FINAL REPORT
A SURVEY OF METAL OXIDES AS SORBENTS
FOR OXIDES OF SULFUR
Contract No. 86-67-51
AVSSD-0043-69-RR
February 1969
Prepared by
AVCO CORPORATION
Space Systems Division
Advanced Chemical Processes Section
Lowell, Massachusetts 01851
-------
FINAL REPORT
A SURVEY OF METAL OXIDES AS SORBENTS
FOR OXIDES OF SULFUR
Contract No. 86-67-51
AVSSD-0043-69-RR
February 1969
Prepared by
AVCO CORPORATION
Space Systems Division
Advanced Chemical Processes Section
Lowell, Massachusetts 01851
-------
TABLE OF CONTENTS
I. SUMMARY.
A. Background 1
B. Experimental Program j_
C. Results and Recommendations j_
II. INTRODUCTION 13
A. General Considerations 13
B. Sorbent Evaluation 13
Determination of Physical and Chemical Properties 13
Experimental Procedure 13
Procedure for Sorption Rate, Sorption Capacity,
and Regenerability 17
Weight Gain as an Indication of Sorption 20
C. Preparation of Sorbents 20
III. SORBENT STUDIES 22
1. Copper Oxide-Alumina 22
a. Thermochemistry 22
b. Preparation and Evaluation 2k
2. Copper Oxide-Silica 3^
a. Thermochemistry 3^
b. Preparation and Evaluation 35
3- Iron Oxide-Alumina Uo
a. Thermochemistry kO
b. Preparation and Evaluation Uo
k. Iron Oxide-Silica ^3
a. Thermochemistry ^3
b. Preparation and Evaluation ^-5
5. Manganese Oxide-Alumina U5
a. Thermochemistry ^-5
One-Stage Regeneration with Reducing Gas 1+9
Two-Stage Regeneration with Reducing Gas 53
b. Preparation and Evaluation - Manganese
Oxide-Alumina 55
APPENDIX I 58
Summary of Sorbents Exhibiting Marginal Performance 58
ii
-------
TABLE OF CONTENTS (cont'd)
1.
Co-Precipitated Sorbents . . . . .
. . . . .
. . . .
a.
Series I, Al-l through Al-14 .
. . . . . . .
. . . . .
b.
Series II, Al-SA, B, C; Si-SA, B, C,;
Al-9A, B, C; Al-12B, C, D; Si-12B, C, D. .
. . . . . .
c.
Series III - Cu-V-l, 2; Co-Al-l; Cr-Al-2;
Fe-V-l; Fe-Zr-l; Mn-Si-l; Mn-V-l; Ni-Al-l. . . .
2.
Sorbents Made by Impregnation. . .
3.
Commercially Available Sorbents.
. . . . . .
. . . .
APPENDIX II. . . . . . . .
. . . . .
. . . . . .
. . . .
iii
58
58
58
68
75
78
79
-------
FIGURE
LIST OF ILLUSTRATIONS
1
TGA Apparatus. . .
.....
.....
2
Gas Flow Control. . . .
. . . .
......
.......
3
4
Sorption/Regeneration History of Sorbent No. Cu-Al-l. . .
Porosity Determination of sorbent No. Cu-Al-l . . . . . .
5
6
Sorption/Regeneration History of Sorbent No. Cu-Al-2. . .
Sorption/Regeneration History of Sorbent No. Al-8A. ..
7
8
Sorption/Regeneration History of Sorbent No. Al-8B. . .
Sorption/Regeneration History of Sorbent No. Cu-Si-l.
9
Sorption/Regeneration History of Sorbent No. si-8B.
10
Sorption/Regeneration History of Sorbent No. si-8c.
11
Sorption/Regeneration History of Sorbent No. Fe-Al-l. . .
12
Sorption/Regeneration History of Sorbent No. Fe-Si-l. . .
13
14
Sorption/Regeneration History of Sorbent No. Mn-Al-2. . .
Sorption/Regeneration History of Sorbent No. Al-12B
15
16
Sorption/Regeneration History of Sorbent No. Al-12C
Sorption/Regeneration History of Sorbent No. Al-12D
17
18
Sorption/Regeneration History of Sorbent No. Si-12B
Sorption/Regeneration History of Sorbent No. Co-Al-l. . .
19
Porosity Determination of ~orbent No. ~e-Zr-l . . . . . .
20
Sorption/Regeneration History of Sorbent No. Fe-Zr-1. . .
21
Sorption/Regeneration History of Sorbent No. Ni-Al-1. . .
iv
15
16
27
28
31
32
33
36
38
39
44
46
57
64
65
66
67
70
73
74
76
-------
TABLE I
II
III
IV
V
VI
VII
VIII
XII
XIII
XIV
XVI
LIST OF TABLES
Sorption and Regeneration Characteristics. . . .
Sorbents Arranged According to Capacity. . . .
. . . .
Sorbents Arranged According to Percent Utilization
of Stoichiometric Capacity. . . . . . . . . . .
Sorbents Arranged According to Initial Rate. . .
Sorbents Arranged According to Half-Time. .
.....
Precipitation Ranges in NH40H and (NH4)2C03'
.....
Equilibrium Product Distribution for S02
Sorption on Copper Oxide. . . . . . . .
.......
Equilibrium Product Distribution for Regeneration
of CU2S05 in Oxygen-Nitrogen Gas. . . . . . . . . . . .
IX
Equilibrium Product Distribution for Sorption
of S02 on Excess Iron Oxide. . . . . . . . .
Equilibrium Product Distribution for Thermal Regen-
eration of Ferric Sulfate in Nitrogen. . . . . .
. . . . .
X
XI
Sorption of Flue Gas on Mn203'
. . . . .
. . . .
MnS04 Regeneration in Nitrogen
. . . . . .
Summary of Equilibrium Computations on One-Stage
Regeneration. . . . . . . . . . . . . . . . . . . . .
Properties of Co-Precipitates, Series I. . .
. . . . .
XV
Properties of Co-Precipitates, Series II
. . . .
Properties of Impregnates Sorbents . . .
. . . .
v
4
8
10
11
12
2l
23
25
41
42
48
50
52
59
63
77
-------
-1-
1.
SUMMARY
A.
Background
Alkalized alumina has for some time been under consideration as a
sorbent for the removal of sulfur oxides from flue gas. It has the ability
to significantly reduce the level of sulfur oxides concentration in a flue
gas and is capable of subsequent regeneration in a reducing atmosphere.
Some inadequacies have been encountered with alkalized alumina, however.
Most significant of these is the poor physical strength of the particles
which causes large attrition losses in use. In addition, the requirement
of different temperatures for sorption and regeneration steps, nominally
350 and 650oC, respectively, places a burden of additional heat exchange
on the process. The requirement for chemical regeneration (e.g., hydrogen)
also contributes to the cost of the process. Finally, although sorption
and regeneration rates on alkalized alumina appear to be satisfactory, any
improvement of rate could only be considered advantageous. The desirability
of an alternative solid sorbent to overcome these weaknesses is apparent.
This study was thus undertaken to assess the possibility of utilizing one of
several possible chemically active metal oxides to sorb oxides of sulfur
from flue gas.
Experimental Program
B.
In the course of this study seventeen metal oxides, on one or more
supporting oxides, were evaluated for their sorbing and regenerating
characteristics. The sorbents investigated were for the most part made by
co-precipitation of the hydroxides of the "active" and "supporting" oxides
followed by calcining, analogous to the preparation of alkalized alumina.
Several catalyst supports impregnated with the "active oxide" as well as
commercially available materials were also evaluated.
The extent and rate of sorption and regeneration of these materials
was determined by following the weight change of a sample exposed to a
synthesized typical flue or regenerating gas with time (see page 17). A
thermal gravimetric analysis (TGA) apparatus, consisting of an analytical
balance supporting a basket of the sample in a furnace tube with the flue
gas atmosphere (see page 14) was used for these evaluations. Regeneration
of sulfur-loaded sorbents was carried out thermally (in nitrogen) or chemically
(in hydrogen).
Where thermochemical data could be found, equilibrium for sorption
of sulfur oxides was calculated and compared to actual experimental results.
These computations were frequently very helpful in predicting or explaining
the behavior of some sorbents.
C.
Results and Recommendations
The best overall perspective of this survey may be gained from Table I
(pages 4-7) in which every sorbent evaluated is presented. The arrangement of
this table is alphabetic according to the chemical symbol of the cationic
-------
-2-
portion of first the "active oxide" and then the "supporting oxide". The
next columns indicate the weight percent of "active oxide" (page 13), the
stoichiometric capacity (page 13) and the surface area (page 13) of the
sorbent~ The central portion of the table gives a summary 8f the sorption
characteristics of a given sorbent at 300oC, 450oC, and 550 C. Since the
purpose of this table is to summarize the performance of all the sorbents
investigated in this study, the values given for any set of conditions are
the average of the results which are presented in much greater detail in
the body of the report. The sorption characteristics listed are defined as
follows: The initial rate is the sorption rate at time = 0, where the rate
is grams of 802/100 grams sorbent - hour - mole fraction 802 in the flue
gas; the percent utilization of stoichiometric capacity (page 13) is an
indication of the extent of reaction of 802 with the amount of "active oxide"
in the sorbent as found by analysis (page 13); the half times for a sorption
and subseQuent regeneration are the time reQuired to sorb to one half of
saturation for the particular sorbent under the condition indicated (page 19).
A comparison of the relative performance of each of the materials
evaluated is given in Tables II, (page 8) III, (page 10) IV, (page 11) and
V (page 12). These tables list each sorbent according to stoichiometric
capacity, percent utilization, initial rate, and half-time, with the sorbent
showing the best value being placed at the top of the list.
On comparing these four tables one finds that some of the capper
oxide-alumina and manganese oxide-alumina sorbents, namely: Cu-Al-l, Au-Al-
2, Mn-Al-2, and Al-9A occur in high positions in Tables III, IV, and V.
This reflects the generally good performances of these sorbents. Other
sorbents occupy a combination of both high and low positions in these tables.
80rbents with higher stoichiometric capacities, a Quality which is
directly related to the amount of active oxide in the sorbent, were found to
have lower surface areas reflecting a less porous structure. Porosity is the
most important single factor on the extent of utilization of the stoichiometric
capacity of a given sorbent. Thus, the higher the stoichiometric capacity,
the less accessible is that capacity. For this reason, sorbents with high
percent utilization are generally very low on the stoichiometric capacity
list. Again, a higher proport~on of active oxide in the sorbent reduces the
porosity.
Tables IV (page 11), and V (page 12) which are listed according to
initial rate and half-time, respectively, are two different aspects of the
same phenomenon, namely, sorption rate. One would therefore expect fast acting
sorbents to appear in high positions in both these lists. This is certainly
true for Cu-Al-l and Cu-Al-2. That there is not a better correspondence for
more sorbents in these two lists is a result of other factors. A high
-------
-3-
initial rate can decline very rapidly and if the sorption then continues
for a long time at a slower rate, the high initial rate has little
influence on the half-time which will then become large, as was the case
with Mn-Al-2. On the other hand, a sorption which has a moderate initial
rate but sustains a good rate throughout the sorption can have a relatively
good half-time, such as Al-9-A, Al-12-D, and Al-12B. It is interesting to
note that both the copper oxide-alumina sorbents, CU-Al-l, Cu-Al-2, share
high initial rates as well as short half-times, indicating a sustained high
sorption rate.
As was mentioned earlier, a sorbent with a higher proportion of
active oxide generally has a lower porosity. Low porosity would be expected
to result in low sorption rate, both initially and throughout the sorption.
From this it follows that sorbents with high stoichiometric capacity
(resulting from high proportion of active oxide - hence low porosity) would
not show good sorption rates. This is une~uivocally demonstrated in
Tables IV and V.
All of the sorbents prepared and evaluated in this study were found
to be thermally regenerable at a temperature of about 650oC. In this regard,
they have an advantage over alkalized alumina which re~uires a reducing
atmosphere. Thermal regeneration may not be practical, however, where the
concentration of sulfur compounds in the regenerate gas is low. Hydrogen
regeneration was also attempted with one of the copper oxide-alumina sorbents
(Cu-Al-2). With this sorbent, it was determined that complete regeneration
was realized at 4000c to 450oC, a temperature at which sorption of 802 also
proceeds very satisfactorily. This suggests the possibility of a sorption-
regeneration cycle all at the same temperature and obviates the cost of
heat exchange re~uired by the differing sorption and regeneration tempera-
tures necessary with alkalized alumina.
The outstanding performance of the copper oxide sorbents in this
study suggests that further investigation of this material should be under-
taken. Manganese and iron oxide sorbents showed promising rates and
capacities and might well merit further work.
-------
TABLE I
SORPTION AND REGENERATION CHARACTERISTICS
Q) . 300°C 450°C 550°C ~
'd .0 °
Q) .~ H "kl ~ . . .r!
'd ° . . . . . . l! . . ~ -j-J
.~ 0 (/) l! . ~ . ~ H . ~ . ~ . ~ ~ 'g ro
"' p., 'g ~'g ..t:: p., 'g ~ 'g p., 'g ~
Q) 0 Q) . S ro Q)N S Q) N ro Q) N ro
'd I> p.,bO Q) -j-Jo w -j-Jo 0 bO -j-Jo 0 "' ~", .r! H
.~ bO .r! ro ~ J2t!) "' Q)", J2t!) "' ~ Q)"' J2t!) (/) Q)
~ -j-J 00 . Q) p::Q) . Q) . Q) p:: Q) Q) !
0 .r! () 0 ..t:: .~ rl'~ rl* ..t:: .~ rl.~ rl* ..t:: .~ rl.~ ~
-j-J ~ .rl Q) ~ () () () (/)
Q) H .g~ () .r! -j-J ro-j-J .~'N .r! -j-J Qj-j-J .~~ .r! -j-J ro-j-J -j-J ,.!.j
I> ° * Qj .r!N ° ~ ° ~ ° ~ ~ H
.r! p., .r! 0 ~ -j-JO -j-J ~ 'g~ -j-JO -j-J ~ 'g ~ -j-JO -j-J ~ 'g~ Q) Q) Qj
-j-J p., . Ot!) ~ .r! t!) t!) rl .r! t!) t!) rl .r! t!) t!) rl .0 bO S
() ~ ~ -j-J ~bO Qj ° Qj ~bO ro ° Qj ~ bO Qj ° ro H ro Q)
~ t!) S HS * ::q z::q HS * ::q z::q H S * ::q z::q ° p.., p::
bO t!)
Cu Al 37.7 30 242 204 49 3 7 222 56.5 14.5 17.5 198 75 18 26.5 Cu- 24
Al-l
28.9 23.2 184 295 141 6.5 5 384 152 8.7 5 550 142 10 5 Cu- 29 H2 I
AI-2 Regen. +=-
I
11.6 9.03 224 197 157 29 10 166 176 58 12 28 20 3 7 Al- 30
8-A
34.4 27.4 74 76.6 34 25 22 15.3 45 - - - Al- 30
8-B
75.3 62 10.5 1 6 5 Al 30
8-C
Al 16.8 15 175- 40 49 - AI-8 58
220
10 13 137 - 78 Ha~shaw
Al- CuO-Al203
Si - - - 1-6 75
Cu Si 59 97 48 234 11 57 14 557 72 40.5 60 400 73 17 40 Cu- 35
Si-l
4.75 3.8 3.0 - 0 - - - 0 - - - 0 - - Si- 37
8A
14.4 12 102 43 83 24 9 30 61.5 2.75 10 185 23 1.3 - Si- 37
8B
43.5 36 4.5 15.5 21.5 59 18 130 48.5 20.5 38 170 18 9 25 Si- 37
8C
,
~
-------
TABLE I (Cont'd)
SORPTION AND REGENERATION CHARACTERISTICS
':Innoc /. t;roc: t;nnOC
Q) . s::
'tj .0 ~16b °
Q) o~ ~ OM
'tj 0 . . . . . . . . . 1i1
o~ 0 10 l! . s:: . s:: l! . s:: . s:: l! . s:: . s::
"' Po 0g Wog Po OM W 0g ~ 0g W 0g ~
Q) 0 Q) . a Qj Q) Qj Q)C\! ro a Q)
'tj :> Pot\D Q) +' C\! 0 M +'0 0 t\D +' C\! 0 t\D OM ~
o~ t\D OM Qj 1i1 ~g "' ~ 0)"' ~(/) ... Q) "' ~g "' ~ 0)"' 10 0)
s:: +' 00 . Q) . 0) p:: 0) . Q) Q) .o
0 OM C,) 0 r-i~ ,..q o~ r-j o~ r-i~ ,..q o~ a r-i~ ,..q o~ r-i o~ A !
+' ~ .r-i Q) C,) C,) r-i OM C,) 10
Q) ~ ,..q",- C,) ro"'- OM +' ro+' o~~ OM +' ro+' o~~ OM +' ro+' +' ~
:> 0 ~ C,) C\! ro OM C\! 0 s:: 0 s:: 0 s:: s::
OM Po 0MO ~ +,0 +' ~ 0g~ +,0 +' ~ 0g~ +,0 +' ~ 0g~ Q) 0) ro
+' : . O(/) ~ OM (/) (/) r-i oM(/) (/) r-i OM (/) (/) r-i .0 be ~
C,) ~ +' s:: :f Qj ~~ ~:f ro 0 ro s:: t\D Qj 0 ro ~ ro
~ (/)6b H ~ :::x::: ~ :::x::: z:::x::: H a ~ :::x::: z:::x::: 0 p.., p::
(/)
Cu V - - 0 - - - - - - - - - - - - Cu- 68 Unsatisfac-
\1"-1 tory
Co Al 49 125 171 90 4 13 19 125 5 5 8 - Co- 69 '
- - - Vl
Al-l I
Cr Al 52.2 65 - - - - - - - - - - - - - Cr- 69
Al-2
- - 131 Al-6 58
Cr Al-Si 2.4 3.0 24.4 1-4 75
Fe Al 46 55 233 85 40 21 8 65 8 11 8 - - - - Fe- 40
Al-l
4.70 5.65 376 15 30.5 4 14 53 41 7 21 - 16 6 - Al- 62
12-B
13.6 16.3 340 38 33 10 23 23 37 19.5 65 0 0 60 Al- 62
12C
33.4 40.1 274 (6 2 3 54 3 4 31 0.1 3 Al- 62
12D
54.2 73 146 Al- 58
I 12
Al- 9.2 11.3 - .03 0.25 1-8 75
Si
Al- 3 3.7 78 Fluorite
Si
: I ~
-------
TABLE I (Cont'd)
SORPTION AND REGENERATION CHARACTERISTICS
300°C 1..500C 550°C s::
0) .
