle
2 was performed
with the sorbent in pre-reduced form, while in cycles 3 and 4 no oxidation of the bed was performed after the H 2 S desorption step.
Mi
20
The results confirmed
,. F
yr vo N 4J r 00
V>
A
c
O
4J
N M
1
-
N
>
N 00
_
L N
uoLO
41 C
N
.^.
N
1
Lm
c 4-) ^
O N
N
C1
C! >
a ^
C
L ^ U F-
c r a
#
N
>
00 U
Z
Q
^
0
Y Z
^ =
N
o o
L to s ++ >-
N C13 u
a^ Z
xV
i
^i
4
q
L
x
1
M 1
x1 N x
U
X
Io
S
O
t
I
(wdd) S H 1311(10 Z
a
21
P--
•
Z O o fa
Z
ad 7 W tL
UQ^
O
U
N
00
^ W
p
pp .i L^1
U C O
d!
r
%0
N
^
F-
4-
•
N v a >o n
V
0 t7
u
5R d
O
U
a
U
LAJ
bs
g v
•r
W N
'-
N
4/ 34
V _ U 3 C NtD "^ r
W
OE P-4 F--
ri
L aci O ^
V L
n
O
tIm N 00
N L Z
t ^
M i
—
Y Q
Ltd
UU a
\\
p
^o v a^
L
0
co o
'+. L-'Il
oo
hl Z
E!
UU
r
g
^
E Pip N O
OZ
I
a L
LAJ
U_rn LL-
vs
Z 2
N
W LY_
N
•
(wdd) S ZH 1311(10
i 22
the earlier contention that V 2 0 5 is first reduced to a lower oxidation state, which then chemisorbs H 2 S reversibly. Temperature seems to have only a mild effect on chemisorption. However, the determination of the activation energy for chemisorption requires additional work.
Interesting results were obtained in the desorption phase of the above cycles. As shown in Figure 5, the desorption of H 2 S seems to be a strong function of both the temperature and the flow rate of purge N 2 . The plateau region, common to all curves, indicates an equilibrium behavior until most of the H2S is removed from the surface. Some preliminary modeling efforts suggest that Lanymuir-type adsorption is not valid, and a different mechanism may be operative here.
Tests with CAT-3N
This sorbent had a Zn0/V 2 0 5 molar ratio of 51/49 corresponding to a eutec-, tic composition (Table 1). As shown in Figure 6, in sulfidation with a dry feed gas containing 4000 ppmv H 2 S, the pre-breakthrough level of H 2 S was less than 0.5 ppm, with sorbent conversion at breakthrough equal to 0.65 (based on ZnS formation). This high and stable conversion indicates rapid kinetics, no metal loss, and little or no loss of surface area in successive cycles.
Regeneration of this sorbent was conducted with 1 mol% 0 2 in N 2 (no H20) at 700°C. Sulfur products included H 2 S, S0 2 and elemental sulfur. Preceding the oxidative regeneration step, a nitrogen purge was used (also at 100°C), upon which H 2 S and elemental sulfur were produced. The total amount of elemental sulfur in the sulfur products was as high as 50-65 mol%.
Characterization of this sorbent included BET surface area, elemental analysis by AA and X-ray diffraction for identification of crystalline phases. The surface area of the fresh, sulfided and regenerated CAT-3N remained virtually constant (86 m 2 /g). The AA analysis showed no vanadium loss, while the XRD analysis showed a detectable amount of zinc aluminate in the regenerated but not in the sulfided form of the sorbent. This finding suggests that any zinc aluminate that was formed under these operating conditions, was decomposed
23
si a
R r-r O
8 ,^
°' tm L a C CJ . 0 O L 4J •r
Z C
•r
N Z 1
o r^
c
E W ^ O tf1
a v ^
L N N 2 w O C O .r 41 a.
L
O Q
O (U O ^
p M
C) N
O rr
8 8 8 8 8 8 8 8 8 ° (wdd) S ZH Alin0
24
O
L ^ 7 •r
♦I
4-
c
O
Q
M
U
C
•
YI •^
^^
rV^
IT
v n
p
c +J G
00
N X`
4J E
++
fO 1A
v
TO
W
ae N ;
1
U r„►
U
t`
1
a
Z
O 1—
r7
^ 'c v► C4 `^
CL of
u O N N
O
r39
CO
Q Q
E 13 d
^
1=
u ^
E
Z M
a
U
cn o0
N N
.., °
N J
W
^ ^ ►►
L
v
O
LA-
^+
rack ;GN oN O
1
Z J
C
0 1` ►,
to
>— E
1
^ ^
0n
•^
►►
Z
cV
U
Z
i 00 m Vd v
s-
c U
CL
z vi
u
U
^
L
ca
^
C
6 00
^
§
8
S
8S
C
C
(wdd; S Z H 13uno
s 25
during sulfidation and formed again in regeneration, a mechanism consistent with what had been suggested in previous work at IGT (6). A Tests with CAT-3N3
In a series of experiments with this vanadium-rich Zn0-V 2 05 sorbent, the effects of steam addition in sulfidation and regeneration were examined. Jn dry sulfidation with 2000 wpm H 2 S, CAT-3N3 consistently gave breakthrough sorbent conversions of 0.64. However, as shown in Figure 7, in the presence of H 2 O (6.7 mol %), the breakthrough conversion was only 0.15.
In this case,
the total H 2 S retained by chemisorption on the sorbent was greatly reduced.
This was shown by measuring the H 2 S elution in nitrogen purge following sulfidation. As previously discussed (CAT-NV), these results indicate a competitive H 2 0-H 2 S chemisorption mechanism. Table 5 summarizes the data obtained with CAT-3N3 and a similar sorbent, CAT-3N2, in several sulfidationregeneration cycles.
As in the case of CAT-3N, the present sorbent was also fully regenerable as shown by the breakthrough curves for cycles 2 and 5 in Figure 7. These findings suggest that zinc is stabilized by vanadium, an important result from the standpoint of the overall performance of Zn0-V 2 05 sorbents. Moreover, no surface area loss takes place during regeneration of these sorbents, apparently because the molten V 2 05 • Zn0 phase is restored at the end of each sulfidation/regeneration cycle.
Another finding from these tests refers to regeneration with steam. With steam
and nitrogen mixtures ( no oxygen), regeneration was very slow and remained incomplete. With r'r and steam together (1-5 mol % 0 2 ), both H 2 S and S02 were formed, and their yields went through maxima at different times. Elemental n^ilfur formation during wet regeneration was unexpectedly lower (by about ter, Limes) amounting to 5 mol% elemental sulfur. Finally, it was later found that regeneration at the higher temperature of 750°C eliminated the problem of zinc sulfate, which was formed to a limited extent at 700°C.
a
O
LA
4-
o r_
•
r -4
E
a r Iy
• rt+ ^
N ♦^
M
W
E v
Q.
N;
d O "'r+
a co Z•^ v
CD ^^
O L N ^ L V M
N
M ~ Q Q
N ti ^O CD
^ t/'1 N ti +E.' O= Mr-4 N
~
NJ t/^
N
vz
CL
O
J
&A W
N
Z
O> W •
O
C;
N
= N
A
v ^ I.-
ui
NN
^
41
oC aJ In too
00
LA W J
U #
`" ao y)
uu ryc
NWA.^ c 39O •—
O
p W
N
^
N J Q
N O
O
Z
%o
^
rv E a+
w N
r^ c pf s:3 o O In c
L
O O O O O O O O O O O O O O O O 1-^ N L^ 'gr M co
(wdd) S Z H Alin0
27
O O
O
c m
Y Mv d +) z r0
1
N
-um .
a rn
O
re
.
Q
V
w
w
4 W, -
Tests with S-1 and S-2
Sorbent S-i was prepared on a low surface area (< 1 m2 /9) zirconia, and contained 0.38 wt% Zn0 and 1.73 wt% V205.
