EPA-600/2-77-072
April 1977
Environmental Protection Technology Series
           FOAM  FLOTATION TREATMENT  OF  HEAVY
                    METALS AND FLUORIDE-BEARING
                          INDUSTRIAL  WASTEWATERS
                                Industrial Environmental Research Laboratory
                                     Office of Research and Development
                                    U.S. Environmental Protection Agency
                                            Cincinnati, Ohio 45268

-------
                RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:

      1.  Environmental Health  Effects Research
      2.  Environmental Protection Technology
      3.  Ecological Research
      4.  Environmental Monitoring
      5.  Socioeconomic Environmental Studies
      6.  Scientific and Technical Assessment Reports (STAR)
      7.  Interagency Energy-Environment Research and Development
      8.  "Special" Reports
      9.  Miscellaneous Reports

This report has  been assigned  to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental  quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                               EPA-600/2-77-072
                                               April  1977
               FOAM FLOTATION TREATMENT OF

HEAVY METALS AND FLUORIDE-BEARING INDUSTRIAL WASTEWATERS
                           by

                    David J. Wilson
                 Vanderbilt University
              Nashville, Tennessee  37235
                   Grant No. R-803564
                    Project Officer

                    Mary K. Stinson
         Industrial Pollution  Control  Division
     Industrial Environmental  Research Laboratory
                Cincinnati, Ohio   45268
      INDUSTRIAL ENVIRONMENTAL  RESEARCH LABORATORY
           OFFICE OF RESEARCH AND DEVELOPMENT
          U.S. ENVIRONMENTAL  PROTECTION AGENCY
                CINCINNATI, OHIO  45268

-------
                       DISCLAIMER
     This report has been reviewed by the Industrial
Environmental Research Laboratory, U. S. Environmental
Protection Agency, and approved for publication.  Approval
does not signify that the contents necessarily reflect the
views and policies of the U. S. Environmental Protection
Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
                            11

-------
                           FOREWORD
      When energy and material resources are extracted,  pro-
cessed, converted, and used, the related pollutional impacts
on our environment and even on our health often require that
new and increasingly more efficient pollution control methods
be used.  The Industrial Environmental Research Laboratory -
Cincinnati (IERL-CI) assists in developing and demonstrating
new and improved methodologies that will meet needs both
efficiently and economically.

     This report is a product of the above efforts.  These
studies were undertaken to perform a laboratory-scale
investigation of floe foam separation techniques to remove
toxic heavy metals and fluoride from wastewaters produced
at primary aluminum smelters, secondary lead smelters, and
copper and brass mills.  Simulated and real industrial
wastewaters were studied.  Floe foam separation techniques
combine the attractive features of simplicity, economy,  low
energy usage, potential for recovery, and effective removal
of pollutants from diluted wastewaters to concentrations well
below the standards established or anticipated by the regu-
latory agencies.

     Such information will be of value both to EPA and to
the industry itself.  Within EPA's R&D program the infor-
mation will be used as part of the continuing program to
develop and evaluate improved and less costly technology
to minimize industrial waste discharges.  Besides its direct
application to effluents from non-ferrous metals industry,
this technology may find application to treat metal-contain-
ing wastes generated by a host of other industries.

     For further information concerning this subject the
Industrial Pollution Control Division should be contacted.
                       David G. Stephan
                           Director
         Industrial Environmental Research Laboratory
                          Cincinnati
                              111

-------
                                  ABSTRACT


     Laboratory studies demonstrated that floe foam flotation techniques are
 capable of reducing lead, cadmium, mercury, copper, zinc, arsenic, and fluoride
 in dilute aqueous systems to very low levels.  Simulated as well as real indus-
 trial wastes were studied.  Copper, lead, and arsenic are readily removed with
 Fe(OH)3 and sodium lauryl sulfate; fluoride and zinc, with A1(QH)3 and sodium
 lauryl sulfate; cadmium and mercury, with CuS and hexadecyltrimethylammonium
 bromide.  Batch flotation techniques as well as continuous flow systems were
 studied.  Continuous flow systems proved to be more efficient than the batch
 techniques.  Floe foam flotation techniques are not effective for treatment of
wastes containing high concentrations (>10 g/£) of dissolved salts.  Adequate
 control of pH is essential in most of the separations, and proper design and
placement of the influent dispersion head and baffles are necessary to permit
maximum column loading.  The calculation of the isotherms for floe adsorption
 on surfactant films was carried out by means of statistical mechanics, and the
 results were found to be in good agreement with the experiment.  Computations
 of column performance in the steady state were made by means of a mathematical
model, and the dependence of column performance on the parameters of the model
were investigated.

     This report was submitted in fulfilment of Grant No. R-803564 by Vanderbilt
University under the sponsorship of the U. S. Environmental Protection Agency.
This report covers the period February 15, 1975 to December 31, 1975 and work
was completed as of January, 1976.
                                      IV

-------
                             CONTENTS
                                                              Page
Foreword                                                      iii
Abstract                                                       iv
List of Figures                                                vi
List of Tables                                                vii
List of Abbreviations and Symbols                            viii
Acknowledgments                                                xi
I     Conclusions                                               1
II    Recommendations                                           3
III   Introduction                                              5
IV    Literature Review                                         7
V     Objectives of the Research Program                       11
VI    Batch Technique Laboratory Studies                       12
           Foam Separation of Lead(II) by Batch Techniques     12
           Foam Separation of Copper(II) by Batch Techniques   16
           Foam Separation of Mercury(II) and Cadmium(II) by   17
                Batch Techniques
           Foam Separation of Arsenic by Batch Techniques      27
           Foam Separation of Fluoride by Batch Techniques     27
VII   Continuous Flow Studies                                  29
VIII  Discussion                                               33
IX    Theory of Adsorption Isotherms                           36
X     Theory of Stripping Column Operation                     42
XI    References                                               54
XII   Appendices                                               61
           Adsorption Isotherms                                61
           Model of Stripping Column Operation                 66

-------
                               LIST OF FIGURES





Number



   1       The Batch Apparatus                                          ^3



   2       Hg(II)  Removal by Precipitate  Flotation of HgS with HTA     19



   3       The Small Continuous Flow Apparatus                         30



   4       Dependence of Isotherms  on  Ionic  Strength                    38



   5       Dependence of Isotherms  on  Surface  Potential                 39



   6       Dependence of Isotherms  on  Floe Particle Charge              40



   7       Dependence of Isotherms  on  Temperature                       41



   8       Dependence of Separation Factor F on Specific  Foam Area     45



   9       Dependence of Separation Factor F on Surface Velocity        46



  10       Dependence of F on Liquid Velocity                           47



  11       Dependence of F on Effective Diffusion  Constants             48



  12       Dependence of F on Mass  Transfer  Rate Constant              49



  13       Dependence of F on Langmuir Parameter Aj                     50



  14       Dependence of F on Langmuir Parameter A2                     51



  15       Dependence of F on cfeed                                    52




B-l      Material  Balance Diagram of the Column                       67
                                     VI

-------
                               LIST OF TABLES





Number                                                                   Page



   1       Floe Foam Flotation of Lead with A1(OH)3 and NLS               14



   2       Floe Foam Flotation of Lead with Fe(OH)3 and NLS               14



   3       Floe Foam Flotation of Copper with Fe(OH)3 and NLS             16



   4       Hg(II) Removal by Precipitate Flotation of Mercuric Sulfide    18



   5       Hg(II) Removal by Floe Foam Flotation with A1(OH)3             20



   6       Hg(II) Removal by Floe Foam Flotation with Fe(OH)3             20



   7       Cd(II) Removal by Floe Foam Flotation with A1(OH)3             21



   8       Effect of Ionic Strength and pH on Cd(II) Removal by Floe      21



           Foam Flotation with Fe(OH)3



   9       Cd(II) Removal by Floe Foam Flotation with Fe(OH)3             22



  10       Hg(II) Removal by Floe Foam Flotation of HgS with FeS          23



  11       Hg(II) Removal by Floe Foam Flotation of HgS with CuS          24



  12       The Effect of pH Variation on Cd(II) Removal by Floe Foam      25



           Flotation of CdS with CuS



  13       The Effect of Increasing Ionic Strength on Cd(II) Removal      25



           by Floe Foam Flotation of CdS with CuS



  14       Optimal Separations by Batch Floe Foam Flotation Techniques    33



  15       Industrial Wastes Treated                                      34
                                     Vll

-------
                      LIST OF ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS
     a2
     c
     C
     c
      feed
     D

     DS
     e
     F
     HTA
     ID
     k
     K
     ki
     LS"
     m
     min
cell dimension
number of sites containing i particles
first Langimiir adsorption isotherm parameter, also
effective thickness of ionic atmosphere
second Langmuir adsorption isotherm parameter
solute or floe particle number density, cm"3
concentration of 1-1 electrolyte, moles/cm3
concentration of solute in column effluent
liquid solute concentration in compartment i
first element of eigenvector
maximum number density of floe particles, cm~3
dielectric constant of water, approximately 78
effective diffusion constant in the liquid phase
effective diffusion constant in the surface phase
(electronic charge|, 4.77 x 10"10esu
column separation parameter
grams/liter
hexadecyltrimethylammonium bromide
inside diameter
Boltzmann's constant
equilibrium adsorption isotherm
rate constant governing solute transport between the
surface and liquid phases
solubility product
liter(s) or column length
lauryl sulfate ion
number of cells in a surface adsorption site
minute(s)
                                     Vlll

-------
ABBREVIATIONS (continued)
     m/H         -- moles/liter
     N           -- normal
     NLS         --sodium lauryl sulfate
     NQ          -- Avogadro's number, 6.023 x 1023
     OD          -- outside diameter
     Pf(j-)      -- one-particle partition function for a particle in cell j.
     pKg         	log 10 KS
     ppm         — parts per million (in water, numerically equal to mg/£)
     q           -- electric charge of,floe particle
     q(s)        -- site partition function
     Q(N,M,T)    -- canonical ensemble partition function for M sites, N
                    particles
     S           -- specific foam area
     s           -- average number of floe particles per site
     S           -- surface excess of floe particles
     T           -- absolute temperature
     V           -- volume of liquid contained in 1 cm3 of foam
     V1          — flow rate of column effluent/v,,
     V(i)        -- potential energy of a floe particle in the ith cell
     v.          — velocity of liquid downward
      A*
     v           -- velocity of surface upward
      o
     x           -- distance from air-solution interface, or distance from
                    base of column
     x-          -- position of the boundary between compartments i and i+1
     a^-          -- integration constant
     3           - 1/kT
     T           -- surface concentration of solute
     T.          -- surface solute concentration in compartment i
     l4          -- second element of eigenvector
     6           -- film thickness of drained film
     ej(s)       -- energy of site when occupied by s particles in cells
                    ji» J2, '—J
                                      IX

-------
ABBREVIATIONS (continued)
     X
     y
     ^config
     ymhos
     a
     I
-- absolute activity
-- eigenvalue
— chemical potential
— configurational free energy per floe particle
-- micro-ohms"1, micromhos
-- position of the midpoint of the ith compartment
-- grand partition function
-- product symbol
— reduced floe density,
— summation symbol
-- electric potential
-- surface potential
-- factorial symbol
SYMBOLS
Al
As
C
Ca
Cd
Cl
Cu
F
Fe
H
-- aluminum
-- arsenic
— carbon
-- calcium
- - cadmium
-- chlorine
-- copper
-- fluorine
-- iron
-- hydrogen
Hg
Mn
N
Na
0
P
Pb
S
Zn

                                                             -- mercury
                                                             -- manganese
                                                             -- nitrogen
                                                             -- sodium
                                                             — oxygen
                                                             — phosphorus
                                                             -- lead
                                                             -- sulfur
                                                             — zinc

-------
                               ACKNOWLEDGMENTS
     The experimental work presented in this report was carried out by Tony
E. Chatman, Ann N. Clarke, Jo S. Hanson, Shang-Da Huang, Ronald P. Robertson,
and Charles S. Wilson.  The theoretical work was done with the aid of James
H. Clarke, Shang-Da Huang, and John W. Wilson.  The advice and help of Mary
K. Stinson and John Ciancia is gratefully acknowledged.
                                      XI

-------
                                  SECTION I


                                 CONCLUSIONS


     Floe foam flotation techniques were investigated on the laboratory scale,
and demonstrated to be capable of reducing lead, cadmium, mercury, copper,
zinc, arsenic, and fluoride in dilute aqueous solutions to very low levels.
Simulated as well as real industrial wastes were studied.

     Copper, lead, and arsenic are readily removed with Fe(OH)3 and sodium
lauryl sulfate (NLS).  The reagents are stable and low in both cost and toxi-
city.  These separations are rapid in the batch mode, and the red-brown color
of Fe(OH)3 makes it possible to follow the progress of the separation visually
in a glass or lucite column.  The separations of lead arsenate and copper were
also carried out successfully by the continuous flow technique in a 4.8 cm-ID
column; lead was also separated in a 10.2 cm-ID continuous flow column at flow
rates of the order of one £/min.

     Zinc and fluoride are removed with A1(OH)3 and sodium lauryl sulfate
(NLS).  Both reagents are stable and low in cost and toxicity.  Fluoride has
been separated by the continuous flow method as well as the batch technique.
It was found that use of a small amount of Fe(OH)3 in the fluoride separation,
to act as a visual tracer, did not interfere with the ability of the A1(OH)3-
NLS system to remove fluoride by the continuous flow technique.

     Cadmium and mercury are removed with CuS and HTA (hexadecyltrimethyl-
ammonium bromide); cadmium is also removed with FeS and HTA.  Attempts to
obtain adequate foam flotation separation with ferric or aluminum hydroxides
and NLS were not successful, although a combination precipitation-foam flota-
tion method using Fe(OH)3-HTA at a pH of 12 was able to reduce cadmium levels
to 0.01 ppm.  We note that sulfide ion and H2S, present in small amounts at
the initial stage of the process are both toxic.  Copper salts, sodium sulfide,
and HTA are all relatively expensive compared to alum or ferric chloride and
NLS.  The black color of the sulfides, however, does facilitate following the
separation visually when transparent columns are used.

     It was found that pH must be closely controlled.  With continuous flow
systems it was also necessary to design and place the influent dispersion head
and associated foam baffles carefully to maximize the feasible influent flow
rate.

     The method is not effective for separations from wastes containing high
concentrations (>10 g/£) of dissolved salts.

-------
     The dependence of separation efficiency on dissolved salt concentration,
temperature, floe electric charge, surfactant concentration, and floe concen-
tration in the influent has been calculated by means of statistical mechanics
from a Gouy-Chapman type model.  The results are quite consistent with the
experimental data.  Increasing salt concentrations decrease floe adsorption
isotherms markedly; temperature has relatively little effect; increasing floe
electric charge or surfactant concentration below the critical micelle concen-
tration increases the adsorption isotherm; increasing influent floe concentra-
tion decreases separation efficiency drastically above a certain critical
range.

     Column performance was simulated mathematically by means of two simulta-
neous differential equations and their associated boundary conditions; these
were solved by a quasi-linearization method in which the adsorption isotherms
mentioned above were used.  The dependence of column performance on such
parameters as liquid and gas flow rate, bubble size, isotherm parameters,
diffusion and mixing constants, and rate constant for floe transport between
the liquid and the film surface was studied.  This computer model should prove
highly useful in continuous flow column design.

