United States
                     Environmental Protection
                     Agency
Industrial Environmental Research
Laboratory
Cincinnati OH 45268
                     Research and Development
EPA-600/S2-83-093 Dec. 1983
SERA          Project  Summary
                    Control  of  Air  Pollution
                    Emissions from  Molybdenum
                    Roasting
                     N. Masarky, K. Schwitzgebel, C. D. Wolbach, R. D. Delleney, T. P. Nelson, R.
                     L Glover, and J. M. Burke
                       Molybdenum, a relatively rare ele-
                     ment, occurs principally as molybdenite
                     (MoS2) and wulfenite (PbMoO4). Molyb-
                     denite  is the commercial source  of
                     molybdenum.  In 1979, three primary
                     molybdenum mines accounted for
                     about 68% of the domestic production
                     with the balance obtained as a principal
                     byproduct chiefly from 16 porphyry
                     copper mines. This latter form  is pre-
                     sently  our  sole source of  rhenium,
                     which is recovered in the processing of
                     molybdenite.
                       While a minor metal in terms of com-
                     mercial volume, molybdenum is widely
                     used in the  form  of molybdenum
                     oxide as an alloying agent in  ferrous
                     metals, its principal application, and is
                     essential for tool steels. Molybdenum
                     compounds are also used in chemicals,
                     catalysts, pigments, lubricants, and
                     electronics.  In 1979, the value  of U.S.
                     exports of molybdenum ore, concen-
                     trates, and products was about three
                     quarter billion dollars.
                       Molybdenum oxide is derived from a
                     concentrate  of molybdenite via a
                     thermal process known as roasting. In
                     practice, the concentrate is processed
                     in a vertical multiple  hearth type
                     furnace to produce the oxide. The roast-
                     ing results in the generation of particu-
                     late and weak sulfur dioxide emissions.
                     This program  was undertaken to: (1)
                     determine the capabilities of a unique
                     fabric filtration system using Teflon®*
                     coated bags in a hot, corrosive atmo-
                     sphere for particulate and trace element
                     control and (2) explore and evaluate the
                     "Mention of tradenames or commercial products
                     does not constitute endorsement or recommendation
                     for use by the U.S. Environmental Protection Agency.
feasibility of a variety of weak SO2 con-
trol systems for application to the molyb-
denum roaster and potentially to other
smelter weak SO2 off-gases.
  This Project Summary was developed
by  EPA's Industrial Environmental
Research Laboratory, Cincinnati. OH.
to announce key findings of the research
project that is fully documented in
three separate  reports (see  Project
Report ordering information at back.


Introduction
  This project was undertaken jointly by
Molycorp  and  the lERL-Ci because of
mutual interest in addressing the problems
of weak SO2 stream control. At the time
the tests were conducted, Molycorp was
concerned with the problem of controlling
sulfur dioxide emissions and eliminating
a visible plume in order  to meet state
requirements. The IERL objectives were
similar. The first task was to characterize
the  pollution control capabilities of the
Teflon® coated fabric filter for removal of
particulate and trace metals in the flue
gas  prior  to atmospheric discharge or
treatment in a sulfur  dioxide  control
system. The second task was to determine
the technical feasibility of applying a flue
gas  wet scrubbing  system  for weak
stream (about 1%) sulfur dioxide control.
The joint study was subdivided into three
tasks: (1) characterization of emissions
and particulate  control,  (2) a study of
alternatives for control of weak sulfur
dioxide emissions, and (3) a pilot-plant
scale test study of one of the approaches
identified in the second task. The
magnesium oxide system was selected
for this study. The results  of these tasks
achieved our objectives and aided Moly-

-------
corp in developing a plan to modernize
their smelter.
  At the time these tests were conducted,
Molycorp, Inc. operated two multi-level
hearth roasters at its facility in Washington,
Pennsylvania.  The plant processed 10
million pounds of molybdenum per year.
In the roasting process,  molybdenum
disulfide concentrate was oxidized by air
to molybdenum trioxide and  sulfur
dioxide. Particulate control was achieved
by a baghouse, followed by a spray cooler
and a packed bed scrubber. The scrubber
water was recirculated with blow-down.
A diagram showing the process flow, flue
gas handling and baghouse is given in
Figure 1. The baghouse employs Teflon®
coated fabric filter bags, a unique appli-
cation in the nonferrous industry.