Since low loading was obtained, a
concentration of 1000 ppm H2S in the inlet gas was used to ex p end the breakthrough times. In the first cycle, with 9.5 mol% steam in the feed gas, the H20,/u,,S ratio was 67.5. The H2S breakthrough took place fast (2 min.) corresponding to a breakthrough conversion of only 0.064 for zinc. The run was continued, and after eleven minutes, steam was eliminated.
The oulet H2S
level then dropped from 450 ppm to 0 ppm and stayed at zero for 25 minutes, when a second breakthrough was obtained.
These observations support the
earlier conclusions about competitive adsorption between H2O and H2S. The second cycle carried out under dry conditions resulted in a breakthrough conversion of 0.67, a value close to earlier results.
Since adsorption effects are predominant in the Zn-V-0 system, the H 2 0/ H2S ratio rather than the absolute concentration of H2O is the most important parameter.
In previously described experiments, the H2U concentration was
approximately 7 mol% to match the METC gasifier exit cu^^ l itions. However, the law H2S concentrations (2000-4000 ppr) used in those experiments resulted in high H20/H2S ratios.
In the METC fixed-bed-gasifier product- as (0.7%
H2S, 7% H20), the H20/H25 ratio is equal to 10.
This value Afas used in
a following set of experiments conducted with the sorbent S-2.
Unlike sorbent S-1 which was prepared by step-wise impregnation, sorbent S-2 (with 0.5 wt% NO and 1.1 wt% V20 5 ) was prepared by co-'npregnation of the
high surface area zirconia support described before. The surface area of S-2 was much higher than that of S-1, as shown in Table 3. Figure 8 shows breakthrough curves obtained with S-2 in several sulfidation/regeneration cycles. The experimental conditions and results are listed in Table 5. The first cycle resulted in a surprisingly high breakthrough conversion of about 0.80. Following sulfidation, desorption and regeneration steps were carried ut. As in the case of dry sulfidation (e.g. CAT-3N, CAT-NV), the amount of H2S retained by adsorption was about 20% of the total. This indicates that H20/ H 2S ratios up to 10/1 might be allowed before a drastic inhibition of H2S chemisorption (by H20) occurs.
28
i
900 S-2 (ON ZIR CONIA)
o
700'
600
P- latm T S • 650, ?'J0°C INLET H2S • 2280 ppm 700°C) C-3 (650°C) H 2OM 2S - 1011 S.V. • 1655 hr 1 • TREGN • 700*C C-2 (650°C)
E 500
C - CYCLE NO.
N F°-
0
400
300
200
100 i ------------ L ------------ L0 --6= —0 0.6 0.8 0.3 0.5 0.7 0.4 0 0.2 0.1 NORMALIZED ABSORPTION TIME (tlt°)
Figure 8. Breakthrough Curves in Successive Sulfidation Cycles of S-2 Sorbent (0.5 wt% ZnO, 1.1 wt% V205); t* = 31 min (Based on ZnS formation)
29
--i-
___
"70
In the second and following cycles a drop in the absorption capacity of the S-2 sorbent was observed, probably due to a pre-reduction step with H2S-free fuel gas. This pre-reduction was used in the beginning of the second cycle, upon which zinc could have been lost and the compositir2n of the sorbent changed. The zirconia support used in the S-2 sorbent preparation does not appear to have played a key role in the performance of the sorbent.
However, additional
tests on zirconia and alumina supports are necessary to establish the role of
sorbent-support interaction. Summary of Performance Characteristics of Zn-V-0 Sorbents The basic chemistry of the ZnO-V205 sorbents may be summarized with a list-
V
ing of the overall chemical reactions during sulfidation and regeneration.
V
Sulfidation (7000C):
H S + (5-x)H2 = Z,` S(s) + V20x(s) + (6-x)H20
ZnO*V205M + 2
H S = V20x-H2S
V20x(s) + 2
3V 2 0 5
(chemisorption)
(5-x)H 2 0 (side reaction) (1) + (5-x)H 2 S = 3V20x(s) + (5 - x)SO 2 +
Regeneration (750*C):
V 20x*H2S = V20x(s) + (1-y) H2S + YH2 + 2Y_ S2 (N2 purge', O-W) x ZnS(s) + V20x(s) + (3+w- 7) 02 = ZnO ' V 2 0 5 M + wS0 2 + — 2 S2 In the above I s represent melt and solid phases, respectively.
features of the
ZnO-V205
The mai n
sorbents can be summarized as follows:
(1) In each sulfidation, the initially existing molten sorbent phase disperses ZnO, V20xq etc. better in the support.
As a result, reproducible high
breakthrough sorbent conversions (dry: 60-70%; w/steam: 10-30%) are obtained.
30
-
^ .- _^_ __- A. I
(2)
Vanadium stabilizes ZnO in the support.
(3)
H 2 S is chemisorbed up to a level of 20 mol% on V 2 0x (no sulfides are formed). At high H 2 O/H 2 S ratios the H 2 S uptake by vanadium is minimal due to competitive adsorption between H 2 O and H2S.
(4)
Regeneration at high temperatures (up to 750°C) is feasible without any loss of surface area (sintering) or active metal loss, and eliminates the formation of zinc sulfate.
(5)
Dry regeneration produces about ten times more sulfur, and an important step in elemental sulfur formation is the catalytic decomposition of H2S.
(6)
Since vanadium does not form any bulk sulfides, the overall sulfur capacity is low. It may be increased if V 2 05 is replaced by a reactive compound.
In view of the limitations of the ZnO-V 2 0 5 sorbent systems it was decided to briefly explore other possible systems in this class. It was found from the literature (10) that zinc and copper molybdates at 32 mol % form eutectics with M00 3 melting, respectively, at 705°C and 560°C. Moreover, molybdenum is also known to react with H 2 S to form sulfides (mainly MO 2 S 3 ).
Hence, the
overall H 2 S capacity was expected to be higher with these systems than with the Zn-V-O sorbents.
In a preliminary experiment a eutectic melt was prepared by mixing powders of zinc molybdate (ZnMo0 4 ) and molybdenum oxide (Mo0 3 ). After cooling, the solidified crystalline material was crushed to obtain a powder. This powder was mixed with the high surface area UCI-alumina and tested in sulfidation with 20 mol% H 2 , 1 mol% H 2 S, 26 mol% H 2 O, balance N 2 at 538 and 720°C.
A
high breakthrough sorbent conversion of 0.75 (based on ZnS only) was obtained, while the pre-breakthrough level of H 2 S was less than 0.1 ppm. Considering the crudeness of the experiments, these results were encouraging. In sulfidation at 720°C, however, evaporative loss of some unidentified compounds took place from the sorbent.
2.2 Supported Cu-Mo-O Sorbents
Since Cu-Mo-O mixtures form eutectics at lower temperatures (10) than the ZnMo-0 systems, and metallic copper is not volatile under sulfidation conditions, further work with supported mixed oxides was focused in developing and testing CuO-based sorbents.
31
Another advantage of Cu-containing materials concerns their easier regeneration 'a in view of the fact that copper sulfate decomposes at lower temperatures than zinc sulfate. Moreover, while copper metal is thermodynamically inferior to ZnO in terms of sulfidation equilibria, little is known about the removal efficiency of mixed oxides containing copper in a higher oxidation state. In the following, the performance of Cu-Mo-O mixtures (CM-sorbents) supported on the UCI-alumina is discussed in terms of sulfidation efficiency and regenerability in the temperature range of 538-650°C.
The CM-sorbents examined are
i listed in Table 4.
Sorbent CM-2
The CM-2 sorbent was prepared with a molar ratio of CUM00 4 /Mo0 3 = 0.47, equal to the corresponding eutectic (Table .j %n the UCI alumina support by the incipient wetness impregnation technique described before. The total loading of CuO and MoO 3 on the alumina was 23.6 wt% and the surface area of the sorbent 52.4 m2/g.