-------
                                 SECTION II


                               RECOMMENDATIONS


     The results obtained in these studies justify a recommendation that pilot
plant studies be initiated on the following separations: (1) lead with Fe(OH)3
and sodium lauryl sulfate (NLS); (2) copper with Fe(OH)3 and NLS; (3)  fluoride
with A1(OH)3 and NLS; (4) zinc with A1(OH)3 and NLS; (5) cadmium with FeS and
hexadecyltrimethylammonium bromide (HTA); and (6) mercury with CuS and HTA.
The apparatus should be constructed to be readily portable by truck, so that
it could be transported conveniently to the various industrial plants when
wastewaters are to be treated.  The characteristics of the various wastes such
as concentration of toxic substance(s), conductivity, total dissolved solids
should be determined and related to optimum separation parameters such as pH,
flocculent and surfactant concentrations, and influent and air flow rates.
The performance of various types of influent dispersion heads and foam sta-
bilizing baffles should be studied to develop designs permitting maximum column
throughput.  The recycling of surfactant should be investigated as a means of
reducing the cost of these separations, and efforts should be made to minimize
surfactant loss.  Finally, experience gained at the pilot plant level should
permit cost estimates of the various separations.

     In the course of the pilot plant studies, efforts should be made to relate
actual column performance to the parameters used in a mathematical model simu-
lating column operation.  Reliable parameters would facilitate optimal column
design and operation.

     Ore processing and smelting wastewaters should be screened for the pre-
sence of arsenic; those found to contain it should be studied on a bench scale
to determine the feasibility of removing arsenic by means of floe foam flota-
tion with Fe(OH) 3 and NLS.

     The possibility of developing floe foam separations for nickel, antimony,
chromate, chromium(III), selenium, dyes, fibers, and phenols should be investi-
gated by means of batch studies on a bench scale.

     When dealing with more concentrated wastewaters, the compatibility of
precipitation pretreatment [such as with Na2C03, Ca(OH)2, Na2S, or other
precipitating agents] and the above floe foam flotation treatments should be
ascertained.

     Synthetic and actual wastewaters containing mixtures of metal ions should
be studied by means of batch floe foam flotation techniques to determine
whether the removal of several components at once is feasible.  The possibility
of selective removal of single species should also be explored, since this

-------
would facilitate recycling of the material back to the process or recovery
for use elsewhere.

-------
                          SECTION III


                          INTRODUCTION


     One of the current problems in the area of advanced methods of
waste treatment is that of removing low concentrations of toxic
metals.  These can cause serious difficulties in secondary sewage
treatment plants and are not effectively removed by such treatment.
Neither are they removed by conventional methods of drinking water
treatment.  Recent studies, particularly the very extensive work
on mercury, and to a lesser extent the work on cadmium and lead,
have focused concern on the impact of trace amounts of these sub-
stances upon man.  The maximum recommended levels of these metals
in drinking water are quite low (0.005 ppm for lead, and 0.01 ppm
for cadmium) , and industries are under heavy pressure to reduce the
discharge of these elements to the lowest possible level.

     A number of techniques exist to remove metal ions from aqueous
solution:  (1) chemical precipitation as an insoluble salt, hy-
droxide, or oxide; (2) ion exchange; (3) distillation; (4) freezing;
(5) reverse osmosis; (6) electro-osmosis; (7) solvent extraction;
and (8) foam flotation methods.  Of these  techniques, foam flo-
tation appears to possess some distinct advantages when one is
dealing with large volumes of solutions that are quite dilute in
the ions to be removed.  This method for the removal of trace
contaminant heavy metals in (1) rapid;  (2) cheap in terms of labor,
equipment, energy and chemicals; (3) capable of application in
small, intermediate, and large scales;  (4) productive of quite small
volumes of liquid or solids highly enriched in the contaminants;
(5) capable of reducing the levels of these contaminants to con-
centrations well below the standards established or anticipated
by the regulatory agencies.  Data published in the literature and
our own results described herein indicate that adsorbing colloid
foam flotation techniques show considerable promise of meeting
these criteria.

     Substantial volumes of dilute industrial wastes containing
toxic metals or fluoride are of common  occurrence in the United
States.  Several examples are listed:

-------
      1.  Wastewaters from copper and brass mills and casting
 operations and from copper smelters and refineries are a signi-
 ficant pollution problem.  There are about 100 rolling and
 drawing mills and over 500 plants involved in the casting of
 copper and brass in the U.S.; and there are about 30 copper
 smelting and refining plants, with a total annual production
 of some 1.3 million tons.  Rinsewater and acid bath dumps are
 the major pollution problems at the mills and foundries; acid
 baths are usually 5% to 10% sulfuric acid, and are often follow-
 ed by 3% to 8% sodium dichromate bath which may contain hy-
 drofluoric acid.  The major contaminants in these wateswaters are
 copper, zinc, chromium, and acid, together with smaller amounts
 of fluoride and nickel.  Copper smelter waste may contain on
 the order of 0.6 mg/1 of cadmium, 7 mg/1 of copper, 7 mg/1 of
 lead, and 7 mg/1 of zinc.  Electrolytic refinery waste may con-
 tain 50 to 100 mg/1 of copper, 1 mg/1 of lead, and 10 to 20
 mg/1 of nickel.  It was noted in a recent report on the non-
 ferrous metals industry that "most pressing needs... were related
 to  smelter-refinery-type wastewaters which were generally acid
 and contained metals or metalloids in the range ^1 to 100 ppm." (1)
     2.  A large amount of fluoride wastewater results from the
use of wet scrubbers in aluminum refining.  The primary aluminum
smelting industry carries out the reduction of purified alumina
to produce aluminum metal.  This is done electrolytically in
electrolytic cells or pots connected in series in a potline.  Some
30 such reduction plants in the U.S. have a total annual capacity
of about 5 million tons of aluminum.  The electrolysis of
alumina in a salt bath of alumina and cryolote results in off-
gases containing oxygen, CO, C02, fluoride, particulate material,
and other constituents.  Water from wet scrubbers used for air
pollution control in the potline is a major source of waste-
water in the primary aluminum industry; these scrubber waters
generally contain up to several hundred ppm of fluoride, de-
pending upon recycle and/or treatment.

     3.  Wastewaters from secondary lead smelters are a serious
pollution problem.  About 40 smelters in the U.S. account for
over 50% of domestically produced lead; input includes princi-
pally used storage batteries and telephone cables, and the
smelted lead is used primarily in storage battery manufacture.
Major pollutants in wastewaters from these operations are lead
(several hundred mg/1) and antimony (up to 50 mg/1) ;  sulfuric
acid is also present.

     The heavy metals and fluoride present in these wastes are
toxic pollutants that can affect man, animals, and aquatic life,
and can completely disrupt secondary sewage treatment facilities
if discharged to sanitary sewers .

-------
                                 SECTION IV


                             LITERATURE REVIEW


     This review is restricted to ion foam fractionation, colloid flotation,
and precipitate flotation; work on ore flotation, microflotation of bacteria
and algae, foam separation of biological materials, and other applications
has been excluded.  Non-foaming adsorptive bubble separations are excluded
also, since it is felt that these are unlikely to lead to satisfactory appli-
cation in the area of interest.

     The earlier work is already covered by review; Sebba's book (2) gives a
particularly detailed discussion, and Cassidy (3) published a review in 1957.
Articles by E. Rubin and Gaden (4) and by Eldib  (5) discuss foam separations
in some detail, and Grieves (6) published an article in 1968 summarizing his
work on foam flotation methods in waste water purification.  A. J. Rubin's
dissertation (7) contains a discussion of general methods and a lengthy biblio-
graphy through 1965.  Karger and DeVivo's review.(8) gives a detailed and
systematic survey of principles and applications, with 81 references through
1968.  Lemlich (9) published a review at about the same time, and has more
recently edited a book on the subject (10).  Somasundaran (11) published an
extensive review in 1972.

     The basis of all foam flotation methods is  the Gibbs equation, which
relates the rate of change of surface tension with bulk solute activity to
the surface excess of solute.  Adamson's book (12) discusses the Gibbs equa-
tion in some detail, and Karger and DeVivo make  clear its application in their
excellent discussion of the general theory of adsorptive bubble separations
(8).  Newson (13) has shown that for the simple  removal of surfactant by
foaming the Gibbs equation is valid, indicating  close approach to equilibrium
in the foaming process.  Grieves and his coworkers have developed a model which
enables them to use batch foam fractionation rate data from one system to pre-
dict the performances of both batch and continuous systems (14).  Karger et al.
(15) and subsequently Pinfold (16) have discussed and systematized the nomen-
clature in the field.  The analysis of the operation of a foam stripping column
has been attacked by a number of workers (13, 17, 18, 19, 20, 21); the earlier
work is reviewed by Goldberg and E. Rubin (21), who also reference a number of
papers on foam drainage, which has a profound effect on the efficiency of foam
separations.  Jorne' and E. Rubin have used the  Gouy-Chapman model of the
diffuse double layer to explain how the ionic charges and sizes of species
present determine the selectivity of surface adsorption of counterions of foam
fractionation (22).  E. Rubin and Gaden derived  the operating equation for
single-stage foam columns (4).

-------
      Ion flotation and foam fractionation involve the removal from solution
 of dissolved ions  by the  addition of surface active collector followed by
 foaming.  In the first technique, ion and surfactant form an insoluble pre-
 cipitate or scum on the foam surface;  in the second, no solid phase is formed.
 In precipitate flotation  the ion is  reacted with a non-surfactant to form a
 precipitate which  is  then removed by flotation with a surfactant.   In ad-
 sorbing colloid flotation the dissolved material is removed by adsorption
 onto colloidal particles  which are then removed  by flotation.   We next review
 representative recent experimental work in these areas.

      A.  J.  Rubin and  his  coworkers have published papers on the foam separation
 of Pb(II)  (23),  of Zn(II)  (24),  and  of Cu(II)  and Fe(II)  (25)  with sodium
 lauryl  sulfate.  Stearylamine was also used with Cu(II) and Fe(III).   Effects
 of pH and  ionic strength  were studied,  and conditions yielding foam fractiona-
 tion and precipitate  flotation were  employed.  Earlier, A.  J.  Rubin,  Johnson,
 and Lamb made  a study in  which the effects  of a  number  of variables on the
 two processes were  investigated (26).   Recently  Rubin and Haberkost reported
 on the coagulation  and flotation of  titanium dioxide with aluminum hydroxide
 and an ethanol  frother (27).   They found that sodium lauryl sulfate was a quite
 satisfactory surfactant for  this,  and  that  the process  was  extremely efficient
 if run in the pH range  of minimum solubility of  A1(OH)3.

     Early work  includes  studies by Walling and  his  associates  (28),  by Sebba
 (29), and by Schnepf  and his  coworkers  (30).   Banfield  and  his  collaborators
 (31) noted a fall-off of  distribution coefficient with  increasing  surfactant
 concentration above the critical micelle  concentration; Karger and DeVivo make
 this observation, too  (8), as have others  (13, 32).

     Karger  and Miller studied the foam fractionation of  chloro complexes of
 Fe(III), and Hg(II),  and Co(II) with a  cationic  surfactant,  and demonstrated
 that these could be separated by control of chloride concentration (33, 34).
 Karger, Poncha, and Miller showed that use of reflux could yield extremely
 good separations from highly dilute solutions  recovering methyl orange from
 0.17 ppm solution (35).  DeVivo and Karger  (36)  studied the  effects of aggre-
 gation in the flotation of colloidal particulates--kaolin and montmorillonite
were floated with ethylhexadecyldimethylaimionium bromide, and  zeta potential,
 cation concentrations, pH, and ratio of surfactant to initial particulate
 concentrations were varied.

     Talbot  (37) has investigated the foam flotation of Cu(II) with sodium
 lauryl sulfate, and Miller and Sullivan  (38) have studied the foam fractiona-
tion of Hg(II) nitro complexes.  Schonfeld and his coworkers have observed
that surfactants which can act as chelates as well as simple exchange sites
 are unusually effective (39).  Lemlich and others have carried out studies
demonstrating the utility of refluxing in foam separations  (5, 18, 35, 40,
43).  Schonfeld and Kibbey (42) constructed an excellent foam fractionation

-------
apparatus with which they obtained concentration factors of greater than 103
in removing radioactive strontium.  Shakir has reported on the foam separation
of U(VI) from sulfate media (43).  Sheiham and Pinfold (44) have studied the
effects of added electrolytes on the flotation of hexadecyltrimethylammonium
chloride and of dodecylpyridinium chloride.  Maas (45) has shown that addition
of organic vapors to the air stream generating the bubbles may markedly speed
up some foam separations.  Bikerman (46) has studied the foam fractionation
of calcium, iron, and manganese with alkyl sulfates and poly(oxyethylene)
sulfates, and noted that non-ionic surfactants were ineffective.  Dziomko
and Sidenko (47) carried out ion flotation of dichromate using N,N,N',N'-
tetrametJiyl-N,N'-didodecylhexainethylenedianimonium dibromide.  Charewicz and
Niemiec (48) investigated the flotation of perrhenates with cationic surfac-
tants (amine salts), and Podneuk (49) patented a method for ion flotation of
tungsten and molybdenum with amines in acid solution.  Sebba (50) patented a
method for the precipitate flotation of metal hydroxides (Fe, Cu, Zn) from
dilute solution.  Charwicz and Niemiec (51) found that cationic surfactants
yielded good separation of chloroaurate.  Moroi and Matuura (52) investigated
the removal of Cs(I) and a number of Co(III) complexes with sodium dodecyl
sulfate.  Kepak and Kriva (53) used dodecylamine or gelatine to separate ionic
and colloidal forms of Ru(IV) chlorides and RuNO(III) nitrate.  Stachurski
(54) developed a stochastic mathematical model for flotation processes which
he checked against data on molybdenum.  E. Rubin and Jorne' (55) investigated
surface hydrolysis effects in foam separations involving anionic surfactants.
Rabrenovic (56) has carried out ionic flotation of uranium from dilute solu-
tion.  Kim and Zeitlin (57, 58, 59) have used adsorbing colloid flotation
(with ferric hydroxide collector and sodium dodecyi sulfate) to recover traces
of molybdate and uranyl carbonate; they have used ferric hydroxide and dodecyl-
amine to recover trace amounts of zinc and copper.  More recently, Matsuzaki
and Zeitlin published what promises to be extremely useful work on the foam
separation with various surfactants of many of the coprecipitating collectors
commonly used in trace analyses in oceanographic work  (60).

     Grieves, Bhattacharyya and their coworkers have devoted much effort to
applying flotation techniques to waste treatment problems.  They have studied
flotation of dichromate (61, 62), phosphate (63), phenolate (64), chromic
hydroxide (65), and cyanide complexed with Fe(II) (66).  They recently investi-
gated the stoichiometry of the separation of the anions I", HCrOit", S203~, and
Ag(S203)~3 with ethylhexadecyldimethylammonium bromide (67).  Earlier work
concerned the foam separation of active carbon (68), the flotation of dichro-
mate with ethylhexadecyldimethylammonium ion (69), the effect of colloidal
particulates on foam fractionation (70), the foam fractionation of colloid-
surfactant systems (71), and foam separation processes for removing inorganic
and organic ions with surfactants (72).