Emissions Characterization and
Particulate Control
  The  sampling and analysis effort
described in this report was conducted to
characterize the particulate emissions
from the molybdenum roaster  perform-
ance of the associated particulate control
devices consisting of a high-temperature
baghouse and a spray scrubber-packed
bed clean-up facility. The characterization
was accomplished by chemical analyses
of all streams, flow-rate measurements,
grain-loading determinations, and particle-
size distribution measurements under
different  operating  conditions. Spark
Source Mass Spectrometry (SSMS) was
used to  semiquantitatively analyze the
samples. From these results, 15 elements
were selected for quantitative determina-
tion. The selection was based on concen-
tration level, volatility, and toxicity. The
elements investigated more fully were:
Arsenic     Bismuth     Nickel
Lead        Antimony    Copper
Molybdenum Cadmium    Manganese
Mercury    Silver        Iron
Selenium   Zinc         Rhenium
The quantitative analytical results were
combined with the total mass flow in
each  individual  stream to derive an
elemental flow  rate.  These data were
used to establish material balances
around the roaster and the baghouse.
  The major findings were:
   • Semiquantitative survey analyses by
     SSMS for 73 elements indicate that
     Twin Buttes  and Questa  concen-
     trates  contain low  concentrations
     (ppm range) of most metals. Excep-
     tions are: copper, lead, zinc,  iron,
     and  manganese.  The Twin Buttes
     concentrate is high in copper con-
     centrate (24,000 ppm versus 1000
     ppm in Questa). The Questa concen-
Contains 4 ft.
 of Packin
        g*

        Scrubber
   Blowdown Liquid Fed-
      to a Thickener
        Conveyor J
   ' Lead Lined Quench Vessel
"oints
Ambient Air Blended to Maintain Baghouse
     Exit Temperature at 400-450°F
Figure  1.   Dust collection system at Molycorp.

    trate shows a high lead concentration
    (4300 ppm versus 930 ppm in Twin
    Buttes concentrate).
  • Mercury and selenium are volatilized
    in  the  roasting process,  pass
    through  the baghouse  essentially
    uncontrolled,  and  are partially
    removed in the quench scrubber.
    For mercury, a feed rate of 0.104
    gm/hr (2.3 x  10~4 Ib/hr) was
    measured with 0.0014 gm/hr (0.3 x
    10~4 Ib/hr) reporting to the product;
    therefore 0.1026 gm/hr was volati-
    lized in the roaster of which 90% was
    removed in the quench scrubber.
    For selenium, a feed rate  of 68.1
    gm/hr (0.15 Ib/hr) was measured,
    with 1.4 gm/hr (0.003 Ib/hr) report-
    ing to the product; therefore 66.7
    gm/hr was volatilized in the roaster
    of which 50% was removed in the
    quench scrubber.
  • Pollutant content of the stack  gas
    was, in gm/hr (in Ib/hr): molybdenum
    — 4.5 (0.01),  selenium  —  3.6
    (0.008); lead —  6.8 (0.015); iron —
    3.2 (0.007); organics — 6.8 (0.015).
  • The baghouse was effective for par-
    ticulate  control. The average  inlet
    loading was 10.3 gm/Nm3  (4.51
    grains/DSCF) and 0.1 Mg/hr (223
    Ibs/hr); the average outlet loading
    was 0.091 gm/Nm3 (0.040 grains/
    DSCF) and  1.8 kg/hr (4.04 Ibs/hr).
    This indicates a control efficiency of
    99.1% on a concentration basis and
             98.2% on a mass basis. One would
             expect an efficiency of 99% or better
             for an installation of this type. For
             zero air  inleakagefi.e.,  dilution) the
             concentration  efficiency should
             equal the mass efficiency. It is not
             clear if the difference here is real or
             whether the lower mass efficiency
             value  is attributable  to: (1)  flow
             measurement error as the location
             of the measurement points was less
             than  ideal owing to the limitations
             imposed by the equipment configura-
             tion or (2) interference by acid mist
             formed by condensation at the outlet
             test point.
             The source of the plume had been a
             controversial point. Some observers
             held  that it  was  caused  by the
             presence of organic material intro-
             duced  by flotation agents while
             others  felt that  the cause  was
             sulfuric acid mist formed in the
             interaction  of the roaster gas  with
             the quench scrubber.  It was found
             that sulfate particles  and sulfuric
             acid mist, not organics, were princi-
             pally  responsible for the visible
             plume  problem. The high oxygen,
             high  862  concentration  of the
             roaster off-gases favors the forma-
             tion of S03 as the flue gas cools. The
             SOa reacts  with solids to form sul-
             fates and with water vapor to form
             sulfuric acid mist. This  increases the
             particulate  loading as  the gas tem-