Sulfidation cycles were carried out at 538 and 650°C with an inlet molar gas composition of 20% H 2 , 2S% H 2 O, 54% N 2 , and 1% H 2 S. carried out at 650% with a 70% N 2 -30% air mixture.
Regeneration was
Two sulfidation/regen-
eration cycles were carried out at each temperature. The H 2 S breakthrough curves obtained are shown in Figure 9. The most striking observation from this figure is that the pre-breakthrough H 2 S levels in all cases are less than 0.1 ppmv, and the corresponding sorbent conversion (based on Cu 2 S formation) is close to 100 percent. XRD analysis of the sulfided sorbent has not identified any crystalline phases of molybdenum sulfides (only Cu 2 S was found). However, the role of molybdenum is not clear at this point, the possibility of its forming amorphous, sulfide phases (with or without copper) left open to discussion.
During sulfidation, limited evaporation of a molybdenum compound was observed. The molybdenum loss (measured by AA analysis) was higher at 650°C. At 538°C the loss of molybdenum was very low, however, the loss was not completely eliminated. Further testing of other Cu-Mo-O sorbents was then carried out with higher Cu/Mo ratios to suppress the evaporative loss of molybdenum.
32
41
ro
c Ul (V
E ^
r^v o¢ c
Y-
r-i t `'
U
#
0
U
^ c
W 0 r Cl U -0 Ln >^v to
CD
1 111
^j M
N
g
U +-) CO
r^
c O O
x------
CL c ro LO) E
----------- -----------y+
o r^ v
W
U
(U
r-/ 00 K1 1
NC
O
O
Q
00 O Z C
1 ^ ^
^
w L. M
r ro ^ r
^^
N O +^ E ^ •w > r U •r \ O Vfl^,o MCr LA
N • tD
U o
v u w
7 MU
cn (D O
E CD N
-^ Q `M
O
v^ ^O CA
E
o
N =
S •
,^ 1 L U
'^
U1 N
E ° ° •^, 00 C m u 1 M
N
1
U
N
N
11
tl
f-
N
=
C
° N
z
—
N
N
C
N
N z
c
a
z
•^
Iy
0
W 2
O
LL.
~
f-
J
N
U
r- \ LO mr V) +) o^ ro L 7^
o
U 1T U^ c t +)
c+ c to 7 Od O^ O ^ r c
O
^
Z
Z
N N _
= to
W
O
Z ^--I
E U U
U .^-
^D
cfm
L O 3S4 •r eac% 3 E
Ol 1 et i S- f N O m U — 4-
ON
C
O i
11
U
•r LL-
.J O
CD
40
N
(D
00
(wdd) S Z H 1311(10 ^- 1 r a
11 33
s
16
Sorbents CM-10 and CM-11
An interesting finding from our parallel work with unsupported Cu-Mo-O sorbents described later in this report was the observation of eutectic melt formation during sulfidation even with sorbents of high Cu/Mo molar ratio (up to 6/1). Cuprous molybdates, if formed, are known to melt at 500°C in argon (15). In the reducing atmospheres used in our sulfidation experiments eutectic oxide sulfide mixtures may exist which can have better sulfidation equilibria, thus explaining the high H 2 S removal efficiency observed with all CM-sorbents.
Sorbents CM-10 and 11 were prepared with Cu/Mo molar ratio of 3/1 on the UCI T-2432alumina support as described before. In order to stabilize and fully convert the oxides to the molybdate and aluminate-phases, a 10-12 hour long calcination at 650% was carried out. The Cu/Mo ratio of these sorbents was chosen from work with the bulk sorbents CM-6 and 8, which showed both high sulfidation efficiency and good stability. The porous alumina used in CM-10 and 11 was intended to provide the necessary surface area as well as a support to the intermediate complex eutectics formed in sulfidation.
A total of thirteen sulfidation/regeneration cycles were carried out at 650°C on sorbent CM-10. In the first eleven cycles, the sulfidation gas consisted of various mixtures of H 2 -H 2 S-H 2 O-N 2 • In cycles 12 and 13, the sulfidation gas also contained CO and CO 2 , and had a composition typical of a j;
Lurgi fuel gas.
The space velocity in all runs was 2120 hr -1 .
Table 6
summarizes the data obtained in sulfidation tests with CM-10, while Figure 10 shows the H 2 S breakthrough curves in successive sulfidations.
A striking
feature is lower than 0.1 ppmv H 2 S breakthrough levels up to very high _ sorbent conversions. In Figure 10 t/t* is calculated based on the assumption that t* corresponds to complete conversion of CuO to CuS. This t* is twice as high as what would correspond to Cu 2 S formation (note that Cu 2 S is the main sulfide identified in our other work with CM-sorbents). In all runs with 1.0 t
mol% H 2 S in the inlet gas, the outlet H 2 S level remained
J
F—
cn
E N f^
W
^ Z
11
u
M
O
L
c
7 4j
O CQ-+
3
CL
CD M
a
4m
C^
111
r
^
L
O
ti
1
e
{
Two sulfide' 'on-regeneration cycles were carried out on a fresh batch of copper ferrite at bU0°C. The results are shown in Figure 20. In both sulfidation cycles, the outlet H2S levels quickly rose to 35-38 ppm range and stabilized at that lev=el until final breakthrough which occurred at about 90% sorbent conversion in cycle 1 and about 75% in cycle 2. The regeneration was carried out to completion at 600% with a 70 mol% 14 2 -30 mol% air mixture. Nitrogen purge followed the regeneration to decompose all residual sulfate. It is noteworthy that the breakthrough 11 2 S levels observed in both cycles 1 and 2 at 600°C were lower than the calculated equilibrium (52.9 ppm) of reaction (6).
In other sulfidation tests with copper ferrite, the sulfidation temperature was raised to 650°C. While a similarly high and stable sorbent conversion took place, the pre-breakthrough H 2 S level at 0.80 sorbent conversion was 85-90 ppn, close to the equilibrium for reaction (6). .Still, the exit 11 2 S level remained below its equilibrium value up to — 0.1b sorbent: conversion.
The sharply different outlet H2S concentra + ions observed at 538°C and 650% are discussed in the following in terms of different solid phase transformations occurring at the two temperatures.
Characterization
of
fresh,
sulfided
(in
H2S-H2O-H2-N2),
reduced
(in
H2-H2O-N2), and regenerated (in 02-N2) samples of CF sorbents was performed by SEM and XRD analyses. Figure 21 shows SEM micrographs of the porous copper ferrite sorbent before (fresh) and after sulfidation at 650°C. Both large (>10 Wm) as well as submicron size pores with similar structure before and after sulfidation can be seen on these micrographs. XRD analysis of the sorbent provided the results listed in Tables 7 and 8. The weight percentage figures are only qualitative and refer to the total crystalline phases. As can be seen in Table 8, the reduced CF sample includes CuFe 2 04, hence, reduc-
f
tion is not very fast at 538°C.
The sample sulfided at 538°C contained the
mixed sulfide compound CuFeS2 (chalcopyrite) along with Fel_xS, Fe304, unconverted CuFe204 and some amorphous material. 650°C, the
In the sample sulfided at
phase CuFeS2 was absent, while there was indication of a
poorly crystalline Cu2S phase. As in the case of zinc ferrite, copper ferrite formation from the individual oxides did not proceed to completion during regeneration.
59
. TV
180 160 140
E
fl. o. `r
N
120 100
J p80 H
0 60 40
I
20 0
NORMALIZED ABSORPTION TIME (t/t*) Figure 20. Breakthrough Curves in Successive Sulfication Cycles of Porous Copper Ferrite at 600°C; t* = 64 min (Based on Cu2S and FeS Formation).