     Ferguson and Wilson have recently developed methods for the removal of
lead and cadmium from industrial wastes by precipitate flotation and by

-------
adsorbing colloid flotation (73).   Nguyen and Phillips have made a
rather detailed study of equilibria and mass transfer rates in the
continuous foam fractionation of phosphate (74).   Ervin and Banner
investigated the effects of a number of parameters, including reflux
ratio, on the continuous foam fractionation of phenol (75).  Clarke
and Wilson have demonstrated the feasibility of floe foam flotati9n
of fluoride with Al(OH)o and sodium lauryl bromide to remove cadmium
and mercury (77).       J

     A streaming potential phenomenon predicted recently by Wilson
and Wilson (78) was observed by Clarke (79).  Floe foam flotation
isotherms and local rate effects were investigated by means of a
modified Gouy-Chapman model (80,  81, 82, 83).  Recently Wang and
his coworkers have worked out a theory for continuous bubble frac-
tionation columns; their approach requires an assumption of local
equilibrium between surface and bulk phases (84,  85, 86).  Goldberg
and Rubin have reviewed a number of models for foam flotation column
operation and analyzed stripping columns without solute transfer in
the countercurrent region (21).  Cannon and Lemlich have given a
detailed analysis based on the assumption of linear isotherms (87),
and Lee has given a somewhat similar treatment (88).  Wilson et al.
have analyzed stripping column operation with axial diffusion, non-
linear isotherms,  and finite rate of solute transport between suf-
face and bulk liquid phases (82).
                                10

-------
                                  SECTION V


                     OBJECTIVES OF THE RESEARCH PROGRAM


     The objective of our work was to carry out bench scale studies using
adsorbing colloid flotation techniques to remove copper, lead, cadmium, and
fluoride from actual and simulated industrial wastewaters, in order to lay
the groundwork for the development of economic and simple methods to concen-
trate these toxic substances from industrial wastes.  Time permitted us to
study techniques for the removal of mercury, arsenic, and zinc, as well.  The
apparatus is rather simple, consisting of an air compressor, foaming column
and sparger, metering equipment for introducing the collector surfactant and
the reagents which form the adsorbing colloid, a foam breaker, and (if it is
desired to reclaim the surfactant) a filter or centrifuge for recovering the
surfactant solution from the collapsed foamate.

     We desired to determine the lowest concentration levels to which these
toxic substances could be feasibly reduced by adsorbing colloid foam flotation,
and what variables significantly affect the speed of removal and the ultimate
concentration of contaminant in the column effluent.  Although continuous flow
studies were not originally within the scope of this project, time and manpower
permitted us to undertake some continuous flow studies and to construct a
column simulator computer program.

     Availability of a computer program simulating column operation would
greatly facilitate the design of continuous flow columns for industrial use.
We developed the theory of adsorption isotherms and rate effects, and con-
structed a column simulator computer program which takes into account axial
mixing, finite rate of solute transport between the surface and bulk phases,
and realistic adsorption isotherms.  We also made use of the Gouy-Chapman model
and the Smoluchowski equation to determine the magnitude of the rate of solute
transport to the surface phase due to driven diffusion.  We were also able to
use the Gouy-Chapman model, modified to take into account the finite volume of
the floe particles, and the methods of statistical mechanics to calculate
improved adsorption isotherms for charged floes on interfaces occupied by ionic
surfactant molecules.  This theoretical work gives us a clear understanding of
(1) the various factors governing adsorption of floes on surfaces of films,
and (2) the factors governing continuous flow stripping column performance.

     The possibility of recycling surface active agent was also established.
                                      11

-------
                                 SECTION VI


                     BATCH TECHNIQUE LABORATORY STUDIES


     The bulk of the experimental work described in this report was carried
out in three essentially identical batch type apparatuses, one of which is
diagrammed in Figure 1.  The column was made of glass, 3.8 cm outside diameter
(OD) by 90 cm in length.  A 2.6 on-OD side arm 3.5 on from the top of the
column permitted discharge of foam to a container; a 1.0 cm-OD side arm with
a standard taper joint 12 cm from the bottom of the column provided a port
for insertion of a miniature combination electrode for measuring pH.  The
bottom of the column was closed with a rubber-stopper through which were
inserted a stopcock for sampling, a "fine" glass gas dispersion tube for
introduction of air, a drain, and a small tube to a 10 cm3 hypodermic syringe
for introduction of reagents during the course of a run.  House air was used
to generate the bubbles; it was passed through ascarite, water (for rehumidifi-
cation) and glass wool.  A sulfuric acid scrubber and silica gel used in some
of the work were found to be unnecessary.  Airflow was controlled with a small
needle valve and measured with a soap film flowmeter and a stopwatch.

     Laboratory grade sodium lauryl sulfate (NLS) was used as the anionic
surface active agent; hexadecyltrimethylammonium bromide (HTA) was used as the
cationic surfactant.  Reagent grade chemicals were used throughout, except for
chemicals present in the industrial wastewater samples.

     The basic procedure was essentially the same for all the batch separations
on synthetic samples.  Stock solutions (1.000 g/l of the metal ions, fluoride,
and surfactant) were mixed in the desired amounts and diluted to nearly 200 ml
with deionized water, sodium nitrate was added if desired to increase the ionic
strength, the pH was adjusted to the desired level by addition of 1.0 M NaOH
and 0.10 M HN03, and the solution diluted to 200 ml.  The solution was then
added to the column and the airflow rate measured several times with the soap
film flowmeter.  The pH was monitored during the course of the run, and approx-
imately 7 ml samples were withdrawn from the bottom of the column at 5-minute
intervals.  The samples were analyzed by atomic absorption spectrophotometry
(lead, cadmium, mercury, copper, and zinc) or by colorimetric methods (fluoride
by the SPADNS method, arsenic by ammonium molybdate-stannous chloride).


FOAM SEPARATION OF LEAD(II) BY BATCH TECHNIQUES


     Previous work by Ferguson (74) has demonstrated the feasibility of re-
moving lead from dilute solutions by means of floe foam flotation with ferrous
sulfide and HTA.  The relatively high cost of HTA compared to NLS and the


                                      12

-------
Figure 1.  The batch apparatus.
1-  air needle value
2-  ascarite for C02 removal
3-  humidifier
4-  glass wool column
5-  fritted glass sparger
6-  reagent syringe
 7-  drain
 8-  pH meter
 9-  pH electrode port
10-  foam discharge port
11-  discharged foam
12-  soap film flowmeter
                                          10
                                              11
                                                      12
                                    5Q
                               13

-------
deterioration and toxicity problems inherent in the use of sulfides led us to
seek alternative methods employing NLS and floes other than sulfides.  Two
systems which have yielded rather good results with lead are Fe(OH)3-NLS and
A1(OH)3-NLS.  Ferric chloride, alum, and technical grade sodium hydroxide are
all readily available in quantity at relatively low cost, and NLS is one of
the cheapest of the commercially available surfactants.

     Floe foam flotation separations are often markedly dependent on pH and
on ionic strength, for reasons discussed in Section VIII and Appendix A.  We
therefore systematically varied these parameters for both the Fe(QH)3-NLS and
the A1(OH)3-NLS systems.  The results for A1(OH)3-NLS are summarized in Table
1; those for Fe(OH)3-NLS, in Table 2.


	TABLE 1.  FLOG FOAM FLOTATION OF LEAD WITH M(OH)3 AND NLS*	^

                                    Residual lead (ppm)
PH
6.0
6.5
7.0
7.5
8.0
At 0 M
NaN03
2.4
2.3
1.2
2.2
8.2
At .025 M
NaN03
6.1
6.2
2.9
6.5
15
At .050 M
NaN03
7.6
6.4
4.4
7.2
14
At .075 M
NaN03
9.1
8.3
6.2
16
23
At .100 M
NaN03
10.0
10.0
21
20
26
*A11 runs made with 50 ppm Pb(II), 100 ppm Al(III) and 30 ppm NLS initially;
 initial volume = 200 ml, air flow rate = approx. 85 ml/min; duration of
 run = 20 min.

          TABLE 2.  FLOG FOAM FLOTATION OF LEAD WITH Fe(OH) 3 AND NLS*
Residual lead (ppm)
PH
6.0
6.5
7.0
7.5
8.0
At 0 M
NaN03
1.4
.0
.1
.0
.1
At .025 M
NaN03
1.4
.0
.14
.20
1.1
At .050 M
NaN03
1.5
.10
.42
.56
2.8
At .075 M
NaN03
1.3
1.0
.7
2.2
6.3
At .100 M
NaN03
2.1
1.1
5.8
2.6
19
*A11 runs made with 50 ppm Pb(II), 100 ppm Fe(III) and 30 ppm NLS initially;
 initial volume = 200 ml, air flow rate = approx. 85 ml/min, duration of
 run = 15 min.
                                      14

-------
     The tabulated data indicate that: (1) increasing ionic strength does
interfere with the separations; (2) pH must be controlled to within roughly
one pH unit; and (3) ferric hydroxide is a more effective adsorbing floe for
lead than is aluminum hydroxide.  The volume of foamate resulting from a
typical run was of the order of 1-2 ml.  A few runs were made using sodium
carbonate as the neutralizing agent; it was found to yield poorer results
than sodium hydroxide, probably due to stronger adsorption of this divalent
anion on the positively charged floe particles.  This would reduce the elec-
tric charge of the floe particles, thereby weakening their binding to the
negatively charged bubble surfaces.  The data also indicate that the Fe(OH)3-
NLS system may well be capable of reducing lead levels down to 0.1 ppm in
continuous flow systems; this is confirmed by the results presented in
Section VII.

     A few runs were made with wastes from a secondary lead smelter (battery
scrapping wastewater) containing 3.4 and 3.8 ppm lead; foam flotation of these
samples resulted in reductions down to about 0.1 ppm when the pH was adjusted
to 5.0.  Apparently the high concentrations of sulfate present in these wastes
shift the optimum pH to a somewhat lower value.  Conductivities of these sam-
ples were 2.5 x 10 ^ and 3.5 x 10 3 micromhos (ymhos) ; they contained 1.53 and
2.03 g of dissolved solids per liter.  A few runs were also made with samples
containing 2.0 g/£ of Na2SOtt and 50 ppm Pb; 10-minute runs at pH 6.0 reduced
the Pb concentration to 0.2 ppm.

     Calcium carbonate was used as the adsorbing colloid with either NLS or
HTA in a few runs.  HTA did not remove CaC03 or lead in the pH range 9.0-10.5.
NLS reduced the lead concentration from 50 ppm down to about 0.2 ppm (100 ppm
Ca(II), 30 ppm NLS) at pH 7.0, but the efficiency of removal decreased very
markedly as pH was changed (14 ppm residual lead at pH 6.5, 8.6 ppm at pH 7.5).
We felt that the extremely precise pH control required for this separation
would be very difficult to maintain in industrial practice, and therefore did
not pursue the use of CaCOa any further.
     A few runs were made using Fe(OH)3 and gelatin for lead removal.  Occa-
sional good separations were obtained  (reductions from 50 ppm lead down to
approximately 0.1 ppm), but the results were erratic and seemed to depend
fairly critically on the details of .the preparation of the gelatin solution.
A pH of 7.0, 250 ppm gelatin,  and 100 ppm Fe(III) gave the best results.  In
view of the difficulties in obtaining reasonably reproducible results and the
large quantity of gelatin needed, we did not feel that further work on this
technique was warranted.

     Batch runs with Mn.02 and  NLS rapidly removed lead, reducing lead concen-
tration from 25 ppm down to approximately 0.2 ppm in 5 to 10 minutes; 25 ppm
NLS and 100 ppm Mn(II) were used, and the optimum pH was approximately 8.0.
In this procedure the Nfti(II) is rapidly air oxidized to Mn02 under alkaline
                                      15

-------
 conditions, and then the surfactant is added to effect removal.  Increasing
 ionic strength interferes with the separation.  Our results did not indicate
 that this method was superior to the use of Fe (OH) 3 and NLS, and manganous
 salts are a good deal more expensive than ferric chloride, so further work
 with manganese was not done.


 FOAM SEPARATION OF COPPER(II) BY BATCH TECHNIQUES


     The ease with which lead was removed by the system Fe(OH)3-NLS and the
 low cost of the reagents needed with this system dictated that it be investi-
 gated first as a technique for removing copper(II).  The effects of varying
 ionic strength and pH are summarized in Table 3.


	TABLE 5.  FLOG FOAM FLOTATION OF COPPER WITH Fe(OH)3 AND NLS*	

                                Residual copper (ppm)

  pH                               Added NaN03 (M)

5.5
6.0
6.5
7.0
7.5
8.0
0
3.0
.10
.02
.02
.01
.02
.025
6.4
.37
.11
.11
.07
.10
.050
7.2
.61
.17
.07
.13
.77
.075
6.2
1.4
.17
.06
.55
.71
.10
5.5
2.7
.17
.22
1.4
>13
.15
>13
1.0
.28
2.8
13.1
>13
.25
>13
6.5
1.5
1.1
5.2
>13
.50
>13
>13
>13
>13
>13
>13
*A11 runs made with 50 ppm Cu(II), 100 ppm Fe(III) and 50 ppm NLS (25 ppm
 initially, 25 ppm 5 min after the start of the run).  Initial volume = 200 ml,
 air flow rate = approx. 60 ml/min, duration of runs = 25 min.


     In these studies the details of the experimental method were changed
slightly; the NLS was added in two equal portions, one at the start of the
run, and one five minutes later.  This resulted in visibly marked improvement
in the removal of the floe, and also in lower residual copper levels.  The
effect is presumably due to the maintenance of the surface concentration of
NLS on the bubbles (and therefore the surface electric charge density) at a
high level for a longer period of time than is possible when all of the sur-
factant is supplied at the start of the run.
                                      16

-------
     The residual copper levels obtained with Fe(OH)3-NLS at pH's of approx-
imately 7 and ionic strengths of 0.1 mole/liter or less were sufficiently
low that it was felt unnecessary to examine other, more expensive techniques.
As will be seen later in this section, it is possible to remove copper as
CuS with HTA; we have preliminary data indicating that it can be removed
with A1(OH)3-NLS and Mn02-NLS, but have not determined the optimum pH for
these separations.

     A number of samples of copper-containing wastewaters were treated by
the Fe(OH)3-NLS technique.  One sample contained 190 ppm Cu(H), and its elec-
trical conductivity of 3.35 x 103 ymhos and total dissolved solids of 1.3 g/£
indicated an ionic strength of about 0.04 m/£ as Na2SOi+.  Treatment with 250
ppm of Fe(III) and 100 ppm NLS (50 ppm initially, 25 ppm after 6 min, 25 ppm
after 11 min) resulted in final Cu(II) concentrations of about 0.15 ppm after
25 min at a pH of 7.0.  Another sample containing 23.5 ppm Cu(II) and having
an ionic strength of about 0.04 m/£ yielded residual Cu(II) concentrations of
about 0.25 ppm with the same treatment at pH 6.0-6.5.  A third sample contain-
ing 5.8 ppm Cu(II) was reduced to about 0.05 ppm by this treatment at pH
6.0-6.5.


