&EFK
             United States
             Environmental Protection
             Agency
              Municipal Environmental Research EPA-600/2-80-1 53
              Laboratory          August 1980
              Cincinnati OH 45268
             Research and Development
Selenium
Removal From
Ground Water Using
Activated Alumina


-------
                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-600/2-80-153
                                          August  1980
         SELENIUM REMOVAL FROM GROUND
         WATER USING ACTUATED ALUMINA
                        BY

    R. Rhodes Trussell, Albert Trussell and Peter Kreft
     Oames M. Montgomery, Consulting Engineers, Inc.
             Pasadena, California 91101
               Contract No. 68-03-1515
                   Project Officer

                   Richard Lauch
           Drinking Water Research Division
      Municipal Environmental Research Laboratory
                Cincinnati, Ohio 45268
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
      OFFICE OF RESEARCH AND DEVELOPMENT
     U.S. ENVIRONMENTAL PROTECTION AGENCY
              CINCINNATI, OHIO 45268

-------
                                 DISCLAIMER

     This  report  has been  reviewed  by  the  Municipal 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.
                                      ii

-------
                                  FOREWORD

     The U.S.  Environmental Protection  Agency was created because of increasing
public  and government concern  about the dangers  of  pollution to the health and
welfare of the American people.  Noxious air, foul  water, and spoiled land are tragic
testimonies to the deterioration of our natural environment.  The complexity of that
environment and the interplay of its components require a concentrated and integrated
attack on the problem.

     Research  and development  is that necessary first step in problem solution; it
involves defining the problem, measuring its impact, and searching for solutions.  The
Municipal Environmental Research Laboratory develops new and improved technology
and systems to prevent, treat, and manage wastewater and solid and hazardous waste
pollutant  discharges from  municipal  and community sources,  to  preserve and  treat
public  drinking  water supplies, and to minimize the adverse economic, social, health,
and aesthetic effects of pollution.  This publication is  one of the products of that
research and provides a most vital communications link between  the researcher and
the user community.

     This report  addresses the feasibility  of removing selenium from drinking water
supplies using activated alumina.  This information is helpful to determining operating
parameters for the  activated alumina process, which can also remove  arsenic and
fluoride from drinking water supplies.
                                              Francis T. Mayo, Director
                                              Municipal Environmental Research
                                              Laboratory
                                       ill

-------
                                   ABSTRACT

      Selenium is a contaminant found in trace quantities in some ground- and surface-
waters in the  United States. Currently, the National Drinking Water Regulations limit
total  selenium to .01 mg/1.  Even though the health affects of selenium in "trace"
amounts are still highly debatable, it is prudent to explore the technology of removing
selenium from drinking water.

      Two  species of  inorganic selenium, with  valence states of  +4 and  +6,  are
typically found in selenium-contaminated waters.  Se(IV) and Se(VI) act very different,
chemically. Se(IV) occurs as HSeO3~ in the pH range of 2.7 to 8.5.  Se(VI) occurs as
SeOj? above pH 1.7. The valence of either of these species is thought to be determined
by the oxidation-reduction potential of a water at a certain pH.  Knowing the oxidizing
state of a water, one can predict whether Se(IV) or Se(VI) should be present.

      Historically, only the total  selenium present in a  sample has  been able to be
determined by atomic  adsorption  spectroscopy.   With  the incorporation of  a
fluorometric technique to determine the Se(IV) in a sample, this  study  was able to
differentiate between the two species of selenium.

      Initial batch studies indicated that Se(IV) was preferentially adsorbed over Se(VI)
in side-by-side tests.   The  isotherm capacity  of activated alumina for  Se(IV)  was
roughly three  times  the  capacity of  Se(VI).   Other  studies  indicated  that while
bicarbonate mildly interfered  with  Se(IV) removal, both  bicarbonate and  sulfate
heavily interfered with Se(VI) adsorption.

      Initial column studies with a three-inch deep bed helped delimit the amounts of
NaOH and  H-SCX to be used during regeneration  of Se(IV)-saturated alumina.  Other
items addressed in the  three-inch  column studies included  how varied concentrations

-------
of NaOH and t^SO^ affected regeneration capabilities and how the varied concentra-
tions affected alumina degradation.

     Deeper (9-inch) column studies showed that capacities for Se(IV) decreased with
increasing influent water pH. pH 5 showed the highest capacity for Se(IV) adsorption.
The  kinetics of regeneration  were  the  most important factors in determining the
capacity of  activated alumina for Se(IV), with pore diffusion seeming to be the rate-
limiting step.  Slow  0.5% NaOH flow  rates  (0.5 gpm/ft2 or less) are necessary to
effectively recover a high percentage of Se(IV) removed during a previous treatment
run.  The following regeneration steps are recommended for Se(IV) removal:

     1.5 lb/ft3 of 0.5% NaOH at 1/2 gpm/ft2, upflow
     5 bed volumes of water at 1-2 gpm/ft , downflow
     0.7 lb/ft3 of 0.25% H2SO^ or HC1 at 1 gpm/ft2, downflow
     5 bed volumes of water rinse at 1-2 gpm/ft  , downflow

     The following bed volumes of treated water can be expected to be produced with
a Se(IV) influent concentration up to 200 ppb:

     pH 5 -    1200 bed volumes
     pH 6 -    900  bed volumes
     pH 7 -    500  bed volumes

     Selenium (VI) is  much more poorly adsorbed than Se(IV). It is suspected that this
low adsorption is due to Se(VI)'s higher solubility with the oxides of alumina.   Because
of this  poorer adsorption, NaOH regeneration  amounts and flow rates are  not  as
important in recovering Se(VI) removed during a treatment run.  Because SO^= heavily
interferes with Se(VI) adsorption by  activated alumina, only HC1 can be used as the
acid rinse during regeneration.

     Capacities of activated alumina for Se(VI) are shown below at the following pH's
for an influent concentration of 50 ppb:

-------
     pH 5  -   100  bed volumes
     pH 6  -    70  bed volumes
     pH 7  -    35  bed volumes

        These capacities are based on a SOf concentration of 100 ppm.  The following
  capacities were developed for Se(VI) at different sulfate concentrations at pH 6:

        SO7  =   5 ppm - 450 bed volumes
        S(X  =  50 ppm- 150 bed volumes
        SO^  = 100 ppm - 70 bed volumes
        SOJj  = 500 ppm- 15 bed volumes

     Variations in capacity for Se(VI) with different amounts of bicarbonate were not
as great.

     Preliminary cost estimates were done to predict costs of treatment for either a
Se(IV) or a Se(VI) contaminated supply. Costs are as follows:

     Se(IV)  -  $ 75/acre-foot
     Se(Vl)  -  $255/acre-foot

     If  a mixture of the two species is present, Se(VI) concentrations will determine
the cost of treatment.

     This  report  was  submitted  in fulfillment  of  Contract  No. 68-03-1515  by
James M.  Montgomery, Consulting Engineers, Inc. under the  sponsorship of  the U.S.
Environmental Protection Agency.   This report covers the period April 1,  1979 to
March 21, 1980, and work was completed as of April 21, 1980.
                                       vl

-------
                                  CONTENTS
Foreword	    iii
Abstract	    iv
Figures    	    ix
Tables    	    xi
Acknowledgment	   xii

      1.    Introduction	«...     1
                Background	     1
                Selenium	     1
                Prior Research	     7
                Activated Alumina	     7
                Project Approach	    13
      2.    Conclusions	    13
                Analysis	    13
                Batch Studies	    15
                Column Studies	    18
                Costs		    21
      3.    Recommendations	    23
      4.    Selenium Analysis Studies	    25
                Organic Selenium	    25
                Atomic Absorption Spectrophotometry	    26
                Colorimetric Determination of Selenium (IV)	    32
                Fluorometric Determination of Selenium (IV)	    32
      5.    Selenium Batch Studies	    35
                Regeneration	    35
                Kinetics	    40
                pH Effects	    41
                Mesh Size	    46
                Interfering Ions	    49
                Selectivity Series	    52
                Capacities of Activated Alumina  .	    56
      6.    Selenium Column Studies	    64
                Introduction	    64
                Initial Testing	    67
                Experimental Summary	    69
                Chronological Summary of Se(IV) Removal
                Tests Using 3-Inch Columns	    69
                Chronological Summary of Se(lV) Removal Tests
                Using Nine-Inch Columns	    92
                Chronological Summary of Se(VI) Removal Tests
                Using Nine-Inch Columns	117
                                       vii

-------
                         CONTENTS (CONTINUED)


     7.    Preliminary Cost Estimate  .................   139
References -
Appendix A -
                                viii

-------
                                   FIGURES


Number

1.    Concentration - pH Diagrams for Inorganic Selenium
     Species	     3
2.    E, - pH Diagram for Selenium	•  •     ft
3.    Pfiotograph of A.A.S	    28
ft.    Photograph of A.A.S	    29
5.    Influence of Se(VI) on Fiuorometric Determination
     of Se(IV), 0-7 ppb	    30
6.    Influence of Se(Vl) on Fiuorometric Determination
     of Se(IV), 0-70 ppb	    31
7.    Photograph of Fluorometer	•    3ft
8.    Photograph of Shaking Apparatus	    36
9.    Initial Preparation of Fresh Activated Alumina with NaOH	    38
10.  Initial Preparation of Fresh Activated Alumina with Acid
     After NaOH Rinse  . -	    39
11.  Effect of Contact Time with Activated Alumina on
     Selenium Adsorption	    ft2
12.  Effect of Initial pH on Adsorption of Selenium	    ft3
13.  Effect on Constant pH on Adsorption of Selenium	    ft 5
1ft.  Effect of Mesh Size on Removal of Selenium (IV)	    ft7
15.  Effect of Mesh Size on Removal of Selenium (VI)  .  .  .	    ft8
16.  Influence of Various Anions on Adsorption of Selenium (IV)	    50
17.  Influence of Various Anions on Adsorption of Selenium (VI)	    51
18.  Influence of Various Cations on Adsorption of Selenium (IV)	    53
19.  Influence of Various Cations on Adsorption of Selenium (VI)	    5ft
20.  Freundiich Isotherm Plots of Se(IV) and Se(VI) Adsorption
     in Deionized Water Matrix	    60
21.  Freundiich Isotherm Plots of Se(IV) and Se(VI) in Synthesized
     Well Water Matrix	    61
22.  Schematic Diagram of Testing Equipment During Removal
     Run and Regeneration	    65
23.  Photograph of Column Testing Apparatus	    66
2ft.  Run 1 (3-inch) - Effect of Contact Time	    7ft
25.  Runs 2-6 (3-inch) - Regeneration Tests	    76
26.  Runs 7-12 (3-inch) - Regeneration Tests	    78
27.  Runs 13 and 1ft (3-inch) - Regeneration Tests	    82
28.  Runs 15-17 (3-inch)-Regeneration Tests	    8ft
29.  Runs 18 and 19 (3-inch) - Regeneration Tests	    88
30.  Runs 20-22 (3-inch) -  Regeneration Tests	    90
31.  Photograph of "Wastewatcher" Automatic Sampler	    93
32.  Runs 1-3 (9-inch) - Capacity Tests	    95
                                      ix

-------
                            FIGURES (CONTINUED)
Number                                                                Page

33.  Runs 4-6 (9-inch) - Capacity Tests	    99
34.  Runs 7-9 (9-inch) - Capacity Tests	102
35.  Runs 10-12, Capacity Tests	103
36.  Neutralization of Activated Alumina by 0.05N H-SCX
     and Effluent pH of Subsequent Treatment Run	106
37.  Runs 13-15 (9-inch), Capacity and Saturation Tests	108
38.  Elution Curves for  Regeneration After Run 13	110
39.  Runs 16-18 (9-inch), Capacity Tests	115
40.  Runs 1-3 (Se(VI)), Initial Tests	118
41.  Runs 4 and 5 (Se(VI)), Regeneration Tests	120
42.  Runs 6-8, (Se(VI)), H-SO. vs. HC1 Regeneration	122
43.  Runs 9-11 (Se(VI)),  Regeneration Tests	125
44.  Runs 12-14 (Se(VD), Regeneration Tests	126
45.  Elution Curves for  Regeneration After Run 14	130
46.  Neutralization of Activated Alumina by 0.05N HC1 and
     Effluent pH of Subsequent Treatment Run  .	131
47.  Runs 15-18, SO.= and HCO,~ Interferences	133
48.  Se(VI) Removal vs. SO..~ ana Alkalinity Concentrations	134

-------
                                    TABLES
 1.    Constants Used to Construct E.-pH Diagram	     5
 2.    Verification of Reproducibility^or Regenerant Analysis	    27
-3.    Resuits of Selectivity Series Tests	    57
 4.    Average RMWD Well Water Composition	    68
 5.    Amounts of Various Reagents Added to Make Up
      Synthesized Well Water	    68
 6.    Summary of Results	    70
 7.    Removals and Recoveries of Se(IV) for
      Runs 10, 11, and 12	    81
 8.    Removals and Recoveries of Se(IV) for
      Runs 12, 13, and 14	    83
 9.    Removals and Recoveries of Se(IV) for
      Runs 1*, 15, 16, and 17	    85
 10.   Degradation of Activated Alumina by
      Various Concentrations of NaOH	    86
 11.   Removals and Recoveries of Se(IV) for NaOH and
      H2SO^ Rinses During Runs 19, 20, 21, and 22	    91
 12.   Degradation of Activated Alumina by Various
      Concentrations of H-SO^	    91
 13.   Summary of Removals and Recoveries of Se(IV)
      During all Runs with 9-inch Columns	    97
 14.   Comparison of Removals of Interfering
      Anions and Cations During Run 12	104
 15.   Removals and Recoveries of Se(VI) for Runs 2 and 3	119
 16.   Removals and Recoveries of Se(IV) for Runs 3, 4, and 5	121
 17.   Removals and Recoveries of Se(VI) in Tests Using
      9-inch Columns, Runs 6-18	   127
 18.   Degradation of Activated Alumina by Varied
      Concentrations of NaOH and by 0.05N HC1 Acid Rinse	137
 19.   Cost Estimate for Se(IV) or Se(VI) Removal Facilities	140
                                       xi

-------
                            ACKNOWLEDGMENT


We would like to express our gratitude to Mr. Richard Lauch, Project  Officer, and

Mr. Thomas Sorg of  the Water Supply Research Division of the  U.S. Environmental

Protection Agency in Cincinnati, Ohio, for their technical assistance and interest in

this project.   We would  also  like to thank the following members  of James M.

Montgomery, Consulting Engineers, Inc. staff for providing assistance  and support

during the project:

                               ENGINEERING
                             Michael Kavanaugh
                                 Carol Tate

                   LABORATORY SUPPORT AND ANALYSIS

                            Lawrence Y.C. Leong
                          Raymond G. Zehnpfennig
                                Douglas Peitz
                              Keith Mainquist
                                Gregg Oelker
                               Glynis Coulter

                                 GRAPHICS
                                 Larry Quay
                              Deborah Shibata
                              Susan Chapman

                          REPORT PRODUCTION
                                 Rita Clark
                              Patricia Stewart
                              Roberta Bullock
                             Alberta Alexander
                                Judith Doocy

                         REPORT REPRODUCTION
                              Alfred Robinson
                               Jack Bencomo
                             Dennis McFadden
                                    xii

-------
                                   SECTION 1
                                INTRODUCTION

BACKGROUND

     Selenium  is a controversial contaminant found in trace quantities in some ground
waters and surface waters in the United States. Selenium is known to be an essential
nutrient  in  minute quantities, but prolonged exposure to higher concentrations has
been known  to bring about respiratory disease and death.  Investigators have reported
that selenium is  a carcinogen and others have claimed that it  has anti-carcinogenic
effects.  ,

     The current Federal  limit for selenium in drinking water is 0.01 mg/1. While
toxicological research continues  looking  into  the health effects of  selenium, it  is
prudent to investigate concurrently the technology for removing selenium from water.
Little  accurate  data on selenium  in drinking water is currently  available.   As
analytical techniques for selenium  become more  accurate,  more systems  may be
affected.

SELENIUM

     Unfortunately, data on the  presence of selenium in drinking water supplies are
limited.  The data that are available suggests that  one would  rarely find  surface
waters containing appreciable (?-0.01 mg/1) amounts of the element . Its presence in
higher  concentrations (> 0.050 mg/1) appears to be limited to ground waters.  Up to
                                                       2
0.^8 mg/1 has been reportedly found in a  well in Nebraska . Other known wells with
relatively high  amounts  of selenium have been  found  in  Southern and  Central
California, Colorado, South Dakota, and Wyoming.

-------
     Inorganic selenium occurs in valence states of +^ and +6 in aqueous solutions. It
is  assumed that  their  presence in  natural  ground  water  is due to  seleniferous
formations within the aquifer. Selenium (IV) occurs as the anion HSeO^"  in the range
of pH's from 2.7 to 8.5.  Selenium (VI) occurs as  the anion SeO^= above pH 1.7.  The
speciation  of  each oxidation  state is given in more detail in the concentration -pH
diagrams in Figure 1.

     Selenate (Se(VI)) in the form of  selenic acid (r^SeO J is comparable in strength
to sulfuric acid.   In solubility, most  salts of selenic acid are similar  to  the sulfates
of the  same metals.  Selenite  (Se(lV)) in the form  of  selenious acid (H2Se03) is a weak
acid.  Most selenite salts are less soluble than the corresponding selenates.

     The behavior of selenium in various environments may be best  determined by an
examination of  the  reduction-oxidation (redox) potentials for its various oxidation
states  as a function of  pH.  The  E. - pH diagram  for selenium, shown in Figure 2, can
be explained as follows: Each line on  the diagram represents equilibrium between  the
oxidized form written  above  the line  and the reduced form written  below it.  The
space between two lines is the stability field of the ion or molecule shown on the upper
side of the lower line  and the  lower side of the  upper  line.     The dashed lines
represent the  stability  limits of  HUO.   The shaded  area shows the normal range  of
waters from pH 6.0 to 8.5, which  is the region of interest in this discussion.

