EPA-460/3-75-002-a
NOVEMBER 1974
SULFATE CONTROL
TECHNOLOGY ASSESSMENT
PHASE 1, LITERATURE
SEARCH AND ANALYSIS
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ann Arbor, Michigan 48105
-------
EPA-460/3-75-002-a
TECHNOLOGY ASSESSMENT
1,
by
William R. Leppard
Exxon Research and Engineering Company
Products Research Division
Linden, New Jersey 07036
Contract No. 68-03-0497
EPA Project Officer: Joseph H. Somers
Prepared for
ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ann Arbor, Michigan 48105
November 1974
-------
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations as supplies permit from the Air
Pollution Technical Information Center, Environmental Protection Agency,
Research Triangle Park, North Carolina 27711; or, for a fee, from the
National Technical Information Service, 5285 Port Royal Road, Springfield,
Virginia 22161.
This report was furnished to the Environmental Protection Agency by
the Exxon Research and Engineering Company, Linden, New Jersey, in
fulfillment of Contract No. 68-03-0497. This report contains the results
of the Phase I project with Exxon on factors affecting automotive emissions
of sulfates. These results were obtained primarily by an extensive
literature search. The results of the Phase II project with Exxon in this
area will be published later in 1975. The contents of this report are
reproduced herein as received from the Exxon Research and Engineering
Company. The opinions, findings, and conclusions expressed are those
of the author and not necessarily those of the Environmental Protection
Agency. Mention of company or product names is not to be considered
as an endorsement by the Environmental Protection Agency.
Publication No. EPA-460/3-75~002-a
n
-------
TABLE OF CONTENTS
Page
Introduction 1
Summary and Conclusions 2
I. Thermodynamics of Automotive Sulfuric Acid Production 2
A. Thermodynamics of Sulfur Trioxide Production 4
B. Thermodynamics of Sulfur Trioxide Hydration 7
C. Thermodynamics of Sulfuric Acid Condensation 7
II. Reaction of Sulfur Dioxide and Trioxide with Exhaust Gas
Constituents and Exhaust System Components 10
A. Reaction of Ammonia with Sulfur Trioxide 10
B. Reduction of Sulfur Trioxide by Ammonia 12
C. Reduction of Sulfur Trioxide by Carbon Monoxide 12
D. Reduction of Sulfur Dioxide by Carbon Monoxide 13
E. Reaction of Sulfur Oxides with Iron 13
F. Reaction of Sulfur Trioxide with Aluminum Oxide 15
III. Automotive Catalysis of Sulfur Dioxide 17
A. Platinum Catalysis: Industrial Application 18
B. Platinum Catalysis: Automotive Application 27
IV. Sulfate Trap 30
A. Particulate Trap 30
B. Sorbent Trap 30
References 43
iii
-------
- 1 -
Introduction
This literature search was conducted to bring together and examine
the literature pertaining to the fate of sulfur oxides in automotive exhaust
systenn. C^rr^^ly ->v~n'">->K"i ^ ^'"to'Line averaf s ahm sulfur in the
form of organic sulfur compounds. During the combustion process gasoline
sulfur is oxidized to sulfur dioxide. In non-catalyst vehicles, this
sulfur dioxide is emitted to the atmosphere. In vehicles equipped with
oxidation catalysts for control of carbon monoxide and hydrocarbon emis-
sions, further oxidation to sulfur trioxide takes place. This can then
combine with water, forming sulfuric acid. The literature was reviewed to
investigate the thermodynamic potential and kinetics of forming the trioxide
and to examine the fate of both oxides in the exhaust system. Since sulfuric
acid emissions maybe deleterious, the literature pertaining to removal of
sulfur oxides from gaseous streams was reviewed. Stress was placed on the use
of metal-oxide sorbents for this purpose.
To cover these subjects, the body of this report is divided into
four sections. The first section details the thermodyanmics of sulfur
trioxide formation, reaction with water, and condensation. The second section
examines possible reaction with materials in the exhaust gas or system. The
third system reviews the catalytic oxidation of sulfur dioxide on platinum
catalysts. The last section examines possible means of removing sulfur
trioxide from the exhaust stream.
-------
- 2 -
Summary and Conclusions
Thermodynamics of Sulfuric Acid Production
1. At typical oxidation catalyst temperatures, conversions of
sulfur dioxide to sulfur trioxide greater than 507= are
thermodynamically possible.
2. The equilibrium conversion is strongly dependent upon oxygen
concentration. ,At temperatures above 400°C, decreasing the
oxygen concentration decreases the equilibrium conversion,
suggesting a possible control strategy.
3. Thermodynamics and kinetics show that exhaust sulfur trioxide
may hydrate to gaseous sulfuric acid within the vehicles
exhaust system, depending upon driving mode.
4. Thermodynamics show that the gaseous sulfuric acid will begin
to condense at about 150°C which is below the temperature
at the tailpipe exit for all driving modes except startup.
Reaction of Sulfur Dioxide and Trioxide with Exhaust Gas Constituents and
Exhaus.t System Components _ • _ - •
1. Thermodynamics shows that ammonia will reduce sulfur trioxide
to the dioxide. However, exhaust ammonia will be oxidized over
the oxidation catalyst before reaction can take place.
2. The formation of ammonium sulfate is favorable only below 225°C.
3. Thermodynamics shows that the reduction of both sulfur oxides
by carbon monoxide is favorable.
4. Reaction of both oxides with the iron oxide surfaces of the
exhaust system is favorable below 425°C.
5. Reaction of sulfur trioxide with the aluminum oxide catalyst
substrate is possible below 425°C. The presence of carbon
monoxide may lower this temperature by about 50°C.
Automotive Catalysis of Sulfur Dioxide
1. The rate limiting step in the catalytic oxidation of sulfur
dioxide is the surface reaction between adsorbed oxygen and
adsorbed sulfur dioxide.
2. The following rate equation appears to represent best the
available experimental data for industrial catalysis and
should be valid for automotive catalysis
1/2
rate = kl ^PS02 % ~ PSO
1/2 1/2
K02 P02 + KS03
-------
- 3 -
Vr.icrc k, ±z the rate constant, subscripted P's are the
partial pressures of the compounds in the subscripts,
Ke is the equilibrium constant of the oxidation reaction,
and the subscripted K's are the absorption equilibrium
constants for the compounds in the.subscripts. . .
This equation is in accord with the above rate limiting
mechanism. This equation also indicates a possible control
strategy of limiting the amount of oxygen over the catalyst.
The automotive catalysis literature is limited.
In addition the data are confounded by many experimental
problems,notably the storage/release phenomenon. In general,
this literature says that more sulfur trioxide is formed over
catalysts than with non-catalyst vehicles but good
quantitative data are lacking.
Sulfate Traps
1. The most promising means of removing sulfur trioxide from the
exhaust stream is to react it with a basic metal oxide.
2. Based on a selection criterion consisting of seven require-
ments, the most promising sorbent material is calcium oxide.
Other less promising but still attractive sorbents are the
oxides of magnesium, manganese, and aluminum.
-------
- 4 -
I . Thermodynamics of Automotive Sulfuric Acid Production
In assessing the production and fate of sulfuric acid within a
vehicles engine or exhaust system, there are three basic reactions which
must be considered. These reactions are: the reaction of sulfur dioxide
with oxygen to form sulfur trioxide, the hydration of the sulfur trioxide
to gaseous sulfuric acid, and the condensation of the gaseous acid to
produce a finely dispersed aerosol. The first reaction is important as it
shows that the formation of substantial amounts of sulfur trioxide is
thermodynamically possible at typical automotive catalyst temperatures.
The second and third reactions are important since they define the state,
sulfur trioxide, gaseous sulfu.ric acid, or condensed sulfuric acid aerosol,
in which the sulfur trioxide will exist in the exhaust system. This is
important from the point of view of controlling the potential emissions within
the exhaust system. Hydration of the trioxide is possible during some driving
modes. Condensation, however, is unfavorable except possibly at start-up
and idle. '
A. Thermodynamics of Sulfur Trioxide Production
The1 reaction of sulfur dioxide with oxygen can proceed either
homogeneously or catalytically . The homogeneous reaction is extremely
slow i 14 ' and would be neglegible considering the very small residence
times'" within the vehicle, typically on the order of a few seconds. In
the presence of oxides of nitrogen, the rate of sulfur dioxide oxidation
is increased markedly^. A calculation using the rate-constant data from
Duecker and West [" 14 ~" shows that under high NO conditions, such as high
speed or load, the conversion of sulfur dioxide to the trioxide would be
/-^1.0%. Since the reaction rate depends on the square of the NO partial
pressure, the conversion would be much less under low NO conditions. The
' cterogeneous catalysis of this reaction will be discussed later.
No matter how rapid the reaction rate, the maximum conversion
•A be limited by thermodynamic equilibrium. Table 1 [_ 48 j lists the
ree energies and equilibrium constants for the reaction as written
S02(g) + 1/2 02(g) ^ S03(g). (1)
The equilibrium constant for this reaction is
I Pso
Ke
P ' P
fn so.
-------
— 5 —
where P indicates component partial pressure in atmospheres. Figure 1
shows the effect of temperature on the conversion for typical exhaust gas
compositions. Lines of different oxygen concentration are shown to illus-
trate the effect of differing air injection rates upstream of an automotive
oxidation catalyst. As this figure illustrates, increasing the oxygen
content of the exhaust increases the maximum possible conversion to sulfur
trioxide. In the region of typical oxidation catalyst temperatures, 500 to
650°C, the maximum possible conversion is between 90 and 50% when using an
air pump supplying 5% oxygen. Assuming that equilibrium controls the oxida-
tion of sulfur dioxide, operation of the vehicle at stoichiometric or
slightly lean air-to-fuel ratios, either by careful carburetion or by using
a three-way catalyst system, the production of sulfur trioxide could be
TABLE 1
Equilibrium Constants for the Oxidation
of Sulfur Dioxide to Sulfur Trioxide
Temperature
Free Energy
of Reaction
(K cal/mole)
Equilibrium
Constant
(atm -
300
350
400
450
500
550
600
650
700
750
800
850
900
950
1000
1050
1100
-16906.50
-15775.99
-14649.01
-13520.27
-12389.19
-11257.38
-10127.01
- 8999.82
- 7876.73
- 6757.71
- 5642.04
- 4528.59
- 3416.23
- 2304.12
- 1192.02
80.38
+ 1029.68
2.069xl012
7.088x109
1.008xl08
3.682xl06
2.600x10^
2.972x10*
4.892xl03
1.061xl03
2.878xl02
9.313x10J
3.477X101
1.460X101
6.754
3,
1,
1.
389
822
039
6.244x10-!
reduced. For instance, operation at 0.2% oxygen instead of 5% could reduce
the conversion by 25 to 65% over the typical oxidation-catalyst temperature
range. However, at the lower temperatures absolute conversions could still
be as high as 70% at the 0.2% level. Therefore, from an equilibrium stand-
point, reduction of the oxygen content of the exhaust could result in
lowering sulfur trioxide production, however, absolute control would not be
feasible. The kinetics of this reaction will be discussed in the Automotive
Catalysis section.