FOAM SEPARATION OF MERCURY(II) AND CADMIUM(II) BY BATCH TECHNIQUES


Precipitate Flotation of HgS. CdS, and Cd(OH)g


     Both mercuric sulfide and cadmium sulfide are very insoluble in water,
as indicated by the solubility product (89):

                        HgS  t  Hg++ + S=      pKs = 53.8

                        CdS  t  Cd++ + S=      pKs = 28.0

The pKg for HgS suggests that one try to remove mercuric ion by precipitate
flotation in the form of mercuric sulfide.  The experimental results are given
in Table 4.

     It was found that mercuric sulfide could not be removed by (anionic) NLS
at any pH, although it could be removed by (cationic) HTA.  Better results were
obtained under highly acidic conditions; the mercury concentration was 0.1 ppm
after 30 min flotation at pH 0.8.  The sulfide ion added initially was varied
from 10"** m/£ (the amount required to precipitate 20 ppm Hg(II) or 10 ppm
Cd(II) to 2 x 10"3 m/£).  It was found that the removal rate and the maximum
amount removed were not sensitive to the sulfide ion concentrations in this
concentration range.  Figure 2 shows the kinetic study of precipitate flotation
                                      17

-------
      TABLE  4.  Hg(II) REMOVAL BY PRECIPITATE FLOTATION OF MERCURIC SULFIDE
               INITIAL Hg(II) CONCENTRATION = 20 PPM
Sulfide
(m/-e)
0.0001
.001
.0001
.0005
.001
.0003
.002
.002
.002
.001
PH
10,7,4
10,7,4
10,7
10,7
10,7
3.5
1.8
1.8
0.8
0.8
NaN03
(m/£)
0.001
.002
.001
.001
.002
.001
.01
.01
.1
.1
Time
(min)
60
60
30
30
30
60
60
30
30
30
Surfactant
NLS
NLS
HTA
HTA
HTA
HTA
OTA
HTA
HTA
HTA
Residual
mercury
Cppm)
>5
>5
>0.5
> .5
> .5
.2
.12
.30
.10
.11
of mercuric sulfide with HTA.

     The precipitate flotation of cadmium sulfide had been studied by
Ferguson (74).  The cadmium concentrations were reported as 0.1 to 0.23 ppm
after 30 min flotation at various pH values.  No trend with pH could be deter-
mined for the cadmium sulfide removals.

     Precipitate flotation of cadmium hydroxide was tried.  The solubility
product of cadmium hydroxide is 2.0 x 10"lk (89).  The solution was adjusted
to pH 11.5.  The precipitate in the form of cadmium hydroxide was foamed with
HTA.  The cadmium concentration was still 0.8 ppm after 30 min flotation, and
this separation technique was not pursued further.
Adsorbing Colloid Flotation with A1(OH)3 or Fe(OH)a
     Adsorbing colloid flotation with A1(OH)3 or Fe(OH)3 was tried.  The exper-
mental results for the systems containing 20 ppm mercury(II) are illustrated
in Tables 5 and 6.  It was found that A1(OH)3 could be removed by NLS at pH
7 to 8, but it could not be removed by HTA.   Fe(OH)3 could be removed by HTA
                                      18

-------
Q.
Q.
                           10
30
                                     20
                       T I M E . M i N
Figure 2.  Hg(II) removal by precipitate flotation of HgS with HTA.

-------
         TABLE 5.  Hg(II) REMDVAL BY FLOG FOAM FLOTATION WITH A1(OH)3

pH
7.2-7.5
7.5-8.0
7.6-7.5
7.5-8.2*
Time
(min)
60
60
60
45
Surfactant
NLS
NLS
HTA
OTA
Residual
mercury
(ppm)
>5
>5
>5
>5
         *Removal with Na2C03.  In all runs the total salts concentra-
          tion was .006 M and the initial Al(III) concentration was
          50 ppm.
         TABLE 6.  Hg(II) REMDVAL BY FLOG FOAM FLOTATION WITH FeCOH)3*
PH
10,12
10
8
NaN03
(m/£)
0.01
0.005
0.005
Surfactant
HTA
NLS
NLS
Residual
mercury
(ppm)
>5
>5
>5
         *In all runs the run duration was 30 min and the initial Fe(III)
          concentration was 100 ppm.


at pH 10 to 12; it could also be removed by NLS at pH 8.  But even though
A1(OH)3 or Fe(OH)3 was removed, the residual mercury concentrations were always
higher than 5 ppm after one hour flotation.  Adjusting pH with sodium carbonate
did not improve the separation.

     Adsorbing colloid flotation of cadmium with 100 ppm Al (OH) 3 was not suc-
cessful either.  The experimental results are given in Table 7.  Cadmium ion
concentrations were higher than 4 ppm after each run.  If 50 ppm phosphate ion
was added, 2 ppm cadmium ion remained in the solution after a run at pH 7.5.

     When the ionic strength was lower than 0.01 m/£ Fe(OH)3 was very effective
in removing cadmium ion from water at highly basic conditions (pH 11 to 12).
Sanple solutions contained 10 ppm cadmium and 100 ppm ferric ion initially.
                                     20

-------
        TABLE 7.  Cd(II) REMOVAL BY FLOG FOAM FLOTATION WITH A1(OH)3*
Phosphate
(ppm)
0
0
100
50
50
pH
6.5-6.8
7.5
6.5
7.5
7.5
Time
(min)
30
20
25
25
45
Residual
cadmium
(ppm)
>5
4.5
5
5
2
        *In all runs the initial Al(III) concentration was 100 ppm,
         the surfactant was NLS, and the NaN03 concentration was 0.01 M»


The solution was adjusted to pH 12 and then 20 ppm HTA was added.  The super-
natant contained approximately 0.25 to 0.50 ppm after 30 min standing.  The
supernatant was poured into the flotation column and was treated by foam
flotation.  HTA was added in pulses to maintain a stable foam.  Only 0.01 ppm
cadmium ion remained after 30 min flotation.  The maximum removal decreased
with increasing ionic strength.  Table 8 shows the effect of ionic strength
and pH on maximum removal.


         TABLE 8.  EFFECT OF IONIC STRENGTH AND pH ON Cd(II) REMOVAL
                   BY FLOG FOAM FLOTATION WITH Fe(OH)3*
Residual cadmium (ppm)
pH
9.5
10.0
10.5
11.0
11.5
12.0
At .006 M
NaN03
0.10
.06
.05
.01
.01
.01
At .016 M
NaN03
—
—
—
0.10
.07
.07
At .30 M
NaN03
—
—
—
0.20
.10
.08
At .056 M
NaN03
—
—
—
0.22
.12
.10
At .100 M
NaN03
—
—
—
0.22
.22
.22
*The supernatant solution containing . 25 to . 50 ppm of cadmium was treated.
                                     21

-------
     If this method is applied to industrial waste water treatment, filtra-
 tion before flotation is necessary.  Pouring all the precipitate into the
 flotation column yields no separation at all.  The experimental results are
 shown in Table 9.  Adjusting the pH with sodium carbonate was tried; some
 precipitate remained in the solution after 40 min flotation with HTA, and
 the final cadmium ion concentration was approximately 1 ppm.
    TABLE 9.  Cd(II)  REMOVAL BY FLOG FOAM FLOTATION WITH Fe(OH)3 AND Na2C03
pH
8.5,10
8.5,10
(with Na2C03)
11.6
(with Na2C03)
5.5
(with Na2C03)
7.7
Added NaN03
OnAO
0.005
.005
.01
.005
.005
Time
(min)
30
30
40
30
30
Surfactant
HTA
HTA
HTA
MLS
NLS
Residual
cadmium
(ppm)
>5
>5
1
>5
3.5
*Removal with Na2C03.   In all runs the initial Fe(III) concentration was
 100 ppm.


     Ferric hydroxide could be removed effectively by NLS at pH 5, but the
cadmium ion concentration remaining in solution was greater than 3 ppm.
Adjusting the pH with sodium carbonate did not improve the separation.


Adsorbing Colloid Flotation of Mercuric Sulfide and Cadmium Sulfide with
Ferrous Sulfide


     Adsorbing colloid flotation of lead sulfide and cadmium sulfide with
ferrous sulfide was tried by Ferguson et al. (74) and was proved to be very
effective.  The CdS was adsorbed on FeS, then removed by foam with HTA as
collector at pH 9.  The addition of HTA to the solution of colloidal FeS
causes the FeS to coagulate and the CdS apparently adsorbs onto the colloid.
The solution may be filtered to remove the bulk of the precipitate, then
foamed to remove the remaining traces, or it may just be foamed initially,
with all the FeS being removed in the foam.  When the initial cadmium ion
                                      22

-------
concentration was 1 ppm, 20 ppm ferrous ion was added, the sulfide ion
concentration was 1 x 10~3 m/£.  Approximately 0.012 ppm cadmium remained
in the solution after 45 min flotation.

     Filtering the solutions left different amounts of Cd in the solution
depending on the amount of HTA added to coagulate the precipitate, the
length of time the precipitate was allowed to coagulate, and the grade of
filter paper used to filter it.  Typically 0.3-0.4 ppm CdS was not filtered
out from an initial 10 ppm solution by Whatman No. 42 filter paper if 50 ppm
FeS had been coagulated with 40 ppm HTA and then allowed to coagulate for
5 minutes.  If these filtered solutions were foamed for 30 minutes to remove
the final traces of precipitate, typical cadmium ion concentrations in the
bulk solution were about 0.003 ppm (74).

     We applied the same method to remove mercuric ion from solution.  Fil-
tering was not tried, since it would increase costs if applied to industrial
wastewater treatment.  The sulfide ion concentration was 4 x 10"3 m/£ ini-
tially.  The ionic strengths were adjusted by addition of sodium nitrate.
The pH of the solution was adjusted, and then 100 ppm HTA was added to coagu-
late the colloid.  The solution was poured into the flotation column and
foaming was initiated.  The experiments are summarized in Table 10.  After
30 min flotation, the mercuric ion concentration was 0.05 ppm when the ionic
strength was 0.01 m/£ and 0.07 ppm when the ionic strength was 0.1 m/£.  Best
results were obtained at pH 9.  Under acidic conditions FeS could not be
removed effectively.


       TABLE 10.  Hg(II) REMDVAL BY FLOG FOAM FLOTATION OF HgS WITH FeS*

Fe(II)
(ppm)
20
20
50
50

Sulfide
(m/£)
0.0008
.0008
.004
.004

PH
9
5
9
9
Added
NaN03
(m/£)
0.002
.002
.01
.1
Residual
mercury
(ppm)
>0.20
> .30
.05
.07
       *The duration of all runs was  30 min, and the surfactant used was
        HTA.
                                     23

-------
Adsorbing Colloid Flotation of Mercuric Sulfide and Cadmium Sulfide with
Cupric Sulfide


     The most effective method for removing mercuric ion and cadmium ion
was the adsorbing colloid flotation of mercuric sulfide with cupric sulfide.
This method was very effective even when the ionic strength was as high as
2 m/£.  It took only 20 minutes to decrease mercury concentrations from 20
ppm to 0.005 ppm.  The cadmium concentration was decreased from 10 ppm to
0.02 ppm with 30 minutes flotation.  This method was insensitive to variation
in pH, such that we could safely run it near neutral pH without adjusting the
pH before or after the flotation.

     10 ml of 0.2 m/£ sodium sulfide solution was added to the solution con-
taining 100 ppm Cu(II) and 20 ppm Hg(II) or 10 ppm Cd(II).  The ionic strength
was adjusted with sodium nitrate.  The total volume of the solution was 500
ml.  40 ppm HTA was added to coagulate the colloid.  The solution was permitted
to stand for 10 minutes, and then the solution along with the precipitate was
poured into the flotation column.  The gas flow rate was kept at 80 ml/min.
Additional surfactant (HTA) was added in pulses during the run to maintain a
stable foam (approximately 10 ppm surfactant was added every 6 min).  The foam
was quite dry; less than 10 ml solution was carried out by the foam during a
30-minute run.  The experimental results for mercury removal are given in
Table 11.
       TABLE 11.  Hg(II) REMOVAL BY FLOG FOAM FLOTATION OF HgS WITH CuS
Cu(II)
(ppm)
12
12
12
12
12
60
60
60
100
100
100
100
100
Sulfide
(m/£)
0.0008
.0008
.0008
.0008
.0008
.002
.002
.002
.003
.004
.004
.008
.008
pH
9.5
6
4
2
1
1
0.8
0.8
0.8
6-6.7
6-7
6-7
7-8
Added
NaN03
(m/£)
0.002
.002
.002
.01
.1
.1
.2
.6
.6
.01
.1
.5
2.0
Time
(min)
30
30
30
30
30
20
20
60
20
20
20
25
40
Residual
mercury
(ppm)
0.20
.20
.20
.10
.09
.005
.005
.50
.005
.005
.005
.005
.005
                                      24

-------
     Increasing the ionic strength did not affect the maximum ultimate
removal of mercury, but it did affect the rate of removal.  In order to
decrease the mercury concentration to 0.005 ppm, it was necessary to foam
for 20 min when the ionic strength was 0.1 m/£, 25 min when the ionic strength
was 0.5 m/£, and 40 min when the ionic strength was 2 m/£.  Varying the pH
within the range 1 to 8 did not affect the maximum removal or the rate of
removal.  The cupric ion concentrations remaining in the solutions were always
less than 0.5 ppm.

     Varying the pH and the ionic strength has some effect on the maximum
removal of cadmium.  Table 12 shows this effect.  When the ionic strength
was 0.1 m/£ or below, the maximum removal was not sensitive to the variation
in pH (the residual cadmium ion concentration was 0.02 ppm); but when the ionic
strength was 0.5 m/£, better results were obtained at pH 7.0 to 8.5.  The ex-
perimental results under optimum pH and different ionic strengths are shown in
Table 13.  About 0.25 ppm cadmium(II) remained in the solution after a 45-min-
ute run when the ionic strength was 2 m/£.  The cupric ion concentration re-
maining in the solution was always less than 0.5 ppm.


     TABLE 12.  THE EFFECT OF pH VARIATION ON Cd(II) REMOVAL BY FLOG FOAM
                FLOTATION OF CdS WITH CuS
Ionic
strength
(m/r)
0.1
.5

At pH
3.0-3.1
0.02
.07
Residual cadmium
At pH At pH
4.0-4.1 5.0-6.5
0.02 0.04
.07 .12
(ppm)
At pH
6.0-7.8
0.02
.07

At pH
7.0-8.5
0.02
.05
  TABLE 13.  THE EFFECT OF INCREASING IONIC STRENGTH ON Cd(II) REMOVAL BY
             FLOC FOAM FLOTATION OF CdS WITH CuS*
Added
NaN03 (m/£)
0.1
.5
1.0
1.5
2.0
Time
(min)
30
30
30
45
45
Residual
cadmium (ppm)
0.02
.05
.15
.25
.25
Residaul
copper (ppm)
<0.5
< .5
< .5
.5
.5
  *The initial Cu(II) concentration was 100 ppm in all runs; the initial sul-
   fide concentration was 0.004 m/£, and the pH was 7-8.
                                      25

-------
      The effect of interference from doubly-charged ions was examined.
 0.1 m/l sulfate ion  (sodium sulfate) or 0.1 m/l calcium ion  (calcium nitrate)
 had no effect on the maximum removal of mercury, but affected the maximum
 removal of cadmium.  In 0.1 m/£ sulfate solution, 0.08 ppm cadmium ion and
 less than 0.5 ppm cupric ion remained in the solution after a 45-minute run.
 In 0.1 m/£ calcium ion solution, 0.18 ppm cadmium ion and 2 ppm cupric ion
 remained.