     The upper portion of the shaded area represents a  highly oxidized situation. Well
aerated surface waters having a  high oxidation potential would fall in this region.  So
alkaline surface waters with selenium  present should show a  great majority of  Se(VI).
Any Se° could be oxidized to HSeO," in  alkaline or mildly acidic  conditions  or the
HSeCL", in turn, could oxidize to SeO.=.

     Selenite  could be  rapidly reduced to elemental  selenium by mild reducing  agents
such as SO-,.  Selenate has been found to be more stable and  there is less probability
that selenate  could be  reduced  to selenite under mild reducing conditions, but the
diagram shows that selenate is also unstable in a  reducing  environment over the long
term.

-------
          13

          O
          UJ
          0.
          (/)
          o


          O
          O
          Q£
          U
          0.
           a
           o
           u
           0.
           (A
           O

           O


           5
           O
           £E
           Ul
           0.
               100
SE(IV)
                                              9  10   II   12
                                    PH
 SE(VI)
                                            8   9   10  It  12
                                     PH
Figure 1.   Concentration  -pH diagrams for inorganic selenium species.

-------
     1.6

     1.2

     0.8

     0.4
rf
LjJ
   -0.4-
   -0,8-
   - i.2 •
   - 1.6 -
HS«04~
                                            NORMAL EH-pH RANGE IN WATER
                  H2Si
                                                        HSe"
                                         SOLID/SOLUTE INTERFACE = 10 7 M (SOLUTE)
                                                                              02
                                                                                            , OXIDIZING
                                                                                        H20
                                                                                            . REDUCING
                                          6           8
                                               pH
                                                      10
12
        CONSTANTS PER TABLE I-I
                            Figure  2.   Eh - pH  diagram for  selenium.

-------
                                 TABLE 1

                     CONSTANTS USED TO CONSTRUCT
                             Eh-pH DIAGRAM
Half -Reaction
SeO^= + 4H+ + 2e~ = H2SeO3 + H2O
SeO^= + H2O + 2e" = SeO3~ * 2OH"
H2SeO3 + 4H+ + 4e" = Se + 3 H2O
SeO3= + 3 H2O + 4e~ = Se + 6OH"
Se + 2H+ + 2e~ = H2 Se (aq)
Se + 2e~ = Se"
Acid-Base Reaction
H2 SeO3 = H+ + HSeO3"
H SeO3" = H+ + SeO3=
H2SeO^ = H+ + H SeO^~
H SeO^~ = H+ + SeO^=
H2Se(aq) = H+ + HSe"
HSe~ = H+ + Se=
Potential Source of
(Volts) of Constant
1.15
0.05
0.74
-0.37
-0.40
-0.92
pK
2.75
8.50
-3
1.66
3.89
15.0
1. Latimer, W., "Oxidation Potentials" 2nd Ed., Prentice-Hall, N.Y., 1952
2. Baes, C.F. and Mesmer, R.E.,
The Hydrolysis of Cations, Wiley-Ii
1
1
1
5
1
1
Source of
Constant
2
2
4
2
2
2,3
•
rcterscience
    N.Y., 1976.

3.   Sillen,  L.G.  and Martell, A.E., Stability  Constants of Metal-Ion Complexes,
    Met calf and Cooper, Ltd., London, 1964.

4.   Langes' Handbook of Chemistry. 12th Ed., Editor-3. Dean, McGraw Hill CO., N.Y.
    1979.

-------
     In  waters  with low  levels  of  oxygen,  the  oxidation  potential is reduced.
Depending on the actual  oxidation potential  of  the water,  a combination of Se(IV),
Se(VI), and Se°  could  be  present.   A  ground  water  would  be  expected to  offer
conditions where  selenium could be present as  a mixture of Se(IV) and Se(VI).

     There is ambiguity concerning the redox potential between Se° and H-Se. It may
be possible in a  very reduced condition  that selenium  as H5e~ could  be  formed  in
natural waters.   However,  the occurrence of free selenide  (H-Se,  HSe~, and  Se~)  in
ground waters is  unlikely due to its rapid  formation  of precipitates of iron.  Evidence
has been  presented to suggest that  when selenites  react with ferric chloride, a very
insoluble  precipitate that  approximates  the composition of  basic ferric selenite,
               is also formed.   This will  also  happen with  aluminum to a lesser
extent.

     Based on the  information presented, the presence of either Se(VI) or Se(lV) in a
water depends on many factors concerning the chemistry of the particular water of
interest.  From the E.-pH diagram, a poorly aerated ground  water could have a varied
combination of  Se(lV) and  Se(VI).   Because  of  Se(IV)'s lowered solubility, any  iron
present in a  water would  tend  to  form  insoluble  precipitates with it.   Elemental
selenium  presents no problem in aqueous considerations.   Therefore, Se(VI) would
appear to be the most common species present in a generalized groundwater condition.

     Organic  selenium occurs in natural aqueous solutions as (CrO Se, (CHj Se_, and
other forms by the means of  microbiological assimilation and degradation.  Although
measurements are not available, organic selenium compounds are not thought  to be a
large component of total  selenium  in  groundwater.  Most of  the knowledge about
organic selenium has been  derived from work involving synthetic organic selenium
compounds. From this  work,  it is  known  that essentially all  of  the  synthesized
compounds contain  selenium in the -2 oxidation state  and that these compounds tend
to oxidize and form elemental selenium.

-------
PRIOR RESEARCH

     In  an  effort  to  provide  information  explaining  how  the  drinking  water
requirement for  selenium could be most effectively met, the Water Supply Research
Division of EPA  conducted a research program consisting of jar test studies and bench
tests.  Ferric sulfate coagulation, alum coagulation, and lime  softening were studied
for Se(IV) removal. Ferric sulfate removed 85% of Se(lV) at a pH of  5.5 or less with an
initial Se(IV) concentration of 0.03  mg/1.   Lime softening  removed 45% with a well
water and Se(lV)  concentration of  0.03 mg/1.  Alum coagulation  was unsuccessful,
                          7 5
removing  only 20% at  best. '    Coagulation  by ferric sulfate,  ferrous sulfate, alum,
and  lime  softening were  unsuccessful for Se(VI),  10%  removal- being  the best.
However, ion exchange and reverse osmosis were found to  both remove greater than
95% of either Se(IV) or  Se(VI) from drinking water.2'5

     Because coagulation and lime  softening were not always  effective or  applicable
to  small  water  systems,  and  because   of  the  great  costs   associated  with
demineralization by ion  exchange or  reverse osmosis  processes, the EPA initiated
investigations of the use of  activated  alumina for removal of selenium.  Activated
alumina has been successfully used in studies to remove phosphate, fluoride,  silica, and
                                                            6789101112
arsenic  from drinking  water  and phosphates from wastewaters. ' '  ' '  '  '    It is
currently  being used in two full-scale fluoride removal facilities.   Alumina  is about
one-tenth the cost of most ion  exchange resins and researchers have reported that it
doesn't remove sulfate, chlorides, and other anions that would compete with selenium
                                    2
removal by anion exchange treatment.

ACTIVATED ALUMINA

     Activated  alumina  is produced  by thermal treatment of  hydrated alumina, the
alumina being extracted  from bauxite ore.   It has a  great affinity for water and is
generally  used as a dessicant  for  drying gases  and liquids.   Activated alumina is
produced  mainly by the Aluminum  Company of America (ALCOA)  and the Reynolds
Aluminum Company.   The  activated alumina produced by  ALCOA is available in
several grades:  F-l, F-5, and F-6.   Type F-6 is a chromatographic  alumina and costs
about eight times as much as type F-l.  Type F-5 is considered a specialty product and

-------
is also more costly than F-l.  Type  F-l  is the general grade of  activated alumina
available and is practically 100 percent alpha alumina.  It is the most inert of all the
alumina  oxides.  F-l  is the grade used in these studies and some of the technical
aspects are shown below:

    Chemical constituents:    A12 0,                     92.2%
                             Na20                      0.9%
                             Fe2O3                     0.08%
                             Loss on ignition            6.5%

    Physical Properties:       Surface Area           =    250 m2/g
                             Loose Density          =    52 lb/ft3
                             Packed  Density         =    55 lb/ft3
                             Specific Gravity        =    3.3
                             Graded  Mesh Ranges    =    8-10,14-28,28-48,48-100

      Adsorbents are defined as natural or synthetic materials  of micro-cystalline
structure,  whose internal pore  surfaces are accessible  for  selective combination of
solid and solute.  Usually the attractive forces are weaker and less specific than those
of chemical bonds.  Its selective action is most pronounced  in a monomolecular layer
next to the solid surface, but at times selectivity may persist to a  height of three or
               14
four molecules.    Adsorption capacity of
the fluid-phase concentration of the solute.
               14
four molecules.    Adsorption capacity of a solid for a solute tends to increase with
      As discussed by  Kubli,  adsorption  of inorganic salts  on alumina  is  due to
hydrolytic adsorption associated with aluminum and hydrogen ion exchange.  If alumina
is  treated  with  an aqueous acid solution, the A^O.,  will be charged  to a hydroxyl-
bearing cation capable of binding the anions of various salts as water-insoluble salts.
If, for example, the acid is HC1, hydrogen ions from the acid react with some of the
attached hydroxide ions of the solid alumina to yield  water molecules, which remain
attached to the alumina.  The network of aluminum and water molecules then  acquires
a positive charge.  The anions of the acid become included in the solid as counter-ions,
and electroneutrality occurs.   These counter-ions,  since they are not built  into the
network, will readily exchange and  these ions are expected  to be exchanged when the
                                        8

-------
solid comes in contact with a less soluble  anion.  If  Se(IV) is in solution as HSe03~,
HSeO-~will readily replace the chloride ion (CD provided that  the solubility of the
chloride  complex is  greater than  the selenite  complex.   The mechanism  can be
explained by the following diagrams:

         = Al - O x.                        = Al - O ^
                   Al-OH + H+ + Cl" —»•            ^ Al- Cl + H,0
         = A1-0X                        = Al-0 ^          *
         = Al-0                          = Al-0
provided that the solubility of
         = Al-Ov
is lower than that of
         = Al-0.
                    AI*CI
      Since activated alumina is an amphoteric substance, it will adsorb cations at pH's
above its isoelectric point and anions in more acidic environments.  This phenomona
can  be explained by the net  surface charge on the alumina.  The isoelectric point is
defined as the particular  pH where the net surface charge is zero.  For type F-l
alumina, this point is at pH 9.2.

      A phenomena called  secondary adsorption is known to exist with alumina.  This
occurs  when cations bond to anions and then other anions adjacently link  onto the
cations and vice versa.  This  effects a chain of  ions.  Secondary adsorption occurs as
(1) joint  adsorption of anions with  multivalent  cations and  (2) joint  adsorption  of
cations with multivalent anions.   There appears to be no adsorption with univalent
                           2
anions and univalent cations  .  This may explain why the hardness ions,  calcium and
magnesium, seem to be partially removed during a removal run at pH's  in the range
of 6-8.

-------
     Kubli developed  an  anion  selectivity series for activated  alumina based on his
work with a few anions of  interest.  They are, in order of decreasing preference:
     OH', po^',  F-,  so^-, cr, No3~,  cio4'

It is presumed  that Kubli did his work at  a normal operating pH in the range of 5
to 8 and thus the anion being removed instead of  PCX   would be H^PO^". Part of
the research discussed in this report involves the development of a  broadened series
which includes other  trace inorganics of interest, namely Se(IV), Se(VI),  As(lII), As(V)
and another anion found in large amounts in every water supply, HCO7~
         Kubli suggested that elution of the alumina after adsorption of the species of
interest had taken place could be done by any of three mechanisms:


         (1)   Alkali (OH") elution, where the heavily favored hydroxide ion displaces
         all other species in or on the sites,


         = Al-O^                       = Al-O
                   Al»HSeO3 + OH"	          ^ Al- OH + HSeO "


         (2)   by contact with other anions which form less soluble  basic aluminum
         salts,


         (3)   by contact with higher concentrations of the original anion bound to the
         alumina (in this case CD.
         = A1-O                         = A1-O^
                ^Al'HSeO, +Cf	          ^  Al«Cl + HSeO"
         = Al-O/         J             = Al-O ^               *

         The most  feasible option  would seem to  be elution  with  a concentrated
                                   Q
amount of sodium  hydroxide.   Ames  tried NaOH and  Na2CO3 as regenerants and
NaOH was found to be more efficient. Since optimum removals of anions should occur
at a pH  less than 9.2, the now  basic alumina must be eluted with an acid to lower

the pH. Preferably, this acid's anion will not effectively complete with the species to
                                        10

-------
be adsorbed for sites on or in the alumina. Based on Kublis's results, perchloric, nitric,
hydrochloric, and  sulfuric  acids  would be  the preferred  neutralizers, in decreasing
order of preference.  Unfortunately, the relative costs of these acids are  in  the
reverse order, perchloric being most costly and sulfuric being the cheapest.

         Various  investigators  have  completed  studies  to  determine  activated
alumina's ability to remove the anions fluoride, silica, and arsenic. '''    In batch
                    19
tests, Choi and Chen   determined that to  achieve a final fluoride level of less than
1 mg/1 requires an initial fluoride concentration lower than 40 mg/1 with an adsorbent
dosage of 25 g/1.  They  found that the presence of other chemical species does not
seriously interfere with  fluoride  removal by activated alumina.  pH  was found to be
the most critical factor in  determining the fluoride removal efficiency.  The optimum
pH was around 6.

        ,Gupta and  Chen  investigated arsenic removal by activated alumina in the
batch mode.  As before, pH played a major role in determining the capacity for arsenic
removal. Good removals were achieved in the pH range of 4 to 7.  However, arsenic is
present in  two valence states in water, +3 and +5.  At  pH 6.5,  As(V)  is present as
H2AsO ~, while at pH 9, As(III) is present as H-AsO.,.  Arsenic (V) removals were on
the order of 10-20  times greater  than  were arsenic (III) removals  under  the same
testing conditions.   Another difference from fluoride  is that the presence of other
chemical species in  the matrix  reduced by as much as 80  percent the amount of
arsenic removed.  As(V)  adsorption was affected  much more by chemical composition
than was As(III). Oxidation of As(IH) to As(V) is necessary to achieve effective arsenic
removal and chlorine has been successfully used to accomplish this.

                            12
         Clifford and  Matson    determined that the optimum  pH  range  for silica
removal by activated alumina is 8.0-8-.5. They believe that the ion Si(OH),0~ is being
removed and that this removal is  optimal  near the pK for ionization of Si(OHk, which
is 9.5.  Varied levels of sulfate were found to have  little effect on the capacity of
activated alumina  for silica. However, the presence of fluoride in the water greatly
affected the capacity for silica.
                                        11

-------
     There has been some work using continuous flow activated alumina columns as a
means  of determining the removal capacity of the element of  interest.  Rubel   has
reported the  optimum operating conditions for  a full-scale fluoride removal facility.
He used 5 feet of alumina in a six-inch  diameter column.  After a treatment run was
finished, the following regeneration procedure was as follows:
                                                                   2
     Backwash:            2.5 bed volumes of treated water @ 9 gpm/ft , upflow
                                                                2
     Regeneration:        2.5 bed volumes of 1% NaOH @ 2Yi gpm/ft , downflow
                                                                 2
     Intermediate Rinse:   4 bed volumes of treated water @ 5 gpm/ft , upflow
     Neutralization:       3 bed volumes of 0.5% H25(X @ 2Yz gpm/ft , downflow

     Neutralization is done by continually adjusting the pH of the raw  water with
HLS(X until the pH of the treated water reaches that of the raw water (pH - 5.5). The
above mentioned step is  the equivalent  and is used to determine the actual amount of
       used to neutralize the bed.
     Treatment:  Usable water is produced after the pH of the treated water reaches
     9.0.  Here, significant fluoride removal will start to occur.

     With this  treatment, he is able to produce 1400 bed volumes of treated water
with an average F" concentration of 0.8 mg/1, from an initial concentration of F" of
5.0 mg/1.
                    Q
     In 1970, Bellack  did some testing in the laboratory to see if activated alumina
could remove arsenic from water supplies.  His brief  work suggested the following
procedure:

     Backwash:           15 bed volumes of tap water @ 9 gpm/ft , upflow
                                                             2
     Regeneration:        4 bed volumes  of 1% NaOH @ 1 gpm/ft , downflow
                                                                 2
     Intermediate Rinse:   8 bed volumes  of distilled water @ 1 gpm/ft , downflow
                                                                2
     Neutralization:       4 bed volumes  of 0.1 N H2S(X @ 1 gpm/ft , downflow
     Final Rinse:           4 bed volumes  of distilled water Q 1 gpm/ft , downflow
     Treatment:           Raw water at 2 gpm/ft  , downflow
                                       12

-------
     The final rinse is necessary to remove the excess 0.1  N H-SO^, from the bed.
This was not needed in Rubel's facility  because the raw water's pH was adjusted with
H-SCXto perform the  neutralization step and the pH was stepwise increased as the pH
of the treated  water dropped. Bellack stated that with an initial concentration of 0.10
mg/1  total  arsenic,  approximately 900 bed  volumes  of  water  with  an  arsenic
concentration  less than 0.01  mg/1 can be produced.   However, he failed to  mention
whether  arsenic in the +3 or  +5 valence  state was present.  He probably did not note
the difference in the two species and made no effort to determine the speciation.
          o
     Ames  used fairly large (5.1 cm D x 49 cm H) columns to determine the removal
of phosphates from wastewater  using activated alumina.  He was able to remove
approximately 90% of  all phosphates applied for 400 bed volumes.  He concluded that
varied  amounts of sulfate  present in wastewater  had  little effect  on phosphate
removal.  pH was not selectively controlled in his experiments, but operating above pH
8.0 led to precipitation and calcium carbonate fouling of the alumina.

He prescribed  the following regeneration:

      8 bed volumes of 1 M NaOH @ 3 gpm/f t2
                                           2
     20 bed volumes of washwater @ 3 gpm/f t

     No acid rinse was used after a caustic elution due to his apparent lack of concern
about the operating pH of the treatment cycle.  It was also noted that from 1% to 5%
of the  column bed was lost per elution, making it necessary to replace the activated
alumina often.