-------
Figure 1
Equilibrium Conversion
SO (r;)+l/2 0 (;>) - S0.,(
-d Z j
u
I
u
§
1—I
H
O
ro
O
to
o
H
300
400
500
600
700
300
Temp., °C.
-------
B. Thermodynamics of Sulfur Trioxide Hydration
As- the S03 produced by -the engine-or over an oxidation catalyst
proceeds through the exhaust system, it is cooled significantly. As the
temperature drops, the hydration of sulfur trioxide to gaseous sulfuric
acid becomes more favorable as shown in Figure 2, where the fraction of
sulfure trioxide hydrated to gaseous sulfuric acid is given as a function
of temperature. This plot was generated using equilibrium constants
calculated from Eqn- (17) of Gmetro and Vermeulen [20 ' . Hydration begins
at about 450°C and is complete at about 200°C. Typical tailpipe exit
temperatures range from~200 to over 500°C Tour measurements, 24 " for
conditions from idle to extended high speed "driving. Thus, it is possible
for sulfur trioxide to either survive throughout the exhaust system or to
be completely hydrated depending upon the driving mode. Gillespie .and
Johnstone 19 in studying sulfuric acid formation found that dry sulfur
trioxide would for™ an aerosol instantaneously upon contact with moist air
These results imply that the hydration reaction is extremely rapid and
would be controlled by equilibrium in an exhaust gas environment.
C. Thermodynamics of Sulfuric Acid Condensation
The dew point of gaseous sulfuric was experimentally determined
and compared with calculated -thermodynamic values by Lisle and Sensenbaugh
1 39 j. They found excellent agreement. The calculated dew point as a
function of temperature is shown in Figure 3. More recently, Verhoff and
Banchero [ 56 ' have reviewed the literature concerning sulfuric acid dew
points and correlated the data which they felt were accurate. The resulting
least squares correlation is shown in Figure 3 for a water content of 12%.
As this figure shows, both curves predict dew points within a few degrees
of each other in the range applicable to exhaust sulfuric acid levels.
A vehicle operating at an air-to-fuel ratio of 15:1 using an
average fuel of 300 ppm sulfur would produce exhaust containing approximately
10 ppm H2SO/4 assuming 50% conversion of sulfur dioxide to trioxide. At
has ct.-.'CtHitration, condensation would begin at /^J.33°C which is below
the exhaust temperature at the tailpipe exit. The rate of condensation,
as pointed out previously, is very rapid, therefore, should the exhaust
temperature drop below 133 C, condensation of the gaseous sulfuric acid
would immediately occur in the tailpipe. However, for the majority of
driving situations, the sulfuric acid would exit the tailpipe in the
vapor rather than the condensed state.
-------
oo
cvj
a
co
c
o
•H
J-)
O
n)
Figure 2
Equilibrium Conversion
S03(g) + H20(g) == H2S04(g)
1.0
0.8
0.6
oo
I
0.4
0.2
0.0
100
200
300
400
500
Temp., °C.
-------
— 9 —
Figure 3
Pew Point of H2SO,(g)
100 h
01
w
P-4
W
rt
O
O
w
CM
Pu
CM
10 h-
1.0 I
0.1
0.01'
Lisle & Sensenbaugh
Verhoff & Banchero [56]
(for 12%.water)
90
100
110
120
130
140
150
160
Temp., °C.
-------
- 10 -
II. Reaction of Sulfur Dioxide and Trioxide with Exhaust
Gas Constituents and Exhaust System Components _
There are several exhaust gas constituents and system materials
with which the sulfur oxides can react. The two main exhaust gas constituents
capable of reaction with sulfur dioxide and sulfur trioxide are ammonia and
carbon monoxide, both of which are more basic than the sulfur oxides. The
pertinent reactions are:
2NH3 + S03 + H20 j=«te (NH4)2 SC>4 (3)
2NH3 + 3S03 v=i N2 + 3H 0 + 3SO (4)
S03 + CO s=fe C02 + S02
2S02 + 4CO ^ S2 + 4C02 (6)
The first reaction is thermodynamically unfavorable at exhaust
temperatures and, further, any ammonia would be oxidized over the oxidation
catalyst before reaction with the trioxide could take place. The second
reaction is also very unlikely due to the oxidation of the ammonia. The
reduction reactions by carbon monoxide are both favorable but most likely
limited due to kinetics. The reactions of both sulfur oxides with the iron
or aluminum oxide catalyst support are favorable. The aluminum oxide can
alternately form the sulfate and decompose as the catalyst temperature cycles
during transient driving modes producing a sulfate storage/release phenomena.
A. Reaction of Ammonia with Sulfur Trioxide
The free energies and equilibrium constants for Reaction 3 are given
as functions of temperature in Table II. The thermodynamic equilibrium
constant, given by the relationship
(7)
for an exhaust gas containing 10 ppm NH,, 12% H_0, and 10 ppm SO. is
8.3 x 10 5 atm"^. Comparison of this value with the equilibrium constants
in Table II shows that the reaction is favorable only at temperatures below
225°C. Since the equilibrium constants are such a strong function of
temperature, order of magnitude changes in the concentration of either
ammonia or sulfur trioxide will not appreciably alter the temperature at
which the reaction becomes thermodynamically favorable. Temperatures
below 225°C occur in the exhaust only at the tailpipe exit and only during
start up and some extended idle periods. Thus, the production of ammonium
sulfate in the exhaust system would be possible only during these modes.
Ammonium sulfate may, however, be formed after exiting the tailpipe in either
the atmosphere or a particulate sampling apparatus. The thermodynamic
favorability of this occurence would depend simultaneously upon the
cooling and dilution rates.
-------
- 11 -
TABLE II
Free Energies and Equilibrium Constants for
the Reaction of Ammonia and Sulfur Trioxide
Forming Ammonium Sulfate • 3.3 I
Equilibrium
Temp Free Energy Constant
(°K) (cal/molc) (atm"4)
298 . -63,830 6.18x10^
.400 -49,100 6.75x10^
. 500 -34,900 1.80x10^
600 -21,000 4.47x10
700 - 7,400 5.32
800 + 5,800 2.60x10
i
The possibility of ammonium sulfate production Is most severe in
catalyst equipped vehicles where the potential of forming sulfur trioxide
is greatest. However, in these vehicles the oxidation of ammonia over the
catalyst must be considered. The oxidation of ammonia over platinum^td produce
nitric oxide is a very important and well known industrial reaction ,15 1
The reaction is very rapid with almost complete oxidation. I J
TABLE III
Free Energies and Equilibrium Constants for
the Reaction of Ammonia and Sulfur Trioxide _
to form Nitrogen, Water, and Sulfur Dioxide [33j
Equilibrium
Temp Free Energy Constant
(°K) (cal/mole) (atm2)
298 -105,510 2.21x10^
400 -114,100 2.22x10^
500 -122,500 3.54x10^
600 -131,000 5.26x10^
700 -139,400 3.36x10^
800 -147,700 2.25x10.:"
900 -155,900 7.26x10;:'
1000 -164,100 7.36x10
-------
- 12 -
B. Reduction of Sulfur Trioxide by Ammonia
Reaction 4 illustrates a reaction mechanism whereby exhaust
ammonia can be consumed. The equilibrium thermodynamics for this reaction
are given in Table III. The extremely large equilibrium constants show
that this reaction has the potential of almost complete removal of either
the ammonia or the sulfur trioxide, whichever is initially present at
lower concentration. Assuming that the reaction kinetics are sufficiently
rapid, this reaction would then preclude the formation of ammonium sulfate.
Typical automotive exhaust without catalysts contains from one to
six ppm ammonia depending upon driving mode with an average of 2.2 for a
typical driving cycle j 23 :. - Our results with the dual-catalyst system
show that an active automotive oxidation catalyst will readily oxidize
ammonia resulting in tailpipe concentrations typically less than 1 ppm.
Therefore, even at low exhaust temperatures where ammonium sulfate forma-
tion is thermodynamically favorable, only srnall_ amounts could be made.
This agrees wi.th Ford Motor Company's finding 17 that little or no
ammonium sulfate or bisulfate is found in automotive particulate.
C. Reduction of Sulfur Trioxide by Carbon Monoxide
Reaction 5 is a second possible reaction which would lead to a
reduction in sulfur trioxide emissions. The equilibrium thermodynamics
for this reaction, as given in Table IV, were calculated from data presented
in the JANAF Thermochemical Tables ' 2CT ' The equilibrium constants
indicate that, even for carbon monoxide" concentrations in the ppm range,
the reaction is favorable at exhaust gas concentrations. The extent of
the reaction would, however, probably be limited by the reaction rate.
TABLE IV
Free.Energies and Equilibrium Constants for
the Reduction of Sulfur Trioxide by
Carbon Monoxide
Equilibrium
Temp Free Energy Constant
(°K) (Cal/mole) (d linens ionles s)
500 -44,870 4.11x10 Jj?
600 -45,020 2.51x10,;*
700 -45,160 1.26x10^
800 -45,310 2.39xlO,i
900 -45,450 1.09x10,
1000 -46,670 1.59x10
s-
-------
- 13 -
D. Reduction of Sulfur Dioxide by Carbon Monoxide
The last possible reaction is the reduction of sulfur dioxide
to elemental sulfur by reaction with carbon monoxide. The thermodynamic
data calculated from the JANAF Tables r 30j are presented in Table V.
The equilibrium constant for this reaction is written as,
4
Kp = ?C02 (8)
P 4P 2
^co rso2
In a pre-catalyst vehicle, typical exhaust consentrations would be
PCO = 0.12 atm, PSOo = 2 x *°~5 atm' and PCO = °'01 atm> At th&S& levels'
the reduction of sulfur dioxide by carbon monoxide would become favorable
at temperatures below 600°C. In a catalyst vehicle the partial pressure
of carbon monoxide would be typically less than lO'4 atm in whlch^case
the reaction would become favorable at temperatures less than 375°C. Hence,
for both catalyst and non-catalyst vehicles, the reduction of sulfur
dioxide to elemental sulfur is thermodynamically feasible within the
exhaust system.