      The sulfide ion remaining in the solution after a run was complete was
 determined by titrating the solution (100 ml) with 0.1 normal KMnO^ standard
 solution.  A typical solution contained 10 ppm Cd(II), 100 ppm Cu(II), 0.1
 m/£ Na2SOit, and 4 x 10~3 m/l sulfide ion initially; the flotation was run at
 pH 6 to 7 for 30 min; 8 x 10 ~5 m/£ sulfide ion remained in the solution after
 the end of the run.

     A smelter waste containing 30 ppm Cd(II) was treated with 400 ppm Na2S,
 200 ppm Cu(II), and 160 ppm HTA.  After 15 minutes of foaming, the Cd(II) level
 was  reduced to 0.08 ppm.  A second sample treated with 200 ppm Na2S, 100 ppm
 Cu(II)  and 80 ppm HTA showed reduction from 30 ppm to 0.1 ppm Cd(II) in 30
 minutes.

     It was pointed out by Ferguson et al. (74) that cadmium sulfide was not
 adsorbed to the container walls at any of the pH values studied (pH 6.5 to 9.5).
This was checked again for the solutions containing cadmium sulfide or mercuric
 sulfide.  After runs were made at neutral pH (pH 6 to 8), the solution was
 acidified to pH 2 or below; no increase in the cadmium concentration or the
mercury concentration was found.

     Adsorbing colloid flotation of mercuric sulfide with cupric sulfide turned
 out  to be the most effective way to remove mercuric ion from waste water.  The
mercuric ion concentrations remaining in solution after treatment were about
 0.005 ppm, which was considered to be safe for drinking water (55).  The cupric
 ion  concentration remaining in solution after treatment was less than 0.5 ppm,
which was less than the standard for drinking water (1 ppm).

     The same method was applied to remove cadmium ion from water.  Approxi-
mately 0.02 ppm cadmium remained after treatment; which was also comparable
to the standard for drinking water (0.01 ppm).

     Adsorbing colloid flotation of cadmium sulfide with ferrous sulfide was
more effective than the above method.   Typical cadmium ion concentrations in
the bulk solution after treatment were 0.01 ppm or less.
                                     26

-------
FOAM SEPARATION OF ARSENIC BY BATCH TECHNIQUES


     It was found that arsenic  (as arsenate) could be reduced from initial
concentrations of 10-20 ppm down to 0.1-0.2 ppm by floe foam flotation
with Fe(OH)3-NLS in the pH range 4 to 5 at ionic strengths up to 0.2 m/£.
Two preliminary runs indicated  that arsenic could also be reduced from
10-20 ppm down to about 0.1 ppm by floe foam flotation with Mn02-NLS at
pH's of about 8.5.


FOAM SEPARATION OF FLUORIDE BY  BATCH TECHNIQUES


     Attempts to remove fluoride were made using Fe(OH)3, CaC03, or A1(OH)3
as adsorbing floes and NLS or HTA as surfactants.  Neither Fe(OH)3 nor CaC03
proved successful in removing fluoride from water when foamed with either
HTA or NLS in the pH range 3 to 10.  Aluminum hydroxide-HTA is also not
effective in removing fluoride; A1(OH)3-NLS, however, is quite effective.

     The Al+ 3 (or Al (OH) 3) forms a complex with the F~, making the anion
unavailable for analysis by the SPADNS method.  Foaming with HTA does not
remove the complex at either acidic or basic pH's.  NLS, however, does effect
removal, performing best in the approximately neutral pH range of 7.3 to
7.8.

     At pH's below 7 a hydro-alumino-fluoride complex is formed as evidenced
by the low free F~ analysis before any surfactant is added.  At acid pH's
however, there is no aluminum removal after 30 minutes, as evidenced by
colorimetric aluminum analysis.  At pH's above 8 the complex appears to
decompose, returning free F~ to the solution.  An actual example of this fol-
lows: A solution containing 10  ppm F~ and 40 ppm Al+3 was adjusted to a pH
of 2.0; the free F~ was found to be 1.5 ppm.  The pH was then adjusted to
8.1; after 60 seconds of agitation, the free F~ concentration was 1.9 ppm.
After 20 min at a pH of 8.6, the F~ concentration in the solution had increased
to 8.2 ppm.

     The removal of the colloidal complex is effected over approximately a
0.5 range in pH for two reasons: (1) The C02-free air is mixing with the C02-
containing water, removing C02  and increasing the pH.  This effect was observed
by not adding the surfactant and noting the pH increase as the air was bubbled
through the Al-F solution in the column.  (2) The second and probably dominant
effect is the ion flotation of  H+ by LS~.  A larger pH range, 7.1 to 7.9, can
be successfully employed as long as the majority of the foaming occurs within
the 7.3 to 7.8 range.
                                      27

-------
     Parameters of the experimental runs other than pH were : 10 ppm F~ , 40 ppm
AT1"3, 40 ppm NLS in a total volume of 250 ml, room temperature, 24-25°C, and
an air flow rate of approximately 35 ml/min.  These conditions and the above
pH range will be implied by the subsequent use of the term "optimum system".
Both the fluoride and aluminum concentrations of this system were reduced to
undetectable levels (less than 0.1 ppm)  within a 30-min period, as indicated
by their respective quantitative tests.

     The amount of fluoride which could be removed was determined as follows:
optimum aluminum and NLS concentrations  and pH were maintained, and the fluo-
ride concentration was increased to 19.7 ppm.  The fluoride concentration
remaining in solution after 30 min was 4.0 ppm, a removal of 15.7 ppm.  The
aluminum concentration was <0.1 ppm.  Under the same conditions and an initial
F~ concentration of 27.6 ppm, the residual F~ concentration after 30 min was
11.8 ppm, a removal of 15.8 ppm.  These data correspond to an Al/F removal ra-
tio (ppm Al added/ppm F removed) of 2.54.  Increased NLS concentration (72 ppm)
did not improve the F~ removal.  A system containing 120 ppm Al"1"3, 26 ppm F~
and 43 ppm NLS yielded complete (<0.1 ppm residual) F" removal within 30 min.
           was used to vary the ionic strength of an optimum system.  An
optimum system with no added NaNOa has an ionic strength of 0.01.
     The system's removal efficiency decreases relatively slowly with increas-
ing ionic strength.  At concentrations of NaNOa above 2.5 m/£ the foams were
very poor and there was essentially no aluminum removal.

     Because of its ubiquitous nature and similarity to F~ , chloride ion was
studied as a possible interference to the system.  Addition of 0.5 m/£ NaCl to
an optimum system decreased the fluoride removal by at most 8.3% below a simi-
lar system containing 0.5 m/£ NaNOa.  An optimum system containing 1.0 m/£ NaCl
exhibited decreased fluoride removal by 8.2% below the 1 m/£ NaNOs system.  An
optimum system containing 8.9 ppm Cl" showed <0.1 ppm F~ in solution after 30
min of flotation.

     When the Al-F-NLS solution was stored for more than a week in laboratory
glassware, leaching of Al occurred.  Borosilicate glass (pyrex, kimax, etc.) is
2% by weight A1203.  Concentrations as high as 25 ppm Al were found after an 8-
day delay when solutions initially containing 0 ppm Al were stored at a pH of
about 8.  Analyses within one day of sampling indicated negligible leaching.

     Batch runs were made with two samples of wet scrubber wastewater from an
aluminum smelter.  One sample contained 150 ppm F~; treatment with 600 ppm Al
(III), 50 ppm Fe(III), and 100 ppm NLS at pH 6.0 resulted in a residual F" con-
centration of 0.9 ppm.  A second sample contained 375 ppm F"; treatment with
1500 ppm Al(III), 150 ppm Fe(III), and 175 ppm NLS resulted in a residual F"
concentration of 2.5 ppm.  Precipitation of F~ from this waste with excess
Ca(N03)2 yielded a residual F~ concentration of 55 ppm.
                                     28

-------
                                  SECTION VII


                            CONTINUOUS FLOW STUDIES


     Though batch studies are most conveniently used for preliminary investi-
gation of foam flotation separations, continuous flow methods would be expected
to yield closer control of operating conditions, greater reductions in contam-
inant concentration  (by using the foam itself as a stripping column), and
reduced labor costs when scaling up for industrial use.  As we have seen
earlier, fairly accurate control of pH is necessary in a number of these
separations; in a flow system such -continuous control can readily be achieved
with presently available pH control equipment.  In the batch technique one
is able to obtain only about one equilibrium stage of separation in the column;
in continuous flow stripping columns, one uses the foam itself for much the
same function as a reflux column in distillation, with corresponding improve-
ment in performance.  The batch technique requires frequent preparation of
the column batches and draining and charging of the column; also, for optimal
results, it is necessary to add surfactant from time to time during the course
of a run.  All of these would add to the labor costs of the process, and argue
for the use of continuous flow methods.  Two small continuous flow columns were
therefore built, and lead(II) removal with Fe(OH)3-NLS, fluoride removal with
A1(OH)3-NLS, and arsenic removal with Fe(OH)3-NLS were carried out.

     The smaller column is glass, 4.8 cm-ID by 121 cm in length, with inlet
dispersion head 42 cm from the top of the column, foam discharge port 6 cm from
the top, and pH electrode port 70 cm from the top (see Figure 3).  Bubbles are
generated by admitting air through a "fine" fritted glass gas dispersion tube
inserted in a stopper closing the bottom of the column.  Air flow rates are
measured with a small variable area flowmeter and/or with a soap film flowmeter
and stopwatch; air flow is controlled with a small needle valve.  House air is
humidified and passed through glass wool.  Influent is prepared by adding the
appropriate chemicals to two liters of tap water, adjusting the pH, and trans-
ferring the solution to a 3-£ polyethylene storage reservoir on a magnetic
stirrer.  The influent flow is controlled with a screw-type pinchclamp and is
measured by a small  variable area flowmeter or by simply timing the duration
of the run.  A small auxiliary reservoir serves to maintain a constant head on
the inlet dispersion head, which receives influent via a glass tube passing
through the large stopper at the bottom of the column.  Effluent is drained
from the bottom of the column by means of tygon tubing, the position of which
can be adjusted to control the height of the liquid pool at the bottom of the
column.  The foam is  discharged from the foam discharge port near the top of
the column into a 3-£ funnel, in which the foam is collapsed by a spinning
wire spider driven by a laboratory stirring motor.  The pH can be monitored
continuously by inserting a micro combination glass electrode into the elec-
trode port half-way  down the column.

                                      29

-------
 Figure 3.   The small  continuous  flow  apparatus.
 1-  air needle valve
 2-  humidifier
 3-  air flowmeter
 4-  glass wool  column
 5-  fritted glass sparger
 6-  magnetic stirrer
 7-  influent reservoir
 8-  influent valve
 9-  influent flowmeter
10-  constant head reservoir
11-  influent dispersion
12-  foam baffles
13-  effluent drain
14-  effluent
15-  pH meter
16-  pH electrode port
17-  foam discharge port
18-  foam breaker
19-  collapsed foamate
head
                                  30

-------
     It is quite important to design and place the distribution head in such
a way as to distribute the influent evenly over the cross-sectional area of
the column.  If this is not done, severe channelling and overturning of the
wet foam occur at substantially lower influent flow rates than can be
tolerated without column upset in a properly designed system.  The tendency
to channel and overturn was further decreased by mounting a series of five
copper screen baffles in that portion of the column 9 to 22 on below the
influent dispersion head.  Even with these precautions, column upset remains
the limiting factor in determining the flow rate which can be handled by
the column.

     The problems mentioned above, together with the fact that the columns
must be built so as to handle the maximum flow rate of waste at minimum capital
and operating costs, point up the need for good mathematical models of column
operation and of the processes determining the adsorption isotherm of the floe
on the film surface.  Such models are discussed in the next two sections.

     The results obtained with this column (described below) were such as to
warrant scaling up.  We also felt that it would be difficult to estimate the
effects of channelling and foam overturn in larger systems from only our work
on the 4.8 cm column.  A lucite column 10.2 cm-ID by 186 cm length was there-
fore constructed.  Influent is introduced through a tube passing through the
top cover plate of the column; on the end of this tube is mounted the disper-
sion head, normally positioned about 80 cm below the top of the column.  The
foam discharge port is mounted in the top cover plate.  Bubbles are generated
by admitting air through a cylindrical 2.5 x 2.5 cm bubbling stone mounted
on the bottom cover plate.  Air flow rates are measured with a variable area
flowmeter; influent rates, by measuring the time required to discharge 1.0 t.
House air is humidified and passed through glass wool.  Influent is prepared
by adding the appropriate chemicals to approximately 40 L of tap water in a
large plastic trash container; the solution is stirred continuously while the
pH is adjusted and the run is made.  The effluent is pumped from the storage
reservoir to a smaller constant head reservoir above the column, from which
its flow to the dispersion head is controlled by a stopcock.  Four screen
baffles are mounted 95 to 113 on below the top of the column; these are made
of 0.6 on stainless steel wire mesh.  Effluent is discharged from the column
through a tygon tube the position of which can be adjusted to control the
level of liquid in the bottom of the column.  The foam is discharged to either
a spinning screen and wire spider foam breaker or a hot surface foam breaker.
At the flow rates possible with this apparatus, the foam breakers tend to
become overloaded, and we plan to build a larger one.  Reductions of lead le-
vels from 50 ppm to 0.25 ppm at a flow rate of about 1 £/min have been achieved
during the preliminary studies with this apparatus.

     Twenty-five runs were made using the smaller of the two columns for the
removal of lead with Fe(OH)3-NLS.  Optimum pH for synthetic samples containing
                                      31

-------
0 to 6 g/t of Na2SOit was in the range 6.0-6.5.   Air flow rates were generally
in the range 180-250 ml/min; influent flow rates,  in the range 40-80 ml/min.
Fe(III) concentration was generally set at 100  ppm; NLS, at 50 ppm; and Pb(II),
at 50 ppm.  Lead concentrations in the column effluent of 0.1 ppm or less were
readily attainable at Na2S(\ concentrations of  4.0 g/£, 2.0 g/l, or 0.0 g/£;
a run made with 6.0 g/£ added Na2SOit yielded effluent containing 0.13 g/£
Pb(II).  Decreasing the NLS concentration to 25 ppm or the Fe(III) concentra-
tion to 50 ppm resulted in higher residual Pb(II)  concentrations (11 ppm and
0.3 ppm, respectively, for solutions containing 2  g/£ Na2SOit).  Foamate volumes
were measured on ten runs, and ranged from 45 to 90 ml.  (The volume of
influent was 2000 ml.)  Samples of battery waste containing 3.8 and 3.4 ppm
lead(II) yielded effluents containing approximately 0.1 ppm residual lead.