PROJECT APPROACH

     The approach of  this project to determine  the feasibility of  removing selenium
from drinking  water using activated alumina was developed so that each progressive
phase contributed to the following  phase. The project was divided  into three phases:
I.  Analytical  Techniques, II.  Batch Isotherm  Tests,  and III. Bench  Scale  Column
Studies.   Straightforward techniques for the analysis of Se(IV), Se(VI), and  organic
selenium  in  water and an  important prerequisite  as they helped  facilitate  full

                                       13

-------
concentration on the selenium removal process itself.  Adsorption isotherms were used
to determine the rough parameters for column studies.  The column tests helped to
determine  optimum regeneration techniques,  selenium  breakthrough capacities, and
interferences by other  ions.

      The column studies should  provide relevant information to efficiently test this
process at  the pilot scale.  This should be the next phase in  studying the activated
alumina  process for  seleniumn removal.   A pilot  plant study  can produce  the
engineering data required to determine the feasibility of using activated alumina to
remove selium from drinking water.

      It should  be noted that  the columns studies contain quite  a bit of  information.
Initial testing done with three-inch columns of alumina may not provide the reader
with a good understanding of the actual capacity information that was developed with
the nine-inch columns.  However, the short column runs did provide us  with  enough
information  to  narrow  down  operating  parameters  with regards  to regenerative
processes.

      To develop  a  clear  understanding  of   the  sequential  development  of  the
experimental design,  it  is  suggested  that the "Summary  of Results",  Table 6,  be
reviewed prior  to reading the discussion of the results  of each  successive run.  The
results of all the tests  performed during this work are  presented in this report.  For a
rapid  review of  the data, the  reader may wish to skim through the section discussing
the three-inch column work.
                                       14

-------
                                  SECTION 2
                                CONCLUSIONS

ANALYSIS

     The  Atomic Absorption Spectrophotometer (AAS) can be used to determine the
total selenium present in a sample.  The method is  quick, sensitive, and produces
reliable results.  Selenium (IV) can be differentiated from other forms of selenium by
using a fluorometric technique. This method requires a very small sample,  10 ml, but
its limitation is the amount of time required to complete the analysis.  Depending on
the amount of glassware available to the analyst, 10  samples can be analyzed in
approximately 6 hours.  Although its reliability  has not been proven over time, the
analysis was found to be reproducible in our experiments.

     Methods evaluated for the detection of organic selenium  compounds proved to be
complex and subject  to interferences.   Although the  literature  shows  that  these
techniques can be feasible in a research environment, they were determined unsuitable
for routine monitoring at concentrations less than 100 pg/1.  Samples to be analyzed
for different species  of  selenium were determined as:  (1) total selenium (by AAS),
(2) selenium (IV) (by f luorometry),  and (3) selenium (VI) and  organic  selenium (by
difference).

BATCH STUDIES

     As part of the overall bench scale testing program, batch tests were necessary to
determine the effect of various parameters on the adsorption of selenium by activated
alumina.
                                       15

-------
Preparation (Regeneration)

     Fresh activated alumina must be initially prepared for selenium adsorption, the
preparation being similar  to proposed regeneration techniques to be used in column
studies. Optimum finding, were:

       •    1% NaOH rinse for 50 minutes
       •    5-minute deionzed water rinse
       •    0.05 N HC1 rinse for 10 minutes (to be used when removing Se(Vl)) or
       •    0.05 N H-SO. rinse for 10 minutes (to be used when removing Se(IV))
       •    2 to 4 deionized water rinses for 5 minutes each

     HLSO. worked only slightly  better than HC1 in preparing the alumina for  Se(IV)
removal.  HC1  worked much better than H2S
-------
Mesh Sizes

     Three different mesh sizes were tested as to their capability of quickly adsorbing
selenium once in contact with activated alumina.   The general rule of thumb is the
smaller the mesh size, the more rapid the adsorption. This is true for both Se(IV) and
Se(VI).

Interfering Ions

     Of the anions  tested, bicarbonate had the most pronounced effect  on  selenium
(IV) adsorption.  It reduced removals by approximately 10 percent.   Chloride, nitrate,
and  sulfate only marginally interfered  with selenium (IV) removal.  Both  sulfate and
bicarbonate heavily interfered with selenium (VI) adsorption at concentrations greater
than 100 mg/1.  Greater than 60  percent reduction  in the  adsorptive  capacity of
activated alumina for Se(VI) was noticed with these two anions. Chloride  and nitrate
had no pronounced effect.

     Three  cations;  sodium,  magnesium,   and  calcium   were  evaluated  for
interferences.  None of  these ions in concentrations as  high as 200 mg/1 negatively
effected adsorption  of either selenium (IV) or (VI).  Some enhancement of adsorption
was  noticed at high concentrations.  This is thought to be due to secondary adsorption
effects.

Selectivity Series

     In equi-molar concentrations, a part of Kubli's preferred anion list for activated
alumina was verified, with additional anions of arsenic and selenium tested.  Results
were (at pH 6.5):
     OH" > HP0" >F~ > HAsO~ >HSeO"
      Because of poor detection limits, the following species that are less preferred
than the above  five anions could not be relatively listed.  Based on column studies,
they are listed in assumed order:
                   >HCO3" >C1~ >NO
                                        17

-------
Capacities of Activated Alumina

     In side by side tests with Se(IV) and Se(VI), activated alumina exhibited a greater
capacity for Se(IV)  in matrices of  deionized  water and  a synthetic well water.
Activated alumina was found to have approximiately 3 times the capacity for Se(IV)
than for Se(VI), based on a Freundlich isotherm model.

COLUMN STUDIES

Selenium (IV)

     Using a  synthetic  well water  modeled  after a known ground water with high
selenium concentrations and a 9-inch deep bed of activated alumina, the  following
capacities  for
meters/hour):
capacities  for  Se(lV) were  developed at a surface loading rate of 3 gpm/ft   (7.3
         pH5-    1,200 bed volumes   , 235 „„, Qf a™?vaffalumina
         pH 6 -    900 bed volumes     = 175 mg/1
         pH 7 -    500 bed volumes     = 100 mg/1

     The above breakthrough capacities (the amount of Se(IV) alumina  will adsorb
before the effluent concentration becomes greater  than 0.01 mg/1) are based on an
influent concentration of 200 ppb (=0.20 rng/1), with regeneration by 0.5% NaOH at a
dose of 1.5 //NaOH/ft3 bed (24 g/1) at a flow rate of  0.5 gpm/ft2 (1.2 meters/hour).  A
slower regeneration  rate  may increase the above capacities.  The effect of actual
flow-through contact time between  the NaOH and  the activated  alumina  was the
                                                                         2
greatest of any  parameters tested with regards to capacity.  At 0.5 gpm/ft NaOH
flow rate, roughly 90% of the Se(IV) removed during the  treatment run was recovered
in the regeneration.

     For lower influent concentrations of Se(IV), the actual capacity is assumed to be
a linear function of  the influent  concentration.   Therefore, an influent Se(IV)
concentration of  50 ppb  should  produce  a capacity of  roughly  60 mg/1  at pH  5.
                                       18

-------
Similarly, the capacities at pH 6 and pH 7 showed 45 mg/1 and 25 mg/1, respectively.
However, to reach an effluent concentration of less than 10 ppb, similar bed volumes
of treated water will be produced regardless of the  influent concentration  up to
200 ppb.

     Neutralization of the activated alumina after NaOH rinsing  can be successfully
accomplished with either F^SO^ or HC1.  The  following  regeneration scheme  was
determined for this study:

     5 bed volumes 0.5% NaOH @ 0.5 gpm/ft2, up
                                       2
     5 bed volumes D.I. water (d 1 gpm/ft , up
     6 bed volumes 0.05 N H2SO^ or 0.05 N HC1 @ 1 gpm/ft2, up
     5 bed volumes D.I. water (d 1 gpm/ft , up

     All the regeneration steps were done in the upflow mode to facilitate rapid
testing.   Upflow regeneration  by  NaOH  is recommended, but the remaining steps
should be done downflow in a full-scale removal facility.

     The NaOH and H«SO^ (or HC1) tend to dissolve the activated alumina and  how
fast it degrades is  dependent on acid  or base concentration and flow rate.  Higher
concentrations will  dissolve slightly more alumina and slower flow rates will dissolve
more alumina than more rapid regeneration. The above-mentioned regeneration steps
dissolved, on the average, the following percentages of a nine-inch bed (by weight):

     NaOH - 0.9% per regeneration
            (or  HCL) - 0.08% per regeneration
      Annual media replacement will depend on how often regeneration is done.  pH
adjustment for removal of Se(IV) can be done with either HLSO. or HC1. HC1 was used
for these tests.
                                        19

-------
Selenium (VI)

      Using a similar  synthesized  water as with Se(IV)  the  following capacities for
Se(VI) with a nine-inch bed of activated alumina were developed at a surface loading
rate of 3 gpm/f t  :

      pH 5  -    100   bed volumes =  4.5 mg Se(VI) liter of activated alumina
      pH 6  -    70   bed volumes =3.2 mg/1
      pH 7  -    35   bed volumes =  1.6 mg/1

      The above breakthrough capacities  are based on an influent concentration of 50
ppb Se(VI) with regeneration by 0.5% NaOH at a dose of 1.5 //NaOH/ft3 bed at a flow
                 2
rate  of 2  gpm/ft .   This capacity is  one-twelfth  the  capacity  for  Se(IV) based on
equivalent influent concentrations of Se(IV) and Se(VI).

      Rate of regeneration was not as critical as with Se(IV) and a smaller NaOH dose
could be used to achieve similar  breakthrough capacities.  This is due  to  Se(YI)'s
relatively low position in activated alumina's selectivity  series.  Regeneration is much
less dependent on diffusion-controlled kinetics.

      The other regeneration steps  were the same as with Se(IV), except only  HC1 can
be used to neutralize the bed. Sulfate from  an H2SO^ rinse heavily interferes with
alumina's capacity for Se(VI).

      Due to the noticeable difference in  acid rinses, tests were done to determine the
effect of  varied concentrations  of  two interfering   anions,  SCX"  and  HCO^".
Decreased amounts of  sulfate  in the water greatly increased the capacity of activated
alumina for Se(VI). The following list  shows these capacities at pH 6:

              SO.  (ppm)           Se(VI) capacity (mg/1)          Bed  Volumes
                   500                        0.7                        15
                   100                        3.2              .          70
                    50                        7.0                       150
                     5                       21.0                       450
                                        20

-------
     Similar tests  with bicarbonate  alkalinity interference did not show as great a
difference at pH 6,  as listed below:

                Alkalinity
             (ppm as CaCOJ         Se(VI) capacity (mg/1)          Bed Volumes
                   500                        1.5                        33
                   100                        3.2                        70
                    50                        4.0                        90
                      5                        5.5                       125

      The above capacities are for an influent Se(VI) concentration of 50 ppb.  Sulfate
tests were done with approximately 100 ppm alkalinity, while the alkalinity tests were
done with a sulfate concentration of about 100 ppm.

      pH adjustment must be done with HC1.   H-SO. addition would increase the
sulfate level and interfere with Se(VI) removal.

COSTS
      Preliminary   cost  estimates  were  developed  based  on  the  capacities  and
regeneration techniques developed during this study.  Annual costs were developed for
amortized  capital costs   over  20 years,   which   included   equipment,  piping,
instrumentation, a small building,  clear well, activated alumina,  and land. Annual
operation and  maintenance costs included  chemicals, labor,  media replacement, and
electricity.

      These costs were  developed to treat a water  supply  of 1 MGD  with similar
quality to that tested  in this study contaminated solely with either a Se(IV) or Se(VI)
concentration of 0.10 mg/1 (100 ppb). Costs are:

      Se(IV)   - 23
-------
      Based on an evaluation of the equilibrium between the various oxidation states of
selenium in water, Se(VI) is expected to be predominately in well-aerated waters.  In
waters with lower oxidation potentials, Se(IV) and Se(VI) would be  present  in varied
amounts.
                                        22

-------
                                  SECTION 3
                             RECOMMENDATIONS

     Because the  speciation of inorganic selenium  in water determines the cost of
removal, a survey  of known  selenium-contaminated ground water sources should be
analyzed to determine the relative amounts of Se(IV) and Se(VI).  Since few actual
determinations of the speciation in ground water have been done, this would be helpful
in assessing conditions.

     The inability  to determine the  presence of organic selenium compounds in a
water sample in amounts less  than 100 ppb makes this  problem of great concern.
Efforts should be made to modify the techniques described in Section 4 or to develop a
new  method  for detecting organoselenium  compounds  in  the microgram-per-liter
range,  since  its presence may hinder the  removal  process by activated alumina.
Perhaps a GC/MS sparge and trap method could be used.

     NaOH regeneration of Se(IV)-saturated,  activated alumina should be optimized to
establish actual flow-through contact time and dose of NaOH required.   Work done in
this  study  indicates that  a longer contact time  may increase  the amount of  Se(IV)
displaced during regeneration.  It is possible  that  smaller amounts of NaOH could be
used than  were  used on  a day-to-day  basis  during these studies.  Because Se(VI) is
removed much less than Se(IV) in removal runs, the regeneration did not play as large a
role  in determining its capacity  on activated alumina.

     Based on  Kubli's assumptions,  chloride  (CO  might be  able to  regenerate
selenium-saturated  activated alumina if present  in high enough concentrations.  We
recommend that some testing be done to evaluate this claim, since the cost of NaCl is
much cheaper than  that of NaOH and/or HC1. Sulfate, in the form of Na2SO4,  might
be a better  regenerant  for  Se(IV) than chloride,  but  its interference with  Se(VI)
adsorption  makes it unfeasible.
                                       23

-------
     Pilot-scale  testing  is necessary  to successfully  model  a full-scale selenium
removal facility.  Since a number of factors (SO^2", HCO3", pH) affect the removal of
Se(IV) and Se(VI), the general water quality of the source should be evaluated. Sparse
data gathered from Ramona  Municipal Water  District in a series  of  well samples
showed that selenium concentrations varied quite a bit.  Continuous pumping should be
done to verify the steady-state concentrations  of  Se(IV) and Se(VI)  before designing
pilot-scale tests to evaluate activated alumina's feasibility.

     Actual  pilot testing should be done to refine the techniques  developed in this
study.   Removal  capacities,  regeneration  techniques and  hydraulic characteristics
should all be  carefully studied to allow for an economic design of a full-scale removal
facility, should it be deemed feasible.  Deeper columns may produce more bed volumes
of treated water than is shown in the 9-inch column used in this study.  Determination
of the optimum method for neutralizing the  alumina with acid after the caustic rinse
should  be looked  at closely.   Since  these  tests showed that activated alumina  is
susceptible to large amounts of dissolution during the regeneration, this should be an
item of extreme  importance to evaluate  in larger scale, multi-cycle tests.  Headless
characteristics of various media sizes should be evaluated in order to make decisions
regarding the use of higher capacity, smaller  mesh sizes of alumina.

     A major item that should be investigated is the disposal of regenerant wastes. In
arid or semi-arid  regions,  evaporation ponds may be feasible.  Otherwise, disposal may
involve a separate, on-site treatment or hauling away by contract carriers.  The costs
associated with the disposal may be high.
                                        24

-------
                                  SECTION 1
                         SELENIUM ANALYSIS STUDIES
ORGANIC SELENIUM
     Gas chromatrography research was performed on three organoselenium indicator
compounds chosen  on the  basis  of  their  probability of  occurrence  in  the soil
environment and their commercial availability. Selenium substituted amino acids are
likely to occur in a  selenium contaminated soil but are likely  to be quickly degraded.
The   three  compounds   chosen  were  dimethylselenide,   diethylselenide,  and
dimethyldiselenide.   Aqueous standards of these pure compounds were made up and
they were  analyzed by the dynamic headspace technique  using nitrogen  to strip the
materials from the  water.  The method used was similar to that described by Bellar
for organohalides in water with  the exception that the, analytical column  was five
percent polyphenyl  ether  (five rings)  on 60/80 mesh chromosorb W-HM  and an FID
detector was utilized.  Testing showed  that  these organoselenium compounds were
effectively stripped from the water and adequately captured by the adsorbent trap at
relatively high concentrations.  Interferring peaks which  eluted at similar retention
times  to  the organoselenium  compounds  prevented  the realization of minimum
detection  limits  less than about 100 ug/1.  Extraction of  these compounds by liquid-
liquid extraction with n-pentane  and  analysis by electron  capture detection showed
that only the dimethyldiselenide compound was detectable but the minimum detection
limit for this compound was 10-100 times better than with  the dynamic headspace FID
analysis.

      Because  of  the   poor   minimum  detection  limits  for  these   indicator
organoselenium  compounds, it appears that  the technique  will not  be useful for
drinking water analysis until further optimization of the technique can be performed.
                                        25

-------
ATOMIC ABSORPTION SPECTROPHOTOMETRY

     The AAS technique  remains as the most efficient  and sensitive  means  for
determination of  total selenium.  In the majority of the batch scale tests done, either
selenium (IV) or selenium (VI) was tested by itself.   Therefore, analysis of selenium
could be done quickly on the AAS with the assumption that the total selenium present
was either Se(IV) or Se(VI). In tests involving combined quantities of Se(IV) and Se(VI),
the AAS determined total  selenium  concentrations and the fluorometric technique,
described later, gave concentrations of Se(IV). Se(VI) was calculated by difference.

     All  total selenium  analyses were  done  by  the flameless atomic  absorption
technique on a Perkin-Elmer 305B AAS with a P.E. 2200 graphite furnace utilizing
deuterium ultraviolet background  correction and an electrodeless discharge selenium
lamp.   Standard conditions for the furnace  were: a drying cycle of 30  seconds  at
105 degrees C, a charring cycle of 10 seconds at 1000 degrees C, and an atomization
cycle of 7 seconds at 2200 degrees C.  Pyrolitically coated  graphite tubes and the
"max.  power" function were used  in analyses. This enabled a much lower atomization
temperature and forced all available current into the furnace tube at a very fast rate,
which is almost equivalent to instantaneous atomization.

     An equal volume of 1000 ppm  Ni(NO,)2  was added to the  furnace  after each
sample  was injected.  This method is an EPA standard method.