TABLE V
Free Energies and Equilibrium Constants for
the Reduction of Sulfur Dioxide by Carbon Monoxide
Equilibrium
Temp Free Energy Constant
'K) (cal/mole) (atm"z)
500 -73,390 1.21x10 ,
600 -68,440 8.54x10,
700 -63,560 7.01x10^
800 -58,660 1.06xl013
900 -53,800 1.16x10,
1000 -48,540 4.07x10 u
E. Reaction of Sulfur Oxides with Iron
The most prevalent solid material with which sulfur oxides can
react is the iron of the engine and exhaust system. Since the newer- model
cars and all future oxidation catalyst vehicles are operated net lean,
either by carburetion or by air injection, the internal iron surfaces of
the exhaust system will be oxidized. A typical iron-sulfur oxide reaction
is the sulfation of ferric oxide by sulfur trioxide,
(9)
-------
- 14 -
A similar type of reaction can also be written for sulfur dioxide where
iron sulfite is the product. These types of reaction are industrially
important in the corrosion of iron or steel surfaces in contact with flue
or stack gases containing oxides of sulfur. In this vein, the .sulfation
of iron has been examined by several researchers (35, 36, 24 J .. Warner
and Ing rah am |58, 59 have also investigated the re1, rse reaction in their
studies on the processing of metallic ores.
These studies as well as thermodynamic calculations, show that
the sulfation of ferric oxide by sulfur trioxide in the concentration range
of catalyst vehicles becomes favorable at temperatures below 450°C. Table VI
contains the pertinent thermodynamic data. Therefore, sulfation is possi-
ble over a large fraction of the exhaust system. Furthermore, there will
be a zone in which the temperature will oscillate, as the driving mode
changes, around the temperature at which the reaction is favorable. It is
therefore possible for this zone to either pick up or release sulfur oxides
depending upon the temperature and sulfur oxide concentration.
TABLE VI
Free Energies and Equilibrium Constants for
the Reaction of Sulfur Trioxide with Ferric Oxide
Equilibrium
Temp Free Energy Constant
"' (°K) (cal/mole) (atm-3)
?n
500 -70,730 8.29x10,7
600 -57,790 1.13xlof;:
700 -49,940 3.92x10;.
800 ' -32,170 6.15x10^
900 -19,420 5.20x10?'
1000 - 6,720 2.94x10
Warner and Ingraham P 59 j allude to the mechanism.of ferric sulfate
decomposition. Their work with" decomposition under atmospheres of varying
sulfur dioxide and trioxide compositions indicates that the decomposition
products are more likely to be sulfur dioxide and oxygen than sulfur trioxide.
Therefore, it is possible to store sulfate in the exhaust system
by the reaction of sulfur trioxide with iron and then to release this stored
sulfate as sulfur dioxide and oxygen.
The relative reaction rates of iron sulfation by sulfur dioxide
and sulfur tripxid.e can be assessed from the work of PechkovskyT 34 j and
Krause et. aL(_35 j . Pechkovsky studied the reaction of sulfur dioxide
with various metal"oxides in an oxidizing environment. He showed that
sulfation of some metals, notably magnesium oxide, was very slow with only
sulfur dioxide present. When a small amount of catalyst, such as ferric
-------
- 15 -
oxide, was added, the reaction proceeded much more rapidly presumably due
to the oxidation of sulfur dioxide to the trioxide which u-.tn reacted more
readily. Krause et. al. | 35 , using a radioactive tracer technique,
found sulfur trioxide to ^be on the order of 10^- times more reactive on a
molar basis than the dioxide.
F. Reactions of Sulfur Trioxide with Aluminum Oxide
Another material in the exhaust system of a catalyst equipped
vehicle with which sulfur trioxide can react is the aluminum oxide catalyst
substrate via the reaction
A1203 + 3S03 =^ A12(S04)3. (10)
The thermodynamic data for this reaction [ 33^) are given in Table VII for
gamma form of aluminum oxide. The equilibrium partial pressures for sulfur
trioxide are also presented in this table. These data show that this
reaction is favorable for typical sulfur trioxide exhaust gas compositions
when the temperature is below 425°C. While no quantitive data are available
on the kinetics of this reaction, qualitatively they are rapid enough to
be significant as witnessed by the sulfate storage noted by several
investigators [_ 2, 18, 45~] ,
TABLE VII
Free Energies, Equilibrium Constants, and
Equilibrium Partial Pressures for the Reaction
of Sulfur Trioxide with Gamma Aluminum Oxide
Equilibrium
Free Equilibrium Partial Pressure
Energy Constant of Sulfur Trioxide
(cal/mole) (atm~3) (atm)
111,000 2.34x10^ 7.53x10"^
96,900 8.87x10^ 2.24x10":-"
83,200 2.34x10^ 7.53xlO":J
69,900 2.91x10^ 3.25xlO~*
56,700 5.05x10::, 1.26xlO~?
43,800 9.25x10, 1.03xlo"^
30,900 3.19x10' 3.15x10
18,200 9.50x10 4.72x10
-2
The release of this stored sulfur can also play an important part
in automobile sulfate emissions. As the above reaction indicates, the
aluminum sulfate can decompose back to the oxide with the release of sulfur
trioxide. Such a release would become thermodynamically favorable above
425°C. Wagner and Ingraham have examined the thermodynamics f 58 : and
kinetics f59 J of the decomposition of aluminum and ferric sulfates. As
discussed~above, they found that the rate of ferric sulfate decomposition
was proportional to the sulfur dioxide driving force. That is to say that
the decomposition favors the formation of sulfur dioxide rather than the
trioxide. Although they did not investigate the mechanism of aluminum sulfate
-------
- 16 -
decomposition, by analogy with ferric sulfate decomposition, aluminum
sulfate could decompose into oxygen and sulfur dioxide. Thus, it could
be possible for sulfur trioxide formed on the catalyst to react with the
substrate only to be released later at a higher temperature as sulfur
dioxide. With this possibility, care must be exercised in designing
and interpreting experiments examining the fate of gasoline sulfur
over an oxidation catalyst.
p ~?
Kelley 33 ; presents an alternate route for aluminum sulfate
decomposition in the presence of carbon monoxide. The reaction is
SCO A10 + 3S0 + 3C0. . (11)
The equilibrium constant for this reaction, K ' = ^SOo PCO I ^CO *~s on
the order of 3 x 10^ atm for all temperatures. Since this equilibrium
constant is very large, the partial pressure of carbon monoxide, must be
extremely small before this reaction becomes unfavorable. Thus this
reaction may provide a route for sulfate release. -Kelley feels that the
actual mechanism of this reaction is the decomposition of aluminum sulfate
to sulfur trioxide which is subsequently reduced to sulfur dioxide by
carbon monoxide. He feels that the addition of carbon monoxide may reduce
the decomposition temperature of aluminum sulfate by about 50°C.
-------
- 17 -
III. Automotive Catalysis of Sulfur Dioxide
The only oxidation catalysts presently envisioned for automotive
application are platinum and platinum-palladium catalysts. This section
will, therefore be limited tb these catalysts. The bulk of the literature
covering platinum oxidation of sulfur dioxide pertains mainly to use in
sulfuric acid plants and, tb a much lesser extent, in flue gas sulfur
dioxide control. In these applications, the sulfur dioxide concentrations
are much higher, 5-8% and 1% respectively, than found in typical automotive
exhaust, 20 ppm. Even though the exhaust concentration is orders of
magnitude below the percent level, the kinetic mechanism of catalysis over
platinum should be the same. This suggests that the rate equations would
also be valid. The first part of this section will review this literature.
The second part will review the more limited, recent literature dealing
directly with the sulfur dioxide oxidation over automotive catalysts.
-------
- 18 -
A. Platinum Catalysis: Industrial Application
The use of platinum as a catalyst in the oxidation of sulfur
dioxide was first disclosed in a patent by Peregrine Phillips in 1831. (_41 J
This discovery is considered the foundation of the contact process for
manufacturing sulfuric acid.
It wasn't until the 1900's that quantitative data became
available concerning the rate and the mechanism of the oxidation reaction
over various forms of platinum. This began with the classic work of
Kriietsch [_ 34J in 1901. Knietsch examined the effect of space
velocity and inlet concentrations on the conversion of SC>2 as a function
of catalyst temperature. The catalyst consisted of 0.5 gm. of platinum
supported on 5 to 10 gms. of asbestos. The oxygen concentration was held
constant at 10% and 7 - 20% SCU concentrations were examined over the
temperature range of 300 to 90G°C. For this range of variables, conversions
of from 0 to 99% were obtained. At a given space velocity, the experimental
conversion was low at 300°C. As the temperature was increased, the conversion
increased rapidly reaching a maximum in the temperature range of 400-500°C
depending upon space velocity. Maximum conversions were in the range of
70 to 99% again depending upon space velocity. The lower space velocities
peaked at lower temperatures and at greater conversions. As the temperature
was increased further, the conversions decreased paralleling equilibrium
with the lower space velocities nearer equilibrium. The limited data at
the higher SOo level (20% instead of 7%) showed a decrease in conversion
at identical temperature and space velocity. From these data, Knietsch
duduced that the reaction rate was proportional to the concentrations of
sulfur dioxide and oxygen [_ 61j and is independent of the concentration
of sulfur trioxide. The accuracy of these data have since been questioned
due to Knietsch1s poor temperature control and large conversions.
In another classical study of heterogeneous catalysis, Bodenstein
and Fink {_ 6, 7j investigated the oxidation of SC^ over a platinum
gauze of 0.06 mm diameter wire. This study, conducted over the experimental
temperature range of 150-250°C, produced a reaction rate which is proportional
to the first power of the sulfur dioxide concentration and inversely
proportional to the square root of the sulfur trioxide concentration for
the case where CQQ^/CQ^ = 3. Where this ratio is greater than 3, the reaction
rate is proportional to the first power of the oxygen concentration and
inversely proportional to the square root of the sulfur trioxide concentration.
The usefulness of these rate equations is, however, somewhat limited due to
the very low temperatures investigated.
Lewis and Ries, 6bjecting to the poor control of experimental
conditions by Knietsch and to Bodenstein1s low temperature range, obtained
catalytic data under carefully controlled conditions approximating -actual
contact plant operations. Their kinetic experiments were designed and
conducted so as to approximate a differential reactor, that is, a reactor
in which the conversion is allowed to change only to a small extent. Thus,
the reaction is taking place throughout the reactor with approximately the
same reactant and product concentrations. In addition, the temperature
does not change due to reaction exo- or endothermicity since conversions
are differential. This is the classically accepted method for obtaining
accurate kinetic data.
-------
- 19 -
Lewis and Ries performed three separate series of tests. In the
first series the effect of inlet sulfur dioxide concentration was examined
by varying the amount of sulfur dioxide in an air-sulfur dioxide feed. In
the second series, the amount of oxygen was varied by dilution of the air-
sulfur dioxide feed with nitrogen. The final series examined the effect
of sulfur trioxide on the reaction kinetics. For this series, a platinum
preconvertor was used to generate feeds with varying sulfur dioxide and
trioxide concentrations.
An attempt was then made to fit these data with an equation
obtained directly from the law of mass action
r = k(S02)2 02 - . . (12)
This equation, which they found to fit Knietsch's data, would not fit
their data. They then tried' Bodenstein' s equation,
1/2
r = k S02/S03 (13)
which was somewhat better than the law of mass action but still incapable
of interpreting the data. Several other forms of rate equations were
tried with th6 best results being obtained with the form
r = k PS02 In PeSQ3 PS02 ' (14)
where the superscript e denotes the equilibrium partial pressure.