     Eleven runs were made with the small column separating arsenic (VI) with
Fe(OH)3-NLS.  Samples containing 20 ppm As(VI), 100 ppm Fe(III), and 80 ppm
NLS and 0.1 N NaN03 consistently yield residual arsenic levels of less than
0.1 ppm at pH's below 4.4.  If the inert salt added is Na2S04 (0.1 N), it is
necessary to go to a pH of about 3.5 to obtain  a residual arsenic level of
0.1 ppm; the procedure fails completely at a pH of 4.4 with 0.1 N Na2S04.

     One run with the small column was made with a sample of aluminum smelter
wet scrubber wastewater containing 150 ppm fluoride; after treatment with 600
ppm Al(III) and 100 ppm NLS at pH 6.5, the residual F~ concentration was
1.5 ppm.

     Twenty-nine runs were made with the small  column separating copper (II)
with Fe(OH)3-NLS.  Four samples containing 200  ppm Cu(II) were treated with
400 ppm Fe(III) and 95 ppm NLS at pH 6.5 to yield 0.39, 0.39, 0.24, and 0.12
ppm residual copper.  Air flow rates of about 200  ml/min and influent flow
rates of about 50 ml/min were used.  At these high copper and iron concentra-
tions difficulty with clogging of the screen baffles occasionally occurred;
this results in channelling and breakthrough of iron and copper in the efflu-
ent.  Actually, 200 ppm Cu(II) is far in excess of what we expected could be
handled at all by foam flotation methods.  Industrial waste samples containing
82, 23.5, and 5.8 ppm Cu(II) on treatment by the above method yielded efflu-
ents containing 0.45, 0.08, and 0.05 ppm residual Cu(II).
                                      32

-------
                                 SECTION VIII


                                  DISCUSSION


     The data summarized in the preceding two sections demonstrate the ability
of floe foam flotation techniques to reduce the concentrations of lead,
cadmium, mercury, copper, zinc, arsenate, and fluoride to quite low levels.
The feasibility of some of these separations when used in the continuous flow
mode is demonstrated in Section VII.  The results of the batch studies are
summarized in Table 14.


    TABLE 14.  OPTIMAL SEPARATIONS BY BATCH FLOG FOAM FLOTATION TECHNIQUES
Pollutant
Cu(II)
Pb(II)
Arsenate
Zn(II)
Fluoride
Cd(II)
Hg(H)
Flotation
system
Fe(OH)3-NLS*
Fe(OH)3-NLS
Fe(OH)3-NLS
A1(OH)3-NLS
A1(OH)3-NLS
CuS-HTAt
CuS-HTA
pH
•x.7.0
^6.5
4-5
8.0-8.5
6-7.5
3-8.5
7-8
Added Residual
NaN03 pollutant
(m/£) (ppm)
<0.05 0.1
<.l .1
<.2 .1
<.15 .2
<.2 1.5
<.l .02
<2.0 .005
    *NLS - sodium lauryl sulfate.  tHTA - hexadecyltrimethylammonium bromide.


     To provide practical tests of these separations a number of samples of
industrial wastewaters were obtained and treated.  The small quantities of
wastewater available precluded careful optimization of conditions of separa-
tion, but, as is seen in Table 15, levels of separation comparable to those
obtainable with synthetic samples are obtained.  We note that two wastewater
samples were obtained which could not be satisfactorily treated as received.
One contained copper, the other lead, and both contained very high concentra-
tions of sulfate (in excess of 15 g/£ as sodium sulfate).  Dilution of these
samples made it possible to treat them, but we wish to point out that floe
foam flotation generally is not well-adapted to the treatment of wastes of
high ionic strength.
                                      33

-------
            TABLE 15.  INDUSTRIAL WASTES TREATED
Wastewater
Pollutant identity
and concentration
      (ppm)
Residual pollutant
   concentration
       (ppm)
Copper
Brass mill waste #1
Brass mill waste #2
Brass mill waste #2
Brass mill waste #3
Brass mill waste #3
Brass mill waste #4
Brass mill waste #4
Brass mill waste #5
Brass mill waste #5
Brass mill waste #5
Brass mill waste #6
Brass mill waste #7
Brass mill waste #7
Wire plant waste #1
Wire plant waste #8
Secondary lead
Smelter waste ffl
Smelter waste #1
Smelter waste #1
Smelter waste #2
Smelter waste #2
Smelter waste #3
Smelter waste #4
Smelter waste #5
Smelter waste #1
Smelter waste #1
Aluminum
Smelter waste #1
Smelter waste #1
Smelter waste #1
Smelter waste #2
Smelter waste #2
Smelter waste #2

Cu
Cu
Cu
Cu
Cu
Cu
Cu
Cu
Cu
Cu
Cu
Cu
Cu
Cu
Cu

Pb
Pb
Pb
Pb
Pb
Pb
Pb
Pb
Cd
Cd

F
F
F
F
F
F

82
23.5
23.5
5.8
5.8
95
95
100
100
100
18
4.3
4.3
50
190

3.4
3.4
3.4
3.8
3.8
3.8
1.5
1.9
30
30

150
150
150
375
375
375

0.45
.12
.04
.05
.05
.37
.28
.39
.09
.13
.23
.04
.00
.10
.15

.10
.10
.11
.0
.12
.27
.12
.0
.1
.08

1.3
1.2
.9
1.75
2.5
2.5
                            34

-------
     Two continuous flow runs were made  on synthetic samples employing recycled
surfactant (actually decanted collapsed  foamate).  Lead was removed as usual
from these samples, indicating the feasibility  of  cost reduction in these
separations by reuse of surfactant.  Our data as yet are not sufficient to
indicate the fraction of surfactant which is  lost  at each cycle.
                                     35

-------
                                  SECTION IX


                        THEORY OF ADSORPTION ISOTHERMS


     Most precipitate flotation and adsorbing colloid flotation processes
depend on electrical forces between the charged particles of precipitate being
removed and the oppositely charged surfactant layer at the film surface.  It
is shown in reference 78 that, provided the film is not too thick, the rate
of driven diffusion of charged floe particles to the interface is quite rapid,
and the time constant for this process is of the order of a few microseconds.
We noted in our experimental work on sulfides, Mn02, Fe (OH) 3, and Al (OH) 3 with
NLS and HTA that the surface films were often quite densely occupied by floe
particles, and that any theory assuming sparse occupancy (79) would run into
difficulty in calculating isotherms for these very useful systems, and might
well grossly over-estimate column efficiencies.

     Vie here present the physical model and results for a theory of floe foam
flotation isotherms which is applicable to surface densely occupied by floe
particles.  The mathematical analysis is given in Appendix C.

     The electric potential ty is given in the vicinity of a charged surface
on an electrolyte by formulas developed from the Gouy-Chapman theory.  We
assume that a floe particle occupies a cubical volume a. cm on a side, and
therefore that the maximum number density c^^ of floe particles is I/a3.
The potential energy of a floe particle having electric charge q and located
a distance x from the surface is given by qip(x).  The configurational free
energy per floe particle is

                                               c
                           yconfig = kT loge —
                                                 -c
                                              max
where c is the concentration of particles in the bulk liquid, k is Boltzmann's
constant, and T is the absolute temperature.  The number of floe particles per
cm2 of surface within a distance ma of the surface is then given by
                        m
                            1 +
                       1=1 '
                                         1            1
                                1-c
a    -   .     kT
where a = c/Cj,,.^, and the surface excess of particles per unit area in this
same boundary layer is given by
                                      36

-------
                                 m
               Q           n  , V* 	1 - exp gV(i)	
               ^excess ~ o; practically,
this can be accomplished by using surfactant concentrations which are nearly
as large as the critical micelle concentration (at which concentration further
increases do not result in any increase in surface concentration or in the
magnitude of the surface potential).

     Figure 6 illustrates  the fact that increasing the charge on the floe
particles increases the efficiency of separation by increasing the adsorption
isotherm.  This parameter  can very often be controlled by adjusting the pH, a
procedure which is very effective with both Fe(OH)3 and A1(OH)3.  The theory
predicts that the temperature dependence of the adsorption isotherms should
be rather slight, as seen  in Figure 7.  This is in agreement with the small
amount of rather qualitative experimental data we have.  We conclude that,
within reasonable temperature limits, say 0-60°C, floe foam flotation separa-
tions are not likely to be greatly affected by temperature.
                                      37

-------
CO
CO
01
                    ,1
                                                                     o
       Figure 4.   Dependence of isotherms on ionic strength (moles/cm).
                  q =0-4.77 x ID'10 e.s.u. \f>0 = -50 mn., cell length =
                  10 A, T = 298°K.
                                  38

-------
 CO
 
-------
 CO
 (/>
 (L)
 o
 X
 
-------
 d)
 o
 X
 
-------
                                   SECTION X


                     THEORY OF STRIPPING COLIMI OPERATION
     If foam flotation techniques are to be used on an industrial scale,
highly accurate mathematical models for column design will be needed to
facilitate the construction of optimal columns at minimum cost.  We present
here a mathematical model for a continuous flow stripping column; the model
includes the effects of axial dispersion, rate of solute transport between
the surface and liquid phases, and non- linear adsorption isotherms.  We
discuss the model and results here; the mathematical analysis is given in
Appendix B.

     We take the following steady-state mass balance equations as our
starting point:
                   0 = D.V -- + VPV -- +   *    (r-kc)
                        *   dx2     *   dx     V+SK
                   0 = D s      _ - v s -	i    (r-kc)
                        s   dx2     S   dx     V+SK
Terms are defined as follows:

     D»    -- effective diffusion constant in the liquid phase.

     D     -- effective diffusion constant in the surface phase.
      o

     V     -- volume of liquid contained in 1 cm3 of foam.

     S     -- surface area contained in 1 cm3 of foam.

     y»    -- velocity of liquid downward.

     v     — velocity of surface upward.

     K     -- equilibrium adsorption isotherm; K = r/c at equilibrium, and
              is a function of c.

     r     -- surface concentration of solute.
                                    42

-------
      c      --  concentration of solute  in  liquid.

      cfeed  ""  concentration of solute  in  column influent.

      k.^     --  rate  constant governing  solute  transport between the surface
               phase and the liquid -phasev

      x      --  distance  from base  of column.

      £      --  column length.

      6      --film  thickness of drained film.

      v^V   --  flow  rate  of  column influent.


      Solvent material balance  at  the top  of the column yields
                                   = vgS6
Solute material balance at the top of the column yields
         Cfeed - VsSr(£) = V£VC(£) ' vsSr W + °£V         + DsS
                                                     dx            dx


Solute material balance on the liquid phase at the top of the column yields
Solute material balance at the bottom of the column yields
           v.Vc(O) = v.Vc(O) - v Sr(0) + D,V         + D S
                                               dx       s    dx


Lastly, the requirement that there be no net flux of solute out of the bottom
of the column via the surface phase yields
                                     43

-------
                           o = vsr(0)  - DSS
dr(0)
 dx
     The two non-linear differential equations with their associated boundary
conditions are then solved by iterative use of a quasi-linearization method
as presented in Appendix D.

     We found that the adsorption isotherms obtained by the method presented
in Section VIII and Appendix C could be approximated rather well by a non-
linear least squares fit of simple Langmuir isotherms of the form
                                K =
                                         c/a2
In computations modeling column behavior we have therefore used Langmuir
isotherms throughout.

     We define the separation parameter F as Vc(0)/Vf  cfe«d.  High column
separation efficiencies then correspond to small values of F.

     Figure 8 exhibits the anticipated decrease in solute concentration in the
column effluent as the specific foam area S increases; this parameter increases
with decreasing pore size in the frit of the gas dispersion device generating
the bubbles at the bottom of the column.  The effect of varying the surface
flow belocity vs is shown in Figure 9; this quantity is increased by increasing
the gas flow rate through the gas dispersion device; it is approximated in
these calculations that increasing the gas flow rate does not affect the wet-
ness of the foam.  The expected increase of F with increasing liquid flow
velocity v» is shown in Fig. 10.  Increasing axial mixing (increasing Ds and
D£) results in increasing F, as seen in Figure 11.  DX and D£ are minimized
by design of the influent dispersion head to distribute the influent as uni-
formly across the cross-section of the column as possible and by the use of
screen baffles to break up incipient channelling in the foam.  Overloading
the column with influent has been found experimentally to result in very serious
axial mixing and degradation of column performance.

     The magnitude of the rate constant for solute transfer between the liquid
and the surface phases decreases with increasing bubble wall thickness (pro-
portional to foam wetness, or V, and increasing with increasing feed rate).
The effect of increasing ki in improving column performance is seen in Figure
                                     44

-------
       15
       10
           .
 F
xlO3
           Figure 8.   Dependence  of  separation  factor  F on  specific foam
                       area  Stem'1).   Ds  =  D£  =  500,  vs =  v£ =  1, V =  .0025,
                       k-j =  .1,  L  = 50, cpeed  =  1.4 x 10~7 moles/cm3,  a^ =
                       6 x 10"3, a2 = 10~8;  all  constants  in c.g.s. units.
                                     45

-------
CTv
                      Figure 9.  Dependence of separation factor F on surface velocity vc(cm/sec).
                                 S = 60, other parameters as in Fig. 8.                 s

-------
X
      Figure 10.  Dependence of F on liquid velocity (cm/sec).   Other parameters  as
                  in Figures 8 and 9.

-------
oo
                        7,6
                        7,2
                        7,0
                           2,7
3,0
3,3
3,6
                            Figure 11.  Dependence of F on effective diffusion constants D  and
                                        (cm 2/sec). We have here set DS =  D».  Other parameters
                                        as in Figures 8 and 9.

-------
X
    1 -
     -3
-2
-1
      Figure 12.   Dependence of F on mass  transfer rate constant 1^ (cm/sec),
                  Other parameters as in Figures 8 and 9.

-------
t/1
o
               X
                         3,0
                      Figure  13.
                                       1,5
Dependence of F on Langmuir parameter

as in Figures 8 and 9.
                                                                                                 6,0
                                                                           (cm) .   Other parameters

-------
 0
-1
-2
   -9,2
    Figure 14.
              -8,6
-8,0
                        LOR10A2
Dependence of F on Langmuir parameter a  (moles/cm3)
Other parameters as in Figures 8 and 9.

-------
       -6,55
-5,95
-5,35
Figure 15.   Dependence of F on c_  ,  (moles/cm^).   Other
            parameters as in Figures  8 and 9.
                       52

-------
12; beyond a certain value additional increase in kx does not significantly
improve performance because equilibrium characteristics and axial diffusion
become the limiting factors, as is seen on the right side of the figure.

     The dependence of the separation parameter on the constants aj and a2
in the Langmuir adsorption isotherm is exhibited in Figures 13 and 14, and
are qualitatively what one could intuitively expect.  The effect of such
a non-linear isotherm is also shown in Figure 15, in which the dependence of
separation parameter on feed concentration is shown.  As the feed concentra-
tion increases, the surface film becomes saturated; at this point the separa-
tion parameter increases rather rapidly to unity, corresponding to negligible
separation.
                                      53

-------
                                SECTION XI


                                REFERENCES
 1.  Hallowell, J. B., Shea, J. F., Smithson, G. R., Jr., Tripler, A.  B.,
     and Gonser, B. W., "Water Pollution Control in the Primary Nonferrous-
     Metals Industry," Environmental Protection Agency EPA-R2-73-247a,
     U. S.  Government Printing Office, Washington, D. C.  (1973).