60
I 1
OF POOR QU..^
It
le
1
. 1
lir-
LEII
0rV Figure 21
`)canning Electron Micrograph ,, of Porous Copper Ferrite (a) Fresh (h) Sulfided at 650'C
OF POOR QUALITY 61
Pi{ The pre-breakthrough H2S levels and the phases observed in the XRD analyses may be discussed qualitatively in terms of the following
thermodynamically
feasible phase transformations: 3CuFe204 + 2H2 = 3CuFe02 + Fe304 + 2H 2 O 3CuFe02 + 2H 2 = 3Cu + Fe 304 + 2H2O CuFe02 + 2H 2 S = CuFeS 2 + 2H 2O 2CuFe0 2 + 3H 2 S + H 2 = Cu 2 S + 2FeS + 4H2O 2Cu + H 2 S = Cu2S + H2 i Fe304 + 3H 2 S + H2 = 3FeS + 4H 2 O
(7) (8) (9) (10) (6) (5)
with equilibrium constants and corresponding H 2 S concentrations listed in Table 9.
Reactions (7) and (8) represent the stepwise reduction of copper ferrite CuFe294 via the intermediate compound CuFe02 to the final products Cu and Fe304 which constitute the stable phases of Cu and Fe in the presence of the H 2 -H20 mixture employed.
The equilibrium H2S levels for reaction (9)
.' were calculated using thermodynamic values for the compound Cu20.Fe2O3 (19). The absence of CuFe0 2 from the reduced and sulfided samples indicates that this intermediate compound is depleted relatively rapidly by reactions (8)-(10) to either the final reduction products Cu, Fe 30 4 (reaction 8) or to the sulfided products CuFeS 2 , Cu 2 S, FeS. The products of reduction, Cu and Fe 3 04 , are also sulfided to Cu 2 S and FeS by reactions (6) and (5), respectively.
The pre-breakthrough level of H 2 S is determined by the thermodynamically most favorable among the sulfidation reactions (5), (9), (10), and (6) assuming that such reaction is sufficiently rapid. We note that the sulfidation of the intermediate compound CuFe02, containing copper in a higher than the ground oxidation state is more favorable than that involving sulfidation of metallic copper, Cu°.
Reaction (9), although thermodynamically less favorable than
(10), is kinetically much more favorable, for it requires no change of oxidai
tion state and only minor rearrangement of the two cations. This hypothesis is
62
-4.4
1
ii consistent with phases found in the sample sulfided at 538°C and with the prebreakthrough concentrations of H 2 S in Figure 19. Chemisorption can be invoked once more to explain the sub-equilibrium H 2 S concentration during the
first sulfidation. ij
At 650°C, the compound CuFeS 2 was no longer present in the sulfided sorbent, possibly due to the very rapid consumption of the precursor CuFe0 2 by reaction (8). In the absence of this intermediate product, sulfidation would proceed by reactions (5) and (6), of which (6) is more favorable and would control the pre-breakthrough level of H 2 S. This hypothesis is again consistent with observed H 2 S concentrations of 90 ppm upon sulfidation of copper ferrite at 650% -- not included in Figure 19. r
Copper Aluminate
In view of the superior performance of copper ferrite in comparison to its constituent pure oxides, another mixed oxide, copper aluminate (CA), possessing the spinel structure was prepared and tested under the same conditions. The results presented in Figure 22 reveal some interesting features. In three sulfidations at 538°C the conversion at breakthrough was higher than 0.7, while
}
the pre-breakthrough H 2 S level was lower than 10 ppm as compared to 46.0 ppm i for the equilibrium of reaction (6). At 650°C, on the other hand, the breakthrough curve has an odd shape. An initial near-zero level is succeeded at 0.15 sorbent conversion by a plateau at about 100 ppm, which terminates at a second breakthrough at 0.80 conversion. Unlike copper ferrite, the structural change that occurred at 650°C raising the H 2 S level to the equilibrium of reaction (6) was reversed when the temperature was again lowered to 538°C
(cycle 5, Fig. 22). In the case of copper ferrite the structural change caused by sulfidation at 650°C was found to be ir. • eversible. As shown in Tables 7 and 8, the fresh and regenerated sorbents contain the crystalline phases
raiai —n. _
ri,o
anti
cnmc
amnrnhmic
matari al
nrnhahl v
A1_0
Tha
i
^
s O 1
L
.4
O d
4-O c O
of •^
r-
rEEo
L U O >1
4-
°
N ^ O A
tp
U
y •r C
v ^ Q
W ^^ ^-
Na N a ea > co N
^p
N
C w
u
C
Ln N E an n
Q cko p W
N Q O
•r
II •K u, +^
a
.w
i OJ O eo U C eo
z > > O ri Q t
}^ L l a
to CL
a n
CV C^
mi UO
N N as.
a rn
•r
LA-
^I
►° r%
I
g Ul%
I
I
g
I i I Ul%
(wdd) S ZH 1nno
16 -
64
rim
I
l
I l
o
xCuAl204 + (x-1)H 2 = CuxU-Al203 + (x- 1)Al 2U3 + (x-1)H2O, 1
o
O
co ^
Q ^
O
o E^L r--i N t " "
U
N
U E
U
N
o
O °' CL
ON O
N `Y. N
C:)
^'
z
C
Z
O
LM 49
N
J
U 0 O
L
^
N z
ZU U W
7 U.w
o 4J i rp
r
+-) . .
Y ,-r ^O
L IIi mu
—4
U H co
(wdd) S Z H Blinn
61
a 4J
O
M N G!
L-
C Q `ONO N N
r.
N7 •r E C • r r-U
W
O
J
N
0
v
sLr) rnm
U
= =
Wv V1
^ v u
Q
L.L
O
a
'A
to ro w m
If
^^
Z
.A,c q* d
:3 Cf U-
3
!a- tnx -- • mmv vOtnNN00 OOmmo ntD%b4Df^-CTLO0 Ow m0 f-v O W^-+ F N r 1
•
A
O
• L
U-
u >1
CN tOCO 00 f^ NO^MCIp OM +0+ V C9 O •r CT Lp w c N r0 —.--- ----------- =ow • N CL— aj CL ^- O 00 0 0 0 0 00 p 0 0 0 0 Cl O U to to Ln M Ln Ln Ln Ln Ln Ln Lfi Ln '•n Ln 4 fE C S W r U 9 O t0 t0 t0 tD t0 t0 t0 WD %D t0 t0 t0 LO t0 N Q
r• ...
U
W J V
s +
0
>- Z U
(
"NM^ LC* t0f` MM O•--L .r
68
1J
Met
ri • 4 .4 .^
41
N
dk
•-L C 4-
sulfidation/regeneration conducted with CFA. After the fifth cycle, gas mixtures containing also CO and CO 2 were used in sulfidation to examine the potential effects of these compounds on the sorbent performance. A mixture with a molar composition of 17% H2, 24.1% H2O, 1.0% H2S, 13.0% CO, 10.1% CO2 and 34.8% N 2 was used in cycles 8 and 9. As can be seen in Table 10, similar results to those discussed above for cycles 1-5 were obtained, clearly indicating no effect (of CO, CO 2 ) on the sulfidation characteristics of CFA. In cycles 6 and 7 with 0.5 molt H2S in the gas mixture, an earlier breakthrough of H 2 S was observed. At that point it was not clear whether this was an effect of the lower H 2 S concentration used in the feed gas or related to incomplete sorbent regeneration.
Further experiments addressed the effect of inlet H 2 S concentration on the sorbent performance (Table 10). The H 2 S removal efficiency of CFA II dropped in tests with 0.5 molt H 2 S in the feed gas, while it did not change appreciably when the inlet H 2 S concentration was only 0.2 molt. It appears from these results that the ratio of H 2 (+CO)/H 2 S is important in determining the stable intermediates and associated sulfidation equilibria of this sorbent system. Another parameter studied briefly in this work was the space velocity. No effect on the sulfidation performance of CFA was observed in changing the space velocity from 2200 to 1000 hr-1.