This rate equation also correlates the data of Bodenstein and Fink
better than the rate equation proposed by Bodenstein and Fink.
Unfortunately., as pointed out by Uyehara and Watson [55J >
the total amount of platinum used by Lewis and Hies was not determined.
The amount used was constant for all of the experiments making the rate
data consistent. However, it is impossible to derive absolute reaction
rate constants since the amount of catalyst is unknown.
Taylor and Lenher ^J used a static platinum hot-wire
technique to examine the approach to equilibrium from both sides at a
temperature of 665°C. They found that the following rate equation best
represented their data
r = ki !g°2 - 2 _ k2(pes03 - PS03) (15)
1/2
PS03
In 1937,Salsas Serra ^O examined the experimental work of
Knietsch and Bodenstein and Fink in light of the law of mass action. In
opposition to the findings of Lewis and Ries discussed above, he found
-------
- 20 -
that the law of mass action adequately represented these data. The law
of mass action for the oxidation of sulfur dioxide yields the following
rate equation,
r = kl p2S02 P02 - k2 p2S03 •
Hougen and Watson [20 j developed general rate equations
for heterogeneous catalysis where one of the elementary reaction steps
in the overall series of steps is assumed to be rate controlling. In
general, the elementary reaction steps, neglecting mass transfer steps,
of the general reaction of A + B to produce C are:
1. The adsorption of either or both reactants
on the catalyst surface.
2. The reaction of A + B either both as
absorbed species or as one absorbed specie
and one gaseous specie.
3. The desorption of the produce from the
catalyst surface. . .
Assuming that one of these steps is rate controlling, i.e., much slower
than the other steps, while the remaining steps are at equilibrium, a
general rate equation can be written for each case. Uyehara and Watson,
in a companion paper [ 65 J , applied this procedure to the catalytic
oxidation of sulfur dioxide. The data of Lewis and Ries were selected
for analysis since they were obtained in an apparatus reasonably
resembling a differential reactor. The data of Knietsch, Bodenstein
and Fink, and Taylor and Lenher were unsatisfactory since they were all
obtained in static systems in xvhich concentration and temperature gradients
were present. Uyehara and Watson examined each possible rate limiting
case and determined that the limiting step is
1/2 p
r=kl(pS02p02 - S°3 )
(1 + KQ2 Pc>2 + KS03 PS03^
where KO~ and KgQ~ are the absorption equilibrium constants for oxygen and
sulfur trioxide respectively. It should be noted that Uyehara and Watson
have dropped the sulfur dioxide absorption term from the denominator of
the general rate equation. Based on Lewis and Reis' first experimental
series looking at the effect of sulfur dioxide concentration, Uyehara and
Watson concluded that this absorption term is negligible. Expressions for
the coefficients appearing in this equation were updated by Hougen and
Watson [ 25 ] by the inclusion of Hurt's data [28.]. Since the mass of
Hurt's catalyst was known, an absolute rate constant could be determined.
These coefficients are:
-------
K03 = exP
- 21 -
(18)
20,360 23.0 , and (19)
—- - --
:- 16,800 17.51 . (20)
KS03 = exp RT - R
Boreskov [ 8 ] found that the data of Bodenstein and Fink,
Taylor and Lenher and Pligunov (unpublished) agreed very well with
the rate equation 0 25 05
PS02 P02' pS03 (2i)
r = k> p 0,5 ~ k2 I0725~
S03 P02
He further states that this .equation can be accounted for by assuming
that the rate limiting step is the absorption of S02 on the catalyst.
This equation was also successfully used by Chesalov and Boreskov £ll,12j in
the study of catalytic activity; of various platinum catalysts.
Roiter et.al [ 47] examined the oxidation near equilibrium
of sulfur dioxide over a platinum screen by means of a tracer technique
using a radioactive sulfur Isotope. The kinetics were also examined
under nonequili.brium conditions using a standard static measurement
technique. The experimental temperature ranged from 600 to 674 °C while
the initial sulfur dioxide level ranged from 2.7 to 6.3%. The data
were then fit to the Boreskov equation with large descrepancies being
noted. A rate equation was then developed assuming that the rate de-
termining step is- the surface reaction of adsorbed oxygen and sulfur
dioxide
2 (22)
This rate equation agreed quite well with the data under equilibrium
and nonequilibrium conditions.
• At this point in the historical review, it is instructive
to examine some of the common features of the studies and rate equa-
tions reported. One important experimental aspect, which is partic-
ularly germane to actual large-scale reactors such as on vehicles, is
the neglect of mass transfer effects in the analysis of the data.
This neglect of mass transfer may be responsible for the difference
in rate equations. For convenience Table VIII summarizes the rate
equations and pertinant experimental conditions. The effect of mass
transfer has been examined in more recent literature and will be dis-
cussed later.
All of the rate equations, with the exceptions of Bodenstein
and Finks second equation and Salsas Serra's equation, predict that the
-------
TABLE VIII
Investigator
Knietsch [ 34 ]
Bodenstein and Fink [6,7 ]
Bodenstein and Fink [ 6,7]
Lewis and Ries [ 37, 38 J
Taylor and Lenher [$Q
Salsas Serra [ 5(7 ]
Uyehara and Watson [ 55.]
Boreskov [ 8 ]
Roiter et al. [ 47 ]
Olsen et al. [ 42 ]
SUMMARY OF RATE EQUATIONS FOR PLATINUM CATALYSIS OF THE OXIDATION OF SULFUR DIOXIDE
Catalyst S02 02 Conversion
Rate Equation . Description Range Range Range
r' = k] Pso2 1?02 Pt on asbestos 7% + 2-% 10% 0-99%
,,-0.5 Pt mesh, 0.06mm _ ,_ ..
r - ki PS02 PS03 diameter PS02/P02 <3
= , p -0.5 " Pt mesh, 0.06mn
r ° ki POZ S03 diameter Pso2/P02 >3
peSOo PS02
rm 1_- i T* 1 11 T°/ T* t- nn n ~Vi r r t" n r -^f} ^v TI i*
0 KJ_ rgQrt JJI /^ r t on asoes tos **\j »j/o air
peso2 pso3
r •= ki (PSo2 - PeS02) PS035 - k2 (peS03 ~ PS03) pt wire
22
•r - ki P S02 P02 ~ k2 PS03 (Analysis of Knietsch [K-8] and Bodenstein [B-4,5]
kl /T>0.5 „ PS03
(1 + PQJ KOO ' + pSOri KS03) '& inclusion of Hurt data) [H-8]
r ° kl PS02 P02 PS03 ~ ^2 P02' PS03 Pt "ire
05 Ps°3
_ t. n\f»J n J TI-. „ __ t o«y /i i n A 1 nnv
r n K "On SO? ~ ^f screen x— JA • wj.y u— IUUA
n ^ S0*3 9
i- m fT* r > /Tl 1 n r "\ • O ow Pi-' nn 1 /° ^ A^ff/ m*T- ' A 70
r D ^"S0o 0? v ' ^ *SOo' • . w.^ rt on i/o D.HJ/. air **— /o
alumina pellets
(50,000 v/v/hr.
Temp.
Range
300-900°C
150-250°C
150-250°C
400-450°C
525-700°C
530-850°C
420-700°C
350-480'C
-------
- 23 -
forward reaction rate is proportional to'the first power of the sulfur
dioxide partial pressure. The functional dependency on the oxygen par-
tial pressure is somewhat less clear. Several of the early equations
are independent of oxygen partial pressure, however, in these cases the
experimental oxygen levels were in large excess. The experimentally
small concentration changes due to reaction were almost insignificant
in comparison with the large unreacted excess. This and other exper-
imental inaccuracies yielded equations which were independent of oxygen
effects. Such was the case of Lewis and Ries' data and analysis. Later
analysis of these data by Uyehara and Watson did however show an oxygen
dependency proportional to the square root of the oxygen partial pressure.
This same oxygen functionality was also determined by Roiter where var-
ious oxygen concentrations were examined.
In essentially all of the experimental work, a marked reduction
was noted with the; presence of even very small amounts of sulfur trioxide.
Two methods of accounting for this retardation were used in the develop-
ment rate equation. The first method was to place the sulfur trioxide
partial pressure in the denominator of the forward reaction rate term.
The second metliod was to include a reverse-reaction term in the rate
equation such as exemplified in the Uyehara and Watson and Roiter
equations. The .second method also provides mathematical tractability
in that the rate can go to zero as equilibrium is approached. The first
method, see for instance the equation of Bodenstein and Fink or Lewis
and Ries, does not account for equilibrium conditions and, therefore,
would be unacceptable.
Of the equations discussed, the best with respect to data
representation, mathematical formulation and generality is that of
Uyehara and Watson. This equation has the general form of the law of
mass action and represents the reliable differential reactor data of
Lewis aiid Ries quite well. In addition, the absolute rate constants
are available where they are not for several of the other equations.
The effect of diffusion on the overall sulfur dioxide conver-
sion of a large-scale reactor can be determined from the work of
Hurt j_28j Hurt used the oxidation of sulfur dioxide over platinum
as an example of his procedure for correlating the performance of
small-scale with large scale reactors. An estimate of the influence of
mass transfer on the overall sulfur dioxide conversion in a GM reactor
can be made using Hurt's kinetic data and mass transfer correlations.
At 40 mph cruise conditions, a typical superficial mass
velocity would be 248 Ib/hr-ft2. This represents a particle Reynolds
number of 30 for 1/8" pellets. Using'Hurt's kinetic correlation
and mass transfer correlation, the overall conversion is influenced
approximately 25% by mass transfer and approximately 75% by kinetics.
Although these values may be in error due to kinetic differences in
Hurt's catalyst and automotive catalysts, it does illustrate that
-------
24 -
mass transfer effects can play a significant role in the formation of
automotive sulfates. It also points out that care should be exercised
in designing and analyzing kinetic experiments particularly when the goal
of the experiment is to predict large-scale reactor behavior from small-
scale reactor experiment's.
Smith et al. [I, 22, 29, 42, 52} completed a very compre-
hensive program involving the design of fixed bed catalytic reactors
taking into account heat and mass transfer. The program involved measurement
of the chemical kinetics in a differential reactor, examination of
diffusional effects, examination of heat transfer effects, and mathema-
tical modeling of an integral reactor incorporating kinetics and heat
and mass transfer. The results are of particular interest since the reaction
used was oxidation of sulfur over a catalyst of 0.2 weight percent
platinum on 1/8" alumina pellets. The platinum was applied such that it
penetrated only the outer 1/32" of the pellet. This platinum loading
and geometry approximates that of the GM catalyst. In addition, the range
of temperature and space velocities spans the range of interest in auto-
motive applications. However, the sulfur dioxide levels were constant
and typical of acid plant operations, 6%, and the oxygen levels were
those of air. Thus, the only variable investigated was conversion. Overall
conversions up to 70% were examined in one paper by using a preconverter
ahead of the differential reactor [42 ]. The maximum conversion over the
differential reactor was in all but two cases less than 30%.