 2.  Sebba, F., Ion Flotation, Elsevier, Amsterdam (1962).

 3.  Cassidy, H. G., Technique of Organic Chemistry, Vol. X  (A Weissburger,
     Ed.),  Wiley-Interscience, New York (1957).

 4.  Rubin, E., and Gaden, E. L., Jr., New Chemical Engineering Separation
     Techniques, (H. M. Schoen, Ed.) , Wiley-Interscience, New York  (1962),
     Ch. 5.

 5.  Eldib, I. A., Advances in Petroleum Chemical Refining, Vol.  7  (K. A.
     Kobe and J. F. McKette, Eds.), Wiley-Interscience, New York  (1963), p.
     98.

 6.  Grieves, R. B., The Foam Separation Process, Brit. Chem.  Eng.  13, 77
     (1968).                                                       ^

 7.  Rubin, A. J., Ph.D. Dissertation, "Foam Separation of Microcontaminants
     by Low-flow-rate Methods," University Microfilms, Inc.,  Ann  Arbor
     (1966).

 8.  Karger, B. L., and DeVivo, D. G., General Survey of Adsorptive  Bubble
     Separation Processes, Separat. Sci. 3,  393  (1968).
                                        *\/
 9.  Lemlich, R., Adsorptive Bubble-Separation Methods.  Foam Bractionation
     and Allied Techniques, Ind. Eng. Chem.  60, 16 (1968).
                                            f\/\>
10.  Lemlich, R., ed. Adsorptive Bubble Separation Techniques, Academic
     Press, New York  (1972).

11.  Somasundaran, P., Foam Separation Methods, Separ. Purif.  Methods 1
     (1), 117  (1972).

12.  Adamson, A. W, Physical Chemistry of Surfaces, Wiley-Interscience, New
     York (1960).
                                      54

-------
13.  Newson, I. H., Foam Separation-Principles Governing Surfactant
     Transfer in a Continuous Foam Column, J. Appl.  Chem. London 16,
     43 (1966).                                                  ™>

14.  Grieves, R. B., Ogbu, I. U., Bhattacharyya, D., and Conger,  W. L.,
     Foam Fractionation Rates, Separat. Sci. 5, 583 (1970).

15.  Karger, B. L., Grieves, R. B., Lemlich, R., Rubin, A. J.,  and Sebba,
     F., Nomenclature Recommendations for Adsorptive Bubble  Separations,
     Separat. Sci. 2, 401 (1967).

16.  Pinfold, T. A., Adsorptive Bubble Separation Methods, Separat. Sci.
     5, 379 (1970).

17.  Rubin, E., Schoenfeld, E., and Everett, R., "The Removal of Metallic
     Ions by Foaming Agents and Suspensions," Report for Oak Ridge
     National Lab. RA-104.

18.  Fanlo, S., and Lemlich, R., Predicting the Performance  of Foam
     Fractionation Columns, Proc. A. I. Ch. E. - I.  Chem. E. Joint Meet.,
     London 1965 (9), 75.

19.  Lemlich, R., "Principles of Foam Fractionation," in Progress, in
     Separation and Purification (E. S. Perry, Ed.), Wiley-Interscience,
     New York (1968).

20.  Haas, P. A., and Johnson, H. F., Foam Columns for Countercurrent
     Surf ace-Liquid Extraction of Surf ace-Active Solutes, Amer. Inst.
     Chem. Eng. 11, 319  (1965).
                'XA,
21.  Goldberg, M., and Rubin, E., Foam Fractionation in a Stripping
     Column, Separat. Sci. 7, 51 (1972).

22.  Jorne1, J., and Rubin, E., Ion Fractionation by Foam, Separat. Sci.
     4, 313 (1969).

23.  Rubin, A. J., and Lapp, W. L., Foam Separation of Lead  (II)  with
     Sodium Lauryl Sulfate, Anal. Chem. 41, 1133 (1969).
                                        'XA,
24.  Rubin, A. J., and Lapp, W. L., Foam Fractionation and Precipitate
     Flotation of Zinc (II), Separat. Sci. 6, 357 (1971).

25.  Rubin, A. J., and Johnson, J. D., Effect of pH on Ion and Precipitate
     Flotation Systems, Anal. Chem. 39, 298 (1967).
                                    'V\j
                                   55

-------
26.  Rubin, A. J., Johnson, J. D., and Lairb, J. C., Comparison of Variables
     in Ion and Precipitate Flotation, Ind. Eng. Chem. Process Design
     Develop. 5, 368
27.  Rubin, A. J., and Haberkost,  D.  C., Coagulation and Flotation of
     Colloidal Titanium Dioxide,  Separat. Sci. 8, 363 (1973).

28.  Walling, C.  T.,  Ruff, E.  E.,  and Thornton, J. L., Jr., Adsorption
     of Cations by Anionic Foams,  J.  Phys. Chem. 61, 486 (1957).

29.  Sebba, F., Concentration by Ion Flotation, Nature 184, 1062 (1959).
                                                       'WY.
30.  Schnepf, R.  W.,  Gaden, E. L., Jr.,  Mirocznik, E. Y., and Schonfeld,
     E., Foam Fractionation-Metals, Chem. Eng. Progr. 55, 42 (1959).
                                                      
-------
39.  Schonfeld, E., Sanford, R., Mazzella, G.,  Ghosh,  D.,  and Mook, S.,
     NYO-9577, US AEC Contract No. AT(30-1)-2093.   Radiation Applications,
     Inc., New York (1960).

40.  Lemlich, R., and Lavi, E., Foam Fractionation with Reflux, Science
     134, 191 (1961).
     OAAi
41.  Brunner, C. A., and Lemlich, R., Foam Fractionation - Standard
     Separator and Re fluxing Columns, Ind. Eng. Chem.  Fundamentals
     2, 297  (1963).

42.  Schonfeld, E., and Kibbey, A. H., Improving Strontium Removal  from
     Solution by Controlled Reflux Foam Separation, Nucl.  Appl. 3,  353
     (1967).                                                   ^

43.  Shakir, K., Studies on the Low Gas Flow Rate Foam Separation of U
     (VI) from Sulf ate Media, Separat. Sci. 8, 345 (1973).

44.  Sheiham, I., and Pinfold, T. A., The Solvent Sublation of
     Hexadecyltrimethylammonium Chloride, Separat. Sci. 7, 43 (1972).

45.  Maas,  K., A New Type  of Adsubble Methods:  Booster Bubble
     Fractionation - Hastened and Improved Bubble Fractionation of
     Low-Foaming Solutions, Separat. Sci. 4, 457  (1969).

46.  Bikerman, J. L., U. S. Office Saline Water, Res. Develop. Progr.
     Rep. No. 510  (1970).

47.  Dziomko, V. M., and Sidenko, Z. S.,  Ion Flotation of Bichromate  Ion
     Using  a Bis-Quaternary Ammonium Cation, Zh. Prikl. Khim. (Leningrad)
     43,  436 (1970).

48.  Charewicz, W., and Niemiec,  J., Flotation of Anions with the Aid of
     Cationic  Surfactants. II, Flotation  of Perrhenates, Nukleonika 1.4,
     607  (1969).

49.  Podeuk, A.  K., Patent U.S.S.R.  236,373, Obogashch. Rud  15, 18  (1970).

50.  Sebba, F.,  Patent U.S.  3,476,553  (1969).

51.  Charewicz,  W., and Niemiec,  J., Flotation of Anions with the Aid of
     Cationic Surfactants. Ill,  Flotation of Chloroaurates,  Nukleonika
     14,  799 (1969).

52.  Moroi, Y.,  and Matuura,  R.,  Mechanism of  Elimination and Selective
     Adsorption of Ions by Sodium Alkyl Sulf ate, Mem. Fac.  Sci.,  Kyushi
     Univ.  Ser.  C 7,  51  (1970).
                                    57

-------
53.  Kepak, F., and Kriva, J.,  The Foam Separation of Radioruthenium,
     Separat. Sci. 5, 385 (1970).
                   'Xi
54.  Stachurski, J., Mathematical  Model for Ion-Extraction Flotation
     Process, Arch. Corn. 15, 219  (1970).
                          w.
55.  Rubin, E., and Jorne1, J., Foam Separation:   Surface Hydrolysis
     Effects, J. Colloid Interface Sci. 33, 208 (1970).
                                        I\AJ
56.  Rabrenovic, R., Ionic Flotation of Uranium from Diluted Solutions,
     Rudy 18, 138 (1970) .
          f\J\j

57.  Kim, Y. S., and Zeitlin, H.,  The Separation of Zinc and Copper
     from Seawater by Adsorption Cbllid Flotation, Separat. Sci. 7, 1
     (1972).                                                     ^

58.  Kim, Y. S., and Zeitlin, H.,  A Rapid Adsorbing Colloid Flotation
     Method for the Separation of  Molybdenum from Seawater, Separat.
     Sci. 6, 505 (1971).
          'v
59.  Kim, Y. S., and Zeitlin, H.,  Separation of Uranium from Seawater by
     Adsorbing Colloid Flotation,  Anal. Chem. 43, 1390 (1971).

60.  Matsuzaki, C., and Zeitlin, H., The Separation of Collectors Used
     as CoPrecipitants of Trace Elements in Seawater by Adsorption
     Colloid Flotation, Separat. Sci. 8, 185 (1973).
                                      'X/
61.  Grieves, R. B., and Schwartz, S. M., Batch and Continuous Ion-
     Flotation of Hexavalent Chromium, A. I. Ch. E. J. 12, 746 (1966).
                                                       'X/X/
62.  Grieves, R. B., Bhattacharyya, D., and Ghosal, J. K., Ion
     Flotation of Chromium (VI) Species:  pH, Ionic Strength, Mixing
     Time, and Temperature, Separat. Sci. 8, 501 (1973).
                                          'X/
63.  Grieves, R. B., and Bhattacharyya, D., Foam Fractionation of
     Orthophosphate with a Cationic Surfactant, Separat. Sci. 1, 81 (1966).
                                                              •X)
64.  Grieves, R. B., and Aronica,  R. C., Foam Separation of Phenol with
     a Cationic Surfactant, Nature 210, 901 (1966).
                          '        'X/XA/
65.  Bhattacharyya, D., Carl ton, J. A., and Grieves, R. B., Precipitate
     Flotation of Chromium, A.  I.  Ch. E. J. 17, 419 (1971).
                                            
-------
67.  Grieves, R. B., and Bhattacharyya, D., Foam Separations of Anions:
     Stoichiometry, Separat. Sci. 7, 115 (1972).
                                  f\j
68.  Grieves, R. B., Ghosal, J. K., and Bhattacharyya, D.,  Ion Flotation
     of Bichromate and of Complexed Cyanide:  Surfactants for Qualitative
     Analysis, J. Amer. Oil Chem. Soc. 45, 591 (1968).
                                       r\j\j
69.  Grieves, R. B. , and Wilson, T. E., Flotation of Bichromate Ion,
     Nature 205 , 1066 (1965) .
            '          V    J
70.  Grieves, R. B. , and Bhattacharyya, B., Effect of Colloidal Particul-
     ates on Foam Fractionation, ibid -207, 476 (1965).
71.  Grieves, R. B., and Bhattacharyya, B., Foam Fractionation of Colloid-
     Surfactant Systems, A. I. Ch. E. J. 11, 274 (1965).
                                         rV\j
72.  Grieves, R. B., Foam Separation Processes, Ind. Water Eng. 2, 11
     (1965) .                                                    ^

73.  Ferguson, B. B., Hinkle, C., and Wilson, B. J., Foam Separation of
     Lead  (II) and Cadmium  (II) from Waste Water, Separat. Sci., 9, 125
     (1974) .                                                     *

74.  Nguyen, Y. V., and Phillips, C. R., Prediction of Equilibria
     and Mass Transfer Rates in Continuous Foam Fractionation of Phosphate,
     Separat. Sci. 8, 11 (1973).

75.  Ervin, E. B.,  and Banner, R. P., The Continuous Foam Fractionation
     of Phenol, Separat. Sci. 8, 179  (1973).
                              a.
76.  Clarke, A. N., and Wilson, B. J., The Adsorbing Colloid Flotation
     of Fluoride  Ion by Aluminum Hydroxide in Aqueous Media, Separat. Sci.
     10, 417  (1975).
     r\J\,
77.  Huang, S.  -B., and Wilson, B. J., Foam Separation of Mercury (II) and
     Cadmium  (II) from Aqueous Systems, Separat. Sci. U, 215
      (1976) .                                          ™

78.  Wilson, J. W., and Wilson, B. J., Electrical Aspects of Adsorbing
     Colloid  Flotation, Separat. Sci. 9, 381 (1974).
                                      %
79.  Clarke, A. N., and Wilson, B. J., Electrical Aspects of Foam
     Flotation:   The Streaming Potential, Separat. Sci. 10, 371 (1975).
                                                        O/Xi
80.  Huang, S.  -B. , and Wilson, B. J., Electrical Aspects of Adsorbing
     Colloid  Flotation. II.  Theory of Rate Processes, Separat. Sci. 10,
      407  (1975).
                                    59

-------
81.  Wilson, J. W., and Wilson, D.  J., Electrical Aspects of Adsorbing
     Colloid Flotation. III. Excluded Volume Effects, Separat. Sci.
     11, 89 (1976).
     
82.  Wilson, J. W., Wilson, D. J.,  and Clarke, J. H., Electrical Aspects
     of Adsorbing Colloid Flotation.  IV. Stripping Column Operation,
     Separat.  Sci. 11,  223 (1976).
                   <\/\,
83.  Wilson, D. J., Electrical Aspects of Adsorbing Colloid Flotation.
     V. Non-Ideal Floes and Salts,  Separat. Sci. 11, 389 (1976).
                                                 'VV
84.  Wang, L.  K-P, Granstrom, M. L.,  and Kown, B. T., Continuous Bubble
     Fractionation:  Part I, Theoretical Considerations, Environmental
     Letters 3, 251 (1972).
            o>
85.  ibid., Continuous  Bubble Fractionation:  Part II, Effects of Bubble
     Size and Gas Rate, 4, 233 (1973).
                        ^
86.  ibid., Continuous  Bubble Fractionation:  Part III, Experimental
     Evaluation of Flow Parameters, 5, 71 (1973).
                                   %
87.  Cannon, K. D., and Lemlich, R.,  A Theoretical Study of Bubble
     Fractionation, Chem. Eng. Prog.  Symp. Ser., 68 (1972)  Water 1971,
     p. 180.                                     ^

88.  Lee, S. J., Bubble Tower Performance in Concentrating Dilute Aqueous
     Surfactant, Ph.D.  dissertation,  University of Tennessee (1969) .