Different gas mixtures were used in regenerating the sulfided CFA sorbents. In all cases the reaction zone temperature was kept at 650°C ±2°C. The feed gas flow rate was typically — 200 cc/min. Table 11 summarizes the different regeneration conditions used, cumulative sulfur products composition, and sulfur balances. The latter were calculated by comparing the total sulfur measured in the product gas to that absorbed (as H 2 S) doiring the preceding sulfidation. The total S0 2 (and H 2 S, when steam was used in regeneration) eluted was measured by absorption in iodine solution followed by titration of the excess iodine. Elemental sulfur collected in the condenser trap was dissolved in a sodium sulfite solution and analyzed by a standard iodometric titration method.
With a regeneration gas containing 30 molt air, 30 molt steam and 40 molt N2, 99.7% of the sulfided CFA II sorbent (after cycle 4) was regenerated in 50 min-
69
t
TABLE 11, REGENERATION TESTS WITH SORBENT CFA II A I
I
EXIT GAS CUMULATIVE
FEED GAS COMPOSITION
CYCLE TEMP. NO.
mol X
02
+
N2
H2O
S02
1(hr-1)
( °C)
TOTAL
SULFUR PRODUCTS MOL%)
SPACE VELOCITY
(S02+H2S
TEST TIME
emen-^u-T u r tal Balance+
(min)
Sulfur
1
650
6.1
62.2
31.7
-
2260
na
na
-
25 5
2
1 650
6.1
62.2
31.7
-
2260
na
na
-
270
3
1 650
6.3
' 93.7
0.0
1
-
2210
na
1.15*
-
340
62.2 = 31.7 1
6.3
93.7 =
6.6
67.1 1 26.3
5.2
94.55
2.6
( 97.18
4
65U
5
, 650
6
65U
7
650
8
I1650
9
) 65U I
I
1 6.3
I
i
` 6.1 j
1
30.01
63.7
i(
2210
97.88
2.12
92.9
170
-
2260
93.45
6.55
101.0
180
-
2210
-
180
-
2100
98.93
1.07
89.0
160
0.0 0.25
2210
99.15
0.85
86.3
200
0.0 0.22
2210
n/a
-
200
0.01
na
na
11
10
1j 650
+10.5
11
1 700
6.3
23.7
12
700
4.2
1 90.8 (
13
1 650
6.1
j
14
650
( 5.3
I
39.5
1
3.52*
50.0
-
2210
96.83
3.17
89.7
290
70.0
-
2210
95.76
4.24
95.1
280
2210
96.90
3.10
-
600
3.84*
-
400
0.015.0 1
62.2
31.7 )
-
2260
-
61.4
33.3
-
2210
-
-
-
200 I
6 Determined together by iodometric titration. # Based on total amount of H2S absorbcd in sulfidation. t Including time of N2 purge. * Calculated in terms of total H2S absorbed in sulfidation. na: not analyzed.
70
tl
F.:;
utes, at which time the effluent gas contained less than 50 ppm S0 2 (as measured by the gas chromatograph). To completely remove S0 2 (to < 2 ppm level) regeneration was continued for another 75 minutes with the same mixture, followed by 40 minutes of N2
purge (140 cc/min) to decompose any remaining
traces of sulfate products. This last step (N 2 purge) was typically used in all regenerations to suppress elution of S0 2 (and early breakthrough of H 2 S) during subsequent sulfidations. The sulfur products in the regeneration off-gas upon completion of this test, consisted of 97.9 mol% (S0 2 +H2 S) and ^-2.1 mol% elemental sulfur. The latter was produced in the first 20 minutes of regeneration. A higher amount of elemental sulfur was observed in the following regeneration (cycle 5) which was run basically at the same conditions (with ^-2.0 mol% higher steam). No explanation for this can be offered at this time. With less steam (26.3 mol%) and about the same oxygen content in the regeneration feed gas (cycle 7), the amount of elemental sulfur produced was lower as shown in Table 11. Still lower was the elemental sulfur content of the regeneration off-gas when a mixture of air (24.75 mol%), N2 (75 mol%) and S02 (0.25 mol%) was used (cycle 8) in the absence of steam. Lowering the oxygen content of the mixture (cycle 9), raised the elemental sulfur product (to —3.5% of the total H 2 S adsorbed, Table 11).
As mentioned above, a total of fourteen cycles of sulfidation/regeneration were run with CFA II at 650°C. The surface area of the sulfided sorbent at the end of the fourteenth cycle was —8 m 2 /g, still a moderately high value, especially after the extended use of CFA, and accidental overheating of the sorbent which took place during regeneration in cycle 12. The crystalline phases of the sulfided CFA (after cycle 14) were identified by XRD. As shown in Table 8, these consisted of a non-stoichiometric mixed sulfide, Cu 9 Fe 9 S 16 , along with CuAl 2 04 , CuFe 2 04 , and some Cu and a -Al 2 0 3 .
There was also a
large amount of an amorphous phase which, presumably contained most of the sulfided compounds (since the overall sorbent conversion was 60%). This sample was also examined by SEM/EDS. The results are revealing in regar
to
pos-
ble mechanisms of sulfidation. Thus, on the surface of the sulfided particles there was uniform distribution of Fe and Cu, with the iron intensity stronger than that of copper, while very little S or Al could be found. SEiVEDS pictures from the surface of sulfided CFA II after cycle 14 are shown in Figures 24a-d. Next, the particles were cross-sectioned and polished, and
71
SEM/EDS maps were taken again. Figures 25a-d show SEM micrographs of a crosssectioned particle (a), and EDS maps of S(b), Fe(c), and Cu(d), respectively. All three elements were uniformly distributed in the porous surface. These data along with the XRD analysis indicate that the most active compounds in CFA are in the underlying amorphous phase, which is comprised of alumina bound Cu (and Fe) oxides. The iron-rich surface of the CFA particles is only partially
sulfided, but may assist in retaining the amorphous phase in an active and t
regenerable state.
In order to elucidate further the sulfidation and regeneration characteristics of the CFA type of sorbent, a set of nine sulfidation and regeneration
cycles was performed on a fresh batch of sorbent CFA II, coded CFA IIb. The results of the sulfidation tests are shown in Figure 26 and Table 12, while the
regeneration results are listed in Table 13. With a sulfidation gas containing 1% H 2 S, 17% H 2 , 13% CO, 10% CO 2 , 24% H 2 O, bal. N 2 , by volume, typical sorbent conversions of 90-95% were achieved in the first two cycles; but these dropped to about 75% in later (7th and 8th) cycles. The H 2 S pre-breakthrough levels stayed below 15 ppm up to 85% conversion in the first two cycles and up to 60% conversion in cycles 7 and 8. The H 2 S levels before the final break-
through were always below 45 ppm in all cases. With a sulfidation gas of nearly the same composition but with 0.5% H 2 S (cycles 3 and 4), the overall sorbent conversion dropped to 70% while the pre-breakthrough H 2 S levels stayed below 15 ppm till about 40 percent sorbent conversion. The H 2 S levels before final breakthrough increased to 60-70 opm. In order to elucidate the role of H 2 /H 2 S ratio, sulfidation in cycle 5 was carried out by decreasing the H2 mole fraction to 11% while maintaining the H 2 S at 0.5%. This lowered the pre-breakthrough H 2 S levels (compared to cycles 3 and 4) considerably. While the final breakthrough still occurred at 70% conversion, the H 2 S level before breakthrough was unly 38 ppm as compared to 60-70 ppm in cycles 3 and 4.
Sr - fidation in cycle 9 was carried out with no steam present in the sulfidation gas, which consisted of 1% H 2 S, 22% H 2 , 17^, CO, 13% CO 2 , bal. N 2 , by
vol ume. In this case, the sorber:t performance was distinctly superior to all cases where H 2 O was present in the sulfidation gas.