Olsen et al. [ 42 ] chose to model their results with the rate
equation of Uyehara and Watson [ 55 ]. Since the inlet concentrations to
the preconverter were always the same, there existed a definite relationship
between PSO?' PSOo> ant* P02* Therefore, the Uyehara and Watson equation
could be simplified, using the relationship, to
1/2 1/2
pS02 P02 - 3 = B + D PS0, (23)
where B and D are constants dependent only upon temperature. Olsen, et al.
found that this equation was an excellent representation of their data
when the component partial pressures were taken at their interfacial
values and not their bulk values, that is to say that diffusion is taken
into account.
In a later paper of this series, Argo and Smith [ 1' ] obtained
differential reaction rates using larger catalyst pellets than Olson, 1/4"
in place of 1/8". These data were then correlated with the Olson data
by an activity factor in conjunction with the rate equation of Olson.
The activity factor and rate equation correlated both sets of data very
well again illustrating that this general equation is capable of describing
the catalytic oxidation of sulfur dioxide by platinum.
-------
- 25 -
Olson also examined five other rate controlling mechanisms. The data were
correlated by rate equations derived from the following rate limiting
elementary steps:
1. Reaction of adsorbed sulfur dioxide and
adsorbed oxygen (this mechanism yields the
Uyehara and Watson rate equation),
2. Reaction of adsorbed sulfur dioxide with
gas-phase molecular oxygen, or
3. Reaction of adsorbed oxygen and gas phase
sulfur dioxide.
Unfortunately, the data could not distinguish among these three steps. The
data were, however, not correlated by any mechanism which assumes that
adsorption or desorption is the rate limiting step.
The differential reaction data.were also analyzed by Hurt's [28 ]
method which also accounts for the effect of diffusion. This method failed
to correlate the data with the primary objection being that the resistance
due to reaction at the catalytic surface was not independent of mass
velocity. Olson found that the diffusional effects were better accounted
for by using the mass-transfer correlations of Hougan and Wilkie [20 ]•
Using these correlations, differences in the sulfur dioxide and trioxide
partial pressures between the bulk gas and the catalytic surface were as
great at 40% depending upon gas mass velocity, degree of conversion, and
temperature. For a GM reactor running at 40 mph, partial pressure
differences on the order of ^ 10% would be predicted for both sulfur dioxide
and trioxide. Therefore, this treatment in qualitative agreement with
Hart's method indicates that mass transfer can be playing a role in the
production of sulfuric acid over the GM type of automotive catalyst.
The mechanism of the oxidation of sulfur dioxide over platinum
was examined by Kaneko and Okanaka J31,32J using a radioactive tracer
technique. The experiments were conducted near 400°C in a recirculating
reactor using a platinum gauze catalyst of 0.1 mm diameter wire. The
kinetic mechanism was examined on both sides of equilibrium, i.e. oxidation
of sulfur dioxide and decomposition of sulfur trioxide. The reaction
mixture contained only oxygen, sulfur dioxide, and sulfur trioxide with
the oxygen and sulfur dioxide always in stoichiometrie amounts. In one
series of experiments, radioactive sulfur was used to follow the reaction.
In the second series an isotope labelling of oxygen was employed. The combined
results of both series show that the rate-determining step is the surface
reaction of adsorbed oxygen and adsorbed sulfur dioadLde.
In summary, the literature covering the mechanism of sulfur
dioxide catalysis by platinum indicates the rate limiting step is the
surface reaction between adsorbed oxygen and adsorbed sulfur dioxide.
-------
- 26 -
The rate equation which produced the best representation of
experimental data was that of Uyehara and Watsonf 55j which is in
agreement with the- above rate controlling mechanism. Most .of the
other rate equations arc special cases of the more general Uyehara-Watson
equation which further.substantiates its validity.
There are two interesting points which deserve mention when
applying this rate equation to the area of. automotive sulfuric acid
production. First, since the rate equation is first order with'respect
to sulfur dioxide partial pressure, the sulfuric acid production will
be linearly proportional to the fuel sulfur level. The second important
point addresses the problem of controlling acid emissions. Since the
rate equation is proportional to the square root of the oxygen partial
pressure, catalytic systems which operate with lower oxygen partial
pressures will produce lower sulfuric acid emissions. Thus, a catalytic
system, such as the three-way catalyst, which, operates with no net oxygen
partial pressure would minimize acid emissions.
-------
- 27 -
Platinum Catalysis; Automotive Application
Although the literature is void of reaction rate and reaction
mechanism studies which "relate directly to exhaust oxidation of sulfur
dioxide, there are several references covering phenomenological observation
of the catalysis of exhaust sulfur dioxide. Chrysler [13], Ethyl [16 ],
Ford [ 17] , and General Motors [ 18 ] have made public their data on
sulfate emissions as requested in the March 8, 1974 Federal Register. In
addition, Pierson et^ _al_. [45] and Beltzer et_ a^. [ 2 .] have presehted
SAE papers dealing with automotive sulfate emissions. All of these
references, with the exception of General Motors, have examined platinum
catalysis only from the standpoint of tailpipe emissions of sulfuric
acid. General Motors has conducted a limited amount of laboratory work
aimed directly at elucidating the kinetics.
In reviewing these references, there is, for the most part, a
common problem when catalysts are employed in not being able to obtain a
balance between the sulfur burned in the fuel and the sulfur emitted from
the tailpipe. When catalysts are not used, balances can be obtained.
Recently, however, some investigations have had problems in closing a
sulfur balance on non-catalyst vehicles. Since the majority of- the tail-
pipe sulfur in a non-catalyst vehicle is sulfur dioxide and a sulfur
balance can be made, the analytical and test procedures for sulfur
dioxide can be assumed to be valid and accurate. The ability then not
to be able to close a sulfur balance on a catalyst vehicle is due either
to an experimental or analytical problem or to the storage phenomena
discussed previously. This problem makes it difficult to assess
quantitative catalytic effects from observed tailpipe emissions.
Ford [ 17, 45] has been able to make sulfur balances for
Englehard IIB-catalyst and non-catalyst vehicles operating under steady
cruise conditions. They found that at 60 mph both the Engelhard IIB
and GM catalysts converted approximately 44% of the sulfur dioxide to
trioxjdc which is close to the equilibrium conversion at the test conditions.
The GM catalyst at 30 mph converted 84% of the sulfur dioxide which is
again very close to the equilibrium value. These conversions were all
calculated on the basis of total sulfur out of the tailpipe, thus
eliminating the storage problem noted with the GM catalysts.
Ethyl Corp. [ 16] found that under 40 mph cruise testing
monolithic platinum catalysts converted 43% of the fuel sulfur to exhaust
sulfate where conversion is based on total sulfur emitted. Their tests
showed that approximately 58% of the fuel sulfur was emitted with the rest
being stored. This conversion cannot be compared with equilibrium since
the catalyst temperature was not given. The magnitude of these results
are in line with Ford's findings.
Beltzer, et al., [ 2 ] have also measured sulfate emissions for
steady cruise conditions." Sulfur dioxide measurements were not made
precluding making a sulfur balance. The sulfate emissions at 60 mph
-------
- 28 -
for the pelletized oxidation catalyst agree fairly well with those
obtained by Ford using approximately the. same fuel sulfur level. If
the storage effects are .similar then by implication, this conversion
is also close to equilibrium. ;
The GM [18 ] extended 60 mph cruise tests on their catalysts
show a large spread both in sulfate conversion and in sulfur'balances.
In general, average conversions based on sulfur emitted were in the range
of 20 to 40%. This is somewhat lower than the conversions discussed
above, but still represents a significant approach to equilibrium.
The Chrysler reference [13 ] did not presen't conversion data
for steady cruise conditions. A consensus of these references shows that
the reaction rate is of sufficient magnitude that the oxidation of•the
exhaust sulfur dioxide approaches equilibrium in actual
vehicle operation. Due to the storage effect together with possible
analytical problems, a more quantitative assessment of the reaction rate
is impossible.
In addition to their vehicle work, GM [ 18 ] also reported some
preliminary parametric studies using a laboratory-scale reactor. They
also caution the reader that the analytical methods caused significant
uncertainties in the sulfur trioxide measurements. The storage problem
was also a large factor. In one instance less than 30% of the inlet
sulfur could be accounted for. . • . •
GM examined the conversion both as a function of temperature
and space velocity using actual Vehicle exhaust. They found that the'
conversions, based on measured sulfur trioxide and average sulfur input,
were between 5 and 15% with only a weak temperature dependency. From
analysis of these data, with respect to temperature and space-velocity
dependency, it was concluded that the reaction is kinetically controlled*
The effect of platinum loading was also investigated by increasing the
loading from 0.1% to 1.0% platinum. The observed effect was only a very
slight increase in measured sulfur trioxide. This finding is in contra-
diction to a kinetically controlled reaction. If the reaction were
kinetically controlled for both loadings, then an order of.magnitude
change in platinum loading would change the rate by an order of magnitude.
Invariance to loading conforms with either a diffusion or equilibrium
controlled reaction.
It is informative to examine these data assuming that the outlet
sulfur trioxide is the difference between the outlet sulfur dioxide and
the total inlet.sulfur. This assumption could be valid if the sulfur
trioxide analyses were in error and no storage was taking place. Under
this assumption, the conversions are all very near equilibrium indicating
that the reaction is equilibrium controlled. Some credibility must be
assigned this probability since GM's platinum loading data also indicate
an equilibrium controlled reaction. However, since the storage problem is
real and the data severely limited, this assumption .may not be -entirely
valid. None the less, the possibility exists that more sulfur dioxide
is being produced than the GM data indicate.
-------
- 29 -
The form of the rate equation describing the catalysis reaction
can, to a limited extent, be deduced from the above studies. The results
of Seltzer, Ford, and GM show that sulfate emissions increase with gasoline
sulfur level. Beltzer and Ford find that conversion appears to be invariant with
respect to fuel sulfur level which means the reaction rate is proportional
to the f^rst power of the sulfur dioxide partial pressure in agreement with
the rate equations previously discussed. The quantitative effect of oxygen
partial pressure can be assessed from GM's vehicle and laboratory studies.
These studies show that reducing the partial pressure of oxygen over the
catalyst reduces sulfate emissions and the reaction rate. The.quantitative
dependency on the reaction rate cannot be determined due to the limited
data of dubious accuracy. " ,
In summary, the use of a platinum automotive-exhaust catalyst
does result in oxidation of fuel sulfur to yield sulfur trioxide. The
extent of this oxidation is clouded by the problems of -sulfur storage and
inaccurate analytical techniques. Results showing anywhere from 10% to
complete approach to equilibrium have been observed. The variability of
these data precluded any quantitative analysis of the reaction rate, however,
the data did indicate the form for the reaction-rate equation.. The rate
equation should be linear with respect to sulfur dioxide partial pressure
and should be proportional to some positive power of the oxygen partial
pressure. This form is in agreement with the rate equations .obtained in
the industrial catalysis section.