89.  Moeller,  T., Qualitative Analysis, McGraw-Hill, New York (1958).

90.  Wilburn,  N. P., Determination of Concentration Profiles in Two-
     Phase Continuous Counter-Current Extractors, Ind. Eng. Chem.
     Fundamentals 3, 189 (1964) .
                  ^
91.  Misek, T., and Rod, V., in C.  Hanson (ed.), Recent Advances in
     Liquid-Liquid Extraction, Pergamon Press, 1971 (p. 216) .
                                   60

-------
                                  SECTION XII


                                  APPENDICES


APPENDIX A.  ADSORPTION  ISOTHERMS


     Here we examine a Gouy-Chapman  type model  for the interaction of a floe
particle and a surfactant-cove red surface  (79,  82, 82).  The electrical
attraction between  the charged  surface of the film and the oppositely charged
floe particles may  be  reduced by the presence of  an ionic atmosphere, the
width and density of which  are  determined by the  dissolved salt concentration
and the temperature.

     For thick films such as  are likely to be encountered in stripping column
operation, the electric  potential i|»(x) and the  distance  from the film surface
are related by Eq.  A-l:
                x =  a.
loge
                               (cosh 2|§. +  1) (cosh 3ta£ -  1)
                                                    2
                               (cosh     -  i) (cosh      +
                                                                          A-l
Here 6   =   (kT)"1

     ij>0  =   electric potential  at x =  0  in  stat volts

     k   =   1.38  x  10"16  ergs/deg

     T   =   absolute temperature

     C   =   concentration of 1-1  electrolyte

     D   =   dielectric  constant of water, approximately  78

     e   =   [electronic charge),  4.77  x  10"   esu
                                         23
     N   =   Avogadro's  number,  6.023 x 10

      -2 _   8ire2N0Ce
     ai  "   	D	
                                      61

-------
     We assume that a floe particle occupies a volume of a3 on3,  and we
examine the rectangular parallelepiped of liquid extending to the right of a
square of film surface of area a2.   This  constitutes a binding site on our
surface film which is capable of binding  1,2...m particles.  We let q(s)  =
Z exp [-BEJ(S)]  be the site partition function when the site is occupied by
                 q(s)  =  2-j   2~i   '"   2-s  2-s   exp c"8e)          A"2
                        Jl=l   J2=2        JS=S  e(js)

If ag sites have  s particles  bound  to them,  then the  canonical ensemble
partition function for the  collection of M sites and  N particles is given by
Q(N,M,T) =  7 J M!  7T  ^  x ,               A-3
                                            Em  r /---via.
                                       M!  7T  L
-------
      m
     s—•
where N   ag = M is our only constraint.  But this is evidently just
      I  i
     s=o
                              H = EC(XT)]M,
where
                                    m
                                   s=o
                                 q(s)xs.
The average number of particles per site is then given by

                                        m
              S = X
                      3 log
                                           sq(s)X£
                                       s=o
                                T      £ q(s)X'


                                        s
     We assume that
                                                                 A-6
                                                                        A-7
                                                                 A-8
                         J2
                                      L - 1

                                      k=l
                                                                 A-9
the energy of any configuration of s particles at the site is simply the
sum of the independent one-particle energies.  This permits us to write
ja-1  ji-1
             ji=2
                        m
                               E
                                     63

-------
                                                                      A-10
where [Pf(ji)]  is  a  one-particle partition function for a particle in
compartment ji.  We write [Pf(ji)] as exp[-gV(jj)]q(jjL), where V(j^) is the
binding energy  of  a particle in compartment Ji to the surface, ~"~4'-  1  n
where Q is the  particle charge.

     On substituting eq. (II) into eq.  (8) we obtain
m     J2-1  ja-l
                                  m
s=o
                        j2-2

                                                                      A-ll
                                                                      A-31
From this result  and eq.  (8)-we  readily obtain
                   s  =
                      k=l
                X qQQ  exp[-3V(k)]
                 + X q(k)  exp[-V(k)]
                                 A-12
^fow q(k)  involves partition  function factors associated with a particle in
a box, essentially,  and with rotational motions.  This is true for all k,
and we therefore make the reasonable approximation that q(k) = q,
independent of k, which yields
                         m
                     s  =
           •Y.
                        k=l
                                 1 +
expCeV(k)]

   Xq
                                                 -1
A=13
                                    64

-------
Now log Xq = gu.  . , - gii.  .    ,
      6  H   MHtotal   MHinternal
             Q, .
           = PMconf igurational,

where the y's represent chemical potentials.   It is shown in ref. 82, that
a configuration^
                    a

                  1 - a
                                                                        A-14
where a is the "density of particles  in the bulk  liquid," the average number
of particles in L compartments  a  large distance away  from the surface,
divided by L.  This  then yields
                         m
                     s  =
                         k=l
                                  .
                                  1-a
                exp[eV(k)]
                                                     -1
                                                                        A-15
     We define  Sexcess  as
    S       = s - Ma,
     excess         »
                                                                        A-16
the net excess  of particles bound per site in the surface  layer.   Substitu-
tion of eq.  (15)  in (16)  then yields Eq.  (17)  as the expression for the
adsorption isotherm.
            excess
                              m
- a)
I
k=l
                                   a[l-exp3V(k)]  + exp3B(k)
                                                                        A-17
                                      65

-------
APPENDIX B.  MODEL OF STRIPPING COLUMN OPERATION


Introduction
     Several mathematical models for foam flotation columns have been
described in the literature; these generally consist of equations derived
from steady state material balance considerations at the ends of the
column, (9, 13, 20).  An integral part of these models is the assumption
of instantaneous equilibrium between the surface and the bulk phases.
Also, they are confined to the plug flow regime.  Goldberg and Rubin (87)
review a number of these models, and also present a model applicable to
a stripping column without solute transfer in the countercurrent region.
Wang et al (84, 85, 86) treat continuous bubble fractionation, and present
a model which includes axial dispersion and the use of equilibrium
absorption isotherms.  Cannon and Lemlich (88) present a detailed analysis
based upon the assumption of linear isotherms, and Lee has given a
somewhat similar treatment (89).

     We here present an approach derived from Wilburn's analysis of two
phase countercurrent extraction (90); he includes effects of diffusion,
convection, and mass transfer, and assumes a linear isotherm.  We analyze
a continuous flow foam flotation column operating at steady state in the
stripping mode, and we include effects of diffusion and turbulence, rate
of mass transfer, and non-linearity of the isotherm.


Analysis


We take equations (1) and (2) as our starting point:

                       d2c          dc      kiS2   (F - Kc)
               0 = D0V	  +  v0V  —  +  	                        B-l
                    *  dx2      *   dx     V + SK
                       d2r          dr      kiS2   (r - KC)
               0 = D S 	  -  v S  —  -  	                        B-2
                    S  dx2      S   dx     V + SK
                                    66

-------



n
n-l

i + l
t
i

1

\

)
                                         \
Figure B-l.  Material Balance Diagram of the Column.
                         67

-------
Terms are defined as follows:


     D^     -  effective diffusion constant in the liquid phase

     D      -  effective diffusion constant in the surface phase

     V      -  volume of liquid contained in one cm3 of foam

     S      -  surface area contained in one cm3of foam

     V.     -  velocity of liquid downward

     V      -  velocity of surface upward

     K      -  equilibrium isotherm;  K = T/c at equilibrium

     T      -  surface concentration  of solute

     c      -  concentration of solute in liquid

     cfeed  ~  concentrati°n °f solute in column influent

     ki     -  rate constant governing mass transport between the surface
               phase and the liquid phase

     x      -  distance from base of  column

     i      -  column length


     The boundary conditions of the problem are determined in the following
way.  First, solvent material  balance at g (see Fig. B-l) yields


                            v^V = vsS6 + VjlV,                           B-3


where 6 is the film thickness  of drained film and v.V is the flow rate of
column influent.   Solute material balance at 3 yielos



   v£V'cfeed -
                                     68

-------
Solute material balance at a yields


           v^Vc(O) = v^Vc(O)  - vsSF(0)  + D£V ^ (0)  + DgS     (0)        B-5
The requirement that there be no net  flux out of the bottom of the column
via the surface phase yields


                          0 = vsr(0)  -  DsS g (0)                       B-6


Solute material balance  on the  liquid phase at the top  of the  column yields



                    V£V'cfeed = v£Vc^  + D£V 3Y W                    B-7


Equations  (4)  through  (7) can be  simplified and combined to yield boundary
conditions  identical to  those used by Misek and Rod for counter-current
extraction  (91):


                      V   C(£) + _*1 ^ ^ =                          B_g
                                V2.   dx        feed
                                      « 0                                B-9
                          vsr(0) - DS     (0) = o                        B-IO
                                   CP)  = o                               B-II

     We  regard K as a function of c (non-linear isotherm) ,  and solve the
differential equations (1)  and (2} as follows.  We subdivide the column
into n compartments as indicated in Fig. 1, and define


                                                                         B-12
                                                                         B-13
                                      69

-------
We shall be constructing an iteractive quasi-linearization method (16)
whereby we make an initial choice of the K=  (based on the  calculation of
c(x) when K(c) is set equal to K(0).  In trie ith compartment our
differential equations are now
                                                     (F1 - ICc1)          B-14
                                           V + SIC
0 = vnV 4£-  +  VD
                               a.
                                          V + SIC
                                                    (r1 -  ice1)
                                                           B-15
We solve.these in the ith region by setting cx(x)  =  c1  exp \\t r1(x) =
T1 exp X^-x to obtain
              0 = r1(SDsXl2
0 =
                        V +  SIC
                                   + c
                  V +
                                                    V + SIC
                                         V +
                                                           B-16
                                                                         B-17
Ihe requirement that T1 and c1 f 0 yields
      0  =
                     SDsX
                                          V + S     V + SKj[
                                              ,.
                                                  V +
                                                           B-18
     We solve this secular equation (which has one  zero root)  to get the
 •    •   •                                     *
X^, (c^, T^), j = 1, 2, 3, 4.  We choose the cj = 1 for all i and j, which

then permits calculation of the T^ from equations  (16)  or (17).   Then
                                     70

-------
        t

Cx)   = y    ou
       £ _ i    J
                                        exp Ax                          B-19
                                        •       •

                                     aJTJ  exp xjx,                       B-20
where the a. are  to be determined from (1)  the four boundary conditions

[equations  (8)  -  (11)], and (2)  the requirements that
                                                                         B-21


                                                                         B-22
                           dx
                                                                         "-24
                              i = 1, 2,...n - 1


This yields a  total  of 4n equations to be solved for the 4n ou.   Wfe
substitute Eq& (19)  and (20)  into (21) -  (24)  to obtain      J
                                     71

-------
1 1
ri ri
ii 12
, i .. i
AI A2
Xiij X2r2
1 1 \ / exp XiX
ri ri \
r3 r, \
x| x£ 1
I
0
0
i i i i / \
Xsi3 Xiil\ / \ 0
i °
^
exp X2x-
0
0
0
0
exp X^x.
0
0 \
i
0
0
exp XitX- /
exp Xi"  x.
  r      i


    0



    0



    0
    0



exp X2  x-



    0



    0
                                             2
                                            ,
                                            X2
                                         0



                                         0
                                     exp
                                                       r3
0



0



0
                                                    exp Xj+1x.
                                                          1



                                                         4+1
                                                         • U



                                                         ,i+l

or, in more compact notation,
                                                                     B-2E
Then
                        D'1 Ci.iDF'1 (i)F(i+l)D(i,i+l)a(i+l)           B_26
                                72

-------
or
So
                         =   T(l)T(2)...T(n-l)a(n)ETa(n)
Our equations (8)  - (11)  become
                         4
              'feed
= V c^ exp(X^)
                                                    ) *n
                                                    Vi
                                                  B-27
                                                                      B-28
                        0 =
                                                  B-29
                        0 -
                                                  B-30
                         0 -
                                                  B-31
or, more compactly,
                        'feed!

                          0

                                     73

-------
                                 =  Aa(l)
                                     'V.
where A and B are matrices defined by eqns.  (28)  -  (31).   On using eq.
 (27) with these we obtain
Then
                    feec
                   0

                   0
                   a(n)

                                 ,AT>
C"1
      a(n)     Ca(n)
B-32
                                                                         B-33
and
                                                                         B-34
as before.  Solute concentrations in the middle of the various compartments
are obtained from
                                     4
                                    3=1
                                                                         B-35
                                     74

-------
where £• is given by eq.  (12).  These c (£ ) are then used to calculate
a new set of K. 's, at which point we go back to eq. (18) and repeat the
calculation up through eq. (34).  We continue these iterations until
results from successive iterations are in sufficiently close agreement.

     The amount of solute which is discharged per second in the column
effluent is given by

                       v0Vc(0) =
the total amount of solute  flowing  into the column per second is given
by v.V c.p  djvie define  a separation parameter, F, as


                             F = Vc(0)/V'cfeed                          B-36


the smallness  of which gives us a criterion of column performance.
                                     75

-------
TECHNICAL REPORT DATA
(Please read laz&uctions on the reverse before completing)
1. REPORT NO.
EPA-600/2-77-072
2.
4. TITLE AND SUBTITLE
FOAM FLOTATION TREATMENT OF HEAVY
METALS AND FLUORIDE-BEARING INDUSTRIAL WASTBvATERS
7. AUTHOR(S)
David J. Wilson
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Vanderbilt University
Nashville, Tennessee 37235
12. SPONSORING AGENCY NAME AND ADO
Industrial Environmental Re
Office of Research and Deve
U.S. Environmental Protecti
Cincinnati, Ohio 45268
RESS
search Laboratory - cin., OH
lopment
on Agency
3. RECIPIENT'S ACCESSION-NO.
5. REPORT DATE
April 1977 issuing date
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
1 BB 610 01-01-03A
11. CONTRACT/GRANT NO.
R-803564
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/12
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Laboratory studies demonstrated that the floe foam flotation techniques are effective
in removing lead, cadmium, mercury, copper, zinc, arsenic, and fluoride from dilute
wastewaters to very low levels. Simulated as well as real industrial wastewaters were
studied. Industrial wastewaters studied originated from primary aluminum smelters,
secondary lead smelters, and brass mills. Copper, lead and arsenic are readily re-
moved with Fe(OH)3 and sodium lauryl sulfate; fluoride and zinc, with A1(OH)3 and
sodium lauryl sulfate; cadmium and mercury, with CuS and hexadecyltrimethylaninonium
bromide. Batch techniques as well as continuous flow systems were used; the latter
proved to be more efficient. Possibility of surfactant recovery was investigated.
Flotation column simulator computer program was also constructed. Floe foam
flotation techniques are not suitable for treatment of wastes containing high
concentrations of dissolved salts and adequate pH control is essential in most
separations .
Foam separation processes combine the attractive features of simplicity, economy,
potential for recovery, and effective removal of pollutants at low concentrations
in wastewaters.
17.
a. DESCRIPTORS
Flotation*
Wastewaters*
Metals*
Surfactants*
Fluorides*
18. DISTRIBUTION STATEMENT
Release to public.
KEY WORDS AND DOCUMENT ANALYSIS
b. IDENTIFIERS/OPEN ENDED TERMS
Floe foam flotation
Batch separation
Flotation column
Continuous flotation
19. SECURITY CLASS (Tills Report)
Unclassified
20. SECURITY CLASS (This page)
Unclassified

c. COSATI Field/Group
13B
21. NO. OF PAGES
88
22. PRICE
EPA Form 2220-1 (9-73)
                                                                      76
                                                                                       . GOVERNMENT PRINTING OFFICt 1977-757-056/5589 Region No. 5-11

-------