-------
     perature decreases. Acid dew points
     were 309°F at the roaster outlet,
     and 269°F at the baghouse outlet.
  Paniculate loading determinations
showed a concentration at the roaster
exit  of 10.3 gm/Nm3  (4.51  grains/scf).
Analysis of the filter for sulfate indicated
a  sulfate contribution of  0.4%. An
average particulate  loading of 0.091
gm/Nm3 (0.040 grains/scf) was measured
at the baghouse exit. Sulfate contribution
accounted for  17% of the particulate
matter. Water droplets and acid mist
were present in the  quench-scrubber
exit gases. Water had to be evaporated in
order to determine a particulate loading.
The EPA method 5 was used. Contributions
of sulfuric acid mist accounted for 35%,
50%, 74% and 91 % at sampling tempera-
tures of 380°F, 365°F, 300°F and 250°F,
respectively. These  results  point to
sulfuric acid mist  as  the cause of the
plume formation. This mist was completely
removed in a bench-scale wet electrostatic
precipitator.

Alternatives for Control of
Weak Sulfur Dioxide Emissions
  Processes  in the primary nonferrous
metals industry produce off-gases  con-
taining  significant quantities  of  SO2.
Typically, these streams are classified as
"strong" or "weak" depending on  their
SO2  concentration. Strong  gas streams
have SOa concentrations greater than 3.5
to  4.0 volume percent while weak gas
streams contain between 0.5  and 3.5
volume percent S02.
  The technology for sulfur  dioxide
control  in emissions from smelters
containing weak concentrations, i.e.,
from about 0.5 to 3.0  percent S02, is a
slowly developing area. This survey of
alternative control technologies  was
undertaken to provide  information to aid
in selection  of  system for pilot-scale
testing.  The  technical feasibility for a
number of scrubbing systems has  been
established through application and full-
scale operation  at certain smelters. It
should be noted that, with the exception
of  cold sea water absorption, each full-
scale scrubber system is unique in the
sense that it has been applied at only one
smelter  site. Some twelve systems were
examined for suitability for application to
nonferrous smelters  and molybdenum
roasting in particular.
  •  CIBA-GEIGY nitrosyl-sulfuric acid
     process
  •  CIBA-GEIGY SO2 sorption-stream
     stripping process
  •  Limestone FGD
  •  Dual Alkali
  •  Magnesium Oxide
  • Wellman-Lord
  • Sulf-X
  • Endako
  • Chiyoda Thoroughbred (2)
  • U.S. Bureau of Mines Citrate
  • Sodium Carbonate Throwaway
  • Dowa Basic aluminum sulfate
The magnesium  oxide  system  was
selected for pilot-scale testing because of
the state-of-development, available
information for  test design purposes,
compatibility with existing pilot plant
equipment,  and the need  to  develop
applicability data to support analysis of
economic feasibility of this system.