Thus, the H 2 S pre-
breakthrough levels stayed below 4 ppm till 92% conversion and increased to
72
ORIGIN'.! Pj°.CE IS OF POOR QUALITY
Figure 24.
(a) S E M Micrugraph c,f Surface of Sulfided CFA II S o r b e n t (650'C, 14 cycles; (b) Corresponding EDS Map of Iron
73
1 ^.m
Figure 24.
(Cons d). ntal D" M,^ps of C-,, t fr (r,j and Su; fur (d) over the surface of I igure 241 r O r ^6 r0
7 }O aJ r• V /.1 N V r 4J
a
r0 O M U C t0 aJ b C
^NMmt
Lo to1^
w O1
U
LJt y ! +- #
C
i{ 79
Q
only 10 ppm before final breakthrough, which took place at nearly
complete
(100%) sorbent conversion. The above observations confirm that the superior performance of sorbent CFA can he maintained in many successive cycles of operation at 650°C. Further, it was confirmed that better sorbent performance is realized at lower H 2 /H 2 S ratios.
A summary of the results of the CFA IIb sorbent regeneration performance in nine cycles is given in Table 13. As can be seen from this table, the element] al sulfur recovery during dry regenerations typically varies from 1 to 2 mol% as seen earlier. With a regeneration gas mixture consisting of 50 mol% air- 50 mol% steam (cycle 8), elemental sulfur is still only 1.20 mol% of the total sulfur products. However, with increased steam in the regeneration gas mixture (70% steam - 30% air), the elemental sulfur recovery increased to about 10 mol% (cycles 7 and 9).
High Temperature Perfomance of Sorbent CFA. Another series of tests with the first batch of CFA sorbent (CFA I) were conducted at the higher temperature of 830°C with a reactant gas mixture containing 30 mol% H 2 1 17 mol% H2O, 1 mol% H 2 S, balance N 2 , simulating the Texaco gasifier-fuel gas. In three cycles, breakthrough of H 2 S took place at complete (100%) sorbent conversion, while the pre-breakthrough H 2 S level was
^a V
1
CD
g —
N
N 11
N
=
i U
^'
o
1
I
I
cri
W
O
u c
7E
N ^C)
C _N
(A^w Lu
I
VO
0
O
O O C
E
U-
E
O
M
O U
U E O 0
N
m M r-4
n I
N m
N
Q o
I
•^
V
L
> co
L11 O
Q
Q
Z
O
O
zW Z= CD w
O O N =
•o
V
g 00
n
^
dW
^'
ti
V)
00
u
oo
0
^1
O
oL b
`p N ^
^ ^ cu o
X11
11
II N
_
=
^j y
yy^ j ^j
a —__
Z
_J
.--r
^ r
o
N
is
• Ift
Co L,
- -_
^Z Z Z
J
^
S:3
a
r a
cm
S f^
Q
V ^O
^ I,1
Q
Q
V CT
V M
Q
V N
L.^
^ r'•4
O
(wdd) s z W 1311no
; 81
Sorbent CM-3. This sorbent was similar to the supported CM-2 sorbent in terms of Cu and Mo contents with Cu0/Mo03 molar ratio equal to 0.32/1, corresponding to a known eutectic of CuM004 with M00 3 (Table 1). However, CM-3 was prepared in bulk porous form accordng to the technique described before for class B sorbents. The aluminum component was introduced in the precursor salt solution in the form of aluminum nitrate.
Sulfidation tests with CM-3 were conducted at 538°C with inlet gas molar composition of 20% H 2 , 25% H 2O, 54% N2, and 1% H 2 S.
Regeneration was car-
The H 2 S breakthrough cur-
ried out with a 70 mol% h2-'O mol% air mixture.
ves obtained at 538% i,: two cycles are shown in Figure 28. As in the case of the supported CM-2 sorbent, it is again noteworthy that the pre-breakthrough H 2 S levels are nearly zero in both cycles, and the sorbent conversion (based on Cu2S formation) is higher than 0.75.
At the end of each sulfidation the appearance of the sorbent through the quartz reactor walls indicated the formation of a molten phase during reaction. This may have consisted not only of mixed oxides but a complex sulfide eutectic as well, which could retain additional amounts of H 2 S. As with the supported CM-2 sorbent, a small loss of molybdenum was Observed during sulfidation of CM-3. This prompted the preparation of the CM-5 sorbent, described below, with lower M00 3 content.
Sorbent CM-5. This sorbent was prepared in dispersed form with a molar ratio Of CuMo0 4 / Al 203 = 1/1.1, as a potentially more stable sorbent than CM-3. XRD analysis of the fresh sorbent identified Cu3Mo 20 9 in the crystalline phase, while the remaining molybdenum and all the aluminum compounds were in a phase amorphous to XRD (Table 7).
Sulfidation tests with CM-5 were conducted at 538% and 650°C. molar
composition
was
15%
H 2 ,
59.5%
fidation cycles were repeated at 538°C. mol% N2-10 mol% air mixture.
N 2 ,
25%
H 2 O
and
0.5%
The inlet H 2 S.
Two
gas Sul-
Regeneration was carried out with a 90
The temperature at
tion was 538°C, then was raised to 650°C.
the
beginning of
regenera-
In the third cycle, *ulfidation was
carried out a^ 650 °C. 1
The H 2 S breakthrough curves obtained for sorbent CM-5 are shown in Figure 29. As can be seen from this figure,
the H 2 S 82
level
in
cycle
I was
close to zero
350 SORBENT CM-3 (BULK CuMoO4 : moo :AI 203) 3 300
C-2
C-1
P - 1 atm T S - 538°C
250
1 NLET H 2
S s 1 mol %, H 2 s
H 2O -
20 mol %,
25 mol %, bal N2
E
S. V. - 2080 h r-1
N _
TREGEN ' 650°C (WITH NJAIR 70130)
a 200
350 r
C - CYCLE NO.
100 t.
50
0 1 0
1
I
0.2
0.4
i..(/
0.6
I
0.8
I
1.0
NORMALIZED ABSORPTION TIME, (t/t•) Figure 28. Breakthrough Curves in Successive Sulfidatiort Cycles of Bulk CM-3 Sorbent (CuMo0 4 :Mo03:Al203=0.41:1:12.52, molar) at 538°C; (Based on Cu2S formation) t* = 23 min
83
180
V
160
140
120
E a 100 r^
2N 80
s 0 60
40 20 0 NORMALIZED ABSORPTION TIME
(tit*)
Figure 29. Breakthroug{ Curves in Successive Sulfidation Cycles of Bulk CM-5 Sorbent (CuM004:Al203 = 1:1.1, molar) A 538°C and 650°C;
t* = 36 min (Based on Cu2S format-:o ►i)
84
4.
till nearly complete sorbent conversion based on copper component alone. The sorbent continued to be active even after complete theoretical conversion of copper (to Cu2S), possibly due to molybdenum contribution. The final H2S breakthrough at--10 ppm took place at a sorbent conversion of about 1.20 (based on Cu 2 S). Direct observation of the sorbent through the quartz reactor wall again indicated the fomation of a molten phase during sulfidation. Since the composition of this sorbent is copper-rich,
a complex sulfide-containing
eutectic is inferred (rather than the CuMo04-Mo03 phase alone). As discussed above for CM-3, this phase may explain the sub-equilibrium H2S levels observed in sulfidation of all CM-sorbents.
The second sulfidation cycle at 538°C, Figure 29, gave a slightly lower pre1
breakthrough sorbent conversion ( 1.15 based on Cu 2 S) and — 20 ppm H2S
1
level. This behavior, typical of all unsupported sorbents, can be attributed to some sintering (loss of surface area) and structural rearrangement during, particularly, the higher temperature (650°C) regeneration conditions.
The
interesting finding from both sulfidation regeneration cycles with CM -5 is that no molybdenum loss took place.