<0 .0 0
0) ] H C\l1E! . . .rl
<0 0 E! bD . . . H . . H . +'
'Q (I) .a . s:: . s:: .£: . s:: . s:: .£: . s:: .. a:!
"' p., 'g s:: .rl p., 'g ~ 'g p., 'g s:: s:: s::
0) 0 0) . E! a:! 0) a:! ~E! 0) C\l rU 0) C\l a:! ~'g bD
<0 :> p.,bD 0) +'C\l 0 +'0 0 bD +'0 0 .rl H
'Q bD .rl a:! !;! &J@ "' 0)", &J(J) 0> ~ 0)"' &J(J) "' 0). (I) 0)
s:: +' 00 . 0) p:: 0) . . 0) P::O) 0) .0
0 .rl t> 0 .£: .~ rl'~ ~ .£: .~ rl.~ rl~ .£: .~ rl.~ A ~
+' rl'-..... t> t> (I)
0) H "2'N t> .~ 'c\1 .rl +' a:!+' a:! C\l .rl +' a:!+' a:!'N 'rl +' a:!+' +' ~
:> 0 ~ a:! 0 S:: .rl 0 0 S:: 'rl 0 0 s:: s:: fi1
.rl p., .rl 0 «+-i +'0 +' «+-i 'g ~ +,(J) +' «+-i 'g ~ +'(J) +' «+-i 'g~ 0) 0)
+' ~ . O(J) H .rl (J) (J) rl .rl (J) rl .rl (J) rl .0 bD ~
t> ~ +' ~ s:: a:! ° a:! s:: bD a:! 0 a:! s:: bD a:! o a:! H a:!
-------
TABLE I (Concl'd)
SORPTION AND REGENERATION CHARACTERISTICS
(]) . . 300°C 1. 150°C l)l)OoC s::
"d P 0
(]) o~ H (\lIE! . . . OM
"d 0 E! till 11 . . H I~ . H . . ~
o~ 0 (/) . s:: . s:: ..q . . s:: ..q . s:: . s:: Qj
... (]) C\ Po 0g ~ 0g Po OM ~og Po 0g ~ 0g g,
(]) 0 (]) . E! Qj Qj (]) (\l Qj E! Q)(\l Qj
"d I> Potill (]) ~O 0 till ~O 0 till ~O 0 till OM H
o~ till OM Qj 1i! ~CI) ... (]) ... ~CI) ... (])... ~CI) ... (])... (/) (])
s:: ~ 00 . Q) 0:: Q) . (]) 0:: (]) . (]) 0:: (]) Q) 1
0 OM t) 0 ' 0 ~ t)(\l Qj oMO 0 s:: OM 0 0 s:: oMO 0 s:: s:: ti!
OM Po oMO r.-t ~CI) ~ r.-t 0g ~ ~CI) ~ r.-t 0g~ ~CI) ~ r.-t 0g ~ (]) (])
~ Po . 000 ~ OM 00 ..-1 OM 00 ..-1 OM 00 ..-1 P till ~
t) Jj ~ ~ ~~ Qj 0 Qj s:: till Qj 0 Qj s:: till Qj 0 Qj H Qj
...: ~ 00 E! ~ ::q Z::q H E! ~ ::q Z::q H E! ~ ::q Z::q 0 p.., 0::
till 00
2.86 2.59 289 42 69 8 18 37 29 4 24 3 17 8 20 Al- 62
9C I
Al- 7.3 6.6 245 1-7 '75 ---.:]
Si I
Mn Si 39 35 36 - 0 - - - 0 - - - 38 50 00 Mn- '72
Si-l
Mn V 62 56 1.1 - 1 10 15 - 7 40 5 - 3 - Mn- '72
V-I
Na Al 8.6 9.4 152 Al~4 58
Na Al- Al- 58
Mn 14
Na Al- - - 20 1-2 '75
Si
Ni Al 26.4 23 306 189 19 15 1 221 11 12 1 106 12 6 1 Ni- '75 H2
Al-l Regen.
24.7 23 275 Al-5 58
Al- - - 24 - - - 1-3 '75
Si
Zn Al 17.0 15 175 AI-7 58
Al- - - 22.8 1-5 '75
Si
- - - - - - - - - - - - - - - - - - '78 I :Ginde Mol
! I Sieve 5A
I
-------
Table II
Sorbents Arranged According to Stoichiometric Capacity
Stoichiometric % Utilization of Initial Rate
Capaci ty Stoichiometric at 300°C Half-Time
(gm S02/ Capacity (gm S02/100 gm Sorbent- at 300°C
Designation 100 gm Sorbent) Realized at 300°C hr.-mol fracto S02) ( min. )
Co-Al-l 125 4 90 13
Al-9 110
Cu-Si-l 91 11 234 51
Al-12 13
Cr-Al-2 65
Al-8-c 62 1
Fe-V-l 60 0
Mn-V-l 56 1 10 I
Fe-Al-l 55 40 85 121 CP
I
Fe-Si-l 48 0 0
Al-12-D 40 2 26 3
Fe-Zr-l 40 22 10 105
si-8-c 36 22 15 59
Mn-Si-l 35 0
Mn-Al-2 34 30 400 13
Si-12-D 32 0 0
Al-l 32
Al-2 31 8
Cu-Al-l 30 49 204 3
Al-8-B 21 34 11 25
Cu-Al-2 23 141 295 1
Ni-Al-l 23 19 189 15
Al-5 23
Al-12-C 16 33 38 10
Al-8 15 40
Al-1 15
Harshaw CuO-Al203 13 1
Si-12-C 13 0 0
si-8-B 12 83 53 24
Al-3 12 21 5
1-8 11 0.03
-------
Table II continued
Stoichiometric % Utilization of Initial Rate
Capaci ty Stoichiometric at 300°C Half-Time
(gm S02/ Capacity (gm S02/100 gm Sorbent- at 300°C
Designation 100 gm Sorbent) Realized at 300°C hr.-mol fracto SO?) (min. )
Al-4 9
Al-8-A 9 157 197 29
I-7 7
Al-12-B 6 30 15 4
Si-12-B 5 0 0
si-8A 4 0
Flori te 4
I-4 3 ,
Al-9-C 3 69 42 8 \0
,
Al-9-B 2 77 38 5
Al-9-A 1 146 61 4
-------
Table III
Sorbents Arranged According to Percent Utilization of Stoichiometric Capacity
% Utilization of Stoichiometric
Stoichiometric Ca'Pacity Initial rate at 300°C Half-time
Capacity Realized (gm S02/ (gm S02/100 gm Sorbent- at 300°C
Designation at 300°C 100 gm Sorbent) hr.-mol fracto S02) (min. )
Al-8A 15'7 9 19'7 29
Al-9-A 146 1 61 4
cu-Al-2 141 23 295 '7
si-8-B 83 12 53 24
Al-12-C '7'7 16 38 10
Al-9-C 69 3 42 8
Cu-Al-1 49 30 204 3
Fe-Al-1 40 55 85 121
Al-8-B 34 2'7 '7'7 25
Al-9-B 33 2 38 5
Al-12-B 31 6 15 4 I
Mn-Al-2 30 34 400 13 I-'
a
Fe-Zr-1 22 40 10 105 J
si-8-c 22 36 15 59
cu-Si-1 11 9'7 234 5'7
Co-Al-1 4 125 90 13
Al-12-D 2 40 26 3
-------
Table IV
Sorbents Arranged According to Initial Rate
Initial Rate
at 300°C Half-Time stoichiometric
(gm S02/gm Sorbent- at 300°C Capacity (gm S02/
Designation hr.-mol fracto SO?) (min. ) 100 gm Sorbent)
Mn-Al-2 400 13 34
Cu-Al-2 295 7 23
Cu-Si-1 234 57 97
Cu-Al-1 204 3 30
Al-8A 197 29 9
Ni-Al-1 189 15 23
Co-Al-1 90 13 125
Fe-Al-1 85 121 55
Al-8-B 77 25 27
Al-9-A 61 4 1
si-8B 53 24 12
Al-9-C 42 8 3
Al-12-C 38 10 16
Al-9-B 38 5 2
Al-12D 26 3 40
si-8c 16 59 36
Al-12-B 15 4 6
Fe-Zr-1 10 105 40
% utilization of
Stoichiometric Capacity
Realized at 3000C
30
141
11
49
157
19
4
40
34
146
83
69
77
33
2
22
31
22
I
I-'
I-'
I
-------
Table V
Sorbents Arranged According to Half-Time
ojo Utilization of Stoichiometric
Half-Time Stoichiometric Initial Rate at 300°C Capaci ty
at 300°C Capacity 300°C (@m S02/100 gm Sorbent- (gm S02/
Designation (min. ) Realized at hr.-mol fracto SO?) 100 gm Sorbent)
cu-Al-1 3 49 204 30
Al-12-D 3 2 26 40
Al-9-A 4 146 61 1
Al-12-B 4 31 15 6
Al-9-B 5 33 38 2
Cu-Al-2 7 141 295 23
Al-9-C 8 69 42 3
Al-12-C 10 33 38 16
Mn-Al-2 13 30 400 34
Co-Al-1 13 4 90 125
Ni-Al-1 15 19 189 I
23 f-I
si-8-B 24 83 53 12 I\)
I
Al-8-B 25 34 77 27
Al-8-11 29 157 197 9
Cu-Si-1 57 11 234 97
si-8c 59 22 16 36
Fe-Zr-1 105 22 10 40
Fe-Al-1 121 40 85 55
-------
-13-
II.
INTRODUCTION
General Considerations
A.
The ability of a material to sorb sulfur oxides depends to a large ex-
tent on the chemistry of the sorbent and sorbate; metal oxide and sulfur oxides
in this case. Chemical equilibrium must, of course, be favorable for both the
sorption and desorption (regeneration) if the system is to be successful, since
it is equilibrium which sets the limits on the extent to which a flue gas can
be cleaned of sulfur oxides and the nature and concentration of sulfur com-
pounds in the regenerate gas.
Of equal importance is the rate at which these sorption and desorption
reactions occur since it strongly influences the economics of the process. The
sorption process may be considered as the sequence in which sulfur oxide is
carried in the flue gas stream to the outside of the sorbent material where it
finds its way to the sorption sites by diffusion through the pores of the sorbent.
Having reached such a site, the actual bonding reaction between sulfur oxide
and sorbent takes place. Each of these steps has a characteristic rate. If
anyone step in the sequence is significantly slower than the other, the rate
for the total processes will be that of the slow step. Thus, the rate limiting
step could be bulk mass transfer or pore diffusion or sorbent-sorbate reaction
rate, for example. In some cases in the discussion of results of this study it
was possible to determine which mechanism was the rate limiting step, by corre-
lating with appropriate mathematical models.
B.
Sorbent Evaluation
Determination of Physical and Chemical Properties
1.
Surface Area
The surface area of each sorbent was determined by the BET method
using adsorption of nitrogen in a Perkins-Elmer 212 C Sorptometer.
2.
Acti ve Oxide Content
Each sorbent was analyzed, using standard wet chemical techniques,
to determine the amount of active oxide, supporting oxide or both. In cases
where only one constituent was determined, the other was taken as the difference,
allowing for volatiles. Since these quantitative analyses could only be made
for the metal and not the oxide, the results have been reported as the metal
corresponding to the specified amount of oxide. The question of which oxide,
in cases where several were possible such as iron and manganese, was resolved
by x-ray analysis or thermodynamics.
3.
Stoichiometric Capacity
Having determined the amount and type of active oxide present in
a sorbent, the capacity for sulfur oxides, based on a reaction between the active
oxide present and sulfur trioxide, was computed. The stoichiometric capacity
considers the reaction of all the active oxides and is expressed in terms of
sulfur dioxide.
-------
-14-
Experimental Procedure
1.
TGA Apparatus
The sorption and regeneration data are obtained by measuring changes
in the weight of a sample of sorbent during exposure to simulated flue and re-
generation gases. The weight is determined in a TGA type apparatus.
The apparatus consists of a means of supporting and weighing a
sample of sorbent which is held at constant temperature in a controllable atmos-
phere. The support and weighing system, the temperature control, and the gas
supply system will be described in turn. Standard operating conditions will
then be given.
2.
Support and Weighing System
The sample of sorbent is supported in a basket suspended from a
Mettler, H-6T balance. The balance is used to determine the sample weight to an
accuracy of ~ 0.2 mg. The sample basket is surrounded by a quartz tube with an
I.D. of 1.5 inches. Gases enter from the bottom of the tube. A two zone fur-
nace surrounds the tube. This equipment is shown schematically in Figure 1.
A light chain is used to connect the sample basket to the balance.
A loop and hook attached to the chain allow positioning of the sample in the
center of the hot zone or the cold zone without opening the apparatus.
3.
Temperature Control
A Lindburgh Hevi-Duty furnace, Model M-I018-s surrounds the
quartz tube. There are two zones in the furnace and the temperature of
each is controlled with an API Model 712 controller. The temperature is
sensed with chromel-alumel thermocouples. The bottom of the quartz tube
is packed with glass beads for uniform distribution and gas preheat.
The sample is supported in the cold zone (3000C) for sorption
and is moved to the hot zone (6500C) for regeneration by shortening the
support chain.
4.
Gas Flow Control
A schematic of the gas flow control system is shown in Figure 2.
Valves S-l and S-2 are three-way solenoids that are simultaneously operated
by a single switch. In the purge position, S-l allows purge N2 to enter
the TGA while the simulated flue gas or H2 entering S-2 is diverted to the
hood. The alternate position closes off the by-pass and purge N2 lines
and allows the flue gas or H2 to flow through S-2 and S-l to the TGA.
To effect a sorption or regeneration, the sorbent is exposed to
flue gas or nitrogen, respectively, for a short period of time. At the end of
the desired interval the solenoids are put in the purge position for 10 seconds
,
thus sweeping all reactive gases from the TGA. The purge N2 is then shut off
and the sample weight taken in static N2. The solenoids are then switched to
the alternate position allowing a second interval of Sorption or regeneration
to begin. Data are generated as tables of weight gain or loss vs. time. The
transitory sorptions were necessary because "noise" on the balances was caused
by slight variations in flow around the particles and hence, the weight could
not be determined with the flue gas flowing through the tube.
-------
Quartz Tube
LD. = 1. 5 inch
Length = 24 inches
Lindberg
Hevi-Duty
Model M-1018-s
Two Zone Furnace
Sample
FIGURE 1 TGA Apparatus
-15-
o 0 0
o 0
I
-------- .~
- ~~~e r
I
I
I
I
I
I I
H
Cool
Zone
I ,
I
i
I
Mettler
H-6T Balance
-v
I
i
I
II I
U'U
. I
I I
: i :
; I
I
I
I
i
I
I t
I
; !
~u
Chromel-Alumel
Thermocouple
-(~1
o 0 I
I
API Model 712
Temperature Controller
o
o
Chromel-Alumel
Thermocouple
Gas Inlet
-------
-16-
FIGURE 2
Flow Control
Gas
Air
R-l
*-l 3/8-25-C
Triflat 0
V-2
f_u-- u~-----
I ->-~st N2 +
-~ ~iL6-CD9' MOl '1
V-3 Triflat I I
-----~
H2
N2
02
R-5
1/16-12-8 0
Triflat
H20 (_~1___{9
-------
-17-
The simulated flue gas for sorption consists of N2, 02' H20,
NOx, and S02. C02 was not included in the simulated flue gas as it was
determined in early experiments that the sorption of C02 on alkalized
alumina was negligibly slow. The proper N2/02 mixture is obtained by
metering dry air and N2 through rotameters R-2 and R-l, respectively. This
mixture is then led to a vaporizer to which liquid water is metered through
rotameter R-5. All lines subsequent to the vaporizer are traced with
heating tapes. NOx and S02 are added to the moist N2/02 mixture downstream
of the vaporizer. Flows of these gases are also metered with rotameters.