Pilot-Scale Test Results for
Magnesium Oxide  Scrubbing
  The  magnesium oxide  (MgO) system
was selected for pilot-scale testing to
generate engineering design data for an
MgO system which, in turn, could be used
as  a  basis for  exploring  economic
feasibility. Specifically,  the  tests  were
designed to quantify the SO2  removal
which  could  be attained by the  MgO
system, and to develop data  on the MgO
system which could be used to design an
absorber for treating similar gas streams
in other smelter applications.
  A schematic of the pilot unit is shown in
Figure  2.  As shown, the pilot unit
consisted of an absorber, reaction  tank,
and solids separator. For the purpose of
these brief tests,  the  solids produced
were disposed of by ponding rather than
being dried and regenerated.
                               The absorber was a 76 cm (30 inch)
                             diameter tower containing two beds of
                             2.5 cm (1 inch) Tellerette® packing. The
                             upper bed depth was 76 cm (30 inches)
                             and the lower bed was 51 cm (20 inches)
                             deep. The piping and valve arrangement
                             was designed to distribute slurry through
                             both beds or the lower bed only. This
                             permitted scrubbing tests at packing
                             depths of 51 cm (20 inches) and 127 cm (50
                             inches). Full  cone spray nozzles were
                             used for liquid distribution and were
                             arranged so that the edge of the spray
                             contacted the tower wall at the level of
                             the packing.
                               Gas flow rates through  the absorber
                             ranged from 2800 to 4000 NmVhr (1700
                             to 2600  scfm)  while slurry  flow rates
                             ranged from 0.4 to 2.4 I/sec(6 to38gpm).
                             The SO2 concentration  of  the gas was
                             controlled by a damper  arrangement on
                             the inlet gas stream so that inlet gas
                             strength and flow rate could be maintained
                             at a constant  value through dilution
                             without interfering with plant operations.
                             During most tests, the pH of the absorber
                             feed was maintained between 7.5 and
                             8.0.  Slurry exiting the absorber was
                             gravity fed to  a 5680 (1500 gallon)
                             reaction tank. The pH of the slurry leaving
                             the absorber ranged from 4.5 to 6.5, and
                             Mg (OH)2 was added to the reaction tank
                             to raise the pH back to 7.5-8.0.
                               The scrubbing slurry contained 4 to 8
                             weight percent Mg SO3 • 6H^O solids. A
                             bleed stream of slurry was removed from
                             the reaction tank and  concentrated  to
                             approximately 35 weight percent solids in
                                    Solids
                                                                       •Gas
  Makeup
   Water
Figure 2.
Schematic of the magnesium oxide scrubber.