Thus, the CM-5 composition provided the
necessary improvement over CM-3 at 538°C. i
F
In the third cycle the sulfidation temperature was raised to 650°C. The corre-
3
i
a
sponding H 2 S breakthrough curve, shown in Figure 29, has the following interesting characteristics. First, nearly zero H 2 S breakthrough up to —0.40 sorbent conversion was observed. Thereafter, the H 2 S level increased slowly to --90 ppm at sorbent conversion of 0.70 (based on copper), and to —110 ppm before the final (third) breakthrough at sorbent conversion of 1.20. The three segments of this breakthrough curve are indicative of different sulfidation mechanisms, the first (at zero H 2 S) being the most interesting.
A sample of CM-5 sulfided at 650°C was analyzed by XRD.
The results are
included in Table 8. The material had high crystallinity and contained
Cut-xS, little unconverted CuM004, and Al203.
Its major components
were M002 (or Cu 6Mo4015) and an unidentified complex phase containing most of the sulfur compounds. Further characterization is clearly necessary to elucidate the sulfidation mechanism in this complex sorbent.
At the end
t
C
—
^
85
r
of sulfidation, a thin layer of a deposit was observed on the reactor wall downstream of the bed indicating that loss of molybdenum had taken place from this sorbent at 650%. In view of this, different CM-sorbents containing higher concentrations of copper were examined next in an effort to suppress the Mo loss at all temperatures of interest.
Sorbents CM-6 through CM-9. The aim with the new sorbents, CM-6 through CM-9, prepared and tested after CM-5, was to identify a sorbent composition and preparation conditions that would stabilize the molybdenum component in the sorbent matrix at the temperature of 650% (for MCFC applications). Table 7 shows the properties of the new sorbents CM-6 through 9. Sorbents CM-6 and 8 were prepared with the same molar ratio of Cu/Mo = 3/1. Different crystalline phases were, however, identified in these two sorbents (Table 7) as a result of different preparation conditions. Sorbent CM-8 was prepared with excess citric acid and, after dehydration in the rotary evaporator and vacuum oven, it was calcined at 650% for 12 hours, while sorbent CM-6 was calcined at 550°C for 4 hours. -
Sorbent CM-7 contained an even higher Cu/Mo molar ratio (=6/1).
Finally, CM-9 was prepared in highly crystalline form with Cu/Mo = 3/2, molar, without aluminum in the precursor solution. In sorbents CM-6, 7 and 8, all containing aluminum oxides, no Al-compounds were detected by XRD. Apparently, under the employed preparation conditions, the aluminum compounds were either very microcrystal line (and, thus, unidentifiable by XRD) or really amorphous. The new CM-sorbents had high surface area (>30 m 2/g) and low density. They were sieved to -20+40 mesh particles, loaded in the reactor without mixing them with inerts (due to their high specific volume), and tested at the same flow rates (200 cc/min) and space velocity (-2000 hr -1 ) as the CFA sorbent, which had approximately seven times hi g her density. Table 14 summarizes the sulfidation data from runs with CM-6,7 and 8. In all tests, the sulfidation gas had a molar composition of 15% H 2 , 25% H2O, 0.5% H 2S and 59.5% N2. The CM-6 sorbent was tested at 538-650°C, Viile CM-7 through 9 were tested at 650°C only. It is noteworthy that no volume shrinkage took place with any of the CM-sorbents even after several cycles of operation.
The CM-6 and 8 sorbents had the same molar ratio of Cu0:Mo03:Al203 (= 1:1/3:1). Both contained a large amount of amorphous material incorporating most of the aluminum compounds. CM-8 had the lowest density of all the CM-
86
4I- W N1 z P-4 O LLA
C I-
F'
E
et
Ln
O
Ln
N
0
N
M
Ln
Q
Ln
CD
O
00
O 00
O
Ln
M
Ln M
N M
c
.r E
L rT O LCN 4 - 10
S
G7
C S
FY
Q W
00 n
n
Ln
O .4
Q
f` n
M Q
n 119
n Ln
m
O
CO t`
00 r\
-4
O•
O
Ln
14
O•
O.
O.
O.
O
O•
O0
O•
r`
O.
r`
O.
r-
O•
r`
M-\
O •
tO
t/1 O N Z*
O
O
119 -4
O -4
O m
O O
Ln 01
O -4
O •••r
N O
O On
119 -4
N •••1
to O
O
O
O
O
O
O
O
O
O
O
tO
-4 r`
tO
t0
OD w
tO 1 S
U G
.r.
E a
N
F-
z W
H
m
N 06 E
SCL
O N i U
C7
S
O
U_ 1
W
a.
a.
O
U
Q W W a' a- cc 1
x
z O_ Fd 0
L N -I W L
t0
.••L
O N
Ln rN
119 Ln
O N
O
00
O
co
111
Ln
tO
Ln tO Lr9
Ln t0
LO
Q
O
Ln
st
O
O
O
O
O
O
N O
Ln
LCt
oI
.1 0I
14 o
Cl
et1 a
to .-•1 i o
1Cf ••-r 1 0
Ln r4 1 0
O
O
O
O
O
O
O
O
O
O N
O N
O N
O tO
N N
O
et
O
.-9
tO
-cr
r`
00
-
Q
d'
O N
O N
tO N N
tD N N
O Ln
N N
t0
M
OD
O OD
119
N
O
O
O
O
Ln
et
Ln
N
O U O 4J
Ln 1 0 -4
O
O
O
tO N
tO N
tD N
Ln
•w r^ N
et
1 0
Ln
-4
4J
U w O C O .y
Ln
L
N
C
O
0C14 S
} Z
O
0• Ln
N
0•
to
N
0.
to
N
0. Lc) N
c,• N
Ln
Ln 01
N
0. to N
0• N N
o Ln
N
O
o
0
0
u
Ln
Ln
Ln
Ln
4J
N
N
N
N
F-
N CL r
N
O E U
J
1- 4 1
Q.
LL
r•.
NN E x
FW L-r U U 0
3 N FN W F-
F' Y
O tD
O
O
_
OC tY
k
O N
Ln 01
Z
Ln
Ln 01
Ln
Ln C1
U9
01
Ln
Ln
Ln 01 Ln
Ln 01
Ln
Ln
Ln
O.
O•
Ln 01
LO
•1.1
04
01
Ln 01
Ln
LO
Ln
O•
O.
O•
O•
r-4
-4
Gl
N
r a E O • u C
Ln
E O
;v S
N Q
O
W
E
N
O•
to
Ln
119
O
N
0:
O.
.--1
N
W
O•
.••L
r-1
O• . -r Ln
Ln
119
O
O
LO
O
0
0
0
O
O O
O Ln
U
M
0
to
%D
tD
O.
O•
111
1n
r-L
.-•t
to
.-•r
Ln r4
O Ln o
O Ln 0
O
o
O Ln O
Cl Ln
00 M
O Ln tO
Ln
t0
Ln
Ln r-L
Ln
r-d
O
0
O Ln 0
Ln
O
O LC)
t0
tO
Ln
Ln
0
O Ln 0
O Ln 0
O Ln
O Ln
Ln
0
0 Ln
tO
tD
tO
LO
—
^—
Ln
r Z
.-1
N
M
qd-
Ln
N
M
t0 1
^ Q1 Lt f 00
1
Cn Ln q1t r-L
OD 1
U
O
U
v
U
-4
-41
N
N
O1 E
O •r- 00
CD ++ i N E r U
u ro u --
0) • r C Z y./ • r Ci E L L O r` R3
tD
C = N
4j -0 4- C
M
U
FZ W m
uu
—
W J
U O
\ 4-
CT1 tD
-i Ln
OJ ^ LO },1 w to L
Q/ ^ to ^ OJ E
Ll(U 4••t U ^
O
s
E
^, 1
87
.
W
sorbents. Figure 30 shows H 2 S breakthrough curves from five cycles of sulfidation/regeneration with the CM-6 sorbent. At all temperatures the pre breakthrough levels of H2S remained less than 10 ppm up to 0.70 sorbent conversion (calculated for Cu2S formation).