5.
Conditions of Operation
Sorption temperature:
Sorption pressure:
Simulated flue gas:
300oC, 450oC, 5500C
1 atm
Component
Mole Fraction
N2
°2
H20
NOx
S02
0.894
0.029
0.075
0.0005
0.00324
Superficial gas velocity
during sorption:
2.4 ft/sec
Total simulated flue
gas flow:
24.3 SLPM
6500c
H2
10.4 SLPM
Regeneration temperature:
Regeneration gas:
Regeneration gas flow:
Superficial gas velocity
during regeneration:
47.3 cm/sec, 1.55 ft/sec
Procedure for Sorption Rate, Sorption Capacity, and Regenerability
Step by step procedure
1.
a.
Fresh sorbent is packed in a tube furnace and activated
in N2 at 6500C for 4 hours. The superficial velocity
is 12 cm/sec.
b.
A sample of activated sorbent weighing approximately
500 mg is charged to the TGA apparatus.
c.
The sample is exposed to an S02 free simulated flue
gas of the following composition:
-------
2.
-18-
N2 89.4% by volume
°2 2.gfo
H20 7.5%
NOx 500 ppm
d.
When the sample is equilibrated with the S02 free
flue gas the sorption cycle is begun. Flue gas
identical to that shown above with the addition of
0.324% by volume S02 is used. Initial sorption
intervals between weighings are 15 seconds. Near
saturation, where the rate of sorption is much slower,
intervals of 10 to 30 minutes are used.
e.
When no further weight gain is observed, the sorption
is complete and the sorbent is saturated. From the
weight gains and time intervals a table of weight
gain vs. time can be constructed.
f.
The regeneration is then started by raising the sorbent
basket to the 6500C zone and changing to nitrogen in
the TGA.
g.
After one minute, the flow of nitrogen is stopped and
the weight loss determined.
h.
step g is repeated at intervals expanding to several
minutes at the completion of regeneration. This
generates the data for a table of weight loss vs.
time exactly analogous to the sorption data.
i.
Steps c through h are repeated to give as many sorption
regeneration cycles as desired.
Basic Sorption Series
To evaluate each sorbent, a basic sequence was devised and
used throughout the study. This consisted of six sorption-regeneration
cycles. The first two sorptions were at 3000C, followed by two at 4500C,
and finally two at 5500C. Regeneration was at 6500C.
3.
Deviations from this sequence are noted where applicable.
Data Reduction and Presentation
a.
Sorption data
The sorption data are presented as sorption capacity
in weight S02/l00 weights of sorbent and as normalized
sorption rate vs. loading.
-------
-19...
The normalized sorption rate is derived from an
assumed sorption rate expression of the form
dW/dt = Y . f (W)
where W ~ sorbent loading, gm S02/gm sorbent
y = mole fraction S02 in gas
t = time
f = function of W
The normalized sorption rate is
(l/Y) (dW/dt)
and is given units of hr-l, and when plotted vs. W the
resulting curve is f(W).
Values of the normalized sorption rate are determined by
numerically differentiating the weight gain vs. time curves using three-
point Lagrangian differentiation formulas.
The half-times for a sorption and subsequent regeneration
are the times at which the weight gain corresponds to one half the satura-
tion weight gain. These half times are presented only to make possible a
comparison of the numerous sorbents evaluated in this study and hence
should not be construed as an indication that these sorptions are strictly
first order processes.
b.
Sorption-Regeneration History Charts
This type figure was devised to present the salient
features of the "basic" sorption series for each sorbent. The description
of Figure 3 (page 27 ) for Cu-Al-l is thus generally applicable to the
equivalent figure for other sorbents. Most of the numerical values given
in the table are self-explanatory or have been explained elsewhere in
this report; however, several remarks are in order. The designations Sl,
Rl, S2, R2, etc., correspond to first sorption, first regeneration, second
sorption, second regeneration, i.e., the actual sequence of the series.
The loading vs. time curve takes as zero loading the weight of the sample
after it has been brought to temperature in the simulated flue gas... less
S02. The time units of the absisca are arbitrary, but the general shape
of the sorption-time curve is preserved. It will be noted that regenera-
tion to constant weight, does not always coincide with the base line. The
difference corresponds to a net gain or loss in the sample weight. In
some cases, as with Cu-Al-2, the regenerated sample weight is considerably
below the base line, but the start of the subsequent sorption is higher.
This results from weight gain on exposure to the S02-free flue gas prior to
sorption.
-------
-20-
Weight Gain as an Indication of Sorption
The increasing weight of a sorbent sample was used as a
measure of sulfur oxide sorption throughout this study. This method has
the advantage of being quick and sensitive, permitting a large number of
samples to be evaluated with good reliability. Certain cautions must be
exercised, however, to be certain that S02 only is contributing to the
observed weight gain. For this reason, prior to each sorption run, the
sorbent sample was saturated with all the constituents of the flue gas -
less S02. The actual sorption run corresponded to the time the S02 was
introduced to the TGA apparatus containing the pretreated sorbent.
As a check, the S02 take-up on
compared to actual sulfur analysis. Duplicate
method gave results which bracketed the weight
within 15%.
one sample (Al-12) was
sulfur analyses by the Leco
gain results and agreed
Preparation of Sorbents
D.
Co-Precipitation
1.
In precipitating the hydroxide or carbonate of a given
substance, consideration must be given to the chemistry involved. Some
cases are relatively simple, and the combination of any amounts of base
and metallic salt will result in precipitation. On the other hand, some
substances exhibit amphoteric properties or have soluble complexes, which
could interfere with or preclude precipitation. To site a few examples
encountered in this study, aluminum hydroxide is amphoteric, copper forms
a soluble ammonium complex, and iron hydroxide freshly precipitated with
ammonium carbonate is soluble in an excess of the precipitant.
In order to avoid difficulties in co-precipitation which might
arise because of these complexes and amphoteric properties, a precipitation
titration was made for each metallic ion of interest. To do this, a solu-
tion was prepared of the same type and concentration as was to be used in
co-precipitation. Using a pH-meter and burette, the pH change with a given
quantity of ammonium hydroxide or ammonium carbonate was noted over a wide
pH range. At the same time, any changes in appearance, such as precipitate
formation, dissolving of the precipitate, or color change, were noted.
Having done this, a permissible pH range for the formation of a given
precipitate was established. These results are presented in Table VI.
The pH ranges indicated in TableVI are useful guidelines
for co-precipitation. To effectively co-precipitate two ionic species, a
pH must be selected which is common to the permissible range for each ion.
For example, if a co-precipitate of copper and aluminum hydroxides is
intended, the range for copper is seen to be 4.6 to 6.8 and that for
aluminum 3.9 to 6.4. The overlapping pH range, and the one in which
co-precipitation must be made is thus 4.6 to 6.4.
filtration.
The co-precipitates were separated from the liquid by
The resulting paste-like filter cake was then formed into
-------
Table VI
-21-
Precipitation Ranges in NH40H and (NH4)2c03
Active Metal Oxides
Metal
CuII
ZnII
CrIII
MnII
FeIII
CoII
NiII
Supporting Oxides
AlIII
ZrIV
SiIV
VV
NH40H
4.6 - 6.8
6.6 - 8.0
> 4.6
~ -3.0
> 2.6
">'-7.5
7.4 - 8.0
3.9 - 6.4
2.0 - 4.0
Acid titrations
<-7,5
7.6 - 6.6
pH Range
( HN03 )
( NH),bC03
4.6 - 6.0
6.5 - 7.2
> 4.2
> 7.1
- 2.5 --6.6
7.0 - 7.8
6.6 - 7.8
4.1 - -7.0+
2.0 - 4.0
-------
-22-
pellets. This pelletizing was done by working the paste into l/8-inch
diameter holes in a l/8-inch thick aluminum plate. On drying in air at
600c the pellets shrank and had sufficient strength that they could be
removed intact from the holes by gentle rapping of the plates. The pellets
were then humidity dried at 900C dry bulb, 700 wet bulb for 8 hours. This
was followed by firing in nitrogen at 6500C for 8 hours.
In this study, the co-precipitated sorbents in the later part of
the program, Series III, were prepared under the conditions indicated in
Table VI. Some of the earlier co-precipitates in Series I and II were,
however, made without the benefit of the guidelines derived from the
precipitation titrations discussed above. Series III sorbents may be
identified by the sample designation which contains the symbol of the
active oxide, the supporting oxide and a sequence number, e.g., Cu-Al-l
for the first series III copper oxide on alumina. Series I sorbents have
designations such as Al-8 or Si-12 indicating an alumina support or a silica
support. Series II uses similar designations followed by a letter, e.g.,
Al-8A, Si-12B, etc.
2.
Impregnation
Oxides of K, Na, Ni, Zn, Cu, Cr, Mn, and Fe were deposited
on a commercial porous carrier, Norton LA-3032 (83% Al203' 15.3% Si02).
The support was boiled in solutions of metal nitrate for 8 hours, washed,
dried and fired at 6500C to decompose the nitrate. These sorbents were
given sample designations 1-0 through 1-8.
Commercially Available Materials
3.
Several commercial materials were tested for S02 sorption
activity. These were Harshaw Catalyst, 10% CuO in alumina, Florite
(Florodin Company) and Linde Mol Sieve 5A.
III.
SORBENT STUDIES
1.
Copper Oxide-Alumina
a.
Thermochemistry
Equilibria for the sorption and regeneration of CuO were
computed. Although copper aluminates and silicates are known to exist
,
no thermodynamic data are available and their influence could not be
considered.
The calculated sorption compositions are given in Table VII.
The data indicate that at temperatures below 6500C, copper sulfate is
formed. The sulfate is CuO.CuS04. Note that in excess S02, the CuO.CuS04
may further convert to cus04. Measured capacities indicate such an event.
It is. apparent that below about 5?00C, dissociation pr~ssure of S02 and
S03 w1ll be less than 0.02% assum1ng that the sorbent 1S unsaturated.
Thus, at temperatures less than 5900C, removal of S02 is 90%. The S03
concentration is a maximum of 0.06% at 6500C and declines at higher
temperatures.
-------
-23-
Table VII
Sorbent Flue Gas
Temperature °c
427 527 627 727 827
A. Mole Fraction in Gas Phase
CO 0 0 0 0 0
cas 0 0 0 0 0
C02 0.1404 0.1404 0.1400 0.1400 0.1400
H2 0 0 0 0 0
H20 0 . 06020 0.06020 0.06013 0.0600 0.0600
H2S 0 0 0 0 0
02 0.02909 0.02909 0.02923 0.02975 0.02989
802 2.310x10-9 1.909x10-6 3.432x10-4 1. 498x10-3 1.785x10-3
803 1. 09Ox10-7 1. 124x10-5 4.073x10-4 5.024x10-4 2. 150x10-4
82 0 0 ° 0 0
N2 0.7703 0.7703 0.7696 0.7682 0.7682
B. Mole Fract ion in Condensed Phase
CuO 2.000xlO-7 0.01292 0.5449 1.000 1.000
CU20 0 0 0 0 0
Cu8 0 0 0 0 0
Cu804 0 0 0 0 °
Cu2S05 1.000 0.9871 0.4551 0 0
H2S04 0 0 0 0 0
8 0 0 0 0 0
-------
-24-
Regeneration calculations were made by computing the equilibrium
products of thermal decomposition of CU2S05 in air and in nitrogen. The
phase CU2S05 represents the phase formed during equilibrium sorption.
The results of the computations are given in Table VIII. The
numbers are meaningful only as long as there is condensed phase CU2S05
present. Under these conditions the mole fractions of S02 and SO~ in the
gas phase are equal to their equilibrium partial pressures over tile sulfate.
The partial pressure of S02 plus S03 is 0.016 atm at 726°c
and 0.3 atm at 827°C. Thus, regeneration temperatures of 7500C and higher
are indicated if a regenerate gas high in SOx is desired. The formation
of aluminate or silicate might improve the regenerability by stabilizing
the oxide form of copper but the absence of thermochemical data leaves
this question unanswered.
The thermochemistry of hydrogen reduction of CuS04 predicts
the formation of CuS or CU2S. Since it has been demonstrated experimentally
(see below) that H2 regenerated copper-alumina sorbents result in conversion
to elemental Cu, it is conjectured that the presence of CuA1204 or CU2A1204
stabilizes the oxide. Again, lack of thermochemical data precludes further
analysis.
Preparation and Evaluation - CuO-Alumina
b.
i.
Cu-Al-l
Sorbent Cu-Al-l was one of the most promising sorbents
made in the course of this program. Sorption rate and capacity were
found to be completely comparable to alkalized alumina. Loaded sorbent
was regenerated both thermally and in hydrogen. The hydrogen regeneration
can be made at a much lower temperature than is possible with alkalized
alumina, which suggests the possibility of an isothermal sorption regenera-
tion process. As a result of these outstanding features, this sorbent
received additional emphasis.
Sorbent Cu-Al-l was synthesized by vigorously stirring
a 1.5 M (NH4)2CO~ into a mixture of 15% Cu(N03)2 and Al(N03)3 at room
temperature, untll pH 5.0 was reached. The blue precipitate indicated
that CU3(C03)2(OH)2 was the predominant copper compound in the precipitate;
however, some green CU2C03(OH)2 and blue-green Cu(OH)2 may have been
present as well. The direction of mixing, (NH4)2C03 into the copper
aluminum nitrate solution was chosen in order to avoid the formation of
the soluble ammonia complex of copper. This complex would have been
formed in the excess of (NH4)2CO~, which would have been present until
just before reaching the endpoint, had the direction of the mixing been
the reverse. The mixture was subsequently "aged" for one hour with frequent
stirring. The precipitate was filtered, pelletized, dried and fired as
usual.
The resulting dark brown sorbent pellets contained copper
reported as 37.7 wt. % CuO, which would correspond to a potential'
-------
Table VIII
E uilibrium Product Distribution for. Regeneration of Cu S05 in Oxygen-Nitrogen Gas
10 Moles CU2S05; 0.1 Mole N2, 0.02 Mole 02)
Temperature °c
427 527 627 727 827 927
A. Mole Fraction in Gas Phase
N2 0.8333 0.8333 0.8332 0.8162 0.4879 6.924xlO-3
02 0.1667 0.1667 0.1668 0.1677 0.2114 0.2994
S02 8.954x10-3 0.2277 0.5961
S03 7.128x10-3 0.07290 0.09758
B. Mole Fraction in Solid Phase I
I\)
V1
2.30Ox10-8 1. 333x10-4 3.940x10-4 I
CuO 3.460x10-7 0.01224 1. 000
CU20 0 0 0 0 0 0
CuS 0 0 0 0 0 0
CuS04 0 0 0 0 0 0
CU2S05 1. 000 1. 000 0.9999 0.9996 0.9878 O
s 0 0 0 0 0 0
C. Heat of Reaction 0 0 0.100 1. 400 41. 80 69.90
(kca1/mole CU2S05)
D. Heat of Reaction 150.0 71.00 67.90 69.90
(Per mole SOr + S03
regenerated kcal/atoms)
-------
-26-
stoichiometric capacity of 30 grams S02/100 grams sorbent. The surface
area of 242 m2/gm was also very attractive. The pellets were not particu-
larly strong, however, yielding to modest thumb-nail pressure.
Sorption studies on Cu-Al-l gave very interesting results.
The first basic series of sorptions and regenerations was carried out at
300, 450, and 5500C with regeneration in N2 at 6500C. Initial rates of
greater than 200 hr-l and capacities of 35 gm/100 gm were obtained. The
sorption capacity is seen to drop with successive runs, indicating aging.
Thermal regeneration rates, however, were higher after the first regeneration,
possibly due to the formation of a CuA1204 phase. X-ray analysis indicated
only CuO in the fresh sorbent, but the aging effects of 6 cycles resulted
in the appearance of CuAl204 and Al20~ phases and slight growth of the CuO
grains. A concise record of this baslc series of experiments is given in
Figure 3.
A porosity analysis was made of fresh Cu-Al-l by the American
Instrument Company using the Aminco 5-7125, 60,000 psi porosimeter. The
results showed a total pore volume (> 30 ~) of 0.58 cm3/gm, a density of
3.92 gm/cm3 (at 60,000 psi) and a 50% porosity of 0.0076 micron. The
distribution of pore sizes is shown in Figure 4. It is interesting to
note that about two-~hirds of the pore volume is in a sharply defined realm
between 30 ~ and 90 A while the other third of the pore volume lies between
0.3 micron and 90 microns. This indicates a possible pore structure in
which the net of very fine pores is connected to the outside of the pellet
by the larger "feeder" pores. This is generally a very desirable type
structure, because of the combination of enhanced pore-diffusion character-
istics and accessability of sorption sites.
In an attempt to reduce aging by lowering the regeneration
temperature, H2 regeneration was tried. H2 regeneration, it was found,
reduces the sorbent 1xJ Cu. This makes actual regeneration rate measurements
based on weight loss difficult to interpret. The half-times, however, are
nonetheless significant.
Temperature
Regeneration, Half-Time, min.
300
350
400
25
9
9
The rapid regenerations at 300-350oC suggest the possibility
of an isothermal sorption-regeneration cycle where the sorbent is trans-
ferred from the flue gas to the regeneration gas with no heating or cooling.