                            3

-------
a clarif ier. Overflow from the clarif ier was
either  returned to the reaction tank or
used to prepare  fresh  Mg(OH)2 in the
slurry tank.
  A well-defined test  program  was
conducted in which the process param-
eters of interest were varied to observe
their effect  on SO2  removal.  The data
collected  during these tests were then
correlated  in the form of a design
equation. As this program was considered
preliminary  effort, long-term testing to
demonstrate reliability and to acquire
operability data was  considered beyond
the  scope of this effort  and was not
attempted.
  The  statistically designed test  plan
consisted  of a matrix of short-term  tests
which included all possible combinations
of three liquid-to-gas ratios, three gas
flow rates, and two packing depths (18
tests).  These tests were supplemented
with a limited number of tests in which
the pH of the scrubbing slurry and the S02
concentration of the off-gas were varied.
In addition, a longer-term, or steady-state
test was scheduled.  This test  was
included to provide data which would be
used to  complete  material balance
calculations  for the system.
  The test results indicate that the  MgO
scrubbing system  is  applicable for
treating gas  streams containing  up to
about one percent SO2. Removal effici-
encies of  over 90 percent of S02  were
possible at gas velocities and absorber
pressure drops similar to those used in
design of utility MgO FGD systems. In the
evaluation of the results  of  the  tests
conducted at Molycorp, a correlation was
developed which can be used to predict
SO2 removal in the pilot plant for a given
set  of  absorber operating parameters.
This correlation was developed based on
absorption theory, using the two-film
mass transfer model. Regression analysis
was used to develop an empirical value
for KgA, resulting in the design equation:
    R = 1-exp[(4.36 x 1CT3) (L/AC)0899
    (pH)286 (W)0284 (SO2)"0621 (P) (D)].
 vhere
  R = fractional efficiency of S02 removal
  L = absorbent flow rate, liters per second
  S02 = SOz concentration of inlet gas, ppm
  W = MgSO3 solids in the absorber feed,
      % by weight
  Ac = absorber cross sectional area,
       square meters
  D = packing depth; centimeters
  P = operating pressure, atmospheres
  pH = pH of the absorber feed slurry
  It is interesting to note that the flue gas
flow rate does not appear. This is due to
an offsetting effect  associated with a
change in gas flow rate into the absorber.
By increasing gas rate, more SO2 enters
the absorber and thus  tends to  lower
removal.  However,  increasing the  gas
flow rate increases the value of Kga  and
thus offsets the effect of additional S02
entering the absorber.  The reader is
cautioned that this is  only true because
the exponent of (G/AC) is 1.0 for this test
program, and should be re-established for
other tests conditions.
  The correlation  between  actual  and
predicted S02 removal efficiency is quite
good with a correlation coefficient of 0.98.
The absorber  design  equation can  be
used to  predict the performance of  a
particular absorber design configuration.
However, this equation has been proved
valid only for specific  conditions  and
ranges of absorber operating parameters
explored  in this program. In using the
equation, the most obvious restriction is
that it is only applicable for use with an
MgO scrubbing system processing gas in
an absorber packed with  2.5 cm (1 inch)
Tellerette® packing. This design equation
is not applicable  to other absorber
configurations (e.g., a spray tower) or other
types or  sizes of packing material.  The
other principal  restriction  is that the
scrubber operating parameters selected
for an absorber design must be within the
range in  which  the  parameters were
tested during the program. Specific
ranges for these parametersare presented
in Table 1. The applicable range for each
parameter represents the  range over
which the parameters were varied during
the test program.
  Another  significant  result of the  test
program  was  determination of  the
absorber pressure drop as a function of
various absorber  operating parameters.
Of the parameters which were examined
during  the test  program, only  packing
depth and gas velocity were found to have
a significant impact on the pressure drop
in the absorber (Figure 3). Based on theo-
retical  considerations  and on  measure-
ments made on the pilot absorber, pressure
drop was found to be a linear function of
packing  depth. That is, a doubling in the
packing  depth results in a doubling of the
pressure drop in the absorber.
  The relationship between gas velocities
and pressure  drop is somewhat more
complex. Pressure drop  appears to
increase exponentially with  increasing
gas velocity. This is an important consid-
eration  in designing an absorber since
the gas  velocity would be selected based
on  a  trade-off  between absorber cross
section and pressure drop. Relatively low
gas velocities result in a larger absorber
with a lower pressure drop. The trade-off
which must be evaluated is an economic
one; the capital costs for a larger absorber
versus the operating costs of overcoming
a high pressure drop.
  The major conclusions of the pilot-
scale  tests are:
 1 )The MgO scrubbing system which was
   tested at Molycorp's Washington,
   Pennsylvania plant  was  capable of
   removing over 90 percent  of the S02
   from  an off-gas stream containing up
   to  9000 ppm SO2. The use of a packed
   absorber in conjunction with the MgO
   sorbent was the major factor which
   contributed to the  good  system per-
   formance. This is  due  to the high
   (relative  to a spray tower) liquid
   residence time and overall  area for
   mass transfer which exist in a packed
   absorber.  The relatively  long  liquid
   residence time in the absorber helped
   promote dissolution of MgSOa solids
   in  the absorber which  effectively
   increased the liquid phase alkalinity of
   the scrubbing slurry.
2) An equationwasdevelopedtocorrelate
   the results of the pilot-scale tests. The
   correlation of experimental data was
   excellent and  the equation can  be
   used to design an MgO absorber using
   identical  packing material treating a
   gas stream which is similar in com-
   position to the one  at Molycorp. In
   using the design equation,  the level of
   absorber operating parameters should
   be in the range of parameters examined
   during the pilot-scale tests.
3) For most operating conditions, plugging
   of  the absorber bed was not a problem.
   However,  at very high SO2 removal
   efficiencies (95 percent) MgSOa solids
   did begin to accumulate  in the absor-
   ber. In order to prevent such accumula-
   tion,   the  S02 concentration at the
   outlet of the packed bed should be
   maintained  above 500 ppm. If addi-
   tional S02 removal is required, a clear
 Table 1.     Range of Absorber Operating Parameters Applicable to Design Equation
  Operating parameter	Applicable range
 Gas Velocity (m/sec)
 pH
 Liquid Velocity (I/see-in )
 Weight Percent Solids
 SOz Concentration (ppm)
                          1.8 to 2.6
                          6.0 to 8.0
                          1.0 to 5 4
                            4 to 10
                        2000 to 9000