-------
- 30 -
IV. Sulfate Trap
There.are two conceivable means of removing the sulfuric acid
from a vehicle's exhaust system. The first is by. tha^use- of paniculate
traps in which the condensed acid is removed as droplets. ; The,-second is
by reaction of the acid either as sulfur trioxide, gaseous su3.fur.ic acid,
or condensed, liquid sulfuric acid with a suitable sorbent. The. capacity
of either type of trap can be conservatively estimated., f or .5Q,QOO miles by
assuming an overall fuel- economy of 10 miles per gallon.; .. If., the ..fve'l sulfur
level is taken as the industry average;.of .0.03 .weight percent.,»and, ifc is
assumed that the worst possible-case of total conversion exists, then. 4.2 kgm
or 131 gm. moles, of sulfur v/ill be consumed and must-be trapped-.
The particulate trapping technique has several serious problems
associated with it which preclude its use in automotive applications. The
sorbent trap method is more attractive. Consideration of sey;er.al important
properties which a sorbent material must process .shews that- the most
promising sorbent material is calcium oxide. Other attractive, -sorbents are
the oxides of magnesium, manganese, and aluminum.
A. Particulate Trap
There are several very serious problems associated with collecting
these sulfuric acid emissions with a particulate trap. The largest
problem is that the sulfuric acid exits the tailpipe in the gas.-phase for
the majority of driving conditions. This has been discussed previously.
Therefore, for a particulate trap to be feasible, an exhaust heat exchanger
would be required. Assuming condensation is possible, then the condensed
acid must be separated from the exhaust gas by some means. This separation
will be exceedingly difficult due to the extremely small particle size of
the condensed acid. Typically, the particle sizes for the effluent of
acid plants is less than 2 microns for 85 to 90% of the acid by weight [53]
Fprd [17 ] found that >90% of the exhaust sulfate mass was less than 0.25
microns. The problems associated with this separatioa would be almost
unsurmountable given the space limitations and low back-pressure requirement. Assumi
these problems can be overcome, the problem then becor.es one of containing
the liquid acid. Assuming the acid would be diluted by 50 weight percent
water, the volume of acid solution collected over 50,000 miles would be
18.5 liters. Not only must this be contained, it must be protected from
further dilution with water during cold starts and shut downs. Based on
all of fhese problems, it is not feasible to use particulate traps to
remove automotive sulfate particulate.
B. Sorbent Trap
The second method of reducing exhaust sulfate particulate is to
trap the sulfur trioxide or sulfuric acid by chemical reaction with a
solid sorbent material. In this manner, the potential sulfate particulate
is trapped and stored in the exhaust system as a solid material. The
obvious sorbents to consider for this application are solids which are
chemically basic. While there are only a few references in the literature
concerning the reaction of basic materials with either sulfur trioxide or
gaseous sulfuric acid, there are numerous references to reactions with
sulfur dioxide. The majority of these references are
concerned with the selection of potential sorbent materials for the
-------
- 31 -
reaction of -either oxide is a gaseous acidic component reacting with a
basic solid component and since the trioxide or gaseous acid is more
acidic than the dioxide, then it would be expected that suitable sulfur
dioxide sorbents would also serve as suitable trioxide sorbents. In fact,
several researchers [ 3, 44 ] have found that the reactivity of the
trioxide is orders of magnitude greater than that of the dioxide. Thus,
the literature of sulfur dioxide removal will be used as a basis for
selecting potential sulfate trap sorbents.
In selecting potential sorbents for automotive exhaust application,
there are several important factors which must be considered. These factors
are, 1 -• activity for sulfur trioxide
or sulfuric acid removal, 2 - thermal stability, 3 - volume and weight
restrictions, 4 - potential side reaction, 5 - water solubility, 6 - cost
and availability of sorbent, and 7 - toxicity of fresh or sulfated sorbent.
The importance of the first factor goes without saying, a sorbent must
be able to sorb over the temperature range encountered in automotive
exhaust • The thermal stability of
the sulfated sorbent is also vital to potential sorbents. The sulfated
sorbent must not begin to decompose at temperatures below the maximum
expected operating temperature of the catalyst. With this restriction,
sulfafed sorbents which decompose at temperatures below 800°C will not be
considered,
Since the sulfate trap must be located in the exhaust system on
the underside of the vehicle where space is limited volume and weight
restrictions are important. Another important factor to consider in the
selection of sorbent materials is the possibility of side reactions with
exhaust gas constituents such as water or carbon dioxide to form stable
compounds. Such reactions would use potential sulfate sorbent and decrease
the sulfation capacity of the trap.
The water solubility of the fresh and sulfate sorbent material
must be considered. There are several conditions in which part or all of
the sulfate trap could be exposed to liquid water. If the sorbent is
appreciably soluble, then the potential sulfate capacity can be decreased.
This leaching of material can also lead to problems in maintaining the
structural integrity of some possible trap configurations.
Since a successful sulfate trap has the possibility of being
installed on millions of vehicles, the cost and availability of the
sorbent must be considered. Thus, expensive metals such as gold or silver
or materials available in limited quantities such as various rare earth
elements cannot be considered.
Since the trap represents a potential source of particulate
emissions due to fresh or sulfated sorbent attrition, the toxicity of the
fresh and sulfated sorbent as well as other possible sorbent compounds
must be considered. For instance, beryllium would be ideal from the
volume-weight; aspect due to its low molecular weight and divalency.
However, it ant} its compounds are extremely toxic [ 51 ] and, hence,
cannot be considered as potential sorbents.
-------
- 32 -
With these factors in mind, the literature has been reviewed
to select possible candidate sorbents to be used in a vehicle sulfate
trap. There are several comprehensive references [ 8, 4, 21, 4,0
44, 59, 60 ] in the literature in which a wide variety of potential
sulfur dioxide sorbents have been examined. These studies were all aimed
at rempving sulfur dioxide in the concentration range of 1 to -5 .percent
from a stack gas. Although these studies .are for a higher concentration
and a different oxide of sulfur, the activity results can be; .readily
extrapolated to the sorption of sulfur trioxide from automotive exhaust.
The major chemical classification of the sorbent cpmpoun'ds
investigated are the metal oxides. The sorbent material is; envisioned
as being installed in the system either as. the oxide or as the metal which
will readily oxidize in the oxidizing exhaust atmosphere. The-, general
sorption reaction can be written as
MeO + S03 (g) ^ Me S04 (24)
where Me is a general symbol representing any sorbent metal. Lowell, et al.
[ 40] evaluated the oxides of 47 elements for potential use,as~sulfur
dioxide sorbents. Their thermodynamic calculations show that-the;formation
of sulfates for nearly all of the elements considered is favorable.
Pechkovsky [44 j examined the rate of sulfation of several metal oxides
in powder form over a temperature range of 400-1000°C. He found that
calcium oxide was more active than magnesium oxide which was more active
then zinc oxide.
The Bureau of Mines [ 4 .] conducted a program looking at the
activity of several bulk oxides with respect to their ability to sorb
sulfur dioxide. Experimentally, a bed of sorbent, 8-24 mesh .particlesj
was exposed to a synthetic flue gas with 0.3 volume percent, sulfur dioxide.
The space velocity was maintained at 1,050 v/v/hr and temperatures- of 130
and 330°C were examined. Typical results showed that 'active, sotp'tion ' .
materials would remove essentially all of the sulfur dioxide for a period
of time. As sulfation of the sorbent preceded, a point .was -reached where
breakthrough would occur. From this point, the fraction of. su'lfur dioxide
removed was found to decrease linearly, with time.
The sorbent materials were rank ordered with respect':to* the
amount of sulfur dioxide removed per unit mass of .sorbent at the point
of 90% removal of sulfur dioxide. The most active.sorbents inr.order of
activity were the oxides of manganese, cobalt, and copper.
The temperature study showed that for most sorbents approximately
twice as much sulfur dioxide had been sorbed up to the 90% breakthrough
point at 330°C than at 130°C. At the higher temperature, it was also
noted that the sulfur dioxide had made a significant penetratidn into the
particles of the more active sorbent materials.
-------
- 33 -
Vogel, e^ al. evaluated the activity of several metal oxides
supported OH alumina. All test samples
were made with the same metal equivalents so the various materials could
be directly rank ordered as to activity. Samples of each
supported sorbent were e-xposed to a synthetic stack gas and the outlet
was continuously analyzed for sulfur dioxide. From these data, the
sulfate loading of each sorbent was determined for the conditions where
the outlet sulfur dioxide concentration was five percent of the inlet
concentration. In addition, the maximum loading was determined by
extrapolation of the data.
The materials were then rank ordered as to the percent of
sorbent reacted at the five percent breakthrough point. This ranking
also agreed with the ranking based on percent of sorbent reacted at
maximum sulfation capacity with one exception. The one exception was
the Bureau of Mines alkalized alumina sorbent which was included in this
study as a benchmark. This material has been extensively studied
[!4, 5, 43, 40 ] as a sulfur dioxide sorbent but would not be
acceptable as a vehicle trioxide sorbent due to its low sulfation capacity
per unit volume.
The activity of the sorbents in order of decreasing activity is:
the oxides of sodium, strontium, copper, calcium, and chromium. All of
these materials had in excess of 50% of the sorbent reacted at the 5%
breakthrough point. These were followed by the oxides of: barium, lead,
cadmium, manganese, magnesium, iron, cobalt, nickel, and zinc. Tin
and vanadium oxides showed no apparent activity. The alumina
substrate was also tested and found to be completely inactive. In general,
these results show that the alkali end alkaline earth metals exhibit
highest reactivities.In addition, copper and chromium showed good
reactivity.
Welty [ 60 ] conducted a theoretical study of the reactivity of
potential sorbent cations based on a characterization factor consisting
of the cation radius, electronegativity, and valence. Using this factor,
the alkali metals are the most promising sorbents followed by the alkaline
earth metals, then by various transition metals. This reactivity scale is
in general accord with the experimental and theoretical works described '
previously.
The second necessary property a sorbent must have is thermal
stability of the sulfated material. Although the temperature regime
the sorbent sees in vehicle use can to some extent be controlled by
location in the exhaust system, the sulfated material must have a higher
decomposition temperature than it is expected to experience. Temperatures
at or near the exit of the exhaust pipe can be as high as 800°C under
sustained high speed driving of vehicles equipped with oxidation catalysts.
Therefore, the sulfated sorbent must have a decomposition temperature at
or above 800°C.