     When analyzing samples using the furnace technique, matrix interferences played
an important role  in determining the selenium present.  The analyses of regeneration
samples, which had high concentrations of sodium hydroxide and either sulfuric acid or
hydrochloric acid,  were greatly affected by the presence of  these salts in the samples.
Normally, standards of selenium were made using deionized water and  concentrated
nitric acid (the normal preservative for selenium).  These standards were not usable
with the regeneration samples, due to the marked difference in results when compared
with  standards  made up  from  an approximate  0.05% sodium  hydroxide selenium
standard. Therefore, regeneration samples were diluted about 10:1 to 0.05% NaOH to
negate  any other interferences and then were analyzed using  a comparative selenium
standard made with 0.05% NaOH that was acidified with HNO3 to a pH of less than 2.

                                        26

-------
     To  verify  reproducibility and  accuracy  of these results, known amounts  of
selenium were spiked into regeneration samples.  The analysis of the spiked samples
showed concentrations that were equal to the sum of the known amount spiked and the
unspiked  sample within  10%.   This accuracy is within  the  limits of the furnace
technique.  Table 2 shows these results.

     The correction for matrix interferences was not necessary when concentrations
of selenium exceeded 1.0 ppm.  Above this level, selenium can be detected using the
flame atomic adsorption technique and the flame technique is not  subject to matrix
interferences.

     At  the  onset of  testing,  it appeared as  though a  double  peak phenomena,
presumed to be selenium (IV) and (VI),  occurred  when  using uncoated  tubes with no
"max.  power" input.  Further work  was conducted with more concentrated standards
and the phenomena was  not reproducible. It was suspected that some  contamination
was present in the reagents  that was responsible for  some of the peaks observed.
Although this technique would have been the simplest method for differentiating Se(IV)
and Se(VI), the idea was  abandoned due to the inconsistent results. See Figures 3 and 4
for photographs of the AAS and associated equipment.

                                   TABLE 2
     VERIFICATION OF REPRODUCIBILITY FOR REGENERANT ANALYSIS*

               Initial            Spiked
              Sample         Concentration     Concentration       Difference
  Date          I.D.             (ppm)            (ppm)             (ppm)

  9-14         Col.  1             0.23             0.63             0.40
               Col.  2             0.28             0.70             0.42
  9-26         Col.  1             0.46             0.82             0.36
               Col.  2             0.44             0.83             0.39

         *24 mi's of sample were added to 1 ml of 10 ppm selenium
         standard  to give a spiked concentration of 0.40 ppm. The
         differences noted above are with  10% of this value.
                                      27

-------
-
                             Figure 3.   Injecting a sample into the A.A.S. furnace.

-------
-
:
                                    Figure  4.   A.A.S.  strip chart recorder.

-------
       100
       90
       80
       70
   6
   z
   D
   £   60

   55
   I   50

   U
   o

   5
   o
   £
   O
       40
       30
       20
       10
FLUOROMETER


TURNER* 110-850


PRIMARY FILTER - 369 NM

SECONDARY FILTER -522 NM

SCALE -SOX
                          2345


                           SELENIUM (IV)  (Ppb )
Figure 5.  Influence of selenium  (VI) on  fluorometric determination of

           selenium  (IV), 0-7 ppb.
                                   30

-------
  100
   90
   80
*—\
P
z
   70
55  60
I
z
o
s
o
I
o
2
    50
40
                               4+W/0 SE6+
                          (A)SE4+W/SE
                                 FLUOROMETER
                                 TURNER#IIO -850
                              •  PRIMARY FILTER  -369  NM
                                 SECONDARY FILTER-522 NM
                                 SCALE -3X
              10       20      30       40       50
                        SELENIUM  (IV)  (ppb)
                                                       60
                                                            70
Figure 6.  Influence of selenium  (VI) on fluorometric determination
           of selenium  (IV), 0 -  70 ppb.
                                 31

-------
COLORIMETRIC DETERMINATION OF SELENIUM (IV)

      The Standard Methods diamino  benzidine method for selenium is  selective for
selenium  (IV) if the  permanganate  oxidation  and subsequent  reduction  steps  are
omitted from the procedure.  This technique has shown to produce precise results, but
the presence of  selenium (VI) produces a slight interference.  The accuracy of this test
with selenium (VI) present is somewhat less than that of the fluorometric  technique
(discussed  next).  But, this  technique is useful in  the selective  determination  of
selenite. It was  not used in this study because the analysis requires approximately one
liter of sample and is much more time consuming than the fluorometric technique.

FLUOROMETRIC DETERMINATION OF SELENIUM (IV)

      Selenious acid (HjSeO,) reacts with  2,3 diaminonapthalene in an acid solution to
form  the strongly fluorescent naptha-(2,3-d)-2-selena-l,3-diozole.   Selenates (Se(VI))
are not reduced to selenites (Se(IV))  under the conditions of this analysis, hence this
method is specific only for selenium in the four-valence state.

      This technique, as described by Raihle  , was  tested as a means to differentiate
the two  forms  of  inorganic selenium  in removal  tests performed with  combined
selenium (IV) and (VI) present and to  determine  the amount of Se(IV) present in some
well water samples that were analyzed for selenium.  Plots of fluorescent  intensity
versus concentration of selenium (IV) in the range  of 0 to 70 ppb are linear and are
practically free from selenium (VI) interferences, as  shown in Figures 5 and 6.

      Standards of known concentrations of selenium (IV), as verified  by the AAS, were
analyzed using this technique and the calibration curves (Figures 5  and  6) were  then
drawn.  Similar  standards of Se(IV) were spiked with varied amounts of Se(VI) and the
results are plotted alongside  the curves without  Se(VI) present.  The curves are
identical within experimental error.

      This technique is much more convenient to use than the colorimetric technique.
10 ml  of  sample  is  required,  compared to  1 liter  needed to  analyze samples
colorimetrically. Also, to analyze 10  samples fluorometrically  requires approximately

                                        32

-------
six  hours, which  is about one-quarter of the time  that it takes to do 10 samples
colorimetrically.  All fluorometric analyses were performed with a Turner Model 110
Filter Fiuorometer with a standard lamp (//110-850).  The samples were analyzed at an
excitation frequency of  369 nm  (primary  filter  //110-811) and had a  fluorescence
emission maximum of  522  nm  (secondary filter #110-822).   See Figure 7 for a
photograph of the f luorometer.
                                        33

-------
-
£
                                             Figure 7.  Fluorometer.

-------
                                   SECTION 5
                           SELENIUM BATCH STUDIES
     As described  in  the  introduction, a  series of bench scale adsorption  isotherm
studies  were undertaken to analyze the various parameters  involved  in  the  use of
activated alumina to remove  selenium.  Studies of regeneration techniques, kinetics,
pH effects, mesh size,  selectivity effects of various anions and cations, and capacities
were performed to  develop the conceptual information necessary for  the  efficient
design of activated alumina contactors.

     All  batch scale studies were  performed with  deionized  water spiked  with
selenium (IV) and/or selenium  (VI) as the stock solution.  No other species were added
to the  matrix,  except in the  case of studies involving pH effects, ion interferences,
and relative selectivity.

     All  batch studies,  with the exception of some pH tests,  were performed in
250-ml  ehrlenmeyer flasks with  glass stoppers.   Contact was achieved by mixing with
a Labllne "Junior Orbit Shaker" at 200 rpm.  Control flasks  containing the selenium
stock solution without  activated alumina showed no loss of selenium during the tests.
It was  assumed that  the  glassware utilized did  not absorb  significant amounts of
selenium within the timeframe  of the tests.  See Figure 8 for a photograph  of the
shaking apparatus.

REGENERATION

     Various regeneration techniques  involving  the  use of acids  and bases  were
employed to prepare virgin activated  alumina  for selenium  sorption.   In all  of the
batch studies,  only virgin  activated alumina were used.  No actual regeneration of
selenium-saturated activated alumina was done during this phase of the study.  Rather,
                                        35

-------
•
                                          Figure 8.  Shaking apparatus.

-------
the expected regeneration techniques to be used in column studies were modeled by
preparing fresh  out-of-the-can  activated  alumina with bases and  acids  to develop
optimal adsorption characteristics.

     Figure 9 shows the relationship of selenium removal versus activated alumina
contact time with  a one percent solution  of sodium  hydroxide (1%  NaOH).  Fresh
activated alumina was rinsed with this basic solution  for various amounts of time to
test the effect of NaOH rinse time on capacity for selenium removal.  This technique
has been suggested  by various authors who have used activated alumina to remove the
chemically similar anions, arsenic, and fluoride  '  ' '  '.

     Activated  alumina  exhibits the highest selectivity for hydroxide (OH~) ions. This
caustic rinse should  remove  all other anions  present in  or on active sites on the
alumina and replace them with hydroxide ions.

     As is  seen from the  graph,  selenium removal  efficiencies of  90 percent are
achieved with as little  as five minutes  of  rinsing.  In all the  following batch  scale
tests, a 50 minute rinse with 1 percent NaOH was used with fresh activated alumina to
ensure optimum  removals of selenium.

     Following  the caustic rinse, a 5-minute rinse with 150 ml of deionized water was
found to be necessary as an intermediate step between base and  acid rinses.  Without
this rinse, the exothermic acid-base reaction tended to alter the activated alumina and
bind it in clumps.

     The next step in the regeneration procedure is to rinse the now basic alumina bed
with an acid to lower the pH, essentially to remove the OH" ions from  the sites on the
                                        2
alumina.  As previously  suggested  by  Ball , hydrochloric acid (HC1) and sulfuric acid
(H2SO^) were utilized to compare, their abilities to prepare the activated alumina for
selenium (IV) and (VI) removal.   Figure  10 plots the relationship of selenium ((IV) or
(VI)) removal versus activated alumina contact time with 0.05 N solutions of HC1 or
H2SO^.  This  concentration of  acid  was taken from  Ball's work.   Prior to this acid
rinse,  the fresh activated alumina had  been initially rinsed for 50  minutes with  1
percent NaOH, then with deionized water for 5 minutes.
                                         37

-------
                                                    SE4+
   1  40
      30
      10
                                                         W/H2SO4
EXPERIMENTAL CONDITIONS
ACT. ALUMINA: l.25o, 28-48  MESH
PREPARATION:
      NAOH; I  •/. ,TIME VARIES
      DEIONIZEDH2O: 5 MIN.
      HG-:O.O5N FOR 10 MIN. [Se(VI)]
      H2SO4 :  O.O5 N FOR 10 MIN. [S
      DEIONIZED H2O: 5 MIN.
VOLUME:  ISO ml
INITIAL [SE (IV)]  - 33 ppb
INITIAL OPE (Vljj  - 55 ppb
CONTACT TIME:  10 MIN.
pH: ^ 3.5
                20      40       60      80       100      120
                     CONTACT TIME W/ ACT. ALUMINA (MlN.)
                                        140
Figure 9.  Initial preparation of fresh activated alumina with NaOH.
                                   38

-------
        100
         90
         80
                      SE 4+W/H2SO4
         20
         10
SE  ' W/ M2SU4 >w
                                                   SE  T W/H2S04/
          EXPERIMENTAL CONDITIONS
          ACT. ALUMINA: 1.25 G, 28 - 48 MESH
          PREPARATION:
                NAOH: I % FOR 50 MIN.
                DEIONIZED H20:  5 MIN.
                HC.-: TIME VARIES, 0.05N
                HaSO. : TIME VARIES, 0.05N
                DEIONIZED H20: 5 MIN.
          VOLUME ;   150 ml
          INITIAL  [Se (IV)J  =  -~ ppb
          INITIAL  [Se  (VI)J =  — ppb
          CONTACT TIME : 60 MIN.
          pH ~3.5
                   10       20       30      40       50       60
                    ACID CONTACT TIME W/ACT. ALUMINA (MIN.)
                                                70
Figure 10. Initial preparation of fresh activated alumina with  acid after
           NaOH rinse.
                                     39

-------
      It appears that there are no differences in the removal capabilities of selenium
(IV)  with  activated  alumina rinsed with HC1 or H2SCV   Because  slightly  better
removals  of  Se(IV) were  achieved with H2SO^, this  acid rinse was used  in all the
following  tests involving selenium (IV) removal.

      Figure  10 also shows that the HC1  rinse  works better in preparing  activated
alumina for selenium (VI) removal than the  FUSCX rinse. This is thought to be because
the sulfate ion (SO^  ) from the H-^SO^ actively competes for sites on the activated
alumina with the aqueous form (Se(k ") of selenium (VI).  The chloride ion  (CO does
not compete  like the sulfate with selenium  (VI).  Therefore,  an  HC1 rinse was used in
all the following tests involving selenium (VI) removal.

      On the  basis of these experiments, a 10 minute rinse with either 0.05 N HC1  or
0.05 N HLSCX was used in all the  batch scale tests from this point on.  This contact
time allows for optimum selenium removals.

      Following the acid rinse, the pH of the activated alumina bed was adjusted  to
achieve a consistent pH of 4.0 during  the  removal tests, except  to determine the
relative selectivity of competing ions.  Depending on the amount of activated alumina
used and the type of  acid rinse used, two to  four 5 minute rinses with deionized water
were required to  reach pH = 4.0.   For  most of the tests,  1.25 grams of  activated
alumina were used and, with either H-SO^ or HC1 rinses, three five-minute rinses were
required.   Five  or more deionized water rinses  did not produce any appreciable  pH
change above pH  =  4.0.   The greatest pH change occurred within the first three
deionized  water rinses.

KINETICS

      The  kinetics of sorption of the two selenium forms were determined to delimit
the range of flow rates suitable for the activated alumina process.  This testing also
helped to  determine  the contact times  to be  used  in batch scale studies.  Individual
flasks, each with the same amount of prepared activated alumina, were used for each
separate time period  tested, between 0 and 120 minutes.
                                        40

-------
     As seen in Figure 11 a great majority of selenium removal occurs in the first ten
minutes of contact with activated alumina.  Rubel13 suggests this optimum contact
time in his work on fluoride removal.  However, remoyal still seems to be taking place
for upwards of  60 minutes in the case of selenium (IV).  For this reason, a 60-minute
contact time with the stock selenium solution was employed in  all batch scale tests.
This ensured equilibrium between the remaining selenium in solution and the activated
alumina.

     The 28-48 mesh size activated alumina was used in these tests. It was assumed
that the  different  mesh sizes would remove  selenium at different rates, but that at
equilibium, the mesh size should not affect the total amount of selenium removed in
the batch scale.

     Previous investigators have conducted isotherm tests to determine the capacity
for removal of  fluoride and  arsenic anions using activated alumina *'.  They have
shown that contact times on the order of 48 hours are  required to reach equilibrium.
Gupta and Chen used an initial concentration of 4.0 mg/1 arsenic  with a dose of 2 g/1
activated alumina.  Choi and Chen  used an initial concentration of 25 mg/1 fluoride
with a dose of 25 g/1 activated alumina to determine kinetics. Both of these tests used
an initial concentration much greater than was used in this study. With C  = 0.10 mg/i
and  a  dose  of 8.3 g/1  activated  alumina,  the  ratio of  initial concentration of
contaminant to dose of activated alumina is 0.012  mg/g. This is nearly  100 times less
than the  ratio of 1 mg/g for the fluoride study.  Since  the  intensity of adsorption  is
considered to be very dependent on the concentration of the contaminant, the results
of this study appear to be well within reason.

pH EFFECTS

     The influence of pH on the sorption of selenite and selenate was investigated in
batch tests covering the range of pH values commonly occurring n\potable waters. As
a  first step in determining the effect of  pH  on selenium adsorption by activated
alumina,  the pH of the stock selenium solution was adjusted to the desired value  with
NaOH  or HC1  and then put in  contact with the  prepared  activated alumina  for 60
minutes.  Figure 12 shows the results. As can be seen, the final pH (pHf) of the
                                         41

-------
                                                       •••••••••••
      30
      20
       10
EXPERIMENTAL CONDITIONS
ACT. ALUMINA: 1.25 o,  28-48 MESH
PREPARATION:
      NAOH:  I  % FOR 5 MIN
      DEIONIZED H2O:  5 MIN
      HCu:  0.05 N FOR 10 MIN. [Se(VI)]
      H2S04: 0.05 N FOR 10 MIN. [ SE(IV)]
      DEIONIZED H2O:  3 Q 5 MIN EACH
VOLUME:  ISO ML
INITIAL [SE(IV)J -  96 ppb
INITIAL [ SE (VI)] -99  ppb
CONTACT TIME: VARIES
         0       20       40        60       80       100      120
             SELENIUM CONTACT TIME WITH ACT. ALUMINA  (MIN.)
Figure 11. Effect of  contact time with activated alumina  on selenium
           adsorption.
                                  42

-------
  100
   90
   80
                                             SE 4+

3.4 _J**^  4.2	M	1i	-it/	i
      3.4
<
   70
   60


a
o  50
I-
   30
   20
    10
                pHF
                  EXPERIMENTAL CONDITIONS
                  ACT. ALUMINA: 1.25 G, 28 - 48  MESH
                  PREPARATION:
                        NAOH:  I  % FOR 50 MIN.
                        DEIONIZED HgO: - 5 MIN.
                        HCi_: 0.05N FOR 10 MIN.-SE(VI)
                        H2S04 : O.05 N FOR 10 MIN. - SE (IV)
                        DEIONIZED  HgO; 5 MIN.
                  VOLUME: 150ml
                  INITIAL [Se (IV)]  -58 ppb
                  INITIAL [SE (VI)]  -53 ppb
                  CONTACT TIME: 60 MIN.
                  pH:  AS NOTED:  pHp - FINAL pH
                  PH WAS ADJUSTED WITH NAOH OR
                  HCu  BEFORE CONTACT W/ ACT. ALUMINA
                               6        7
                           INITIAL PH
                                                          10
    Figure 12.  Effect of initial pH on adsorption of selenium.
                                43

-------
solution changed radically from the initial pH.  At no time were we able to maintain a
constant pH in the solution during any of the initial pH studies, except in the range of
pH = 4.0.