-------
   72.0-
   70.0-
t
-*e
I
    8.0'
8
    6.0 —
    4.0'
    2.0-
        7.5
                  \
                 2.0
 \
2.5
 \
3.0
Figure 3.
                    Gas Velocity (m/sec)

Absorber pressure drop as a function of gas velocity
                                                                         SOa removal efficiencies and lower
                                                                         absorber pressure drops than those
                                                                         reported by utility applications.  The
                                                                         principal reason for the superior
                                                                         performance is the absorber design
                                                                         used, a packed absorber in the pilot
                                                                         tests versus a venturi type absorber in
                                                                         the  utility  systems.  In general, the
                                                                         packed  absorber appears well suited
                                                                         to treating off-gas  streams which
                                                                         contain high concentrations  of  SO2
                                                                         (up to 9000 ppm).
   liquor spray  installed  above  the
   packing should permit absorber oper-
   ation without plugging.
4) Process control of the pilot-scale MgO
   system was not difficult. Measurement
   of reaction tank  pH provided an
   indicator which could  be used to
   accurately control the Mg(OH)2 feed
   rate to the  system. Also, changes in
   gas flow  rate (over the range tested)
   did not require any  corresponding
   changes  in  other absorber operating
   parameters to  maintain  a  constant
   SOa removal efficiency.
5) Because the pilot  unit was  not
   operated as  a  closed-loop system,
   there were certain issues which were
   not addressed during the test program.
   Of these, it appears that buildup of
                                MgSO4 (aq) will  have the  largest
                                impact on test results. This is due to
                                the fact that as MgS04 (aq) concentra-
                                tion builds up,  MgSOj (aq) solubility
                                decreases, thus decreasing available
                                liquid phase alkalinity in the scrubbing
                                slurry and possibly decreasing the SO2
                                removal  efficiency.  Theoretically, a
                                decrease in alkalinity can be offset by
                                a  corresponding increase in liquid
                                flow rate to the absorber, but tests
                                should be conducted to confirm this
                                fact prior to using the design equation.
                              6) A comparison of the pilot plant test
                                results to  those reported for utility
                                applications of the  MgO  system
                                indicates  that the system  tested at
                                Molycorp was generally superior. The
                                Molycorp pilot tests resulted in higher

-------
     N. Masarky is with Molycorp, Inc., Washington, PA 15301; K. Schwitzgebel, C. D.
        Wolbach,  R. D. Delleney, T. P. Nelson. R. L Glover, and J. M. Burke are with
       Radian Corp., Austin, TX 78759; the EPA Project Officer J. O. Burckle (also the
       author of this Project Summary) is with the Industrial Environmental Research
       Laboratory, Cincinnati, OH 45268.
     The complete report consists of three volumes, entitled "Control of Air Pollution
       Emissions from Molybdenum Roasting,"(Set Order No. PB 83-264 184; Cost:
       $33.00, subject to change)
         "Volume I. Emissions  Characterization and Paniculate Control," (Order No.
         PB 83-264 192; Cost: $11.50, subject to change)
         "Volume II. Alternatives for Control of Weak Sulfur Dioxide Emissions,"
         (Order No. PB 83-264  200; Cost: $13.00, subject to change)
         " Volume III. Pilot Scale Test Results for Magnesium Oxide Scrubbing," (Order
         No. PB 83-264 218; Cost: $14.50,  subject to change)
             National Technical Information Service
             5285 Port Royal Road
             Springfield. VA  22161
             Telephone: 703-487-4650
     The EPA Project Officer can be contacted at:
             Industrial Environmental Research Laboratory
             U.S. Environmental Protection Agency
             Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
  BULK RATE
U.S. POSTAGE
    PAID
Cincinnati, Ohio
Permit No. G35
Official Business
Penalty for Private Use $300
                                                                                   U.S. GOVERNMENT PRINTING OFFICE: 1984-759-102/807

-------