This would eliminate heat exchange equipment and heat losses and would
result in a much lower regeneration cost than shown in Figure 14 of the
phase I final report.l This would be especially true at low loadings and
the economic optimum loading would be shifted downward.
lRemoval of S02 from Flue Gas, Avco Corp., PB 177492, Contract No. PH 86-67-51.
-------
Figure 3
SORPl'IONjREGENERATION HISTORY OF SORBENT No.
Cu-Al-l
Composition
37.7 Cuo, 61.38% Al203
Surface Area, m2jgm
242
Stoichiometric Capacity, gm S02/1oo gm sorbent
30
40 '
Sorbent Loading
(gm S02/l00 gm sorbent)
I
vs.
20
Time
(arbitrary units)
0
Sl Rl S2 R2 S3 R3 s4 R4 S5 R5 s6 R6
Sorption/Regen. Temp., 0
C ~OO 6'10 ~oo 61)0 41)0 61)0 41)0 61)0 1)'10 6'10 550 650
Initial Normalized Sorption
Rate, ~ loading/~ S02 hr. 18'1 - 210 - 260 - 18'1 - 210 - 18') -
Ratio
Measured/Stoichiom. Cap. ~ 120 - 49 - 61 - 52 - 76 - 74 -
Sorption/Regeneration
Half-time, min. 40 100 ~ 7 17 15 12 20 18 2') 18 28
I
I\)
-..:]
I
-------
0.20
0.18
0.16
0.1"
0.12
0.10 I
J
n
0.01 n
0.06
0.04
0.02
0.00
Figure 4
POROSITY DETERMINA TlOH
DATE
August 6, 1968
(By 5-7107 or 5-7108 Aminco-Winslow Porosimeter)
0.58 ee/g.
SAMPLE
Cu-Al-l
Measured Pore Volume:
WT. OF SAMPLE, G.
0.21g4
Measured Density to 60,000 PSI:
3.92 glee.
D=I'OIE D.AMml 'MICIONS,
0-OCD""'00'" ..
0000000
...
o
-00000 0
o:.om'-4 0. in .
Co>
o
Co>
0-0........ 00 '" ..
'"
...
Co>
o
o
!='
Co>
!='
'"
000 0 0 0 0
:... 00 0 0 0 0
o .om...... 0. (.n .a:...
0000 0 0 0
0000 0 0 0
-000 0 0 0
o.om"""OoVw ..
o
o
o
'"
o
o
o
'"
o
o
'"
o
o
'"
I
f\)
CO
I
.. '"
o 0
o 0
'"
o
o
o
o
'"
o
o
o
o
... ColI 0- ".~-
P P 00000
~ go 'boag?
~ g 8808
'"
o
o
o
-------
-29-
In addition, H regeneration appears to increase sorbent
activity, as was demonstra~ed in the following experiment. A sample was
exposed to H2 for 30 minutes at 3000C, re-oxidizing to reconvert the copper
to copper oxide, and then sorbed to saturation in the standard flue gas at
300oC. Subseqgent regene:ation in H2 consisted of 2~ hours at 3000C,
3 hours at 325 C and 45 mlnutes at 350oC. After subsequent re-oxidation,
a second sorption at 3000C was made. Again, the sorbent was regenerated
in H2, this time for 50 minutes at 3500C followed by 50 minutes at 3750C.
The sample was again reoxidized and a third sorption made at 3000C. The
initial rates in these three sorptions increased from about 200 hr-l to
400 hr-l, and finally to over 850 hr-l.
X-ray analysis of the H2 regenerated material showed no sulfide
phases, as would be predicted by thermochemistry, albeit without any
knowledge of the influence of the aluminates. There was however a strong
, ,
Cu phase and a weaker phase which might have been CuA1204 or CU2Al204. One
explanation of this behavior might be the postulated two-step regeneration:
1.
CuO.CuS04, A1203' and H2 go to CU2A1204,
CuA1204, H2S, and H20.
CU2Al204, CuA1204, and H2 go to Cu and A1203'
The second step is presumed to be very much slower than the
first and on this basis no Cu will be formed until evolution of H2S has
ceased.
2.
ii.
Cu-Al-2
In the preparation of sorbent Cu-Al-2, the intent was
to repeat in every detail the preparation of Cu-Al-l. The one exception
was that four times as large a quantity of all ingredients was used.
Small differences became apparent, however, which deserve some consideration
because they are probably significant for all co-precipitations.
Again, 1.5 M (NH4)zC03 was stirred vigorously into a
mixture of 15% Cu (N03)2 and Al(NO~)3 at room temperature. Even though
the magnetic stirrer and rate of aadition were the same, the equivalent
amount of carbonate only brought the pH to 4.80 instead of 5.0. Since 4.8
is within the established precipitation range for copper-aluminum, no more
carbonate was added. All subsequent treatment was the same as given for
Cu-Al-l. The analysis of the finished sorbent Cu-Al-2 showed it contained
copper corresponding to 28.9 wt. % CuO and had a surface area of 184 m2/gm.
Although these differences are not so great that drastically
different performance characteristics might be anticipated, they do raise
the question of their origin. It is conjectured that the differences
stem from subtle differences in the co-precipitation. The mixing character-
istics of the 4 times larger volume of liquid resulted in slight differences
in the instantaneous pH at the specific point of mixing. This could, for
-------
-30-
example, have a profound effect on the relative amounts of CU3(CO~)2(OH)2
and CU2C03(OH)2 and Al(OH)3 in the co-precipitate. If this hypotnesis is
true, the significance of this mixing phenomenon for all co-precipitates
becomes apparent.
Because of the improved rate and capacity resulting from
hydrogen regeneration, it was decided to run the basic sorption series this
way for Cu-Al-2. The results of this series are summarized in Figure 5.
The unusually high rates, t~ = 6.5 min at 3000C) and saturation loadings
( 33gm S02/100 gm sorbent) ~ake this one of the most promising combinations
of this study. As mentioned earlier, it will be noted in Figure 5 that
regeneration to constant weight brings the sample weight below the "base
line" or original sample weight. This probably is the result of the removal
of water and/or other sorbates during regeneration. The start of the sub-
sequent sorption, however, is again above the "base line" since the sorbent
is resaturated with S02-free flue gas prior to each sorption run.
X-ray diffraction indicated the presence of poorly crystal-
line CuAl204 and CuO in the unused sorbent.
iii.
Al-BA, B, C
These three sorbents, which preceded Cu-Al-2, 2, were
co-precipitated under conditions that were similar but not closely controlled.
The final copper contents of the sorbents differ markedly from the anticip-
ated, however, again underlining the great importance mixing and localized
transitory precipitation conditions. Cu(N03)3.3H20 and Al(N03)3.9H20 were
used as starting material for each of these sorbents. The procedure followed
for all three sorbents is as follows. The amounts of these substances were
computed based on intended levels of copper oxide in the final sorbent.
These quantities of starting material, and the actual CuO levels are presented
in the following table. (For purposes of comparison, the surface area,
which was determined subsequently, is also included in the table.)
Grams of Grams of Surface
Sorbent Cu(N03)2.3H20 Al(N03)~.9H20 wt. % of CuO Area - m2/g
Al-BA 7.56 129 11.6 224
Al-BB 22.6 92.3 34.4 74
Al-Bc 45.2 37.4 75.3 11
The indicated quantities of copper and aluminum nitrates
were dissolved in 200 ml of water at 70oC. AnITnonium hydroxide (approximately
0.5 M) was added until the pH reached 6.3; the mixture was allowed to age
for two hours at B50C, and then filtered. The precipitate was washed with
distilled water. The paste-like precipitate was then pelletized, dried and
fired in the usual manner. The resulting sorbents, Al-BA, B, and C were
subjected to the basic sorption series, with a few additional measurements.
The results of these tests and the important properties of each sorbent are
combined in Figures 6 and 7. The resul~with Al-Bc were sufficiently
disappointing that they are not presented here.
-------
Figure 5
SORPl'IONjREGENERATION HISTORY OF SORBENT No.
Cu-Al-2
Composition 28.9% CuO in Al203
Surface Area, m2 / gm 184
Stoichiometric Capacity, gIn S02/100 gm sorbent
23.2
40 '
Sorbent Loading
(gm S02/1oo gm sorbent)
I
vs.
0
Time
(arbitrary units)
\
"
-40 . ,
H,..., Regeneration Sl Rl S2 R2 83 R3 84 R4 85 R5 86 R6
Sorption/Regen. Temp., 0 450/
C 300 400 300 4'50 450 450 450 550 550 550 550 550
Initial Normalized Sorption
Rate, ~ loading/% S02 hr. 27~ - '111 - 288 - 480 - 859 - 246 -
Ratio
Measured/Stoichiom. Cap. % 10;4 - 128 - 1'1~ - 1'11 - 142 - - -
Sorption/Regeneration <5 10.~
Half-time, min. 6.1 <5 6.3 <5 5.9 11.5 <5 <5 - -
I
W
I-'
I
-------
Figure 6
SORPI'ION/REGENERATION HISTORY OF SORBENT No.
Al-8A
Composition
11.6% CuO in Al203
Surface Area, m2 / gm
224
Stoichiometric Capacity, gm S02/loo gm sorbent
9.03
20
Sorbent Loading
(gm S02/100 gm sorbent) I
vs.
10
Time
(arbitrary units)
0
Sl Rl S2 R2 S3 R3 84 R4 85 R5 86 R6 87 R7
Sorption/Regen. Temp., 0
C 300 650 300 650 300 675 450 675 450 675 550 675 550 675
Initial Normalized Sorption
Rate, ~ loading/~ 802 hr. 197 - - - - - 171 - 161 - 21 - 35 -
Ratio
Measured/Stoichiom. Cap. ~ 192 - 153 - 127 - 163 - 189 - 24 - 15 -
Sorption/Regeneration 65
Half-time, min. 30 12 37 9 20 8 50 10 13 3 7 3 3
I
LA)
[\)
I
-------
Figure 7
SORPl'IONjREGENERATION HISTORY OF SORBENT No.
Al-8B
Composition ~4.4% CuO in Al~03
Surface Area, m2/gm 74
Stoichiometric Capacity, gm S02/100 gm sorbent
27.4
10
Sorbent Loading
(gm S02/100 gm sorbent)
I
vs.
5
Time
(arbitrary units)
°
81 Rl 82 R2 83 R3 84 R4 85 R5
Sorption/Regen. Temp., 0
C 300 620 450 600 450 675 450 675 300 675
Initial Normalized Sorption
Rate, i loading/~ S02 hr. 76.6 0 8.2 '15.4 12.8
Ratio
Measured/Stoichiom. Cap. ~ ~4 11 10 19 2.Q
Sorption/Regeneration
Half-time, min. 25 S SO SO ~g 70 46 2~ s
I
W
W
I
-------
Of the three sorbents, the only one with a high SOo
capacity was A1-8A, the one lowest in CuO. The first five runs at 300 C^
and U50°C had an average saturation capacity of about l8 gms S02/100 gra^s
sorbent. This is twice the loading that would be expected from the CuO
analysis, assuming the sorption proceeds as
CuO + S02 + i 02
However, the twice stoichiometric loading might be explained by dithionate
formation, but this has not yet been verified. Sorption of moisture and
flue gas components other than SOp cannot be responsible as these have
been measured and are zero at ^50DC and less than h% at 300 C.
Note also that very little capacity was observed at 550 C
This is contrary to the results of the thermochemical calculations which
indicate that sorption rather than dissociation should prevail until temper
atures of more than 590°C are obtained. However, if sorption does occur
by formation of bisulfates or dithionates these calculations no longer
apply. No explanation of the "shoulders" in the sorption curves (Fig. 6)
for Bk and S5 is apparent.
The ratio of the measured to the stoichiometric capacity
decreased uniformly with increase in CuO content, averaging over 100$ for
A1-8A, about 20/0 for A1-8B and less than 5$ for A1-8C. Only sorptions
at 300°C and U50°C were considered in these averages. It is postulated
that this decrease in utilization of potential capacity is a result of the
observed decreasing surface area and hence decreasing availability of the
CuO.
Attempts to correlate rate data with reaction rate and
pore diffusion models were inconclusive. Thus, there is no clear-cut
determining process that could be established for sorption on these three
sorbents.
iv. Other Copper Oxide Alumina Sorbents
The following sorbents were made and tested, but their
performances were not of sufficient significance to merit detailed
description: Al-8, 1-6, and Harshaw CuO-A^O^ catalyst. Available data
for these sorbents may be found in Appendix I.
2. Copper Oxide-Silica
a. Thermochemistry
The thermochemistry of this system was considered only in
terms of copper, copper oxides, and sulfates. Thus reduced, the thermo-
chemistry becomes the same as has already been discussed for copper oxide-
alumina, to which the reader is referred.
-------
-35-
b- Preparation and Evaluation - CuO-Silica
Silica supported sorbents can be made by co-precipitation
with silica gel, followed by firing. The silica gel, Si02.nH?0, is
conveniently formed by the reaction
Wa2Si03 + 2H+ (n-l)H20 - >Si02.nH20 + 2Na+
The precipitation of the copper constituent is from a solution of a copper
salt, such as CuNO^, which is acid because of hydrolysis. This acidity
is usually sufficient for the formation of the silica gel. At the same
time, Na2Si03 forms a basic solution which causes the precipitation of
Cu(OH)2. The overall reaction for these hydrolyses and precipitations is
thus:
Cu(N03)2 + Wa2Si03 + (n + l) H20 - >Cu(OH)2 + Si02.nH20 + 2NaNO
Stoichiometric quantities of the reactants yield a sorbent of 57.3 wt. %
CuO in Si02. Some variation in this ratio can be realized by adjusting
the amounts of Cu(WC>3)2 and NigSiCy If wider variation of the levels of
copper oxide is desired than can be obtained in this fashion, slight pH
adjustment can be made. Sorbents with low copper content, for example, can
be made by adding acid to the copper nitrate solution. If either of these
alternatives is to be utilized, however, care must be taken not to exceed
the limits of the pH range for co-precipitation, as discussed earlier.
i. Cu-Si-1
This sorbent was prepared by adding 750 ml of 1 M
CuNOo to 630 ml of 1 M Na2SiOo with vigorous stirring at room temperature.
The final pH of the mixture was 5.0. After aging, with frequent stirring
for 1 hour, the precipitate was separated by filtration. The filter cake
was broken up, slurried in 200 ml 0.5 M WH^KO^ and re-filtered to remove
sodium. After repeating this washing operation three times, the filtrate
showed negligible sodium in a flame test. The paste-like precipitate was
then pellet i zed, dried and fired in the usual fashion.
The finished sorbent, Cu-Si-1, was found to have copper
corresponding to 59.1$ CuO, very close to the 6l$ predicted from the ratio
of reactants in the co-precipitation. The surface area was found to be
97 m2/gm. This is an attractively high surface area for a sorbent so high
in active oxide. As can be seen with other CuO- silica as well as other
sorbents, surface area generally drops off with increasing amounts of active
oxide.
The sorbent was subjected to the usual basic sorption
evaluation series, plus two additional measurements to determine the
effect of hydrogen regeneration. This sequence may be seen in Figure 8.
The initial sorptions at 300°C showed low loading and slow rate. Subsequent
-------
Figure 8
SORPl'ION/REGENERATION HISTORY OF SORBENT No.
Cu-Si-1
Composition
59.1% CuO in Si02
Surface Area, m2 / gm
97
Stoichiometric Capacity, gm S02/100 gm sorbent
47.6
40
Sorbent Loading
(gm S02/100 gm sorbent)
vs.
20
I""
Time
(arbitrary units)
0 ,
Sl Rl S2 R2 S3 R3 s4 R4 S5 R5 s6 R6 S7 R7 s8
Sorption/Regen. Temp., 0 65 0/
C 300 650 300 650 450 650 450 650 550 650 550 650 300 H2 30
Initial Normalized Sorption
Rate, ~ loading/~ S02 hr. 106 - 362 - 894 - 220 - 450 - 350 - -10 - -10
Ratio
Measured/Stoichiom. Cap. ~ 13 - 10 - 74 - 69 - 76 - 69 - 3 - 47
Sorption/Regeneration
Half-time, min. 77 14 37 14 49 54 32 65 20 100 14 180 23 <5 190
I
W
0'\
I
-------
-37-
sorption at 4500 and 5500, while much higher in rate and loading, indicate
that the sorbent is losing activity with age. To determine the extent of
this aging effect, a sorption at 3000C was made subsequent to the basic six.
Both rate and loading were found to be much lower than in the first sorp-
tions at 3000, indicating that the sorbent had undergone changes in the
first six cycles. After the additional sorption at 3000C, a hydrogen regenera-
tion at 6500C was followed by still another sorption at 3000C. This later
sorption showed the same low rate, but the loading was considerably improved.
Thus, the beneficial effect of hydrogen regeneration is again demonstrated.
Unfortunately, however, little can be deduced from these limited data
concerning the cause of these changes in activity.
Attempts to correlate rate data with models for pore diffusion
and chemical kinetic were inconclusive for all sorption measurements with
this sorbent, due to a high degree of scatter, but the rates and loading
were too low to justify an extended treatment of the data.
ii.