     The data does show a trend, though.  In the small variation of pH in the selenium
(IV) tests, pH seemed to have little effect on Se(IV) adsorption.  However, the slight
changes  in  pH noticed in  the  selenium (VI) tests  seemed  to indicate a relationship
between lower pH and increased removal.  But, this difference is probably due to the
interference of OH" ions added  to adjust the pH upward.

     Our inability to maintain a constant pH during the 60 minutes of contact time
with the activated alumina led  us to look at buffers that could maintain a constant pH
without interfering with adsorption.  Different buffers comprised of pthalate, acetate,
borate, and  phosphate were tried for different pH's   and  they  all interfered with
selenium (IV) and (VI) adsorption.  Their presence in the sample also interfered with
the atomic absorption spectrometry analysis of selenium.

     The next approach to maintaining  a constant pH  during  contact with the
activated alumina was to  continously monitor and  adjust the pH  using a jar  testing
apparatus to provide the mixing.  The activated alumina was prepared as usual in the
ehrlenmeyer flasks, then it was transferred to 1000 ml beakers with the  stock selenium
solution. The pH was monitored continously and was adjusted by the dropwise addition
of either 0.1 N NaOH or 0.1 N HC1.  The pH was allowed to deviate plus or minus 0.2
pH units before any adjustment was done.

     Figure  13  shows  the  results of  the testing.   Adsorption  of selenium (IV) by
activated alumina  was not effected by pH less than 7.   At pH 8 and 9, slightly lower
removals  were  observed.  This  was probably related to the fact  that the dominant
species of Se(IV) in solution changes from HSeO,~ to SeO3~ at about pH 8 and greater
amounts of NaOH  were added to  maintain these pH's. As can be seen, pH'had a varied
effect  on selenium (VI) adsorption.  Removal appears to be higher at low pH (3), drops
at slightly higher  pH, then suddently increases  again to reach a  higher removal at
pH 7.  Removal tends to drop off as pH is increased above 7.  It is most  likely that the
addition of NaOH  caused interferences that hindered Se(VI) adsorption.  Two separate
tests were done with Se(VI) and  Figure 13 shows the variability.
                                        44

-------
   100
   90
   80
    70
en
o

S,  50
    40
    ,30
    20
    10
EXPERIMENTAL CONDITIONS
ACT. ALUMINA:  1.25 G,   28 -48  MESH
PREPARATION:
      NAOH: I  % FOR 50MIN;
      DEIONIZED H2O:  5 MIN.
      HCu : 0.05 N FOR 10 MIN.   [ Sc(VI)]
      H2S04 :  0.05 N FOR 10 MIN. [ S*(IV)J
      DEIONIZED H2O:  3 <® 5 MIN. EACH
VOLUME: 150 ml
INITIAL  [SE(|V)]  - 97 ppb
INITIAL  [SE(VI)j  -99 ppb
CONTACT TIME:  60 MIN.
PH :  AS NOTED
PH WAS  ADJUSTED CONTINUOUSLY WITH
        OR HCu
                               6       7
                                 PH
   Figure 13. Effect of constant pH on adsorption of selenium.
                                45

-------
      Because activated  alumina  is  an amphoteric material,  pH  should  be  an
important factor in determining its capacity as an adsorbent. These tests fail to show
this phenomena.

      Ball's  work showed a great dependence on pH for Se(lV) and Se(Vl) adsorption by
activated alumina.   Adsorption  increased  almost  linearly from pH 7  to 4,  with
adsorption "peaking out" at approximately pH 3.  This is the type of phenomena that
would be expected.

      All further batch tests were done at pH = 4.0.  No  adjustment was necessary to
achieve this pH and, from this  data, pH had no effect on selenium (IV) adsorption in
the batch mode.  It was  later determined that the pH of the activated alumina could
be adjusted by rinsing with a less  concentrated acid.  This was  done in the selectivity
series tests and could be used to repeat the pH experiments just mentioned.

MESH SIZE

      The effect of activated alumina mesh size on selenium sorption efficiency was
determined for the following medias:  14-28, 28-48, and 48-100 (arranged in order from
largest to smallest).  These are three of the  four mesh  sizes  typically produced.  A
larger size, 8-10 mesh, is also available, but was not evaluated.

      Figure 14 shows the relationship between selenium (IV) removal and contact time
with the three sizes of activated alumina. As expected, the smaller mesh size, 48-100,
removes selenium (IV) faster than  the other two sizes due to its larger surface area per
unit weight.  In actual activated alumina contacts,  a  5-minute contact time through
the bed  might  be chosen  as  a median value for  a  cost-effective  residence time.
Comparing removals of the three mesh sizes at 5 minutes contact time gives us:

          14-28   -  59% Se(IV) removal
          28-48   -  68% Se(lV) removal
          48-100  -  81% Se(IV) removal

     Figure 15 shows selenium  (VI) removal versus contact time with the  three sizes
of activated  alumina.   As expected,  the smaller mesh  size  removes selenium (VI)
                                       46

-------
                                               48 -100  MESH
  30
  20
  10
EXPERIMENTAL CONDITIONS
ACT. ALUMINA:  1.25 G, MESH VARIES
PREPARATION:
      NAOH : I % FOR 50 MIN.
      DEIONIZED H2O: 5 MIN.
      HCu :  NONE
      H2SO4: 0.05 N FOR 10 MIN.
      DEIONIZED H2O;  5 MIN.
VOLUME : 150ml
INITIAL [SB (IV) ] ~   100ppb
INITIAL [SE (VI)] -NONE
CONTACT TIME : VARIES
 PH : ~  3.5
            20       40      60       80      100      120      140
           SELENIUM CONTACT TIME WITH ACT. ALUMINA ( MIN.)
Figure 14.  Effect of mesh size on removal of selenium  (IV).
                              47

-------
   100
   90
   80
   70
>  60
£
   50
D
i
U  40
U)
   30
   20
    10
48-100 MESH
                                           28-48 MESH
                                 ••••••••••••••••••••••••••
                                              28 MESH
 EXPERIMENTAL CONDITIONS
 ACT. ALUMINA: 1.25 G, MESH VARIES
 PREPARATION:
       NAOH: I %> FOR 50 MIN.
       DEIONIZED HzO;5 MIN.
       HCu: 0.05N~ FOR 10 MIN.
       HzS04: NONE
       DEIONIZED HzO; 5 MIN.
 VOLUME: ISO ml
 INITIAL[SE (iv)]~ NONE
 INITIAL  [Se (Vlj-lOOppb
 CONTACT TIME: VARIES
      0       20       40       60       80      100       120
          SELENIUM CONTACT TIME WITH ACT. ALUMINA (MlN)


 Figure 15. Effect  of mesh size on removal of selenium  (VI)
                               48

-------
faster.  Again, comparing removals of the three mesh sizes at 5 minutes contact time
gives us:
           28-48  - 65%
           48-100 - 80%

While the smaller mesh size (48-100) provides faster removal, one must consider design
constraints in constructing a full scale removal facility.  The larger  mesh size would
be more difficult  to fluidize during backwash, but could allow longer run lengths if
headless is involved as a  limiting factor. A trade-off would have to be made between
removal efficiencies, run length, and backwash volumes in  the final  design of a  full-
scale removal facility. The most practical size is probably a 28-48 mesh.

INTERF^ING IONS

      The effect of various anions and cations on the selenium  sorption capacity of
activated alumina  was  investigated. ,  Specific  anions of interest  were  chloride,
bicarbonate,   sulfate,  and  nitrate.   Specific cations of interest  were  calcium,
magnesium, and sodium.

      Figure  16 depicts the relationship between concentration of  the various anions
and removal of selenium (IV).  No great effect is seen by any of the anions introduced
in concentrations less  than  50 mg/1.  Greater concentrations of the anions produce a
slight decrease in efficiency of  removal, with  the exception of bicarbonate (HCO,~).
This  ion  has a more pronounced effect  on  activated alumina's ability  to  adsorb
selenium (IV), but this may  be an artifact of the effect of  pH on the form of Se(IV).
All anions were added as a salt of sodium  (Na), because sodium is expected to  have
minimal interferences with adsorption.

      Figure  17 shows how the chosen anions effect the removal of  selenium (VI).  For
concentrations of 25 mg/1 of all the anions, an apparent increase in  removal efficiency
was  noted.   It is difficult to say if a real trend  exists at  these low levels, because
certainly one would not  expect  enhancement of adsorption by the presence of other
                                        49

-------
      100
      90
      80
      70
   >  60
   O
   ,*
   c. 50

   2

   LJ
   UJ
   U)
      40
      30
      20
      10
                                                       &.-
                            PHF=6.2  '	
                                          * •
                                                  '\
                                  S04
                                                      HC°3"
                                                      pHF = 6.5
EXPERIMENTAL CONDITIONS
ACT ALUMINA:  0.5 G,  28-48 MESH
PREPARATION:
      NAOH:  I %  FOR 50 MIN
      DEIONIZEDHgO:  5 MIN.
      HCu:   NONE
      H2SO4: 0.05 N FOR 10 MIN.
      DEIONIZED H20: 2^5 MIN EACH
VOLUME:  150 ml
INITIAL  [  Se  (IV)]   -  100 ppb
INITIAL  [ SE (VI)]  -NONE
CONTACT TIME:   60 MIN.
PH :   4.0  EXCEPT AS NOTED
        0       40       80      120      160      200
                       INTERFERING AN IONS   ( mg/1)


Figvire 16. Influence  of  various anions on adsorption of selenium (IV).
                                   50

-------
                                             • ••<•
                                                    -
                                EXPERIMENTAL CONDITIONS
                                ACT. ALUMINA: 0.5 G,  28-48MESH
                                 PREPARATION:
                                       NAOH:  I  % FOR 50 WIN
                                       DEIONIZEDHaO : 5 MIN
                                       HCi.:  0.05 N FOR 10, MIN
                                       HgSO^.  NONE
                                       DEIONIZED H20:3e5 MIN. EACH
                                 VOLUME:  150ml
                                 INITIAL CSs  0V)>NONE
                                 INITIAL CSE  (VI) I]- 100 ppb
                                 CONTACT TIME:   60 MIN
                                 PH :  4.0 EXCEPT AS NOTED
               40      80      120      160      200
                     INTERFERING ANIONS ( mg/l)
Figure 17. Influence of various anions on adsorption of selenium (VI)
                                  51

-------
anions.   Perhaps  the lower concentrations calculated in  these flasks were due to
matrix interferences from the anions added.  However, at high concentrations, sulfate
(SO^  ") and bicarbonate (HCO3~) interfered greatly.  As mentioned previously, sulfate
is expected to heavily compete with selenium (VI) for sites on activated alumina since
they  are  chemically very  similar and  because  of  activated  alumina's  alumina
preference  for sulfate  over selenate.   Other researchers have  noted bicarbonate's
interference with adsorption of ions similar to selenium, but the mechanism is unclear.
The presence of high  alkalinity in raw water  will hinder the removal of both selenium
(IV) and  (VI), though  it will have a  greater effect on selenium (VI) removal.  Nitrate
and chloride exhibit little interference of adsorption.

      In  the next  test, the cations Mg++, Ca++ and Na"1"  were  added to the stock
solutions of selenium  (IV) and (VI) as salts of chloride (Cl) because,  from  the anion
interference tests, chloride seemed to exhibit less interference than the other anions.

      As expected,  the  cations shows  little  tendency to  inhibit  selenium  (IV)
adsorption.  Figure 18 shows little  change in the removal of selenium  (IV)  with varied
doses of  cations.  However, addition of varied amounts of cations to the selenium (VI)
stock seemed to enhance adsorption slightly. As is  shown in Figure 19 removals rise
from  about 90 percent with no cations present to about 98 percent with 200 mg/1 of
calcium and magnesium present.  This phenomena may be an artifact of the matrix
interferences, but  it may  also  be due to  the  "secondary  adsorption"  phenomena
explained in the introduction.

SELECTIVITY SERIES

      Kubli's original  selectivity  series for adsorption of anions by activated alumina
included  many more ions than those  listed  in  Section 1. The complete list includes the
                                                           32          2
following anions, in order of decreasing preference:  OH , PO^  , C-Qj,  , F , (SO-  >
(Fe(Cn)6)^~, CrO^2"),  S2O32", SO^2', ((Fe(CN)6)3",  Cr^of'), (NO2~, CNS"), I",  Br~,
Cl", NO3",  MnO^~, C10^~, CH3 COO", S2".

      His method for  determining the selectivity series was to introduce a solution
 containing two competing anions  into  a  very narrow column filled  with activated

                                       52

-------
100
90
80

70


J
^
> 60
I
K
I 50
2

i
UJ 40
(0
30
20
10
n
-5T-^ *-c»-H'
XM.-H-

EXPERIMENTAL CONDITIONS
ACT. ALUMINA: 0.5 G, 28-48 MESH
PREPARATION:
N*OH: 1 % FOR 50 MIN
DEIONIZEDH2O: 5 MIN

HCi_: NONE
HjjSO^ 0.05 N FOR 10 MIN.
DEIONIZED HgO: 2ft 5 MIN EACH
VOLUME: ISO ml
INITIAL CSe (IV) ] ~* 100 ppb
INITIAL CSe (vi) 3-NONE
CONTACT TIME: 60 MIN
pH: 4.0





                  40       80      120.      160      200
                      INTERFERING CATIONS (mg/1 )
Figure 18. Influence of various cations on adsorption of selenium  (IV)
                                    53

-------
        100
        90
        80
        70
     Q 60
o:
>
G'
2
D

I
        50
        40
        30
        20
        10
            ..>>er^:rrr^s^
           ^
        NA+
                      -H-
EXPERIMENTAL CONDiTIONS
ACT. ALUMINA: 0.5 e,  28-48 MESH
PREPARATION:
      NAOH:  I  % FOR 50 MIN
      DEIONIZED H2O:  5 MIN
      HO. :   0.05 N FOR 10 MIN.
      H2SO4: NONE
      DEIONIZED H2p: 3@ 5MIN. EACH
VOLUME: 150ml
INITIAL  CSE  (IV)>NONE
INITIAL  gE  (VI) J ^ 100 ppb
CONTACT TIME:  60 MIN
pH:  4.0
                 40      80      120      160     200
                       INTERFERING CATIONS ( mg/1 )
Figure 19.  Influence of various cations on adsorption of selenium  (VI)
                                 54

-------
alumina which  had been rinsed with 1:1 perchloric acid.  The solution was "filtered"
through the column and  the relative position in the column was detected by the use of
a  developer  which produced specific color changes depending on the species present.
The  anion found in the top part of  the  column was determined  to  be the  more
preferred ion.

     In this study, equal molar amounts of competing species were put in solution with
identically  pretreated alumina  at a  pH of approximately 6.5 and contacted by the
normal  method.  By determining  the  percent of each  species removed,  the greater
removed species was assumed to be more preferred by activated alumina.  The ions of
interest were:
     HSe03~ (Se(lV)), SeO^ (Se(VI)), H3AsO3 (As(III)), H2AsO^" (As(V)),
     F", H2POif", S0^~, HCO3", Cf, NO3".
     Two and one-half grams of activated alumina, when contacted for 1 hour with
150 mi's of 5e(IV) containing 5.0 mg/1, was found to remove approximately one-half the
selenium.  This  amount of alumina  was chosen to do  the  remaining tests.  5.0 mg/1
Se(IV) corresponds to a concentration of 6.3x10"  M.  This molar concentration was
used for all competing species.  The preparation of each batch of activated alumina
was pretreated the following way:

           50 minutes, 1% NaOH, 200 ml
           5 minutes, D.I. water, 200 ml
           3-10 minute rinses, 0.001 N HC1, 200 ml
           2-5 minute rinses, D.I. water, 200 ml

     In the two competition tests with the chloride ion, 0.001 N HC1CX was used as
the acid rinse.  HC1  was used .in most of the tests  because it would appear that
hydrochloric acid may be the acid used in an actual treatment facility.

     To reduce the number of tests, we  prepared an estimated selectivity series and
tested  two consecutive species in the list to verify our hypothesis,  starting with the
most preferred ion.  If the preferred adsorption  did not occur as expected, then  it

                                       55

-------
would require only one extra test to determine the actual order of preference. Table 3
shows the results of these tests.

     The data indicates that  H2?O^~ and F" were the  two most  preferred anions.
Hydroxide (OH") was not tested as it was assumed that  it was the most preferred
anion.  Tests 3 and it show that As(V) is preferred over Se(IV), but is behind fluoride in
the series. It appears as though Se(VI) is removed approximately the same in all of the
tests with HCO3", SO^2~, Cl", NO3", and H3AsO3.  In the  nitrate-selenate test,  very
little nitrate is removed and more selenate is adsorbed than nitrate.  Therefore, we
can  assume that  Se(VI)  is preferred over NO— With the arsenite-selenate tests, more
Se(VI) is removed than As(III).   Since  As(III) is  in  the unionized form,  H3AsO3, at
pH 6.5,  it  is not expected  to  be removed to any great  degree.   Nitrate was more
preferred over arsenite.

     The data concerning sulfate, bicarbonate, and chloride is hard to interpret.  In
each test involving one  of these species, more was detected in the treated water than
in the stock solutions. Because the detection limits  are fairly high for SO.2" and Cl",
the increases may be more of a reflection on the poor detection limits than anything
                                                                  2
else. From this data, it is impossible to develop a ranking between SO,,  , HCO-", Cl",
          2-2-
and  SeO.  .   Where  SeO^   falls  in preference between these four species is
indeterminate.
     Based on these tests, the following incomplete series is proposed:  OH", H-PO,.",
                          22
F , H2AsO^~, HSeO3", (SO^  , SeO^  , -HCO3", Cl"), NO3", H3AsO3.  The anions in
parenthesis  represent the  four  species  whose direct  selectivity  is unknown.   Kubli
suggested that sulfate was preferred over chloride, but that is all we know about the
interactions of  these ions. H3AsO3, being in an unionized state  in water at pH's less
than nine, will be poorly adsorbed, even less than nitrate.