-------
- 34 -
Table IX lists the decomposition temperatures of several sulfated
sorbents. The temperature ranges and slight disagreement between references
are due to the difficulty in experimentally determining decomposition
temperature and in the different experimental methods employed. These
results show that the sulfated alkali and alkaline earth metals all
show acceptable decomposition temperatures. The transition metals.,
however, show borderline temperatures, particularly aluminum oxide.
If these materials are to be considered, their installation would have
to be limited to points as far from the oxidation catalysts as practical.
As estimated above, the sulfate trap must have the capacity to
react with 131 gm. moles, of sulfur. Table X lists the mass of typical
sorbents which would fulfill this requirement. In addition, the volume
of sorbent as calculated using the crystalline density is included. In
the cases where more than one crystal structure exists, an average
density was used. There are several important general conclusions which
can be arrived at from an examination of this table. Within a given
group of the periodic table, the required amount of the lower molecular
weight sorbents are lighter and of lower volume. For instance, the
required mass and volume of sodium oxide are 8.12 kgm and 3.58 liters
whereas the higher molecular weight cesium oxide requires 36.9 kgms and •
8.69 liters. -
Another important consideration is the valence of the>sorbent
cation. For instance, two univalent cations are required for each sulfate
anion whereas only one divalent cation is. If these cations have .
approximately the same molecular weight, then the divalent one would be
preferable. An excellent example is the third period of the periodic
table. The first three members of this period, sodium, magnesium, and
aluminum, have nearly the same atomic weight but they are, respectively,
univalent, divalent, and trivalent. The required amount of sodium oxide
is 8.12 kgm, of magnesium oxide is 5.28 kgm, and of aluminum oxide is
4.45 kgm. The volumes likewise decrease. Therefore, to minimize•trap
in';- . and volume, lower cation molecular weight sorbents with higher valence
states are preferred.
There is another problem which must be considered in the selection
of a sorbent material and in the engineering of the sorbent structure to
be installed in a vehicle. This problem is the increase in volume of
the sorbent as sulfur is picked up. The magnitude of this problem, is shown
in Table X where the volumes of the totally sulfated sorbents are given
along with the ratio of the sulfated volume to the fresh sorbent volume.
This volume increase is a result of two compounding factors. First, the
mass of the trap is continuously increasing due to the pickup of sulfur
trioxide and dioxide. Second, the crystalline density of the sulfate is
always less than the corresponding oxide.
-------
- 35 -
Table IX
DECOMPOSITION TEMPERATURES OF SULFATED SORBENTS
Compound
Decomposition
Temperature
Determination
Method
Reference
Sodium
Potassium
Cesium
Magnesium
Calcium
Barium
Aluminum
Manganese
Iron
>800°C
>800°C
>800°C
750°C
890-9 72°C
1180
>1200°C
>800°C
f 652
/ 590-639°C
|650-950°C
699-790°C
880-1100 °C
630
781-810°C
702-736°C
700-840°C
Thermodynamic Calculation
In air flow
Thermodynamic Calculation
In air flow
Vacuum
In air flow
Inert gas flow
In air flow
In air
Vacuum
Inert gas flow
In air flow
In air
40
40
40
60
3
60
3
40
58
3
40
3
40
58
40
3
40
-------
- 36 -
Table X
MASS AND VOLUME REQUIREMENTS OF
Sorb en t
Material
Na2°
V
Cs20
MgO
CaO
BaO
A12°3
MnO
Fe2°3
CuO
ZnO
MgC03
CaC03
BaC03
Na2C°3
K CO
Mass
Req'd.
(kgm)
8.12
12.34
36.9
5.28
7.35
20.1
4.45
9.29
6.97
10.4
10.7
11.0
13.1
25.9
13.9
18.1
Vol.
Req'd.
(lit)
3.58
5.32
8.69
1.48
2.21
3.51
1.12
1.70
1.33
1.63
1.90
3.73
4.65
5.84
5.48
7.46
Sulfated
Mass
(kgm)
18.6
22.8
47.4
15.8
17.8
30.6
14.9
19.8
17.5
20.9
21.1
15.8
17.8
30.6
18.6
22.8
VARIOUS SORBENTS*
Sulfated
Vol.
(lit)
6.94
8.58
11.2
5.93
6.46
6.79
5.51
6.09
5.64
5.80
5.97
5.93
6.46
6.79
6.94
8.58
Sulfated Volume,
Expansion Ratio
1.94
1.61
1.29
4.00
2.92
1.94
4.92
3.58
4.24
3.56
3.14
1.59
1.39
1.16
1.27
1.15
*Assuming 131 gm. moles of sulfur to be trapped.
-------
- 37 -
Unfortunately, the magnitude of this volume increase problem
increases for the more desireable (from initial volume and weight con-
siderations) sorbents. For instance, aluminum oxide has the lowest volume
required, yet has the highest volume increase ratio, to the other
extreme, cesium oxide has the largest volume requirement, yet the lowest
expansion ratio.
One means of minimizing this problem is to provide for this
expansion chemically by using a sorbent which gives up some mass as it
picks up the sulfur oxides. This exchange reaction must release a compound
which itself is not a deleterious emission. One class of ideal exchange
sorbents are the metal carbonates which release carbon dioxide in exchange
for sulfur oxide . Several potential carbonates are listed in
Table X.. where their volume and mass requirements are shown. It is
immediately obvious that the volume and mass of the carbonate sorbents
is much greater than for the oxide, however, the volume expansion is
drastically reduced. Use of the carbonates would simplify the engineering
of the physical sorbent structure by eliminating the need for large volume
expansions. This simplification more than outweighs the initial increased
volume of the sorbent carbonates over the oxides.
There are two side reactions of the sorbent with exhaust gas
constituents which need to be considered in selecting potential sorbent
materials. The first side reaction is the reaction of the sorbent oxide
with carbon dioxide to form the sorbent carbonate. This reaction is
important for several reasons. If the carbonate is unreactive towards
sulfur trioxide, then formation of carbonate decreases the potential
sulfation capacity of the trap. If the sorbent carbonate is reactive,
it is possible that the reactivity will be lower than for the sorbent
oxide which would lower the reactivity of the trap. Another problem
is that the formation of carbonate causes a volume expansion. This
expansion within a sorbent particle causes the pore volume to decrease
which, in turn, can hinder diffusion of the sulfur trioxide into the
particle. Therefore, the potential amount of internal sulfation will
.be decreased.
The best example of carbonation is the reaction of calcium
oxide. Thermodynamic calculations show (see for instance Ref. [ 46^)
that the carbonation of calcium oxide is favorable in an exhaust gas
environment at temperatures below 760°C. Thermodynamic calculations also
show that the calcium carbonate formed will react with the oxides of
sulfur. Thus, the possibility of carbonate formation exists and will
be competitive with the sulfation. If the carbonate is formed, it can
also reaqt with the sulfur oxides. However, the possibility exists
that this reaction will be significantly slower than the reaction with
the oxide which would lower the overall activity of the trap. There
are no rate data available which are applicable to exhaust gas con-
centrations, and temperatures. Therefore, the possible activity reduction
cannqt be determined or estimated.
-------
- 38 -
Table XI
SOLUBILITIES OF VARIOUS FRESH AND SULFATED SORBENTS
SOLUBILITIES IN GRAMS PER 100 ML. OF COLD WATER
Sorb en t
Material
Sodium
Potassium
Cesium
Magnesium
Calcium
Barium
Aluminum
Manganese
Iron
Copper
Zinc
Oxide
Solubility
d.*
v.s .
v.s .
0.00062
0.131
3.48
i
i
i
i
0.00016
Carbonate
Solubility
7.1
112
260.5
0.0106
0.0015
0.002
0.0067
i
0.001
.Sulfate
Solubility
S.
12
167
26
0.209
0.0002
31. .3
52
si, s.
1^.3
S
* d. - dissolves
v.s. - very soluble
S. - soluble
sl.s - slightly soluble
i - insoluble
-------
- 39 -
The amount of pore volume reduction due to volume increase upon
carbonation can be estimated from Table X. For the same equivalent of
calcium, the carbonate requires 2.10 times the volume of the oxide.
Therefore, if the initial particle of calcium oxide had a porosity of
50% ,complete carbonation would fill all of the pore volume assuming all
expansion is internal and no reaction occurs with sulfur oxides.
These same problems are also present with essentially all
potential sorbent oxides. Since the desired type of sorbent reaction, a
basic sorbent with the acidic sulfur oxides, is the same as for car-
bonation, potential sulfur oxide sorbents will also react with carbon
dioxide.
The second detrimental side reaction is that of either
the fresh or sulfated sorbent with water. The main problem associated
with the reaction of water is the volume expansion of the bed at low
temperatures such as during startup. Several of the sorbent oxides can
react with water to form hydroxides and most of the sulfated
sorbents are highly hydroscopic and form hydrates with as many as 18
molecules of water per molecule of sulfated sorbent (i.e., Al2(S04)3'18 H20)
These added water molecules cause a volume increase which can plug pores
and even increase pressure drop across the catalyst bed. Since the water
of hydration is driven off even at very low temperatures, the problem of
volume increase would be applicable only at startup.
Both possible, side reactions apply to almost all potential
sorbents. The degree to which these reactions would decrease the activity
and/or capacity of a sulfate sorbent is a very complex question which
cannot be answered without direct experimentation under exhaust gas
conditions. Since all potential sorbents could exhibit these problems
to some unknown degree, a selection of potential sorbents using the
criterion of deleterious side reactions is impossible.
Since the sulfate trap must operate in an environment where
contact with liquid water is highly probable at many times during the life-
time of the trap, the fresh and sulfated sorbent should be insoluble.
Dissolution of the fresh sorbent would decrease the total sulfation
capacity of the trap. Dissolution of either or both the fresh and sulfated
sorbent would decrease the structural strength of the sorbent particles
possibly leading to increased attrition and/or channelling.
The solubilities of several potential sorbent oxides, carbonates
and sulfates are given in Table VL These values were all obtained from
reference [10]. The table shows that the oxides, carbonates, and
sulfates of the alkali metals are all quite soluble with the solubility
increasing with increasing molecular weight. The alkaline earth oxides
show more favorable solubilities. Magnesium and calcium oxides have
acceptable solubilities. Barium oxide is more soluble and could present
-------
some problems. The carbonates of the alkaline earth series all have
acceptable solubilities. Thr- solubilities of the suifates decreases
with increasing molecular weight. The solubility for magnesium sulfate
would be borderline while the higher molecular weight members would be
acceptable.
The oxides and carbonates of the transition metals, are insoluble,
or, at worst, only slightly soluble. The suifates are, however, quite
soluble with the exception of iron which is only slightly soluble.
Based on solubility restrictions, none of the alkali metals
would be acceptable. Calcium of the alkaline earth metals would be
acceptable, and iron of the transition metals listed would be acceptable. •
Another consideration of somewhat lesser importance-,.in selecting
potential sorbents is the cost and availability of materials ..• -The. real
importance of cost and availability would come if; two sorbents were to
show equal potential. However, for screening purposes,.only those materials
which are in short supply or are very costly should be. eliminated.