Si-8A, B, C
These three sorbents were made with three different levels
of copper oxide in order to test the effect of this variable. The co-
precipitations were made by quickly adding 200 ml of acidified copper
nitrate solution to 200 ml of sodium silicate solution at 700C. The
quantities of nitrate, silicate and acid were determined for the desired
copper level and a 20 gram yield. After mixing, the pH was adjusted to 5.0
with dilute HN03 or NH40H and the precipitate "aged" for two hours. Filter-
ing, pelletizing and drying followed the same scheme as outlined in the
Introduction. No attempt was made to wash excess sodium from the cake
which resulted in 0.7 to 1.5% sodium content in the sorbent.
The amounts of copper nitrate, sodium silicate and con-
centrated nitric acid used for each sorbent, the CuO level, together with
the surface area, are presented for comparison in the following table:
Gram of Grams of cm2 conc. Surface
Sorbent Cu(N03)2.3H20 Na2Si03.9H20 HN03 wt. % CuO Area, m2/g
si-8A 3.03 89.192 38.1 4.75 3.0
si-8B 9.09 80.5 31.1 14.4 102
si-8c 27.3 52.1 8.9 43.5 4.5
These sorbents were subjected to the basic sorption series. si-8A showed
essentially zero capacity. The other two sorbents gave good results which
are summarized in Figures 9 and 10.
As with alumina based sorbents, the capacities at 3000C
and 4500C are good but drop rapidly at 5500C. In contrast with the alumina
based sorbents however good utilization of the active oxide is obtained
, ,
at high oxide concentrations, especially at 4500C. Thus, the 45% CuO sorbent
-------
Figure 9
SORPrION/REGENERATION HISTORY OF SORBENT No.
si-8B
Composition
15% CuO in Si02
Surface Area, m2/gm
102
Stoichiometric Capacity, gm S02/100 gm sorbent
12
-
,
20
Sorbent Loading
(gm S02/100 gm sorbent)
vs.
10
Time
(arbitrary units)
0 . ,
Sl Rl S2 R2 S3 R3 s4 R4 S5 R5 s6
Sorption/Regen. Temp., 0
C 300 650 300 650 450 650 450 650 550 650 550
Initial Normalized Sorption 140
Rate, ~ loading/~ S02 hr. 55 - 51 - 32 - 27 - 230 -
Ratio
Measured/Stoichiom. Cap. ~ 89 - 76 - 63 - 80 - 23 - 23
Sorption/Regeneration 1.6
Half-time, min. 24 6 24 12 2.5 - 3.0 10 - 1.0
I
W
())
I
-------
Figure lO
SORPl'IONjREGENERATION HISTORY OF SORBENT No.
si-Bc
Composition
45% CuO in Si02
Surface Area, m2 / gIll
4.5
Stoichiometric Capacity, gIll S02/100 gIll sorbent 36
20
Sorbent Loading
(gIll S02/100 gIll sorbent)
I
vs.
10
Time
(arbitrary units) II
0 . ,
Sl R1 S2 R2 S3 R3 s4 R4 S5 R5 s6 R6
Sorption/Regen. Temp., 0
C 300 650 300 650 450 650 450 650 5 50 650 550
Initial Normalized Sorption
Rate, ~ loading/~ S02 hr. 21 - 10 - 140 - 120 - 210 - 130
Ratio
Measured/Stoichiom. Cap. ~ 22 - 21 - 48 - 49 - 16 - 20
Sorption/Regeneration
Half-time, min. 51 - 61 18 21 42 20 34 9 21 9 23
I
W
\0
I
-------
-40-
(si-8c) achieves about half of stoichiometric loading or 18 grams S02/100
grams sorbent at 4500C.
Note that the surface area for si-8B (102 m2/gms) is
higher than for si-8c and that the ratio of measured to a stoichiometric
capacity is also higher, a correlation which has been observed for most of
the copper oxide sorbent studies. Thus, as with the copper oxide-alumina
sorbents, a higher surface area seems to indicate a higher availability of
the active metal oxide.
Increasing sorption temperature from 3000 to
the rate about tenfold. This is equivalent to an activation
13 kcal which might correspond to a mixed diffusion-reaction
sorption or a chemisorption rate controlled sorption.
4500C increased
energy of about
rate controlled
3.
Iron Oxide-Alumina
a.
Thermochemistry
Equilibrium computations were made for the sorption
(Table IX) and thermal regeneration (Table X) of iron oxide (Fe203) sorbent.
Thermodynamic data for ferrous aluminate were found which permitted its
consideration in the computations. The results for the sorption, given in
Table ~ show, however, that FeA1204 does not form under the conditions
chosen.
The calculations show that the concentration of combined
sulfur oxides reaches 0.02% at 475°C. Thus, to achieve 90% removal of
sulfur oxides, sorption would have to be carried out at less than this
temperature. This constraint is in accordance with the experimental
evidence on iron oxide sorbents which indicates a high sorptive capacity
at 4500C and almost no capacity at 5500C.
b.
Preparation and Evaluation
1.
Fe-Al-l
This sorbent was synthesized by co-precipitating with
1.5 M (NH4)2C03 from a solution containing 125 gm Fe(N03)3.9H20 and 184 gm
Al(N03)~.9H20 in one liter of water. Because it was determined in the
precipitation studies that freshly precipitated Fe203.nH20 is dissolved in
excess C03' it was necessary to make the co-precipitation by adding the
(NH4)2C03 to the mixture of iron and aluminum to a final pH of 6.6, as
opposed to adding the iron and aluminum to the carbonate. A large flocculent
precipitate was obtained. Before filtering, the precipitate was "aged" for
1 hour, during which period the pH of the mixture drifted to 6.0. This
might indicate that more hydroxide was precipitating, or that the initial
precipitate was undergoing an exchange reaction from the carbonate to the
hydroxide. The material was then filtered, pelletized, dried and fired as
usual.
-------
-41-
Table IX
Temperature °c
327 427 527 627
A. Mole Fraction in Gas Phase
CO ° ° 0 0
C02 0.1404 0.1404 0.1402 0.1401
H2 0 0 0 0
H20 0.06018 0.06018 0.06009 0.06003
H2S 0 0 0 0
02 0.02909 0.02909 0.02914 0.02914
N2 0.7703 0.7703 0.7692 0.7684
S02 1.103xl0-l0 4. 213xl0-7 1.983xl0-4 9.145xl0-4
S03 8. 726xl0-8 1. 986xl0-5 1. 166xl0-3 1. 087xl0- 3
S2 0 0 0 0
B. Mole Fraction in Solid Phase
FeO 0 0 0 0
FeO.940 0 0 0 0
Fe304 0 0 0 0
Fe203 0.9000 0.9010 0.9681 1.000
FeS 0 0 0 0
FeS2 0 0 0 0
FeS04 0 0 0 0
Fe2(S04)3 0.1000 0.0990 0.03195 0
H2S04 0 0 0 0
S 0 0 0 0
* Both Fe2Si04 and FeAl204 were found not to form.
-------
-42-
Table X
E~uilibrium Product Distribution for Thermal Regeneration of Ferric Sulfate in Nitrogen
( 1.0 Fe2( S04)3 + 1.0 N2)
0
Temperature C
427 527 627 727 827
A. Mole Fraction in Gas Phase
N2 0.9999 0.9968 0.9246 0.2070 0.1949
02 1. 094xlO- 5 6.622x10-4 1. 589xlO-2 0.1719 0.2203
S02 2.200xlO-5 1. 325xIO-3 3.177x10-2 0.3439 0.4407
S03 2.00lxlO-5 1.17lxlO-3 2. 770xlO-2 0.2772 0.1441
S2 0 0 0
B. Mole Fraction in Solid Phase
FeO 0 0 0 0 0
FeO.940 0 0 0 0 0
Fe304 0 0 0 0 0
Fe203 1. 399xlO-5 8. 346xlO-4 2. 144xIO-2 1.000 1.000
FeS 0 0 0 0 0
FeS2 0 0 0 0 0
FeS04 0 0 0 0 0
Fe2(S04)3 1.000 0.9992 0.9786 0 0
S 0 0 0
A HR -0.03 +0.17 +4.00 173.13 185.79
(kcal/mole
Fe2(S04)3)
-------
-43-
The sorbent was subjected to the first three sorption-
regeneration cycles of the "basic sorption series", after which the
sequence was interrupted, the sorbent removed, and a fresh sorbent sample
(N2 activated but not previously exposed to flue gas) was introduced to
the TGA apparatus. The series was terminated after an additional two cycles.
Figure 11 shows this sequence and the results of these measurements. The
initial sorption reached a saturation loading of 26.8 grams S02/100 grams
sorbent which is attractive, but the rate was rather slow - the half-time
being about 88 minutes. Subsequent sorption showed a decreasing saturation
loading which could be attributed to imperfect regeneration, hydrothermal
aging, or the transition to 0< Fe203' On observing this loss of activity
after three cycles, a fresh sample was taken and lower regeneration temper-
atures, below the 0{ Fe203 transition, were tried. The first sorption
with this fresh sorbent (84) showed a higher rate and saturation loading
than had been observed in the third sorption (S3). Subsequent sorption,
however, again showed a similar loss of activity, even with the lower
regeneration temperature. Thus, this effect cannot be entirely ascribed
to the 0(. Fe203 transition.
Analysis of the data of the first sorption (Sl) shows
that there is a high initial rate, up to the loading of 5 gm/100 gm,
followed by a lower rate. The rates at low loading (N5 gm/100) correlate
fairly well with the kinetic model suggested in the Phase I, Final Report
(ref. 1, pg. 26) for a reaction rate controlling mechanism. For higher
loadings the data fit a pore diffusion model. These observations permit
two differing explanations. One interpretation is that the pore structure
of the sorbent has some large open pores and some very small pores. The
larger pores fill quickly and at a rate determined by the chemical reaction;
the smaller pores are more slowly filled by a diffusion controlled mechanism.
The other interpretation is that if the cause of the loss in activity is
related to the structure of the sorbent, some influence might be expected
after a few minutes of exposure to flue gas. Further conclusions on this
activity decay and possible solutions to the problem would require further
experimentati on.
ii.
Other Iron Oxide Alumina Sorbents
The following sorbents were made and tested but their
performances were not of sufficient significance to merit detailed
description: Al-12, Al-l2A, Al-12B, Al-12C, Al-12D, I-8, and Florite
(Floridin Company, Pittsburgh, pa.).
Additional information may be found in Appendix I.
4.
Iron Oxide-Silica
Thermochemistry
a.
Equilibrium computations for iron oxide-silica sorbents
indicated that no appreciable ferrous silicate formed. The thermochemistry
then becomes that of iron oxide-sulfur oxide system. This is treated in
the section on iron oxide-alumina sorbents.
-------
Figure 11
SORPl'ION/REGENERATION HISTORY OF SORBENT No.
Fe-Al-1
Composition
45.6% Fe203 in Al203
Surface Area, m2/gm.
233
Stoichiometric Capacity, gm. S02/100 gm. sorbent
55
40
Sorbent Loading
(gm. S02/100 gm. sorbent)
vs.
20
Time
(arbitrary units)
0 . ,
Sl R1 S2 R2 S3 R3 ~ s4 R4 S5 R5
Sorption/Regen. Temp., 0 ~
C 300 700 300 700 450 700 ::q 450 590 450 590
Initial Normalized Sorption ~
IIi
Rate, ~ loading/~ S02 hr. 89 - 80 - 65 - 116 - 38 -
Ratio
Measured/Stoichiom. Cap. % 49 - 30 - 6 - 16 - 2 -
Sorption/Regeneration
Half-time, min. 88 -9 154 .-6 8 8 14 12 12 8
I
+=-
+=-
I
-------
-45-
b.
Preparation and Evaluation
The precipitation of silica supported sorbents has
been discussed in detail in the section on copper oxide-silica
For the specif1c case of iron oxide-silica co-precipitation the
reaction may be considered to be:
already
sorbents.
overall
2Fe(N03)3 + 3Na28i03 + (n + m)H20
.Fe203.nH20 + 8i02.mH20 + 6NaN03
As with copper co-precipitates, the proportion of iron in the sorbent may
be varied by pH adjustment during precipitation.
i.
Fe-8i-l
To ~ne liter of a 0.25 M solution of Na28i03' increments
of a 15 wt. % solutlon of FeN03 were added, with vigorous mlxing, until
the final pH was exactly 7.0. A total of 865 ml of the Fe(N03)3 solution
was used. The precipitate was then aged for an hour at room temperature,
filtered, and washed five times to remove sodium by re-slurrying in 1 M
NH4N03 and refiltering. The resulting paste was pelletized, dried, and
fired in nitrogen as usual.
The finished sorbent was found to have iron corresponding
to 42.5 wt. % Fe203' which is close to the 47% predicted by the stoichio-
metry of the chemical equation above. The surface area was quite good,
248 m2/gram.
The sorbent was subjected to the "basic" sorption series,
except that a second sorption at 3000C was omitted since the first one
showed no 802 sorption. This sequence may be seen in Figure 12. As
predicted by the equilibrium computation, 802 sorption at 5500C was very
low, and at 4500C fairly high, measuring 3 and 22 gram S02/100 gram sorbent,
respectively. The rates were slow, especially at 4500C where the half-
times were about 2 hours. Attempts to correlate the rate data with pore
diffusion and reaction models were inconclusive.
ii.
Other Iron Oxide Silica Sorbents
The following sorbents were made and tested but their
performances were not of sufficient significance to merit detailed discussion:
Si-12B, 8i-12C, 8i-12D.
Available data may be found in Appendix I.
5.
Manganese Oxide-Alumina
a.
Thermochemistry
The equilibria for sorption and regeneration of manganese
oxide sorbents have been computed for a wide variety of conditions and
the results compared with available data. The conclusions are as follows:
-------
Figure 12
SORPTION/REGENERATION HISTORY OF SORBENT No.
Fe-Si-1
Composition
Fe203 in 57.5% Si02
Surface Area, m2 / gIll
248
Stoichiometric Capacity, gIll S02/100 gIll sorbent
51
20
Sorbent Loading
(gIll S02/100 gIll sorbent)
I
vs.
10
Time
(arbitrary units)
0 . ,
Sl 82 R2 83 R3 84 R4 85
Sorption/Regen. Temp., 0
C 300 450 650 450 650 550 650 550
Initial Normalized Sorption
Rate, ~ loading/~ S02 hr. 0 --50 - -50 - 25 - 23
Ratio
Measured/Stoichiom. Cap. ~ 0 42 - 39 - 7 - 6
Sorption/Regeneration
Half-time, min. - -120 18 120 39 17 11 25
I
+-
0\
I
-------
-47-
1. Mn ° is thermochemically approximately equivalent to
Fe203 and CuO as. a2s6rbent. That is, the partial pressures of 802 and 80~
in a flue gas stream in equilibrium with the sorbents are all similar. Tne
product of sorption is Mn804'
2. Mn804 may be thermally regenerated but a temperature on
the order of 8000c is required for a partial pressure of 802 and 803 of
0.1 atm. This is to be compared with 6500C for regeneration of Fe2(804)
and 7500C for CuO.Cu804. Regeneration in steam is essentially equiValen~
to thermal regeneration.
3. Chemical regeneration of Mn804 with excess H2, CO H2 + H20,
and CO + H20 all yielded a solid with residual Mn8 as well as MnO or Mn20
as the regeneration product. Great excesses of H20 + H2 or H20 + CO shouid
complete the regeneration, but at the cost of a very dilute off gas.
4. MnO or Mn203 cannot be stabilized relative to Mn8 in the
presence of a reducing gas by reaction with 8i02 to form Mn8iO. This
results in the formation of Mn8 during regeneration. Other siiicates or
zirconates, vanadates, chromates, etc. might stabilize the oxide, but we
have been unable to find thermochemical data which would allow checking
this point.
5. A two-stage regeneration process was considered in which
Mn8 from H2 or CO reduction of Mn804 was exposed to C02, H20, or C02 + H20.
This scheme does not seem practical as large volumes of gas would be
required to convert the sulfide to the oxide, resulting in a very low
sulfur concentration in the regenerator gas.
Equilibrium computations were made for a sorbent of 0.1
mole Mn203 in contact with a flue gas consisting of the following components:
C02
802
14.0 mole
°2
3.0 mole
0.2 mole
H20
6.0 mole
N2
76.8 mole
The distribution of products, computed for the temperature
range 327°C to 727°C may be found in Table XI.
It is
results and those for
temperatures required
are 590°C, 590°C, and
interesting to note the similarity between these
copper and iron oxide sorbents. For comparison, the
for 90% removal of the 802 initially in the flue gas
475°C for manganese, copper, and iron oxide, respectively.
The possibility of formation of MnSi03 in a silica-based
sorbent was considered but the formation of this silicate was shown to be
, 0 °
thermodynamically unfavorable over the temperature range 327 C to 927 C.