CAPACITIES OF ACTIVATED ALUMINA

     A primary step for  determining adsorption  capacities of activated alumina for
selenium (IV) and (VI) is to conduct batch scale isotherm  tests.   These are done for
three reasons:  (1) to give a general idea of how effectively  activated alumina will

                                        56

-------
                              TABLE 3
                RESULTS OF SELECTIVITY SERIES TESTS*
Test
No.
1
2
3
4
5
6
7
Species
F"
F"
H SeO3"
• HSeO3"
F"
H Se03"
H SeO3"
H CO3"
HCO3"
*Minimum detection
Minimum detection
Initial
Concen-
tration
(mg/1)
2.0 (asP)
0.93
1.19
4.6 (asSe(lV))
3.6 (asAS(V))
3.9
1.15
4.7
6.8
4.75
6.5 (as CaCO3)
4.0
5,4
limits for Cl~ =
limits for 5
-------
                                  TABLE 3
                               (CONTINUED)
Test
No.
8
9
10
11
12
13
14

Species
HC03"
secy
ay
secy
secy
cr
Secy
Secy
"f-03
H3As03
Cl"
*Minimum detection
Minimum detection
Minimum detection
Initial
Concen-
tration
(mg/1)
6.7
4.1 (asSe(VI)
5.5
4.2
4.2
2.3
4.4
0.9 (as N)
4.3
3.9 (asAS(lII)
3.8
0.9
4.0
2.5
limits for Cl" =
limits for SCy =
limits for HCO3" =
Final
Concen-
tration
(mg/1)
9.8
3.5
6.4
3.4
3.3
2.5
3.5
0.85
3.5
3.8
3.6
0.8
3.7
2.8
2.0 mg/1
2.0 mg/1
0.5 mg/1
Removal
15
19
21
20
6
19
3
5
11
8

(as CaCO3)
                                                               6.4      6.5
                                                               6.1      6.2
                                                               6.2     6.3
                                                               6.3      6.5
                                                               6.3      6.3
                                                               6.4     6.5
                                                               6.4
6.4
The anions were all added in conjunction with sodium (Na), except for As(V), (ASjO J
and N03" (KNO3).
                                     58

-------
adsorb different forms  of selenium present in water. This  relatively short screening
procedure  can usually  show  whether  it is worthwhile  to  conduct time-consuming
column studies, (2) to predict the maximum quantity of  selenium activated  alumina
will adsorb, and (3) to obtain "ballpark" data to judge whether activated alumina may
be an economic way to  purify a given raw water.  Data from isotherms should allow a
rough estimate of the size and cost of contactor units.

      Figures 20 and 21 show plots of the logarithms of the two variables frequently
calculated in  isotherm data; the  equilibrium concentration of  the contaminant
(selenium) and amount of contaminant removed per unit weight of activated alumina.
Figure 20 is a comparison of the capacities  of activated alumina for selenium (IV) and
(VI) without any matrix interferences.   Either selenium (IV) or selenium (VI) only were
added to deionized water and then contacted with the activated alumina.

      The activated alumina  was prepared  as in  previous  tests; for adsorption of
selenium (IV), H2S(X was used for the  acid rinse, and for adsorption of selenium (VI),
HC1 was used for the acid rinse. The general trend  shows  that if you projected the
straight lines for selenium (IV) and selenium (VI), the "Y"-axis intercept for selenium
(IV) would be greater  than that for selenium  (VI).   This  would indicate  a greater
capacity for  selenium (IV) adsorption ^than  for  selenium  (VI) adsorption by activated
alumina.  Note on the  plot of selenium (IV) that the sloped  line  degenerates into a
vertical  line  with decreasing x/m.   This indicates that after a  certain equilibrium
concentration of selenium (IV) (~3 ppb) was attained, no further removal occurred even
by adding more activated alumina.  This may be an artifact  of the decreased accuracy
of the AA5 to determine amounts less than  5 ppb, or it may be due to the  decreased
adsorption capacity at low concentrations.

     Figure 21 shows the capacity of activated alumina for removing either selenium
(IV) or selenium (VI)  from  a water  containing  a  synthesized  mixture  of typical
constituents of a well water. The composition of this water is described in  Section 6.
The resulting capacities are lower than in the test with a deionized water matrix.
                                        59

-------
(3
     2.4
     2.0
  3  16
     1.2
     0.8
     0>
  o
                                                                          SE 44-
       / EXPERIMENTAL CONDITIONS
     /   ACT. ALUMINA: WEIGHT VARIES,28-48MESH
 ,        PREPARATION:
  '            NAOH: I  %  FOR 50 MIN
/              DEIONIZED H20:  5 MIN.
               HCi.: 0.05 N FOR 10 MIN. [Se  (VlQ
               H2SO4: 0.05 N  FOR  10 MIN. [SE  (IVj]
               DEIONIZED Hgp:  3 @ 5 MIN EACH
         VOLUME: 150ml

         INITIAL L"SE(IV) U~100ppb
         INITIAL CSe (Vl)!]^ 100ppb
         CONTACT TIME:  60 MIN
         pH :  4.0
                             0.5
                                             LOG
       1.0
      CE
(ppb)
                  1.5
2.0
           Figure 20. Freundlich isotherm plots of Se (IV) and Se  (VI)  adsorption
                      in deionized water matrix.

-------
Q
til
m
K
d
(0
0
      2.0-
       1.6
 S  u
 CD*
I
3
      0.8
      0.4
EXPERIMENTAL CONDITIONS
ACT. ALUMINA: WEIGHT VARIES, 28-48 MESH
PREPARATION:
         N*OH: I  •% FOR 50 MIN.
         DEIONIZED H2O: 5 MIN.
         HG_: 0.05N FOR 10 MIN. [Se (VI)]
         H2SO4: 0.05N FOR 10 MIN.  [SE(IV)]
         DEIONIZED H20: 3 @ 5 MIN. EACH
VOLUME: I50ML
INITIAL [SE(IV}]~ 100 ppb
INITIAL [SE(Vli~ lOOppb
CONTACT TIME: 60 MIN.
pH JS 4.0
                                                                      SE
                                                                         6+
                          0.5
                                          1.0
                                            1.5
2.0
2.5
                                                LOG Ce ppb
           Figure 21. Freundlich isotherm plots of  Se (IV)  and Se  (VI) adsorption
                       in synthesized well water matrix.

-------
     Both of the results were approximated by Freundlich isotherms.  The adsorption
values plotted were calculated according to the Freundlich equation:

     x/m =KFCe1/n

     or

     log x/m  = log Kp +  (1/n) log Ce

where x/m = the amount of selenium adsorbed per unit  weight of  adsorbent (ug/g),
C  = the concentration of selenium in the solution after 1 hour of contact time, and
n and Kp are constants.  Normally, selenium concentrations and other trace amounts
of chemical species in waters are presented as ug/1 or  ppb. Thus, mass units are more
convenient than  mole units in the  comparison of  data,  and their use makes no
difference in the use of the Freundlich equation except for the constant, Kp:

     KF = KF . Mw

     KP is on the molar basis, KP  is on the mass  basis, and Mw  = the molecular
                      10
weight of the adsorbate   .

     Kp can be determined from the intercept of the straight line with the "y"-axis.
1/n will  be the slope  of  the straight line.  Kp  is roughly an indicator of sorption
capacity and 1/n  of adsorption intensity.  As seen in Figure  20 neither of the lines
intersects with the "y"-axis. It is shown merely as a comparison with Figure 21.  From
Figure 21, values of Kp, Kp, and 1/n are tabulated below:
                          KF'Gig/g)           KpOumoles/g)              1/n
       Se(VI)                1.41                  0.069                0.31
       Se(IV)                1.66                  0.756                0.82
                                        62

-------
     On the basis of the selectivity series, the  ions present in  the synthesized well
water  that  could be interfering with selenium  (IV) and (VI) removal are  fluoride,
sulfate, and bicarbonate.  The presence of SO^~  and HCO^~ may cause interferences
due to increased adsorption intensity at high concentrations.

     Thus,  under the  same  matrix conditions, we can  predict  the  approximate
capacity of activated aluminum for selenium from the isotherms.   For an initial
concentration of 100 ppb  and an effluent concentration of 10 ppb (90% removal),
approximately 9.5  g of  Se(IV) will be  removed  per gram of activated  alumina.
Similarly, approximately 3  g of Se(VI) will be removed per gram of activated  alumina.
                                        63

-------
                                  SECTION 6
                         SELENIUM COLUMN STUDIES

INTRODUCTION

      Following  the  batch tests, continuous flow  column studies  were initiated to
analyze various operating procedures  for  the  use  of activated  alumina to remove
selenium.   Figure 22 shows a schematic diagram of the testing equipment. Figure 23
shows a photograph of the actual apparatus.  A list of materials and instruments is in
the Appendix.

      The  columns were set  up  as pressure flow devices.  The flow was controlled by
the variable speed control (SCR) for the motor. The  feed was pulled out of the tank
(or  tanks)  through teflon tubing.   The pump  tubing was tygon  and was connected
directly to the teflon tubing on the suction and discharge sides of the pump.  Teflon
tubing then carried the stock solution  into the top of the ground glass joint, which was
adapted with Swagelock fittings. At the bottom of the column, above the stop cock, a
small amount of fine stainless steel wire mesh was inserted. This was done to prevent
alumina from  leaking  out of  the  column  during  regeneration and the subsequent
treatment  run.  Once through  the column,  the effluent  was carried via tygon tubing
into a waste trough to drain.

      Caustic and acid were mixed in  two-liter beakers. Since all the regeneration,
neutralization, and rinse steps were done  in the upflow  mode, the  pump connections
were merely reversed, with the suction line  being fed  from either the caustic, acid, or
deionized water beaker. The eluants were collected  in 2000-ml graduated cylinders to
measure the regenerant  volumes.

      In all of the sample containers,  enough concentrated  HNO, was  added  as  a
preservative to  keep the pH  of the  sample  <2.   This is  an  EPA  recommended

-------
                                                              55 GALLON
                                                               STOCK
                                                              TANK(S)
       TO SAMPLERS
        AND DRAIN
MOTOR WITH
VARIABLE SPEED
CONTROL
WEIGHT
                            DURING REMOVAL RUN
       2000 ml
  GRADUATED CYLINDERS
                         DURING REGENERATION
                                2000 ml
                                BEAKER

                               (NAOH OR
                               ACID OR
                               D.I. H20)
Figure  22.  Schematic  diagram of  testing equipment during removal run and
            regeneration.   (Not to  scale.)
                                      65

-------
3
3
                                                                                     U S ENVIRONMENTAL
                                                                                     PROUCtiON AGENCY
                                         Figure  23.  Column testing  equipment.

-------
 preservation method.  Even in the regenerant samples with a  high  pH due to large
 volumes of NaOH, not enough HNCX was added to significantly dilute the sample.

 INITIAL TESTING

      It  was decided to  test the ability of activated alumina to remove Se(lV) from
 deionized water. Three different depths of alumina (3, 6, and 9 inches) were chosen to
 run  side-by-side as a means of comparing the effect of residence time on Se(IV)
 removal. The  respective empty bed contact times were .62,  1.25, and  1.87 minutes.
 34, 68, and 102 grams, respectively, of activated alumina were added to achieve those
 depths.

      When  preparing  fresh out-of-the-can (virgin) activated alumina,  the following
 steps were followed.  The alumina was  added to a column partially filled with water to
 prevent  solidification of the material.  The alumina was then backwashed at a high
 enough rate (22.0 - 24.5  m^frs =9-10 gpm/ft2) to expand the bed and  rinse most of
 the  fines out of the column.   Complete  regeneration of the virgin media was  not
 required. An initial acid  rinse and subsequent deionized (D.I.) water rinse were used.

      The removal run with deionized water after the acid rinse  showed that no Se(lV)
 appeared in  the effluent  after 6,200 bed volumes in the three-inch column and after
 3,600  bed volumes and  2,000 bed volumes in  the six-inch and nine-inch columns,
 respectively. This was done at a loading rate of 3 gpm/ft  (= 62 ml/min) and a Se(IV)
 stock concentration of 45 ppb.  It was  decided that no useful breakthrough data could
 be obtained  from this information,  so a synthetic well water was used  to effectively
 model a  real-life situation.  Previously, work had been done for the Ramona Municipal
 Water District  (RMWD) in San Diego  County to evaluate their  problem of selenium
contamination  of a  few  wells and a  synthetic water was  prepared to match  this
quality.  Table  4 lists the average composition of the major components of those well
waters.

      In order to match this quality, a  combination of reagents had to be mixed with
 D.I.   water.  Table 5 shows the different amounts of these chemicals added to 200
liters of  water to achieve the above concentration.  The amounts for Se(lV) and Se(VI)
are for 0.050 mg/1.
                                       67

-------
                      TABLE 4

     AVERAGE RMWD WELL WATER COMPOSITION

                                        Concentration
      Species _                 _ (mg/1)
Sodium
Potassium                                    3 . 0
Caicium                                     90
Magnesium                                   40
Bicarbonate Alkalinity                        335(asCaCO,)
Sulfate                                     105         J
Chloride                                    230
Nitrate                                      20
Fluoride                                     0.90
Selenium                                  0.005-0.050
                     TABLE 5

AMOUNTS OF VARIOUS REAGENTS ADDED TO MAKE UP
             SYNTHESIZED WELL WATER

   Reagent                     Weight Added (grams)

  CaCl2                             50.0

  MgCl2 6H2O                       67.0

  NaHC03                          113.0

  KF 2H2O                           0.9

  Na2SO^                            31.0

  KNO3                             13.0

  Na2Se03 (Se(IV))                      0.022

  Na2SeO^ (Se(VI))                      0.024
                         68

-------
      Freshly mixed, the  synthetic solution had a pH of about  8.3.  Most of the runs
 were done at either pH 5, 6, or 7.  Based on the batch tests, Cl~ interfered much less
 than SO. = in removal of Se(VI) by activated alumina, hence concentrated HC1 was used
 to adjust  the pH instead of H2SO^.  An analysis was  made of the various  stock
 solutions to determine the change in alkalinity and chloride  after reducing the pH.
 The results were:
                               Alkalinity
                            (mg/1 as CaC03)       Cl" (mg/1)
                   pH5          22.8                402.6
                   pH6          90.6                359.5
                   pH7         200.0                312.4
      Except for varied amounts of selenium added, these were the only variables in
 the  Se(IV) stock solution  during the testing program.  In the Se(VI) tests, the matrix
 remained the same, except for the tests on alkalinity and sulfate interferences.

 EXPERIMENTAL SUMMARY

      The remainder of this chapter will be a chronological description of each of the
 column experiments  that  were conducted during this  study.   Table 6 provides a
 summary  of each of the experiments along with its objectives and findings so that the
 reader may easily develop an overall perspective.  This table will also serve as an
 effective  reference once the reader has reviewed the ensuing discussion.

 CHRONOLOGICAL SUMMARY OF Se(IV) REMOVAL TESTS
 USING 3-INCH COLUMNS

      The  data  presented  in this  phase of the column studies  does not  draw  any
 conclusions as to Se(IV) capacity nor does it define optimum regeneration techniques
 to efficiently recover Se(IV).  The three-inch tests helped to delimit the  range of
amounts  of NaOH  and H2SO^  in  a fairly short period of time.  The  suggested
techniques were  then  tried with nine-inch deep alumina beds to verify results and
optimize operating parameters.  The reader may wish to skim through this section to
get a general feeling about the process.
                                       69

-------
         Experiment
SELENIUM (IV) - 3 INCH
COLUMN

     1. Effect of contact
        Time

   2-4. Repeatability tests

   5-6. Regeneration tests:
        16, 40, 80 bed
        volumes of 1%
        NaOH

   7-9. Fresh media -
        Repeatability runs

   10. Regeneration tests:
        1.6, 16, 80 bed
        volumes of 0.5%
        NaOH
 11-12. Regeneration tests:
        1.6, 3.2, 4.8
        bed volumes of
        0.5% NaOH (//I)
 13-14. Regeneration tests:
        1.6, 3.2, 4.8
        bed volumes of
        0.5% NaOH (//2)

 15-17. Regeneration tests:
        0.5%, 1.0%, 2.0%
        NaOH
18-19. Regeneration tests:
        1.6, 3.2, 4.8
        bed volumes of
        0.25% H
       TABLE 6

SUMMARY OF RESULTS

 	Objective	
         Result
  Determine minimum depth  Use 3-inch bed
 Verify 3" depth
 Determine amount of
   NaOH
 Start over - prepare to
   test regeneration

 Determine amount of
   NaOH
 Fine tune NaOH require-
   ment
 Fine tune NaOH require-
   ment
 Test alumina degrada-
   tion, selenium (IV)
   recovery


 Determine optimum
         amount
20-22. Regeneration tests:
        0.25%, 0.50%, 1%
 Test alumina degrada-
  tion
3 runs were reproducible

No apparent difference,
 leakage increased
16 bed volumes of 1%
 NaOH worked

Lower amount seemed
 most promising
Too rapid breakthrough-
 reduce stock feed   -
 flow rate to 3 gpm/ft


Use 4.8 bed volume
 of 0.5% NaOH
Higher concentrations
 dissolved media more,
 didn't improve
 regeneration

Use 4.8 bed volumes
 of 0.25% H2SOff
Higher concentrations
 of acid dissolved
 media faster
                                      70

-------
                                   TABLE 6
         Experiment
SELENIUM (IV) - 9 INCH
COLUMN

   1-3. Capacity tests:
         pH 5, 7.9

     4. Capacity test:
         pH 5, 6, 7
   5-6. Capacity tests:
         C. - iOO opb,
                  NaOH
  7-12 Capacity tests:
         C. = 200 ppb,
         1 gpm/ft* NaOH


 13-14. Capacity tests:
         C. = 200 ppb,
         l/2gpm/ftz
         NaOH

    15. Saturation test:
         C. = 18 ppm

 16-18. Capacity tests:
         C = 200 pph,
         l/2gpm/fr
         NaOH

SELENIUM (VI)

     1. 3-inch  column
                         SUMMARY OF RESULTS
                              (CONTINUED)

                          	Objective	
                          Determine optimum pH
                           of stock

                          Determine optimum pH
                           of stock
                          Try to speed up rapid
                           breakthrough and re-
                           cover 100% of
                           selenium

                          Try to speed up rapid
                           breakthrough and re-
                           cover 100% qf
                           selenium