For reference, the costs of several potential sorbent materials
are listed in Table VIT. The cost listed is for enough material to- trap
131 gm. moles, of sulfur. All material costs were obtained from Reference
[ 9 ] and pertain to bulk quantities f.o.b. New York. The first cost
column assumes that the trap material can be prepared by same-simple
mechanical treatment, e.g. pelletization. The costs for the/transition
metals are based on the bulk metal rather than the oxides. .The, second
cost column was developed assuming that a soluble salt, in this pase. the
nitrate, would be required as the raw material in a precipitation-procedure.
The first cost column shows that the least expensive sorbents
would be the oxides of the alkaline earth metals followed by the transition
metals. In all cases, the cost of the soluble nitrate salts.were higher
than for the oxides. Also, the cost of preparing the sorbent from the
sale would be greater than from the oxide. ••,;.•
Ideally, the construction of the trap would preclude: fresh or
sulfated sorbent attrition. However, in practice attrition and- emission
of the trap material will occur. Therefore, emissions of the fresh and/or
sulfated sorbent cannot be harmful. Two examples immediately stand out:
beryllium and lead. Beryllium, Llie lowest molecular weight .alkaline.>,-. •
earth metal, would be an ideal sorbent cation from, many of the-, consid-, .
erations discussed above. However, beryllium oxide and sulfate are. .-
extremely toxic [ 57 ] and, therefore, must be eliminated from consideration.
The same is true for lead. The lead oxide candle procedure for quantitative
sulfur dioxide measurement shows an excellent sorptive. activity, but lead
oxide and sulfate emissions are harmful. Hence, lead must also be
eliminated as a potential sorbent.
-------
- 41 -
Table XII
ESTIMATED COSTS OF VARIOUS SORBENTS
Sorbent
MgO
CaO
BaO
Ni
Cu
Zn
Mn
Molecular
Weight
40
56
153
59
64
65
55
Kgm of Sorbent
per trap
5.2
7.3
20.0
7.7
8.4
8.5
7.2
Estimated
Sorbent Cost
0.50
0.20*
6.80
27.00
10.00
'6.50
5.20
Sorbent Cost,
Nitrate Salt
14.00
11.00
23.00
23.00
7.70
___
* from limestone
-------
In selecting potential sorbents based upon these considerations,
there is no single material which stands out in all categories. The
one cation which is consistently near the top in all categories is
calcium. This cation could be employed either as the oxide or as the
carbonate. Other potential sorbents which show promise are magnesium,
manganese, and aluminum. Again, the most probable sorbent would be the
oxide of these materials. The remaining potential sorbents have
deficiencies in one or more categories.
-------
- 43 -
References
1. Argof W. B., and J. M. Smith, Ind. Eng. Chem. , 45, 298 (1953)
2. Beltzer, M. , R. J. Campion, and W. L. Peterson, SAE Paper 740286,
Detroit, Michigan, 1974
3. Bienstock, D. , L. W. Brunn, E. M. Murphy, and H. E. Brown, U.S.
Bur, Mines Inform. Circ. 7836, 1958
4. Bienstock, D., J. M.. Fields, and J. G. Myers, U.S. Bur. Mines Rep.
Invest., 5735, 1961
5. Bienstock, D., J. H. Fields, and J. G. Myers, J. Eng. Power, 86
353 (1964)
6T Bodenstein, M. and C. G. Fink, Z. Physic. Chem., 60, 1 (1907)
(C,A. 1-2849)
7f Bodenstein, M. and C. G. Fink, Z. Physik. Chem., 60, 46 (1907)
-------
- 44 -
19. Gillespie, G. R. and H. F. Johnstone, Chem. Eng. Prog., 51, 74-F
(1955)
20. Gmetro, J. D. and T. Vermuelen, A I Ch.E J., 10, 940 (1.964)
21. Haas, L. A., U.S. Bur. Mines Inform. Circ. I.C. 8608 (1973)
22. Hall, R. E. and J. M. Smith, Chem. Eng. Prog., 45, 459 (1949)
23. Harkius, J. H. and S. W. Nicksic, Environ. Sci. Technol.,
1, 751 (1967)
24. Heinen, C. M., SAE Paper 486J, Detroit, Michigan (1962)
25. Hougen, 0. A., and K. W. Watson, "Chemical Process Principles Part III:
Kinetics and Catalysis", John Wiley and Sons., Inc. New York (1957)
26. Hougen, 0. A. and K. W. Watson, Ind. Eng. Chem., 35, 529 (1943)
27. Hougen, 0. A. and C. R. Wilkie, Trans. .Am. Inst. Chem. Engr,s,,,
45, 445 (1945)
28. Hurt, D. M., Ind. Eng. Chem., 35, 522 (1943)
29. Irvin, H. B., R. W. Olson, and J. M. Smith, Chem. Eng. Prog., 47,
287 (1951)
30. JANAF Thermochemical Data, Dow Chemical Co., Midland, Mich.
31. Kaneko, Y. and H. Odanaka, J. Res. Inst. Catalysis, Hbkkai'ifo Urvi'v.,
13, 29 (1965)
32. Kaneko, Y. and H. Odanaka, J. Res. Inst. Catalysis, Hokkaidq TIni'v,., 14,
213 (1966)
33. Kolloy, K. K., C. H. Shomate, F. E. Young, B. F. Naybor,
A. E. Salo, and E. H. Hoffman, U.S. Bur. Mines Tech. Paper, 685,
(1949)
34. Knietsch, R., Chem. Ber., 34, .4069 (1901)
35. Krause, H. H., A. Levy, and W. T. Reid, J. Eng. Pow., 90, 38 (1968)
36. Levy, A., E. L. Merryman, and W. T. Reid, Environ. Sci. Technol., 4,
653 (1970)
37. Lewis, W. K. and E. D. Reis, Ind. Eng. Chem., 17, 593 (19'25)
38. Lewis, W. K. and E. D. Reis, Ind. Eng. Chem., 19, 830 (1927)
39. Lisle, E. S. and J. D. Sensenbaugh, Combustion, 36, 12 (1965)
40. Lowell, P.S., K. Schwitzgebel, T. B. Parsons, and K. J. Sladek,
Ind. Eng. Chem. Process. Des. Develop., 10, 384 (1971)
41. Lange, G. , A. C. Cummings, and F. M. Miles, "The Manufacture of Acids
and Alkalies, Vol. IV. The Manufacture of Sulfuric Acid (Contact
Process) D. Van Nostrand Co., Inc., New York, 1925.
-------
- 45 -
42T Olson, R. W., R. W. Schuler, and J. M. Smith, Chem. Eng. Prog., 46,
614 (1950)
43, Paige, J. I., J. W. Town, J. H. Russell, and H. J. Kelly,
Bur. Mines Rep. Invest. 7414 (1970)
44. Peqhkovsky, V. V., J. Appl. Chem. (Russ.), 30, 1643 (1950)
45. Pierson, W. P., R. H. Hammerle, and J. T. Rummer, SAE Paper 740287,
Detroit, Michigan, Feb. 1974
46. Reid, W. T., J. Eng. Power, 92, 11 (1970)
47. Roiter, V. A., N. A. Stukanovskaya, G. P. Korneichuk, N. S. Volikovskaya,
and G. I. Golodets, Kin. i Katal., 1, 408 (1960)
48. Ross, L. W., Sulfur, 65, 37 (1966)
49. Russell, J. H., J. W. Town, and H. J. Kelly, U.S. Bur. Mines Rept.
Invest. 7415 (1970)
50. Salsas - Serra, F. , Chem. et Ind., 37, 1056 (1937)
51, Sax, N. I., "Dangerous Properties of Industrial Materials" Reinhold
Book Corporation, New York, 1968
52. Schuler, R. W., V. P. Stallings, and J. M. Smith, Chem. Eng. Prog.
Symp. No. 4, 48, 19 (1952)
53. Sittig, M. , "Sulfuric Acid Manufacture and Effluent Control", Noyes
Data Corporation, 1971
54. Taylor, G. B. and S. Lenher, Z. Physic. Chem. Bodenstein-Festbend, 30
(1931) (C.A. 25-5341)
55. Uyehara, 0. A. and K. M. Watson, Ind. Eng. Chem., 35, 541 (1943)
56. Verhoff, F. H. and J. T. Banchero, Chem. Eng. Prog., 70, 71 (1974)
57, Vogel, R. F., B. R. Mitchell, and F. E. Massoth, Environ. Sci.
Technology 8, 432 (1974)
58. Warner, N. A. and T. R. Ingraham, Can. J. Chem., 38, 2196 (1960)
59. Warner, N. A. and T. R. Ingraham, Can. J. Chem. Eng., 40 , 263 (1962)
60. Welty, A. B., Hydrocarbon Processing, 104 (1971)
61. Weychert, S. and A. Urbanek, Inter. Chem. Eng., 9, 396 (1969)
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
i. REPORT NO.
EPA-460/3-75-002-a
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Sulfate Control Technology Assessment Phase 1
Literature Search and Analysis
5. REPORT DATE
November 1974
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
William R. Leppard
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Exxon Research and Engineering Co.
Products Research Division
Linden, New Jersey 07036
1O. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-03-0497
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Protection Agency
Emission Control Technology Division
2565 Plymouth Road
Ann Arbor, Michigan 48105
13. TYPE OF REPORT AND PERIOD COVERED
Task 1
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The report covered the following four areas: 1. Thermo-dynamics-of Sulfuric
Acid Production. At typical catalyst temperatures, conversions of-SO^ to SO- greater
than 50% are thermodynamically possible. Equilibrium conversion is strongly dependent
upon oxygen concentration. Exhaust SO., may hydrate to gaseous sulfuric a-cid within
the vehicle's exhaust system,:arid the gaseous sulfuric acid will begin to condense at
about 150 C. 2. Reaction of Sulfur Dioxide and Trioxide with Exhaust Gas Constituent
and Exhaust System Components. Results show that the formation of ammonium sulfate is
favorable only below 225 C, the reduction of both sulfur oxides by. CO Is favorable,
reaction of both oxides with the iron oxide surface of exhaust system is favorable
below 425 C and the reaction of SO- with the aluminum oxide catalyst substrate .is
possible below 425 C. 3. Automotive Catalysis of Sulfur Dioxide. The rate limiting
step in the catalytic oxidation of S(L is the surface reaction, between 'adsorbed
oxygen and adsorbed SOp. Literature search indicates that automotive catalysis
literature is limited and inconclusive. 4. Sulfate Traps. The most promising means
of removing SO, from the exhaust stream is to react it with a basic metal oxide.
Based on a requirement selection criterion, the most promising sorbent material is
C_0. Other less promising but still attractive sorbents are the oxides of magnesium,
M||, and A
r
.17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
18. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
48
2O. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
46
------- |