This does not rule out the possibility that some other silicate such as
Mn2Si05 or MnSi205 might form, but no thermodyn~ic data c?uld be f?und for
these species. For the same reason, the formatlon of alumlnates, zlrconates,
or vanadates has not been checked. Both thermal regeneration and regeneration
-------
Table XI
(0.1 Mn203 Sorption of Flue Gas on Mn~03 76.8 N21
+ 6 H~O + 14 C02 + 0.2 SO~ + 3 O~ +
0
Temperature, C
327 427 527 627 727
A. Mole Fraction in Gas Phase
N2 0.7698 0.7698 0.7698 0.7695 0.7682
C02 0.1404 0.1404 0.1404 0.1403 0.1400
H20 0.06019 0.0619 0.06017 0.06003 0.06001
02 0 . 02961 0.20961 0.02961 0.02962 0.02970 I
.j::""
2. 813xl0-12 7. 589xl0-9 2. 722xl0-6 2. 536xl0-4 0.001498 co
S02 I
S03 2. 18lxl0-9 3.590xl0-7 1.607xl0-5 3.018xl0-4 0.000502
B. Mole Fraction in Solid Phase
Mn203 5.470xl0-7 9. 139xl0-5 4.708xl0-3 0.1613 1.0000
Mn2S04 1.000 0.9999 0.9953 0.8387 0.0000
-------
-49-
by reducing gas were considered. For regeneration by reducing gas, both
one-stage and two-stage procedures were considered.
i.
Thermal Regeneration
The equilibrium pressures of S02 and S03 over MnS04
were computed for the temperature range of 427°C to 826°c. A mixture of
10 moles MnS04 and 1 mole of N2 was considered.
The results of the computations are given in Table XII.
The partial pressure of S02 plus S03 is 0.016 atm at 7l8°c, 0.1 atm at
800oc, and 0.3 atm at an estimated 560oC. Thus, regeneration temperatures
of about 8000c would probably be required to obtain a reasonable regenera-
tion rate. This is higher than the 7500C estimated for thermal regeneration
of CU2S05 and 6500C for Fe2(S04)3'
It was found experimentally that thermal regeneration
does not proceed to completion at 600oc. This result is consonant with the
computations.
Regeneration with Reducing Gas
ii.
Single-stage regeneration with H2, CO, H2 + H20 was
considered. In addition, two-stage regeneration in which the MnS04 was
first reduced to MnS and then reacted with H20, C02 or H ° + C02 was
considered. Equilibrium computations were made for the %emperature range
427°C to 927°C.
One-Stage Regeneration
Regeneration of MnS04 according to the following reac-
tions was considered.
1.
MnS04 + 4H2
MnS04 + 4co
. MnO + 3H20 + H2S
.. MnO + 3C02 + COS
2.
Computations were made with 50% excess H2 and CO and
with a lOX excess of H2. The solid phase was greater than 95 percent MnS
at all temperatures when H2 was the reducing gas and greater than 99 percent
when CO was the reducing gas. The remainder of the solid was MnO. There
was no difference in the solid composition between a 50-percent excess of
H2 and a lOX excess. Thus, regeneration cannot be effected by great
excesses of reducing gas. It appears then that H2 and CO are unsatisfactory
as regenerating gases due to sulfide formation.
The addition of water to the regenerating gas to provide
for the conversion of MnS to MnO was also considered. Computations were
made for the following cases:
3.
4.
5.
MnS°4 +
MnS04 +
MnS04 +
3(H2 + H20)
9(H2 + H20)
20(H2 + H20)
6.
7.
8.
MnS04 + 3(CO + H20)
MnS04 + 9(CO + H20)
MnS04 + 20(CO + H20)
-------
Table XII
MnSO~ Regeneration in Nitrogen
----(10 MnS04 + 1.0 N2)
427
527
Temperature, °c
627
727
826
A. Mole Fraction in Gas Phase
N2 1.0000 1. 0000 0.9988 0.9788 0.7835
02 0.2226xlO-9 0.5383xlO-5 0.1999xlO-3 0.00350 0.03487
S02 0.8152xlO-6 0.2341xlO-4 0.8864x10-3 0.01584 0.1605 I
Vl
0.3346xlO-8 0.1867xlO-5 0.8669xlO-4 0
S03 0.001822 0.02107 I
B. Mole Fraction in Solid Phase
Mn203 4.106xlO-8 1. 269xlO-6 4. 870xlO-5 9.028xlO-4 1.173xlO-2
MnS04 1.000 1. 000 1. 000 0.9991 0.9883
-------
-51-
For case 3, H2 + H20, the solid was 54 to 73 mole
percent Mn8 with the remainder being MnO and MnS04' The presence of Mn804
was not observed at temperatures equal to or greater than 527°C. For case
6, CO + H20, the solid phase was 60 to 73 mole percent MnS with the remainder
being MnO and MnS04. The Mn804 was again a low temperature phase forming
between 427°C and 527°C.
For cases 4 and 7, greater excesses of regenerating
gas were supplied, and for cases 5 and 8 even more gas was supplied. Case
4 gave a solid phase consisting of 82 mole percent MnS at 927°C increasing
to 99 percent at 427°C with the remainder being MnO. No Mn804 was computed
to be formed. The greater excesses of Case 5 gave a solid consisting of
67% MnS at 927°C and 98% at 427°C. Again, the balance of the solid was MnO.
Case 7 gave a solid phase of 90% MnS at 927°C and 99%
at 4270C with the remainder as MnO. The greater excess gas in Case 8 gave
a solid of 82% MnS at 927°C the bal~nce being MnO. Below 670°C the equilibrium
computations indicate the presence of solid carbon.
Thus, increasing the regenerating gas from 6 to 18
moles reduced the conversion to MnO while further increases in the gas to
40 moles effected an increase in MnO. Additional excesses of H2 or CO plus
steam would be expected to completely convert the solid to MnO, at least
if the temperature were high enough to prevent Mn804 or C formation. How-
ever, the maximum partial pressure of sulfur compounds for any of the above
cases was computed to be less than one mole percent. The regenerate
gases would be very dilute in sulfur compounds.
considered.
The possibility of regeneration in steam alone was also
Computations were made for the following case:
9.
MnS04 + 6H20
Above 8000C, the regeneration should proceed well.
At 827°C the combined pressure of 802 and SO~ is 0.14 atm. However, at
727°C, the combined pressure is only 0.018 atm. These results differ very
little from thermal regeneration in N2, and the additional hazard of hydro-
thermal aging would seem to make N2 regeneration preferable to steam
regeneration.
A summary of the results is given in Table XIII. In
all cases there is either sulfide formation or, in the case of steam
regeneration no advantage over thermal regeneration. These computational
results should be compared with available data and the two reconciled if
possible. The experimental work has been largely carried out by the Bureau
of Mines at Bruceton.
Bu Mines found that when H2 or CO was used as the
regenerating gas sulfur removal increased as the temperature was increased
from 4500e to 9000C. At 600°C and higher, the solid phases were identified
as MnO and MnS. These results are in accord with the equilibrium computa-
tions. When regenerating at 4500C, the additional phases MnS04 and
-------
-52-
Table XIII
Summary of Equilibrium Computations for One-Stage Regenerationf
Case Results
1. MnS04 + 6 H2, 40 H2 95 mole percent MnS
2. MnS04 + 6 CO 99 mole percent MnS
3. MnS04 + 3(H2 + H20)* 54-73% MnS, bal. MnO, MnS04
4. MnS04 + 9(H2 + H20) 82-99% MnS, bal. MnO
5. MnS04 + 20(H2 + H20) 67-98% MnS, bal. MnO
6. MnS04 + 3(CO + H20)* 60-73% MnS, bal. MnO, MnS04
7. MnS04 + 9(CO + H20) 90-99% MnS, bal. MnO
8. MnS04 + 20(CO + H20)** 82+% MnS, bal. MnO, C
9. MnS04 + 6H20 Mn203, MnS04
*
MnS04 present at 427°C calculations but not at 527°C and higher
** C present at 427, 527°C but not at 627°C and above.
f When a range of sulfur removal is given the removal increases with temperature
-------
-53-
MnO.Mn203 were also formed. The
by equillbrium and they probably
reduction of MnS04 and Mn203'
presence of these phases was not predicted
result from kinetic limitations on the
When H2 + H20 and CO + H20 were used as regenerating
gases, the sulfur removal again increased with temperature. At 6000c
and above, the phases MnO and MnS were found. These results are in accord
with the e~uili~r~um calculations. At 450oC, the phases MnS04 and MnO.Mn203
were also ldentlfled. The presence of MnS04 was predicted by the equilibrium
calculations, but the Mn203 is probably present due to the slow rate of
reduction at 450oC.
When steam alone was used as the regenerating gas, the
sulfur removal again increased with temperature, significant removal occurr-
ing at temperatures greater than 7500C as indicated by the equilibrium
calculations.
It would appear that the regeneration schemes proposed
are all equilibrium limited at reasonable temperatures, either by the
formation of MnS or by the low partial pressure of H2S or cas over the solid.
This suggests that it is still important to find a carrier that will react
with and stabilize manganese oxide relative to the sulfide, thus allowing
the regeneration with H2' CO, etc. As has been pointed out in previous
reports, Al203 serves this function with an Na20 sorbent, and in the absence
of A1203' Na2S04 regenerates to Na2S rather than NaA102'
Two-stage Regeneration with Reducing Gas
All of the one-stage regeneration calculations indicated the
presence of residual sulfur as MnS. Thus, it was decided to compute the
equilibria for various possible schemes for a second-stage reaction to
convert the MnS to MnO or Mn203' The following second-stage gases were
tried: H20, C02' H20 + C02, and air.
Two systems were studied to simulate air regeneration.
These
were:
10.
11.
MnS + (6 N2 + 1.5 02)
MnS + (12 N2 + 3 02)
For case 10, at temperatures of 527°C and above, the solid
phase was entirely MnO and the sulfur was present in the gas entirely as
S02 at a partial pressure of 0.143 atm. At 427°C, the solid phase has
become largely (90 mole percent) MnS and MnS04 and the partial pressure of
S02 has dropped to 0.123 atm.
For case 11, in which there is an excess oxygen beyond the
requirements for forming SO~ and Mn203, the equilibrium is quite different.
At temperatures of about 9000C and higher, the solid is entirely Mn203 and
sulfur is present in the gas phase as S02 and 803 at a combined partial
pressure of 0.07 atm. At lower temperatures, the solid becomes largely
Mn804 with decreasing traces of Mn203' Correspondingly, the partial
-------
-54-
pressures of S02 and S03 become very small.
The conclusion to be drawn is that, below 900°C, conversion
of MnS to Mn203 in air will be very slow due to the low equilibrium partial
pressures of S02 and S03' Also, great vol~es of air would be requir~d .
and the sulfur-bearing stream would be as dllute as the flue gas, obvlatlng
sulfur recovery. These conclusions are partly confirmed by the BuMines work
in which exposure of MnS sorbent to air at 600°C removed only about one-
half the sulfur in ten hours.l
Mixtures of C02 and H20 have been used to convert alkali
sulfides to their oxides. It was felt that this gas mixture should also
be considered for conversion of MnS. Equilibria were computed for the
following two cases:
12.
MnS +
(C02 + 2 H20)
3(C02 + 2 H20)
13.
MnS +
For case 12, the mole fraction of MnO in the solid phase
decreased uniformly from 0.026 at 927°C to 0.816 x 10-3 at 427°C. The
balance of the solid was MnS. The partial pressure of sulfur bearing
gases also decreased from 0.017 atm at 927°C to 0.538 x 10-3 atm at 427°C.
For case 13, the mole fraction of MnO in the solid phase was
higher than in case 11 and decreased uniformly from 0.077 at 927°C to
0.0024 at 327°C. The balance was MnS. Again, the pressure of sulfur
bearing gases decreased from 0.017 atm to 0.538 x 10-3 atm. Clearly, the
use of a sufficient quantity of the H20-C02 mixture could further convert
the MnS to MnO, but, as in the case of air, the sulfur values would be
very dilute and non-recoverable.
Finally, computations were made for MnS exposed to pure H20
and pure C02' The following three cases considered for steam regeneration:
14. MnS + 1. 5 H20
15. MnS + 4.5 H20
16. MnS + 45 H20
For case 14, the solid phase was 3.1 mole percent MnO at
927°C and 0.12 percent at 427°C. The remainder of the solid was MnS.
For case 15, the solid phase was 9.4 percent MnO at 927°C and 0.36 percent
at 427°C. For case 16, the solid was 93.6% MnO at 927°C and 3.6% at 4270C.
In all cases, the partial pressures of sulfur bearing compounds were 0.0206
atm at 927°C and 0.00081 atm at 427°C. Thus, a sufficient excess of steam
should completely remove the sulfur from the solid. This would be done at
the cost of a very dilute off gas, however. For the.cases at 927°C which
gave the highest concentration of sulfur bearing gases, the combined volume
percent of H2S and S02 was 2.07. The additional threat of hydrothermal
Ref.l-Bu Mines, Pittsburgh, Quarterly Report to PHS, September 30, 1966.
-------
-55-
aging makes steam regeneration appear very unattractive.
Two cases were considered for C02 regeneration:
17.
18.
MnS + 1. 5 C02
MnS + 15 C02
For case 17, the solid was 1.49 mole % MnO at 927°C and
0.0023% at 427°C. For case 18, the solid was 14.9% MnO at 927°C and
0.023% at 427°C. The balance was MnS. For both cases, the partial pressure
of sulfur bearing gases was 0.0098 atm at 927°C and 0.000016 atm at 427°C.
Thus, the tendency of MnS to convert to MnO in C02 as measured by the partial
pressure of these gases is less than one tenth that in steam.
In summary, none of the routes examined for the conversion
of MnS to MnOx shows appreciable promise. This result again underlines the
importance of discovering a carrier for Mn0x which will stabilize the oxide
form relative to the sulfide during regeneration by a reducing gas.
b.
Preparation and Evaluation - Manganese Oxide Alumina
i.
Mn-Al-l
A solution composed of 150 ml of 15% Mn(N03)2 and 275 ml
of 15% Al(N03)~ was slowly added to 600 ml of 1.5 M (NH4)2C03 at room
temperature. The final pH of the mixture was 8.4, somewhat on the high
side of the permissible pH range for aluminum precipitation. A substantial
precipitate did form, however, which was aged, with frequent stirring, for
one hour before filtering. The cake was then pelletized and fired in nitrogen
in the usual manner.
The resulting sorbent was extremely weak and tended to
powder if handled. For this reason, it was decided to try this synthesis
again.
ii.
Mn-Al-2
This sorbent was made in exactly the same way as Mn-Al-l
except that the carbonate was added to the mixed nitrates. After mixing
in 190 ml of the (NH4)2C03 solution, the pH had reached 5.0 and a very
heavy precipitate had formed. Subsequent treatment was the same as with
Mn-Al-l.
The resulting sorbent was somewhat stronger than Mn-Al-l,
but still showed a tendency to powder. Manganese content corresponded to
37.8% MnO and the surface area was 198 m2/gram.
The sorbent was subjected to the "basic" sorption series,
with the exception that hydrogen was used for regeneration rather than
nitrogen. The decision to use hydrogen was based on the results of the
-------
-56-
thermochemical computation and the poor results with attempts to thermally
regenerate previous manganese sorbents.
The sorption-regeneration sequence for Mn-Al-2 may be
seen in Figure 13. Good saturation loadings and rates were found at all
three temperatures. After the third cycle, however, the slight decline in
the initial rates might be an indication of the onset of aging effects.
Most of the rate data are sufficiently scattered that no clear-cut inter-
pretation can be made concerning the rate determining mechanism. For both
of the sorptions at 550oC, however, a plot of the sorption rate versus
sorbent loading is clearly concave downward, indicating that pore diffusion
is governing the sorption rate.
iii.
Other MnO - Alumina Sorbents
The following sorbents were made and tested but their
performances were not sufficiently significant to merit detailed discussion
here: Al-9, Al-9A, Al-9B, Al-9C, and 1-7.
Additional information may be found in Appendix I.
-------
Figure l3
Composition
SORPl'IONjREGENERATION HISTORY OF SORBENT No.
MnO in 62.2% Al203
Mn-Al-2
Surface Area, m2/gm
198
Stoichiometric Capacity, gm S02/100 gm sorbent
34.0
2n. '
Sorbent Loading
(gm S02/100 gm sorbent)
I
VB.
F'
Time
(arbitrary units)
0 . ,
...
H0 Regeneration 81 Rl 82 R2 83 R3 84 R4 85 R5 86
Sorption/Regen. Temp., 0
C ~OO 6')0 ~OO 650 450 650 450 650 550 650 550
Initial Normalized Sorption
Rate, ~ loading/~ S02 hr. 350 - 445 - 652 - ~45 - 286 - 270
Ratio
Measured/Stoichiom. Cap. ~ ~I') - 26 - ~4 - ~~ - ~q - ~q
Sorption/Regeneration
Half-time, min. 12.1 - 13 .5 - 10.7 <5 11.1 <5 9.4 <5 9.5
I
\.J1
~
I
-------
-58-
APPENDIX I
Summary of Sorbents Exhibiting Marginal Performance
For purposes of clarity these sorbents are grouped according to the
method of preparation, co-precipitation, impregnation, and those available
commercially.
Co-Precipitated Sorbents
1.
A large number of sorbents was prepared by this method. These lend
themselves to subdivision into three groups, designated Series I, II, III,
which correspond to the chronological sequence in which they were made.
a.