                          Try to speed up rapid
                           breakthrough and re-
                           cover 100% of
                           selenium

                          Saturate media with
                           Se(IV)

                          Try to achieve steady
                           state, evaluate
                           "worst case" capaci-
                            ties
                          Determine approximate
                           breakthrough
   2-3. Repeatability tests    Verify 9-inch depth
4-5.  Regeneration tests:
      0.5, 5.3, 53.0
      bed volumes of
      1% NaOH
                            Determine amount of
                              NaOH
         Result
pH 5 best, pH 9 forms
  CaCO3

pH 5 best, leakage
  increased, must in-
  crease stock con-
  centration

Must increase stock
  concentration again
Regeneration recovers
 30-40%, reduce
 regeneration rate

Recovery 80%
Achieved total break-
 through, recovered
 80-100%

Steady state not
 achieved, capacity:
 pH 5 - 235 mg/liter
 pH6-175mg/l
 pH 7 - 100 mg/1
Bed too shallow, use
 9-inches
Reproducible runs, must
 reduce NaOH amount
0.5 bed volumes of 1%
 NaOH worked as well
 as 53
                                      71

-------
                                   TABLE 6
         Experiment
SELENIUM (VI) (Continued)
SUMMARY OF RESULTS
     (CONTINUED)

 	Objective	
   6-8. Regeneration tests:    Determine if H7SO.
        H2SO/f vs. HC1

    9. Regeneration test:
        pH5, 7
 10-11. Regeneration tests:
         1/2 gpm/ft^ NaOH
 12-13. Regeneration tests:
        2 gpm/fr NaOH
   14. Capacity test:
        PH5,7


15-16. Sulfate inter-
        ference
 17-18. Alkalinity inter-
        ference
   rinse interferes
   with Se(VI) removal
  Determine optimum pH
  Effect of slower re-
   generation on re-
   covery of Se(VI)
  Effect of faster re-
   generation on re-
   covery of Se(VI)
  Run another cycle,
   verify previous
   results
  Determine effect of
   SOj,-~ concen-
   tration on Se(VI)
   capacity
  Determine effect of
   HCO," concen-
   tration on Se(VI)
   capacity
                                                               Result
                                                      Use HC1 rinse,
                                                       interferes
                                                      pH 5 best, 1
                                                       NaOH recovered 100%
                                                       of Se(VI)

                                                      No increased recovery
                                                       versus 1 gpm/ft
                                                      No decreasecLrecovery
                                                       vs 1 gpm/ft
                                                       Approximate capacity:
                                                        pH 5 - 4.5 mg/liter
                                                        pH - 1.5 mg/liter
                                                         ^~ heavily inter-
                                                        feres, greater capacity
                                                        with reduced sulf ate
                                                      Not as interfering as
                                                       SO.=, slight im-
                                                       provement with reduced
                                                       HCO3
                                       72

-------
      After terminating the first try at selenium removal in a deionized water matrix,
the 3 inch, 6 inch, and 9 inch columns were all regenerated as follows:

      20 bed volumes of D.I. water at 9 gpm/ft , upflow
                                                2
      80 bed volumes of 1 percent NaOH at 3 gpm/ft , upflow
                                           2
      20 bed volumes of D.I. water at 6 gpm/ft , upflow
                                              2
      20 bed volumes of 0.05N H2SO^ at 3 gpm/ft , upflow
      20 bed volumes of D.I. water at 6 gpm/ft , upflow

      A liberal amount of NaOH was used to ensure optimum  recoveries of selenium
removed during the previous run.  Comparisons of Se(IV) removed during a run versus
Se(IV) recovered in the regenerant were not conducted until after run 6.

Run No. 1; 'Effect of Contact  Time

      Run  1 was  done  under operating conditions  similar to the initial removal run,
except that the synthesized well water was  used and pH was  adjusted to 6.5.  As is
seen in  Figure  24, the  three-inch column produced water  with less than 10 ppb Se(IV)
for 600  bed volumes, the six-inch column  produced 750 bed volumes, and the nine-inch
column  produced 1,100 bed volumes.   Due to  the  length of the exhaustion cycle,
operation  was  not  continuous.   Later  on  automatic samplers  were used  so that
breakthrough curves could be monitored through the night. In  run 1, two overnight
shutdowns  were required.   The results of this can be seen most dramatically in  the
three-inch column  data.   The discontinuities were  just  after  1,000 and  2,150 bed
volumes.   Upon  restarting in the  morning,  the effluent concentration markedly
decreased from the night before, and thereafter, the slope of the original breakthrough
curve was resumed. This would suggest that the  adsorption process is  "slow" and may
be diffusion limited. Notice also that there seem to be plateaus in effluent concentra-
tions.  Refer to the discussion after runs 13 and 14  of the Se(IV) nine-inch column
experiments for a more detailed explanation of this  phenomena.

     On the basis of  these  tests, three-inch  columns  were  chosen to do  some
preliminary testing  regarding various parameters of interest.   The presupposition was
that the length of a three-inch column run  would be six  to seven hours, whereas a

                                       73

-------
   40
   30
UI
   20
M
U)
   10
                                    9"
                                                                       STOCK-45
                                                                       pH-6.5
                  500
1000
1500           2000
   BED VOLUMES
2500
3000
                 Figure  24.  Run 1 -Effect of contact time on Se(IV)  removal.

-------
 nine-inch  run  would last  about  30 hours.  Also,  the  three-inch columns seemed  to
 provide  enough contact time to achieve quality goals at the beginning  of  the  run.
 Information pertaining  to amount and concentration of NaOH, rate at which NaOH is
 applied, amount and concentration of  H2SO^, rate at which H2SO^ is  applied and
 degradation of the activated alumina by various amounts and concentrations of NaOH
 and HUSO,, can be quickly determined using these  short columns.   Understandably,
      £   *r»
 capacity information that is generated should not be taken as complete in light of the
 fact that the deeper columns were able  to produce acceptable water for more bed
 volumes. For all subsequent tests, the removal run was completed during  the day, the
 columns were regenerated, and then sat overnight until the following removal run was
 started the next morning.

 Run Nos. 2-fr;  Repeatability Tests

      Figure 25  shows these  runs.  The media from the three columns of run 1  was
 mixed together  and three inches  were put into each column.  Prior to each  run, the
 columns were regenerated as described before run 1.  The operating pH was 6.5.

      The breakthrough  curve  for each run is actually  a composite of the three
 columns' data.   Since the three columns' effluent qualities were  practically identical,
 the data was averaged and the resultant graph was plotted.

      From these tests,  it appeared that the three columns would, produce long enough
 runs of high quality water to continue testing with them. The overall removal was still
 as good as  seen in the three-inch column from run 1.  A slight decrease in removal was
 noted from run  3 to run 4.

 Runs Nos. 5 and 6; ReReneration Tests, Varied Amounts of NaOH

     These runs were done to test the effects  of large differences in volume of 1%
 NaOH in the regeneration.   The remaining steps  in the regeneration were kept the
same. The  surface loading rate was increased to 6 gpm/ft  because the preceding runs
lasted on the order of 10 to  12 hours and we wished to decrease the  duration of the
runs to six  or seven hours.   At a stock pH of 6.2, run 5 shows a definite difference

                                       75

-------
   20 -
ft
20 -I    RUN 5
       STOCK =60 ppb
        • 16 B. V. I %
        • 40B. V.  1 %
        A 80 B. V.  I %
a-
LL
UJ
<>  5 "I
 lit
W  «
20 -

15 -

10 -

 5 -
          RUN 6
          STOCK =60 ppb

           • 16 B. V. I %
           * 40 B. V. 1 %
           4 80 B. V. I % NAOH
               l
              200
                      400       600

                           BED VOLUMES
 T
800
1000
1200
             Figure 25. Runs  2-6, Regeneration Tests.
                                76

-------
 between the three columns.  The leakages, defined as the amount of Se(IV) present in
 the column effluent immediately upon the start of a removal run,  were 3, 2, and 0 ppb
 for columns 1, 2, and 3.  600, 800, and  1,000 bed volumes of  treated  water were
 produced with  a concentration less than 0.010 mg/1.  After another similar regenera-
 tion,  run 6 failed to show the same difference.  Figure 25  shows that the effluent
 concentrations for run 6 were much closer to each other, all having an initial leakage
 of 2 or 3 ppb and 700-800 bed  volumes until a 0.010 mg/1 breakthrough occurred.

 Run Nos. 7-9;  Fresh Media, Repeatability Runs

      After run 6, it was decided to remove the old media and start with a fresh batch.
 Because the  history of each column varied  quite a bit, it was difficult to predict how
 each compared with the  others.  A main point that should  be  brought out here is that
 the proces^ shows a lot  of hysteresis and  the  history of each bed's exhaustion and
 regeneration plays an important role in determining how  it will operate in a removal
 run.   Inefficient  regeneration  in one column  and not the  others  will allow more
 selenium to remain on the alumina, thus diminishing its  capacity for removal. Even in
 side-be-side comparisons, this  may lead to inaccurate conclusions.

     Although  experiments 2  through 6 develop useful  data on relative performance,
 they also  demonstrated  that the experimental  conditions  used did not  result  in
 reproducible  data  and did not  allow reasonable projections  of adsorption capacity.  As
 a result, several changes were made in the  operating conditions to  match them to
 conditions shown earlier  to be suitable for  fluoride removal in large scale facilities.
                                            29
 These conditions were summarized by  Trussell  following  an extensive review of data
 from the literature and the field. It was presumed that the nature of  the process of
 adsorption/elution of Se(IV) and Se(VI) on activated alumina did'not necessite operating
 conditions different from  fluoride.

     Runs 7-9  (Figure 26) represent three  consecutive runs done with initially fresh
activated alumina in 3-inch columns.  Preceding run 7, only a 0.05N  H-SCX  rinse for
 10 bed volumes at  3 gpm/ft   followed by  20 bed  volumes of D.I.   water rinse  at
6 gpm/ft2 was  done.  Both of  these rinses were upflow. Preceding runs 8 and 9, the
following regeneration took place:
                                        77

-------
                                                  RUN 7-VIRGIN MEDIA
    20-
    15-
    10-
     5-
           RUN 10
           STOCK = 50 ppb
S
                                      •  1.6 B. v.  0.5 % NAOH
                                      •  16 B. V.  0.5 % NxOH
                                      *  80 B. V. 0.5 %
fa   2°-l
           RUN II
           STOCK =50 ppb
in
    15-
    10-
     5-
                                       •  1.6 B. V.  0.5 %
                                       •  3.2 B. V.  0.5 %
                                       A  4.8 B. V.  0.5 % NAOH
    20-

    15-

    10-

     5-
RUN 12
STOCK = 50 ppb
                                       • 1.6 B. V.  0.5 %
                                       • 3.2 B. V.  0.5 %  NAOH
                                       * 4.8 B. V.  0.5 %
                                      600
                                 BED VOLUMES
                                                800
                                               1000
1200
           Figure  26. Runs  7-12,  Regeneration Tests.
                                   78

-------
      10 bed volumes of D.I. water at 9 gpm/ft , up
      16 bed volumes of 1 percent NaOH at 3 gpm/ft , up
      20 bed volumes of D.I. water at 6 gpm/ft , up
                                           2
      8 bed volumes of 0.05 H2SO^ at 3 gpm/ft , up
      10 bed volumes of D.I. water at 6 gpm/ft , up

                                                2
      The treatment run was operated  at 6 gpm/ft .  Seen from the figure, run 7 is
 much poorer  than runs 8 and 9.  Again, the runs are composites of  the three columns
 effluent curves for each run.   Only  slight differences were found between any of the
 columns in any run. 8 and 9 show a  breakthrough capacity of 1,100  bed volumes. The
 following  chart shows the amounts  of  Se(lV) removed during the run and recovered
 during the following regeneration. Calculation of the removal during the run was done
 by integrating the area above the breakthrough curve using the stock concentration of
 65 ppb.   Regenerant amounts were  determined by  knowing the volume of regenerant
 and analyzing the regenerant for  Se(IV).

                                          Se(IV) Recovered
                    Se(IV) Removed       During Subsequent          %
        Run         During Run (mg)       Regeneration (mg)       Recovery
        7                 2.87                    1.50                52
        8                 3.25                    1.48                46
        9                 3.20
     Approximately  50  percent of the  selenium removed is recovered  by the
 regeneration.  If continued  like this for many repetitive  cycles,  the percent recovery
should approach 100 as the Se(IV)  removed during each run grows progressively less.

 Run No. 10;  Regeneration Test. Varied Amounts of NaOH (//I)

     Following run 9, it was decided to test various amounts of 0.5  percent NaOH on
the three columns.  0.5% NaOH was  chosen because a review of the current operating
procedure at a full-scale fluoride removal  facility used this concentration.   It was
assumed  that  each  column had  undergone  the  same  history  since using the virgin
alumina in run 7.   Column 1 was  given  1.6 bed  volumes of 0.5 percent NaOH at
3 gpm/ft ,  upflow,  and Columns   2 and 3  were   given  16  and  80  bed  volumes,
respectively at the same concentration and rate.  Recoveries were as follows in mg:

                                      79

-------
     Column 1 - 0.32 = 10 percent recovery
     Column 2 - 1.82 = 57 percent recovery
     Column 3-2.4  = 75 percent recovery

                                                                2
     As  Figure 26 depicts, run 10 at a surface loading rate 6 gpm/ft  showed obvious
differences.  The leakage for  columns 1,  2, and 3 was 4, 3, and 2 ppb respectively.
Removals of Se(lV) during the run  were  1.48 mg, 1.72 mg,  and 1.86 mg with break-
through occurring at 250, 400, and  650 bed volumes. In looking at the removals  of
Se(IV) during regeneration, column 2 used  ten times as much regenerant as column 1,
but recovered only six  times as much selenium.  Column 3 used fifty times as much
regenerant, but recovered only 7.5 times as much selenium.  This fact led us to believe
that the optimum recovery of Se(IV)  would  fail in the range of 1.6 to 16 bed volumes  of
0.5 percent  NaOH,  with a better chance  of the  range  being 1.6  to  8 bed volumes.
Hence, runs 11 and 12 were done with  1.6, 3.2, and 4.8  bed volumes of 0.5 percent
                 2
NaOH at 3 gpm/ft   as  the  regenerant.  The other steps in the regeneration remained
the same as described in run 7.

Runs Nos. 11-12; Regeneration Tests; Small Amounts of NaOH

     Figure 26 shows runs 11 and 12.  Run  11 shows leakage in the range of 6 to 8 ppb
and a very rapid breakthrough  to 10 ppb.  Run 12 had leakage from 0 to 1 ppb and the
change from the previous run is probably due to lesser amounts of selenium present  to
be  regenerated.  However, the slope of  the breakthrough  curve increased  rapidly,
resulting in a very sharp decrease in adsorption capacity.  Table 7 shows the removals
and recoveries for the two runs.

     As  a matter  of  reference,  1.6,  3.2,  and 4.8 bed  volumes of 0.5%  NaOH
correspond to the following doses, respectively:

                                  grams  NaOH                       -
           8.0, 16.0, and JM    ,iter act. alumina    = 0.5, 1.0, 1.5 lb/ft3

Pounds of regenerant per cubic foot of  bed and grams per liter are typical ways  to
discuss regerant doses in ion exchange literature.
                                      80

-------
                                   TABLE 7
                    REMOVALS AND RECOVERIES OF Se(IV)
                           FOR RUNS 10,  11, AND 12
                               Se(IV)  Recovered   Amount of 0.5%
            Se(IV) Removed      in Subsequent     NaOH Used (Bed       %
     Run   During Run (mg)    Regeneration (mg)      Volumes)      Recovery
                 1.48                0.15               1.6            10
      10         1.72                0.20               3.2            12
                 1.86                0.27               4.8            15
0.15
0.20
0.27
0.15
0.19
0.25
1.6
3.2
4.8
1.6
3.2
4.8
                 1.36               0.15               1.6           11
      11         1.41               0.19               3.2           13
                                                                     17
                 1.05
      12         1.21
                 1.18
Run Nos. 13-14; Regeneration Tests, Smaller Amounts of NaOH (//2)

     Due to the relatively poor breakthrough capacity exhibited by the previous runs,
an adjustment had to be made. The empty bed residence time in the activated alumina
during the previous runs was 0.31 minute (38.7 ml/124 ml/min.).  This was determined
to be too  short.  This prompted  us to  return  to  a  lower  surface loading rate of
3 gpm/ft  (=7.3 meters/hour).  This doubled the previous residence time and hopefully
would show more promising results.

     Runs 13 and 14 were conducted under .the same regenerative scheme as runs
                                                             2
11 and 12, the only difference being the treatment rate of 3 gpm/ft .  Figure 27 shows
leakage  was less than the other two runs and that the breakthrough capacity increased
to 600-700 bed bolumes for run 13, but dropped to 300-400 bed volumes for run  14. It
is evident that the larger the amount of 0.5 percent NaOH used, the less leakage and a
greater  capacity in the following runs were produced.   Table  8 shows recoveries and.
removals for runs 12, 13, and 14.