At 538°C (cycles 1 and 5) the exit H2S Higher than 100% conversion at
level was nearly zero until breakthrough.
breakthrough indicates that sulfides of copper other than Cu 2 S are also - j
formed, perhaps incorporating some molybdenum. Sulfur loadings at breakthrough were-0.076 gS/g sorbent. In all cycles steep H 2S elution profiles were obtained, see Figure 30. Cycles 3 and 4 were run at 650°C. Still, the same high absorption capacity was observed, particulary in cycle 4. However at the end of this sulfidation (but not at 538 or 600 0 C), a thin film of a deposit was observed on the cooler end of the quartz tube, later identified to contain molybdenum. After cycle 4, a sulfidation was run at 538°C with
r-4-)
O
U, a
N
^-
O O W N
>o
C%4
Z Z E 75 UU rl N M O N UU UU U Up U= N O ° II II w p O J 00 ^ ^ NO
= cn n
^ J
CD M
C O Z
N
W
W
¢ C
•o
E
V cr
O Om
Ln
1 ^
1
ti O
> •r
140
o
00
r^
OD
M
ON
ZW
c
rho
LLJ
e-
N
vr o
00
U
^
4- O O •r
z Ln ^ 1D
v
0 M
0 L s +) +) rO
Y R1 l0
^ 1 s_
C7
O
U
0
Z Q
rl
O
O
M
S-
:3
•r
O CD O N °O D ° O ^ p N OW ^ ° M M N N N r"4 '""1
(wdd) S Z H 1311(10
89
O
I
cm00
LL-
U
280
C -4 240
SORBENT CM-7 (CuO : 116 Moo
: 3
C-3 C-1
AI 203)
P = 1 atm 200
E a _
S = 0.5mol %, H 2 = 15mol % INLET H 2 H S = 25mol %, N 2 = 59.5mol % 2
160
NN S
tz
T S = 650°C
120
^C-2
S. V. = 2260 hr-1 6500C T REGEN = C = CYCLE NO.
1
0
i 80 I. 40
Q^ 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.21.3 NORMALIZED ABSORPTION TIME Wt*) Figure 31. Breakthrough Curves in Successive Sulfidation Cycles of Sorbent CM-7 at 650°C; t* = 68 mi;i (Based on Cu 2 S formation)
90
(as in CM-6 and 8), the sorbent performance was optimized. With still higher Cu/Mo ratios (e.g., 6/1 as in CM-1), the excess Cu0 present appears to reduce the sulfidation efficiency of the sorbent. Another test was performed with sorbent CM-9, which contained no aluminum oxides and had a Cu/Mo molar ratio of 3/2. This sorbent was found by XRD to be highly crystalline (Table 1). After mixing it with inerts CM-9 was tested at 650°C. A eutectic was formed during reaction (clearly visible through the quartz tube). Still, this basically unsupported material gave better than 25% sorbent conversion (based on Cu2S formation) at breakthrough. No deposit was found on the cooler part of the reactor tube after this experiment.
The complexity of CM-sorbents may be in part explained in tersas of what is known from the literature (15, 22) about cupric and cupro^-s :i1olybdates. The subsolidus phase diagram of the Cu 20-CuO-Mo0 3 system was given in a recent publication (15).
Five chemical compounds, CuM004, Cu3Mo209, Cu2Mo3010•
Cu6MO4015• and Cu4-xMo3012 (0.10< x < 0.40), were identified in this system. The cupric molybdates CuM004 and Cu3Mo209 are stable in air up to 820°C
and 855°C, respectively, melting at these temperatures with simultaneous decomposition (oxygen loss). Congruent melting points of cuprous molybdates Cu 2Mo 30 10 and Cu 6 Mo 4 0 15 , in argon, are 532%
and 466°C, respectively.
The
non-stoiciometric phase Cu4-xMo3012=Cu32+Cu°1-003 6+0 12, melts in argon
4
between 630°C and 650°C depending on the values of x and at 525-530°C undergoes polymorphic transformation. Several areas of coexistence of the above mentioned phases have been determined. In the reducing atmospheres used in our sulfi-
4 1
dation experiments, eutectic melt formation was observed with all CM-sorbents in the temperature range of 538-650°C. These eutectics possibly include sulfide compounds (as well as cuprous-cupric-molybdates). While the composition of these complex intermediates has not yet been identified, the h4gh H2S removal efficiency of the CM-sorbents may be attributed to better thermodynamic equilibria associated with these eutectics.
In view of the eutectic formation, it then became apparent that further work with CM-sorbents should ;nvolve preparation of alumina-supported copper molybdates with Cu0:Mo03 molar ratio of 3/1 (as in CM-6,8). The porous support would provide the necessary surface Brea and stability to the intermediate coma
plex eutectics formed. Accordingly, the alumina-supported CM-10 and 11 sor• bents were prepared and tested as described in a previous section of this
report. 91
i
Experimental work under this project has aimed at developing new improved sorbent materials for high-temperature desulfurization of NO gas streams. Two classes of sorbents were investigated far this purpose. Class A consisted of alumina-supported mixed oxides that form eutectic melts coating the pore surface of the support during operation. The sorbent. systems Zn-V-O and Cu-Mo-O were examined in this class. In parallel development, class B sorbents were prepared and tested. These sorbents consisted of bulk porous mixed oxides, e.g., Zn-Fe-O, Cu-Fe-U, Cu-Fe-A1-0, Cu-Mo-A1-0, prepared from homogeneous organic precursors. Both classes of sorbents are characterized by high surface area and porosity that enhance reaction kinetics. The sulfidation/regeneration performance of these sorbents was examined in many cycles of operation in the temperature range of 538-100°C. High H2S removal efficiency, improved sorbent utilizi^%*A on and good regenerability were desired features of the sorbents developed in this work.
Class A Sorbents. The alumina-supported Zn-V-O sorbents exhibit several
attractive properties, e.g., (a) high H 2 S removal effic-'-ncy (to less than 0.5 ppm H 2 S at 700°C), (b) no loss of surface area, and (c) ZnO stability against reduction/volatilization. These are all related to the presence of vanadium (and eutectic formation). However, the chief drawbacks of these sorbents, namely their low conversion and the release of some S0 2 during sulfidation, are also caused by their vanadium oxide components.
Better overall performance is displayed by the alumina-supported Cu-Mo-O sorbents. For example, sorbent CM-10 containing 11 wt% loading of CuO and M003 on the UCI T-2432 alumina gave very reproducible (over 13 cycles) high sulfidaZion performance at 650°C. This is summarized in Table 15. The observed high H 2 S removal efficiency of this and all other Cu-Mo-O sorbents is attributed to the presence of a complex eutectic phase during sulfidation with much better sulfidation equilibria than each of the constituent sorbent oxides. The single limitation of CM-sorbents concerns loss of M00 3 at a temperature of 650°C or higher during regeneration with steam-air mixtures. In dry regeneration no loss of M003 was detected. .I 92
TABLE 15. SULFIDATION PERFORMANCE OF NOVEL SORBENT MATERIALS
rl
PRE-BREAK-
SORBENT T
FEED GAS COMPOSITION —7mo X
THROUGH H S P pm
AT BREAKTHROUGH
SORRM CONVERSION
-7tj7_
j ETD " 'SJ LOADING
i
g sorbent ZF (ZnO:Fe 20 3)
600
15 H2, 25 H 2 0 , 1 H2S S,, 59 N22
CF (CuO:Fe 2 0 3)
538
20 H2, 6.8 H2O, 0.26 H2S, 72.94 N2
650 600
CA 2 (CUO:A1 0 3)
538
CM-6 (CuO:1/3MoO3:Al203)
0.318
5-lU
0.80
0.267
90
0.90
0.300
38
0.80
0.267
0-10
0.75
0.066
100
0.80
0.070