Series I, AI-I through Al-14
Oxides of Na, K, Li, Ni, Zn, Cu, Mn, Fe, and Cr were co-precipitated
with Al2~~. A solution of Al(N03)~ and the desired metal nitrate was pre-
pared. 'fl1is was heated to rOoC ana mixed with a solution of (NHld2 C03 at
the same temperature to cause co-precipitation of the oxides. Great excesses
of carbonate ranging from 30 to 150 percent were used. The pH after carbon-
ate addition is estimated to be about 9. The proportions of metal nitrate
and Al(N03)~ were chosen to match the metal to aluminum ratio desired in the
product soroent. The resulting precipitate was filtered, washed, and the
paste spread into a layer on a glass plate, and dried in air for 24 hours
at l25°C. The dried material was crushed to approximately the desired 8
mesh size and fired at 6500C in H2 or N2.
Note that there was no pH control during precipitation, no uniform
mixing procedure, no casting of the precipitate paste before drying to give
uniformly shaped pellets, and a rather severe drying schedule.
Chemical analysis, X-ray analysis, and surface area measurements
were carried out on sorbents Al-l through Al-14. The results of these tests
are given in Table XlV. In general, the stoichiometric capacities of the
sorbents are interestingly high, and with the exception of Al-2, the sur-
face areas are quite high. There is a marked tendency to form aluminates
as indicated by the X-ray analysis. Nickel oxide, which showed only a
weak aluminate phase as fired (6500C) developed a strong phase where ignited
at 1000oC.
The sorbents were tested for S02 capacity and regenerability in the
TGA apparatus. None was found to have sufficient activity to be of interest.
b.
Series II, AI-8A, B, C; si-8A, B, C; Al-9A, B, C; AI-l2B, C, D;
Si-l2B, C, D
On the basis of the results of Series I sorbents, the more promising
oxides of copper, manganese, and iron were chosen for further evaluation.
-------
Table XIV
Properties of Co-Precipitates, Series I
Stoichiometric
Preparation Conditions Capacity
Drying Firing Surface on Volatile X-Ray
Temp. Time Temp. Time Composition Area Free Basis Diffraction
Material ~ (hrs) (OC) (hrs) Atm. ( approx. wt. %) (m2/grn) (grn S02/ grn sorb) Analysis
Al-l Li20-Al203 24 14.8% Li20
(LiN03 raw material) 125 650 8.8 H2 135 0.32
Al-2 Li20-Al203 650 5 H2 13.3% Li20 9.6 0.31 Li2S04.H20,
(Li2S04 raw material) 77.0% Al203 LiAl508
Al203
Al-3 K20-Al203 125 24 650 10 H2 16.7% K20 32.3 0.12
77.1% Al203 I
\J1
6.2% volatiles \.()
I
Al-4 Na20-Al203 125 24 650 10 H2 8.6% Na20 152 0.094
85.5% Al203
5.9% volatiles
Al-5 NiO-A1203 650 4 N2 24.7% NiO 275 0.23 NiO poorly
65.2% Al20) A1203 crystal-
9.1% volahles NiAl204 line
Al-6 Cr203-Al203 125 24 650 4.5 N2 131
Al-7 ZnO-Al203 125 24 650 4 N2 71.6% Al203 175 .15 ZnAl206 poorly
650 4.8 H2 17.0% ZnO -Al2 3 crystal.
11.5% volatiles line
Al-8 CuO-Al203 125 24 650 5 N2 72.1% Al203 175-220 .15 CuAl204 spinel &
16.8% CuO -Al203' CuO(weak:
13.5% volatiles poorly crystalline
-------
Material
Al-9 Mn02-Al203
Al-12 Fe203-Al203
Al-14 Na20-Mn02-
Al203
Preparation
Drying
Temp. Time
~ (hrs)
125
24
24
125
125
24
Table XIV (Concl' d)
Conditions
Firing
Time
(hrs)
Temp.
t:9.L
650
650
650
650
5.5
5
8
8.5
Surface
Composition Area
Atm. (approx. wt. %) (m2/gm)
N2
N2
98+% Mn203
54.2% Fe203 146
34.,% Al20~
11.1% volatiles
H2
H2
Stoichiometric
Capaci ty
on Volatile
Free Basis
(gm SO?/gm sorb)
1.1
0.,3
X-Ray
Diffraction
Analysis
Fe203
I
0'\
o
I
-------
-61-
Each of these active oxides was prepared on an alumina and a silica support.
The procedure was as follows:
i.
Alumina Based Sorbents
. To.200 cc of dis~illed w~ter ~ere added Al(N03)3 and the desired
metal nltrate In the proportlons deslred In the product. Sufficient total
salts were dissolved to give 20 grams of product at 100 percent recovery.
The solution was heated to 700C and NH40H solution added to
the metal nitrate solution with stirring until the desired pH (as measured
with pH paper) was obtained. The precipitate was then aged 2 hours at
this pH and 850C, vacuum filtered, and washed with distilled water. The
precipitate, as a thick cake, was pressed into holes of 1/8-inch diameter
drilled in an l/8-inch aluminum plate.
The material was then dried for 1 1/2 hours at 500C and ambient
humidity followed by 2 hours at the controlled humidity of 94°c dry bulb,
74°c wet bulb (in a humidity oven). After this step, the pellets were
about l/16-inch in diameter by 1/16-inch in length. Finally, the pellets
were fired in air, heating to 6500C over a period of 2 hours and maintaining
at 6500C for 6 hours.
The desired pH to be achieved in the precipitation step was
determined ahead of time by separately adding NH40H to solutions of Al(N03)3
and the other metal nitrate and observing the pH at which precipitation oc-
curred. Short range pH paper was used to determine the pH. The final pH
was 6.3 - 6.6 for copper sorbents, 6.0 - 6.4 for manganese, 5.4 - 6.4 for
iron.
ii.
Silica Based Sorbents
Solutions were prepared of sodium silicate in 200 cc of dis-
tilled water and of acidified metal nitrate in 200 cc distilled water. Sili-
cate and metal nitrate were added in the re~uired ratios to give the desired
sorbent composition at complete precipitation. Sufficient material was
added to produce 20 grams of sorbent at 100 percent yield.
Acidification of the metal nitrate solution was with concen-
trated HN03. Sufficient acid was added so that the two solutions combined
would be in stoichiometric balance for the following reactions:
M(N03)x + ~ x Na2Si03 + ! x H20 ~ M( OH)x + ! x Si02 + NaN03
where M
=
metal
2HN03 + Na2Si03 . Si02 + H20 + 2NaN03
The solutions were warmed to 700C and the metal
tion added ~uickly to the silicate. The pH was adjusted to
nitrate solu-
a value of 5
-------
-62-
with dilute HN03 or NH40H and the precipitate digested for 2 hours. Fil~
tering, pelletizing and drying proceeded as with the alumina based materlals.
Metal oxide-carrier combinations were prepared and given the
following designations:
Metal Carrier
Oxide ~03 SiO~
CuO Al-8A, B, C si-8A, B, C
FeOx Al-12B, C, D Si-12B, C, D
Mn0x Al-9A, B, C Not Prepared
Chemical analyses and surface areas are given in Table XV.
Sorption-Regeneration Results - Series II
iii.
Copper oxide-alumina, Al, 8A, B, C
Copper oxide-silica, Si, 8A, B, C
The results for these sorbents are given in the body of this
report (page 37).
Manganese oxide-alumina, Al-9A, B, C
As shown in Table XV, the active oxide content of these sorbents
was very low. This accounts for the very low saturation loadings found
experimentally.
Iron oxide alumina - Al-l2B, C, D
The performances of these sorbents are summarized in Figures
14, 15, and 16. The degrees of utilization demonstrated by these sorbents
were low. This is surprising in view of the high surface areas measured
(376, 340, 274 cm2/gm). The highest capacity observed with these
'sorbents was only about 4 gms S02/100 gms sorbent on Al-12-C.
Iron oxide-silica, Si-12B, C, D
Of these sorbents only the data for Si-12B are given (Figure 17)
as the saturation capacities for Si-12C and D are less than 5 grams 802/100
grams sorbent.
-------
Table XV
Properties of Coprecipitates, Series II
Sorbent
CuO-Alumina CuO-Silica Mn?03-Alumina Fe?03-Alumina Fe?03-Silica
Chemical Analysis
Wt. % Al-8A Al-8B Al-8C si-8A si-8B si-8c Al-9A Al-9B Al-9C Al-12B Al-12C Al-12D Si-12B Si-12C Si-12D
--
Active oxide 11.2 34.1 75.3 4.75 14.4 43.5 1.7 2.1 3.2 4.7 13.6 33.4 4.4 10.8 26.8
Al203 (bydif.) 85.5 65.3 22.7 94.9 76.9 96.0 80.8 77.5 66.4
Si02 90.7 83.5 50.6
Na20 1.5 0.7 1.2
I
Vo1ati1es* 3.3 0.6 2.0 0.1 0.1 2.0 3.4 2.1 0.8 14.5 ~ 0.2 0.9 1.5 4.2 0\
w
100.0 100.0 100.0 97.1 98:7 97.3 100.0 100.0 100.0 100.0 100.0 100.0 I
Stoichiometric CapacitY**
gm S02/100 gm sor- 62 3.8 45 6.5 18 40
bent 9 27 12 2.1 3.2 3.9 5.2 13 32.2
Surface Area,
m2/gm 224 74 10.5 3.5 102 4.5 304 306 289 376 340 274 163 258 174
*
300oC, 1 hr, He atmosphere
** based on sorbent weight after volatiles removed
-------
Figure 14
SORPl'ION/REGENERATION HISTORY OF SORBENT No.
Ai -12B
Composition
4.70% Fe203 in Al203
Surface Area, m2 / gm
376
Stoichiometric Capacity, gm S02/1oo gm sorbent 5.65
2
Sorbent Loading
(gm S02/1oo gm sorbent)
vs.
1
Time
(arbitrary units)
0 . ,
......
Sl R1 S2 R2 S3 R3 s4 R4 S5
Sorption/Regen. Temp., 0
C . 300 750 300 750 450 700 450 750 550
Initial Normalized Sorption
Rate, ~ loading/~ S02 hr. 15 - 26 - 43 - 36 - -
Ratio
Measured/Stoichiom. Cap. ~ 30 - 13 - 26 - 27 - 11
Sorption/Regeneration
Half-time, min. 10 16 15 - 6 22 6 11 6
,
0\
+-
I
-------
Figure 15
SORPI'IONjREGENERATION HISTORY OF SORBENT No.
Al-l2C
Composition
13.6% Fe20~ in Al20~
Surface Area, m2/ gIll
340
Stoichiometric Capacity, gIll S02/100 gIll sorbent
16.3
8 '
Sorbent Loading
(gIll S02/1oo gIll sorbent)
I
vs. 4
Time
(arbitrary units)
°
""
Sl Rl 82 R2 83 R3 84 R4 85 R5 86
Sorption/Regen. Temp., 0 650/
C 300 800 300 750 450 750 450 750 500 - 500
Initial Normalized Sorption
Rate, ~ loading/~ S02 hr. 55 - 10 - 28 - 23 - 0 - 0
Ratio
Measured/Stoichiom. Cap. ~ 22 - 21 - 24 - 25 - 0 - 0
Sorption/Regeneration
Half-time, min. 8 - 8 - 17 60 17 70 - 60 -
I
0'\
V1
I
-------
Figure 16
SORPI'ION/REGENERATION HISTORY OF SORBENT No.
Al-12D
Composition 33.4% Fe~03 in Al?03
Surface Area, m2/f!J1l 274
Stoichiometric Capacity, f!J1l S02/1oo f!J1l sorbent
40.1
1
Sorbent Loading
(gIn S02/1OO f!J1l sorbent)
vs.
0
Time
(arbitrary units)
. ,
81 Rl 82 R2 83 R3 84 R4 85 R5 86
Sorption/Regen. Temp., 0
C 300 700 300 700 4 50 700 450 700 550 700 550
Initial Normalized Sorption
Rate, ~ loading/~ 802 hr. 35 - 17 - 48 - 60 - 20 - 42
Ratio
Measured/Stoichiom. Cap. ~ - - 2 - 3 - 3 - 0.1 - 0.1
Sorption/Regeneration
Half-time, min. - - 3 - 5 - 3 - 3 - 2
I
0\
0\
I
-------
Figure :1.7
SORPl'IONjREGENERATION HISTORY OF SORBENT No.
Si-l2B
Composition
4.35% Fe203 in Si02
Surface Area, m2 / gm
163
Stoichiometric Capacity, gm S02/100 gm sorbent
5.22
,
Sorbent Loading
(gm S02/100 gm sorbent)
vs.
2
Time
(arbitrary units)
0 . .
Sl S2 R2 S3 R3 s4 R4 S5 R5
Sorption/Regen. Temp., 0
C 300 450 100 450 100 550 100 550 100
Initial Normalized Sorption
Rate, i loading/i S02 hr. 0 5 - 51 - 26 - 31 -
Ratio
Measured/Stoichiom. Cap. i 0 16 - 36 - 29 - 35 -
Sorption/Regeneration
Half-time, min. - 16 -10 5 -5 6 -3 4 -2
I
CJ\
~
I
-------
-68-
Series III - Cu-V-l, 2; Co-Al-l; Cr-Al-2; Fe-V-l; Fe-Zr-l; Mn-Si-l;
Mn-V-l; Ni-Al-l
c.
i.
Copper Oxide-Vanadia
The precipitation of eu(OH)2 and H4V2Oy (pyrovanadic acid)
from NaV03 can be represented by the e~uations:
Cu(N03)2
+
20H-
..
Cu(OH)2
- H4 V20y
+
2N0 -
3
2NaV03
+
2W
+
H20
+
2Na:+
Since the acid-base re~uirements of these two reactions are opposite, the
co-precipitation of eu(OH)2 and H4V2?7 poses some difficulty. Keeping in
mind, however, that because of hydrolysis, eu(N03)2 solution is acidic and
NaV03 solution is basic, the mixing of these two solutions can cause the
net reaction:
Cu(N03)2
+
2NaV03
+
3H20
~
Cu(OH)2
+
H4V2Oy
+
2NaN°3
The resultant sorbent, based on this reaction, would have 30.5 wt. % euO.
If sorbents with higher copper content are desired, a larger proportion of
Cu(N03)2 to NaV03 must be used, and base or carbonate solutions added to
precipitate the additional copper. For lower copper content, additional
vanadate and acid must be used.
The first attemp~ Cu-V-l, at a sorbent by this scheme was made
by slowly adding 15% Cu(N03)2 to 1M NaV03 in the indicated stoichiometric
proportions. (1 mole Ca(N03)2 to 2 moles NaV03). The final pH was 4.05
which was more acid than anticipated and it was apparent that precipitation
had stopped before half of the copper solution had been added. In hopes of
enriching the copper content of the precipitate, 1.5 M (NH4)2C03 was added
until pH 5.6 was reached.
This precipitate was stirred for one hour and then filtered. At
this point, the precipitate appeared to have a very small particle size and
very little was retained in the cake on filtering. The small ~uantity and
chalky appearance led to the decision to try a different approach.
The second attempt, Cu-V-2, utilized the same materials but
differed in that the NaV03 solution was added to the Cu(N03)2 solution until
a pH of 6.1 was reached. After one hour of stirring, the precipitate was
filtered. The ~uantity and apparent ~uality of this material was somewhat
improved over the previous attempt but it was difficult to filter. This
finding, and the modest yield of precipitate led to the decision to forego
washing with NH4 N03 solution originally planned to reduce contamination by
sodium. Pelletizing and drying were done in the usual manner. On firing
in nitrogen at 650oC, the pellets softened and fused together. Possibly it was
residual NaV03 (m.p. 6300C) which melted, or the normal melting point of
-------
-69-
V205 (6900c) may have been lowered somewhat by the presence of sodium or
copper salts.
No sorption experiments were attempted with this material.
ii.
Cobalt Oxide-Alumina
Only one attempt was made at a sorbent with this combination
,
which has the designation Co-Al-l. The co-precipitation was made by adding
1.5 M (NH4)C03, to a mixture of 15% Co(N03)2 plus 15% Al(NO )3 to a final
pH of 7.3. Here, as with the precipitation of the copper-a1umina sorbents,
the direction of mixing was chosen in order to reduce possible formation
of Co(NH3)6++. The co-precipitate was, as usual, "aged" for one hour,
pelletized, dried and fired in the usual manner.
The finished sorbent was subjected to part of the basic sorp-
tion seriesj the sorptions at 5500C being omitted because of the less than
encouraging results at the two lower temperatures. Saturation loading varied
between 3.5 and 5.4 gram S02/100 grams sorbent which is somewhat surprising
in light of the high surface area (171 m2/gm) and cobalt content correspond-
ing to 49% Coo.
Figure 18 shows the history of the sorption-regeneration sequence.
The increasing regenerated sample weight, or shifting base line is quite
apparent, and accounts for an increase of over 9% in the four cycles. The
oxidation of CoO to C020~ could at most account for 5.2%. Another possibility
could be the formation of C02(S04)3 which is probably less readily regen-
erated thermally. Unfortunately, there is insufficient information to sup-
port this conjecture.
iii.
Chromium Oxide-Alumina
Sorbent Cr-Al-l was made by adding a solution containing approxi-
mately 15% Cr(N03)2 and Al(N03)3' in a proportion calculated to yield 50 wt.
% Cr203 in the sorbent, to a 1.5 M solution of (NH4)2C03' The final pH
was 8.2. The addition of the mixed metallic nitrate solution to the carbon-
ate was considered advantageous because it provided for an excess of carbon-
ate throughout the precipitation. The blue-gray precipitate was aged for
one hour before filtering and pelletizing. After air drying at 60oc{
humidity drying at 90oC/70oC - dry bUlb/wet bulb (in a humidity oven), the
pellets were dried in N2 at 650oC. The resulting sorbent was extremely weak
and the pellets crushed to a powder under the lightest touch. The color,
a mixture of dark green and black, indicates that the chromia content might
be quite high.
Because of these poor structural properties, a second attempt
was made using half the amount of Cr(N03)3' In precipitating Cr-Al-2,
a 1.5 M (NH4)2C03 was added with vigorous stirring to mixed chromi~ an~
aluminum nitrates until a pH of 7.2 was reached. After the same fllterlng,
-------
Figure 18
SORPl'ION/REGENERATION HISTORY OF SORBENT No.
Co-Al-1
Composition
49% CoO in Al203
Surface Area, m2/gm
171
Stoichiometric Capacity, gm S02/100 gm sorbent
84
10
Sorbent Loading
(gm S02/100 gm sorbent)
vs.
5
Time
(arbitrary units)
0 . ,
Sl Rl S2 R2 83 R3 84 R4
Sorption/Regen. Temp., 0
C 300 650 300 650 450 650 450 650
Initial Normalized Sorption
Rate, ~ loading/% S02 hr. 104 - 74 - 150 - 100 -
Ratio
Measured/Stoichiom. Cap. ~ 5 - 4 - 6 - 4 -
Sorption/Regeneration
Half-time, min. 12 16 14 21 6 7 4 9
I
-...:]
o
,
-------
-71-
pelletizing, drying, and firing se~uence as with Cr-Al-l, the physical prop-
erties of the pellets were not markedly improved. No sorption experiments
were attempted with either Cr-Al-l or Cr-Al-2.
iv.
Iron Oxide-Vanadia
The source of vanadia in this combination makes use of the
reaction
2NaV03
+
2H20
+
2H+
..
H4V2~
+
2Na+
which yields a gel-like precipitate. Since a solution of Fe(N03)3 is ~uite
acid because of hydrolysis, it can effect the above reaction. At the same
time, NaV03 yields a basic solution and hence can precipitate "iron hydroxide",
Fe203.H20. The overall reaction for these hydrolyses and precipitations may
be written
2Fe(N03)3
+
6NaV03
+
(6 + n)H20
... Fe203. nH20
+
3H4V2~ + 6NaV03
Based on this stoichiometry, the final sorbent should have 22.7% Fe20~
by weight. If other levels of iron are desired, addition of acid to the
iron solution (to yield lower Fe203) or base to the vanadate (to yield
higher Fe 03) before mixing would be possible. The iron oxide-vanadia
sorbent, re-V-l was made by adding a 1.0 M solution of NaV03 to a 15% solu-
tion of Fe(N03)3 until a pH of 5.65 was reached. The precipitate was aged
for one hour, then filtered and the cake washed by slurrying in 200 ml of
0.5 M NH4N03 and refiltered. This washing operation was repeated 5 times
even though the filtrate from the third washing showed only very weak flame
tested for sodium.
The washed precipitate was pelletized, dried and fired in the
usual way.
The resultant sorbent showed an iron content on analysis cor-
responding to 52.7 wt. % Fe203. The pellets were extremely strong and would
appear to be highly attrition resistant. Unfortunately, however, the sur-
face area was found to be only 1.1 m2/gm. On exposure to a simulated flue
gas, no weight gain was observed at 300, 450, or 550oC. In the hope that
hydrogen might activate the sorbent, the sorbent was exposed to hydrogen
at 5500C for one hour. Subse~uent exposure to flue gas resulted in a weight
gain corresponding to a low 0.4 gm S02/100 gram sorbent.
v.
Iron Oxide-Zirconia
Hydrated zirconium oxide is precipitated from alkaline solution
as a hydrogel by the reaction
OH-
ZrOC12
+
3H20
.
Zr02.2H20 +
2HCl
This reaction lends itself to the formation of zirconia based sorbents.
-------
-72-
The iron oxide-zirconia sorbent, Fe-Zr-l, was made by slowly
adding 1.5 M (NH4)2C03 to a mixture of 0.3 M Fe(N03)3 and 0.25 M ZrOC12,
at room temperature until the pH reached 6.0. The voluminous, heavy pre-
cipitate was "aged" with freCluent stirring, for one hour before filtering.
The cake was pelletized, dried and fired in nitrogen as usual.
The resulting sorbent was found to have iron corresponding to
33.6% Fe20~ and a surface area of 272 m2/gm. Porosity determination, by
mercury intrusion, showed the sorbent to have an appreciable pore volume,
0.73 cm3/gram, betweenoO.003 and 0.04 microns. Additional pores, smaller
than 0.003 micron (30 A) were also present but could not be measured in the
test eCluipment used. The pore size distribution curve may be seen in
Figure 19.
The sorbent was subjected to simulated flue gas at 300°C,
450oC, and 550°C. The saturation loading at 300°C was not especially good
and declined at the higher temperatures. As seen in Figure 20, the sorbent
also became increasingly difficult to regenerate, and after three cycles
its activity was reduced to nearly zero. No further experiments were
undertaken.
vi.
Manganese Oxide-Silica
The co-precipitation of this combination poses problems similar
to those discussed for CuO-V205. By analogy, the net reaction,
Mn(N03)2 + Na2Si03 + (n + 1)H20 . Mn(OH)2 + Si02.nH20 + 2NaN03
might be useful in making this sorbent. Following this approach, a 15%
solution of Mn(N03)2 was added to a 0.25 M solution of Na2Si03 in a propor-
tion calculated to ultimately yield a sorbent of 50 wt. % MnO in Si02.
The pH of the resulting mixture was 8.9 which is much too alkaline to pre-
cipitate the silica. The pH was, therefore, adjusted to 6.9 with 3 M
HN03. This precipitate was then filtered and washed five times in 0.5 M
NH4N03 to remove traces of sodium. The filter cake was then pelletized,
dried and fired in the usual fashion.
The resultant sorbent, Mn-Si-l, was found to contain manganese corres-
ponding to 39 wt. % MnO and a surface area of 36 m2/gm. In the standardized
sorption experiments, the resulting sorbent showed zero capacity for S02
at 300°C and 450°C. A slight weight gain in a sorption at 550°C could not
be removed by regeneration at 650°C.
vii.
Manganese Oxide-Vanadia
The vanadia support for this sorbent was realized by the preci-
pitation of vanadic acid (H4V2~), as was described for the iron oxide
vanadia sorbent Fe-V-l.
Sorbent Mn-V-l was co-preci~itated by vigorously mixing 700 ml
of solution containing 252 grams of Mn(N03)2 into 1.4 liters containing 172
grams of NaV03. The heavy gelatenous precipitate was aged with freCluent
stirring for one hour and then filtered. To remove sodium, the filter cake
-------
-
0_... 0- '" ~ Co> ...
00000 0 0 0 0
0.20 D'I
0.18
0.16
0.14
0.12
0.10 I
J
n
0.08 n
0.06
0.04
0.02
0.00
... Co> ~ VI 0- 'CD.oO
-
POROSITY DETERMINATION
(By 5-1101 or 5-1108 Aminco--Winslow Porosimeter)
DATE
SAMPLE
August 6, 1968
Fe-Zr-l
WT. OF SAMPLE, G.
0.~020
O..oCD'" 0- '" ~
Co>
...
o
Co>
o
... tn o-.....,m.oo
0000000
D=POIE DIAMml IMICRONS)
...
-000 0 0 0
o~a. " 0. Ot ;...
~
Co>
~
...
P=AlSOLUTE PIESSUIE PSI
II III
...
o
o
Co>
o
o
... lrt 0- """CD...o-
g g ggggg
o
Measured Pore Volume:
0.73 cc/g.
Measured Density to 60,000 PSI:
000 0 0 0 0
:... 00 0 0 0 0
0..0(0 ...... 0- U'I lito.
000 0 0 0
0000 0 .
-000 0
000(1)""'" 00:
o
o
Co>
o
o
...
j'11"
':
i . ~' I
I " 1;1,
r ~ , '.
[
tit, ',: ,
~.j. t-t
j , ." j'
, . ,tJl,: I ~1jI1" ,.
, ; t t~ '; rt ,I '
"I ,..r; ,n Ii
:lim I.,
. I
, . ,
I
t
'I
, ,
2.79 gf
0.219 cc
o
o
~
o
o
o
Co>
o
o
o
...
"
,
-:]
W
I
... Co> ~ '" O-....,Q)-O- ... Co> ~ '" 0."""«»-0-
o 0 0 0 0 0 0 0.0 '" 0 0 0 0 0 0000
o 0 0 0 0 0 000 0 0 0 0 0 0 oc.o?
o 0 0 0 00000 0 0 0 0 0 00000
o 0 0 0 0 0 00000
o
-------
Figure 20
SORPl'ION/REGENERATION HISTORY OF SORBENT No.
Fe-Zr-1
Composition
33.6% Fe203 in Zr02
Surface Area, m2 / gm.
272
Stoichiometric Capacity, gm. S02/100 gm. sorbent
40.4
10
Sorbent Loading
(gm S02/100 gm. sorbent)
I
VB.
5
Time
(arbitrary units)
0 . ,
Sl Rl S2 R2 83 R3
Sorption/Regen. Temp., 0
C 300 600 450 600 550 600
Initial Normalized Sorption 61
Rate, ~ loading/~ S02 hr. 10 - 129 - -
Ratio
Measured/Stoichiom. Cap. ~ 22 - 16 - 6 -
Sorption/Regeneration
Half-time, min. 105 8 14 3 4 8
I
~
+"
I
-------
-75-
was washed five times by reslurrying in 0.5 M NH4N03 and refiltering.
washed co-precipitate was pelletized, dried and fired as usual.
~e
~e resultant sorbent was found to have manganese content cor-
responding to 62.1% MnO and surface area of 1.1 m2/g.
On exposure to simulated flue gas the saturation loadings of
0.83 a~ 3000C, 10.2 and 4.2 at 4500C and 3.4 ~rams 802/100 grams sorbent
at 450 C were measured.
viii.
Nickel Oxide-Alumina
Co-precipitation was made by adding a total of 2000 ml of 1.5 M
(NH4)2C03 to a.solution containing 195 grams Ni(NO~)2'6H20 and 371 grams
Al(N03)3'9H20 In 1500 ml water. The final pH of tne mixture was 6.3. After
aging for one hour, the precipitate was filtered and the cake pelletized.
~~ pellets were then air dried at 600c and humidity dried at 900C dry bulb,
70 C wet bulb temperature, before firing for eight hours in nitrogen
at 6500c.
The resultant sorbent, Ni-Al-l, was analyzed and found to con-
tain nickel corresponding to 26.4 wt. % NiO and surface area of 306 m2/gm.
On exposure to simulated flue gas, saturation loadings of 4.3,
2.6 and 2.7 gram 802/100 gram sorbent at 3300c, 4500C, and 5500C were ob-
served. This se~uence of sorptions and regenerations may be seen in Figure
21.
2.
80rbents Made by Impregnation
Oxides of K, Na, Ni, Zn, Cu, Cr, Mn, and Fe were deposited on a commer-
cial porous carrier, Norton 3032, 83% Al203' 15.3% 8i02' The support
was boiled in solutions of metal nitrate for 8 hours, washed, dried and
fired at 6500C to decompose the nitrate. Nine sorbents were so prepared.
They are designated 1-0 through 1-8.
Chemical, X-ray and surface area analyses were carried out on these
samples. The results of these tests are given in Table XVI. Generally
the surface area of the impregnated sorbent was slightly less than that
of the carrier, but still high. X-ray analysis showed only weak lines at
best for the added metal oxides.
Chemical analyses were carried out on the Cr, Mn, and Fe sorbents. As
shown in Table XVI both the iron and manganese sorbents had sufficient metal
oxide present to have a stoichiometric capacity of more than 10 percent 802.
The sorbents were exposed to standard flue gas as with the co-precipitated
sorbents. In no case was the observed saturation capacity greater than 3
percent 802'
-------
Figure 21
SORPI'ION/REGENERATION HISTORY OF SORBENT No.
Ni-Al-1
Composition
26.4% NiO in Al201
Surface Area, m2 / gm
306
Stoichiometric Capacity, gm S02/100 gm sorbent
22.6
11"\
Sorbent Loading
(gm S02/100 gm sorbent)
vs.
0
Time
(arbitrary units)
-1"
H2 Regeneration Sl R1 S2 R2 S3 R3
Sorption/Regen. Temp., 0
C 300 650 450 650 550 650
Initial Normalized Sorption
Rate, ~ loading/~ S02 hr. 189 - 221 - 106 -
Ratio
Measured/Stoichiom. Cap. ~ 19 - 11 - 12 -
Sorption/Regeneration
Half-time, min. 15 1 12 < 1 6 <1
I
-.:J
0\
I
-------
Table XVI
Properties of Impregnated Sorbents
Preparation Conditions
Drying Firing Surface X-Ray Dif-
Temp. Time Temp. Time Atm. Composition Area fraction
Material ~ (hrs) ~ (hrs) - ( approx.) (m2jgm) Analysis
1-0 Basic Alumina* 83% Al203 25.6 Quartz
15% Si02 Al203
1-1 K20-Alumina* 650 7 H2 6.8 Quart z, KN03
Al203
1-2 Na20-Alumina* 125 24 650 6 N2 20.0 Quartz, NaAlll017(wk)
Al203, - Al203 I
---;]
---;]
1-3 NiO-Alumina* 125 24 650 6 N2,H2, Air 24.0 Quartz, NiO I
Al203
1-4 Cr203-Alumina* 125 (overnight)650 10 N2 2.4% Cr203 24.4 Quartz
Al203
1-5 ZnO-Alumina* 125 (overnight)650 14 ~ 22.8
1-6 CuO-Alumina* 125 (overnight)650 11 N2
600 (O.N.) N2
1-7 Mn203-Alumina* 125 (overnight)650 13 H2 7.3% Mn203 24.5 Quartz
Al203
1-8 Fe203-Alumina* 125 (overnight)650 10 N2 Quartz
650 4.5 N2 9.2% Fe203 Al203
600 5.0 N2 22.4 Fe304 (weak)
* Norton Alumina - Catalyst Carrier No. LA-3032
-------
-78-
Para-toluidene was also impregnated on alumina from a boiling ethanol
solution. There was no sorption of S02 at 40oc, 60oc, or 100oC.
3.
Connnercjally Available Sorbents
a.
Florite (Floridin, Company, Pittsburgh, Pa.)
Florite is an activated bauxite containing largely alumina (75%)
with some silica (15%) and Fe203 (3%). Samples of this material were ex-
posed to flue gas. Saturation weight gains were on the order of 1-2%.
Further study is not recommended.
b.
Harshaw CuO-Al~03 Catalyst
The catalyst was supplied with an. analysis of 10% CuO by weight and
a surface area of 137 m2/gm. The sorbent was twice exposed to simulated flue
gas at 450oC. The S02 concentration of the flue gas was nominally 0.25
percent by volume and the NOx concentration 500 ppm. Sorbent loadings of
over 6 gms S02/100 gms sorbent were obtained and saturation had not been
reached. The stoichiometric saturation capacity is 13 grams S02/100 grams
sorbent. Sorption rates were found to be much lower than with any of the
other CuO-alumina sorbents.
c.
Linde Mol-Sieve 5A
Saturation weight gains of 10.5 and 16 grams/100 grams of sorbent
were obtained at 1500C and 1000C respectively, in a dry flue gas containing
1% S02. Saturation was complete after 40 minutes. A saturation weight
gain of 9 grams/100 grams of sorbent was obtained in a dry flue gas contain-
ing 0.2% S02 at 100oC. Saturation required about 150 minutes. Regeneration
was quickly accomplished in nitrogen at 3500C and activity was restored.
It is not known whether the sorption occurs as 802 or 803 so the
weight gain may reflect 802 or 803 pickup. Capacities are reasonable but
the rates are low. Also, it is not known whether water would preferentially
occupy sites filled by SOg in these tests. C02 sorption at 100oC, however,
was found to be only 1 gm/100 grams sorbent in 30 minutes from a gas
containing 15% C02.
-------
GLOSSARY
Term
Sorbent loading
Saturation loading
Stoichiometric capacity
Normalized sorption rate
Average normalized sorption
rate
-79-
APPENDIX II
Symbol
Meaning
w
Weight S02 sorbed/lOa weights
sorbent
Final value of sorbent loading
after exposure to flue gas for
a sufficient period that
weight gain has ceased.
Sorbent loading stoichiometrically
equivalent to conversion of all
metal oxide to its normal
sulfate.
r(w)
l/y dW/dt
r(w)
=
where y =
t
mole % S02 in gas
time, hours
r(w)
r(w)
l/y wit
where
t
length of time
sorption has oc-
curred to achieve
the loading w
=
------- |