     Because of the  results of runs 13 and 14, 4.8 bed volumes of 0.5 percent NaOH at
3 gpm/ft2 was adopted as the regenerant  dose.   This is equivalent  to 24.0 g/1 or 1.5
///ft3.
                                      81

-------
     30
     25-
     20
     15
     10
5
3

U.
U.
U
 u
tO
     30 i
     25
     20
     15
     10
             RUN 13

             STOCK = 50 ppb
                                    1.6 B. V. 0.5 %
                       3.2 B. V. 0.5 %
                             1.6 B. v. 0.5
             RUN 14


             STOCK =50 ppb
                     3.2 B. V. 0.5 % N*OH
                  200
400        600         800

      BED VOLUMES
1000
1200
             Figure  27. Runs 13 and 14, Regeneration Tests.
                                       82

-------
                                   TABLE 8
                    REMOVALS AND RECOVERIES OF Se(lV)
                           FOR RUNS 12, 13, AND 14
            Se(IV) Removed
Run During Run (mg)
12
13
14
1.05
1.21
1.18
1.87
1.96
2.03
1.06
1.17
1.24
 Se(IV) Recovered
  in Subsequent
Regeneration (mg)
Amount of 0.5%
NaOH Used (Bed
   Volumes)
0.04
0.09
0.15
0.07
0.13
0.22
1.6
3.2
4.8
1.6
3.2
4.8
                                                                  Recovery
                                                                      4
                                                                      7
                                                                     13
                                                                      4
                                                                      7
                                                                     11
Runs Nos. 15-17;  Regeneration Tests; 0.5%, 1.0%, 2.0% NaOH
      i

     Runs 15, 16, and 17 were done with a  1.5 ///ft  NaOH dose, but the concen-
trations were varied  to  see  if they had any bearing on regeneration efficiency or
activated alumina degradation.  The other regeneration steps were kept the same as in
previous runs. For 0.5 percent NaOH, regeneration  time was 3  minutes.  Respective
times for  1.0 percent and 2.0 percent  were  1.5 and 0.75 minute.   Figure  28  shows
virtually  no difference in any of the breakthrough curves between the three columns
for any given run.  The breakthrough capacity for both runs  15 and  16 were 500 bed
volumes,  while it  increased to 800 bed volumes in run 17.  This is probably due to the
decreased stock concentration of 35 ppb Se(IV) for this run.  Table 9 shows removal
and recoveries of Se(IV) in runs 14, 15, 16, and 17.

     Under these conditions, it appears as though 0.30 mg - 0.35 mg of Se(IV) can be
recovered during  regeneration.   Upon a number of  repetitive cycles, this  would be
expected as the equilibrium amount of Se(IV) that could be removed during a removal
run.   The  difference  between the different  concentrations of  NaOH  in recovering
Se(IV) is very small.
                                      83

-------
201

15

10-

 5

 0'

20-

15
         RUN IS
         STOCK - 50 ppb
                                                      • 4.8 B.V. 0.5 %
                                                      • 2.4 B. V.  1.0 %
                                                      A 1.2 B. V.   2.0 %
I
ft
      RUN 16
      STOCK  = 55 ppb
                                                      • 4.8  B. V.  0.5  %
                                                      • 2.4  B. V.  1.0 %
                                                      * 1.2  B. V,  2.0  %
20 i

15

10-

5 -
         RUN 17
         STOCK = 35 ppb
         •4^'
                                                       • 4.8 B. V.  0.5
                                                       • 2.4 B. V. 1.0 %
                                                       A 1.2 B. V.  2.0  % NAOH
                200
                         400         600         800
                                 BED VOLUMES
                                                             1000
1200
               Figure 28. Runs 15 - 17,  Regeneration Tests.
                                      84

-------
      Table 9 shows that increasing concentrations of  NaOH regenerant have little
 effect on the ability to remove Se(IV), providing that the dose of NaOH per volume of
 activated alumina is kept the same. Quite possibly the effect of increasing the driving
 force for elution (increased concentration) is negated by the decreased flow through
 time in the bed.  If the concentration of NaOH is doubled while keeping the dose the
 same, it will take only  one-half the time to complete this.  Actual duration of contact
 between the NaOH and the activated alumina may be very important in determining
 optimum regenerative techniques.
                                   TABLE 9
                    REMOVALS AND RECOVERIES OF Se(lV)
                          FOR RUNS 14, 15, 16, AND 17

                              Se(IV) Recovered
            Se(IV) Removed      in Subsequent                          %
     Run   During Run (mg)    Regeneration (mg)      % NaOH      Recovery
                 1.06                0.32               0.5            30
      14         1.17                0.33               1.0            28
                 1.24                0.34               2.0            27
                 1.31                0.32               0.5            24
      15         1.31                0.34               1.0            26
                 1.32                0.35               2.0            27
                 1.47                0.31               0.5            21
      16         1.45                0.23               1.0            16
                 1.45                0.32               2.0            22
                 1.28
      17         1.29
                 1.28
     Table 10  depicts  the comparative amounts  of  aluminum  removed  during
regeneration  under the above described conditions, 1.5 //NaOH/ft^, operating with
0.5 percent, 1.0 percent, and  2.0 percent NaOH.   This amount was  present  in  a
combined sample containing the initial backwash, the NaOH rinse and the subsequent
deionized water rinse. Samples were analyzed for aluminum on the flame A AS and the
percent of activated alumina removed was calculated by converting from Al to A1-OV
                                      85

-------
     Table 10 gives an indication that higher concentrations of NaOH were able to
dissolve activated alumina faster while keeping the dose the  same.   This will be
verified in future tests.  From this data the following numbers of cycles  could be
completed  at  each  concentration of NaOH  before  all of the alumina would be
dissolved:

     0.5% NaOH - 910 cycles
     1.0% NaOH - 770 cycles
     2.0% NaOH - 625 cycles
                                 TABLE 10
                 DEGRADATION OF ACTIVATED ALUMINA BY
                    VARIOUS CONCENTRATIONS OF NaOH
                 Al Recovered in Subsequent Regeneration ((mg)
                  (% by Weight of 3-inch column in parenthesis)
             Run    0.5% NaOH      1.0% NaOH      2.0% NaOH
             14    18.6 (0.09%)    21.0 (0.10%)    26.5 (0.13%)
             15    24.5 (0.12%)    32.0 (0.16%)    39.3 (0.20%)
             16    22.0 (0.11%)    26.8 (0.13%)    32.6 (0.16%)
        Operating conditions:
        10 bed volumes of D.I. water at 9 gpm/ft , up
        1.5  //Na^H  at 3 gpm/ft2 w/0.5%, 1.0%, 2.0% NaOH, up
               ft
        15 bed volumes of D.I. water at 6 gpm/ft , up

     On another basis, if a column was regenerated once every day, after one year the
following percentages of original weight would need to be replaced.

     0.5% NaOH - 40%
     1.0% NaOH - 47%
     2.0% NaOH - 58%
                                    86

-------
      Remember that these figures are only for the dissolution of activated alumina by
 NaOH. Additional degradation is expected by the acid rinse.  The above differences
 could  mean quite a bit of extra money  being spent on replacing media on an annual
 basis.  Currently, type F-l, 28-48 mesh is selling for $0.60/lb.  For the above figures,
 the  following  annual  costs for replacement  of media  could be expected  for an
 activated alumina contactor with a volume of 250 ft  , with regeneration every  day.
 This size contactor was taken from a typical fluoride removal facility with a capacity
 of 0.7  mgd.

     0.5% NaOH - $3,300
      1.0% NaOH - $3,900
     2.0% NaOH - $4,800

 Run Nos. 18-19;  Regeneration Tests; Varied Amounts of H,

     The next step was to determine the  optimum acid rinse for a given  dose of
     i
 NaOH.  The limits set for testing were again taken from a review  of operating
 procedures at a full-scale fluoride removal facility.  1.6, 3.2, and 4.8 bed volumes of
 0.05N  H2SO^ (= 0.25%  H2SO^) were evaluated on their ability to reduce the pH of the
 bed and its effect on Se(IV) adsorption. The regeneration prior to the acid rinse was as
 follows:

     10 bed volumes D.I. water at 9 gpm/ft  , up
     4.8 bed volumes of 0.5% NaOH at 3 gpm/ft2, up
     10 bed volumes of D.I. water at 6 gpm/ft , up

     Figure  29 shows  that there is no  discernable difference in  the  breakthrough
curves for  the  different acid rinses.  However,  the pH of the activated alumina  was
noted  after  the  final  rinse (which  consisted of 5 bed volumes  of D.I. water at
6 gpm/ft2).  For 1.6 bed volumes of 0.5N HjSO^, the pH was 10.0-10.5.  For 3.2  bed
volumes, the pH was 9.0-9.5, and for 4.8 bed volumes, the pH was 5.5-6.5.

     Note  that the bed volumes until breakthrough increased from 200 in run 18 to
900-1,000 in  run  19. This may be partially due to the change in stock concentration
from 60 ppb to 40 ppb.
                                       87

-------
  30 n
  25 -
  20-
  15-
  10-
RUN 18
STOCK = 60 ppb
• 1.6 B. V.  0.25  % H2SO4
• 3.2 B.  V.  0.25 % H2S04
* 4.8 B.  V.  0.25 °/0 H2S04
h
Z
y  0
UJ 30-i
 u

-------
 Run Nos. 20-22;  Regeneration Tests, 0.25%, 0.50%. 1.0%

      From the above data, it was decided to use the equivalent of 4.8 bed volumes of
 0.05N H2SO^ for the acid rinse.  This is  a dose of 0.75 #H2SO^/ft3 bed.   Runs 20
 through 22 are very similar to the varied concentrations of NaOH runs, 15 through 17.
                                                     Q
 Keeping the other regeneration steps constant,  0.75 ///ft H2SO^ was applied at 0.05 N,
 0.1 ON, 0.20N H2SO^ at 3 gpm/ft2.  The amount of time  it took to do this was 3.0, 1.5,
 and 0.75  minutes, respectively.  No real  difference was expected in the breakthrough
 curves. Regeneration following runs 19, 20, and 21 was as follows:

                                         2
      10 bed volumes D.I. water at 9 gpm/ft , up
      4.8  bed volumes of 0.5 percent NaOH at 3 gpm/ft2, up
                                           •\
      10 bed volumes of D.I. water at 6 gpm/ft  , up
      4.8, 2.4, and 1.2 bed volumes of 0.05N, 0.10N, 0.20N HjSCL at 3 gpm/ft2, up
      10 bed volumes of D.I. water at 6 gpm/ft  , up
     Table 11 gives the removals and recoveries of Se(IV) for the NaOH and
 rinses  for runs 19 through 22 (Figures 29 and 30).  Caustic rinses contained the initial
 backwashes and intermediate rinses. Acid rinses included the final rinses.

     Notice that  the acid rinse removed in the range of 7%-13% of the total Se(IV)
 recovered during regeneration.  There should be some removal expected from the acid
 rinse because it is very concentrated compared to the Se(IV) concentration on or in
 the alumina, even though Se(IV) is preferred over SO** in the selectivity series.

     Figure 30 shows that the bed volumes to breakthrough vary from 200 to 400.  The
 leakage is still significant and reduces the volume capacity until breakthrough.

     Table  12 shows the amount  of dissolved  aluminum  present  in the acid  rinse
 portion of the regeneration for runs 19 through 21.  No regeneration was done after
 run 22, therefore, no data is available.  As before, the aluminum  was analyzed by the
 flame AAS and the present of total weight involved the conversion of AI to A17OV

     Table  12 shows  aluminum  has a  moderately higher solubility  in  the  more
concentrated acid.  Although the  differences aren't that great, there is a trend that
                                       89

-------
     20-
15-
     10-


      5
              RUN 20
              STOCK=60ppb
                                        • 4.8 B. V.  0.25 % H2S04
                                        • 2.4 B.V.  0.50 % H2S04
                                        A 1.2 B. V. 1.0%  H2S04
5

G!
U.
UJ
 111
0)
20-


15-


10-


 5-
              RUN 21
              STOCK = 60 ppb
• 4.8 B. V.  0.25 % H2SO4
• 2.4 B. V.  0.50 % H2S04
A |.2 B. V.  1.0  % H2S04
     20-


     15-


     10


      5H
        RUN 22
        STOCK =60 ppb
                                       • 4.8 B.V.  0.25 % H2S04
                                       • 2.4 B. V.  0.50 % H2S04
                                       A 1.2 B. V,  1.0 °/0 H2S04
                  200         400         600

                           BED VOLUMES
                                                 800
                     1000
            Figure 30.   Runs 20 - 22, regeneration tests.
                                    90

-------
would suggest  using a less concentrated  acid to prevent  rapid dissolution of the
alumina.  In comparison with Table 10, approximately one-third to one-fourth of the
amount removed from the bed during the caustic  rinse is removed in the acid rinse.
This will be verified in deeper column tests.
Run
 19
 20
 21
 22
                                TABLE 11
        REMOVALS AND RECOVERIES OF Se(IV) FOR NaOH AND H9SO,.
                  RINSES DURING RUNS 19, 20, 21, AND 22      *
Removal of
Se(IV)
During Run
(mg)
1.58
1.59
1.68
1.43
1.45
1.42
1.42
1.44
1.41
1.56
1.57
1.56
Recovery of Se(IV) During
Subsequent Concentration
Regeneration (mg) of H2SOj,
NaOH
0.44
0.46
0.44
0.42
0.45
0.44
0.45
0.43
0.44
*•"
H2S04
0.035
0.054
0.037
0.062
0.070
0.054
0.056
0.066
0.041
—
Used
0.05N (0.25%)
0.10N (0.5%)
0.20N (1.0%)
0.05N
0.10N
0.20N
0.05N
0.10N
0.20N
"•
% Total
Se(IV)
Recovery
30
32
28
34
36
35
36
34
35
—
                                TABLE 12
            DEGRADATION OF ACTIVATED ALUMINA BY VARIOUS
                       CONCENTRATIONS OF r
        Run
                Al Recovered in Subsequent Regeneration (mg)
                  (% by Weight of 3-inch bed in Parenthesis)
0.25%
19
20
21
5.2
6.9
6.1
(0.03%)
(0.03%)
(0.03%)
0.50%
                                    7.2  (0.04%)
                                    5.2  (0.03%)
                                    7.6  (0.04%)
1.0%
                                  8.9  (0.04%)
                                  10.8  (0.05%)
                                  8.9  (0.04%)
                                    91

-------
 CHRONOLOGICAL SUMMARY OF Se(IV) REMOVAL
 TESTS USING NINE-INCH COLUMNS

      In  reviewing the  three-inch column  data, a couple of key points were noted.
 When doing countercurrent regeneration (where the direction  of regenerant flow  is
 opposite to that of the treatment cycle), it is important to keep the particles of the
 alumina  in the same place during regeneration.  The strongest portion of regenerant
 will react with the least saturated (in terms of Se(IV)) amount of alumina. At the top
 of the bed, the weakened  NaOH will still  be able to regenerate the more saturated
 alumina.   In  ion exchange processes,  co-current  regeneration sometimes presents
 leakage  problems, but  countercurrent  processes minimize leakage and  the  use  of
 regenerant.  Therefore, a high rate backwash prior to  the regeneration was excluded
 from the regeneration program.  This backwash mixed the bed thoroughly and altered
 the distribution of alumina that was present during the  treatment run. Also, the rates
 at which the intermediate and final D.I. rinses were done were lowered.  Since those
                   2
 high rates  (6 gpm/ft )  caused  some expansion of the bed and mixing, it was thought
 that reducing the rate would increase the rinsing efficiency.

     The majority of the information  we hoped to obtain related to the capacity of
 the activated  alumina for Se(IV) under  varying conditions of pH and regenerant  flow
 rate. Since these tests with 9-inch columns were expected to last from 36 to 48 hours
                                  2
 at a treatment flow rate of 3 gpm/ft  , and each column was going to be operated at a
 different pH, each column had to have  its own separate tank of stock solution. Slight
 variations  in  the synthetic  well water  composition  could occur and  the Se(IV)
 concentration could greatly vary.  However, extra care was taken to ensure that each
 container received the same amounts of each constituent.  After the mixing and pH
 adjustment, the solutions were allowed  to sit for 24 hours to ensure  that bubbles were
 removed from the water and did not enter the columns.

     Since the 9-inch column runs  would  last  overnight, a method was needed  to
 collect samples during  the period when no one was  attending the columns.   We
 obtained three automatic samplers, called "Wastewatchers" that took a sample every
hour up to 24 hours.  See  Figure 31   for a  photograph  of  a  "Wastewatcher."  The
effluent drain lines were put into 250 ml beakers in the drain trough and allowed to fill

                                      92

-------
oo
                                          Figure  31.   Automatic  sampler,

-------
up and  overflow.  The sampler lines were put in the beakers, also.  Every hour the
sampler would take a 200 ml sample, then purge the line of remaining water by reverse
pumping back into the beaker.  Since the beakers were just larger than the volume of
samples  taken, and the detention time in the beaker was approximately 4 minutes
samples were fairly discrete.

                          .   .   ,        (250ml)
                          4 minutes =
                                        (62 ml/min)
The  "Wastewatcher"  sample containers  were analyzed  for  their ability  to  collect
selenium samples and  not  effect the  actual  concentration  of selenium  present.
Previously, known standards of selenium  were stored with HN(X preservation in glass
bottles and in "Qubetainers."  The glass was able to store samples of selenium with ppb
concentrations up to two weeks without any noticeable change, while samples stored in
Qubetainers were reproducible for one week.  The conventional polyethylene  sample
bottles  from the "Wastewatcher" showed no measurable change in selenium concen-
trations in the ppb range for up to 48 hours.  After this time, a loss of selenium was
noticed.  During the  course of the 9-inch column runs, any samples collected in the
"Wastewatcher"  containers were analyzed within 48 hours after taking them.  From
time to time during the testing, duplicate samples were taken in the "Qubetainers,"
glass bottles, and the "Wastewatcher" sample  bottles.  No  difference was  noticeable
between the three containers  as  long as they  were  analyzed within 48 hours after
taking the sample.

Run Nos. 1-3; Capacity Tests at pH 5, 7 and 9

     Runs  1 and 2 were conducted at stock pH's of 5,  7, and 9.  During the initial part
of run 3, there was a considerable amount of calcium carbonate build-up in the pH 9
column.  The pH 9 run was discontinued  after two hours during this run. These tests
were done to check the varied capacities for Se(IV) over a wide pH range.

     Run 1  used virgin media  that had been acid rinsed prior to use.   Figure 32 shows
that  the pH 7 column had the  greatest capacity for Se(IV) in run  1,  but  dropped
considerably in run 2. The breakthrough capacities for a stock concentration of 50 ppb
Se(IV) after run  2 are estimated to  be:   pH 5 - 1500  bed volumes, pH  7-900 bed

                                       94

-------
  30 -
  20 -i
  10 -
          RUN I
          VIRGIN MEDIA
                          PH9
                      /  STOCK = 68 ppb
                                pHS
                               STOCK = 67 ppb
                                                        PH7
                                                        STOCK =67 ppb
  30 -
  20
 u
(0
          RUN 2
                                        .X  PH9
                                     /  STOCK = 51 ppb
                                /"" PH7
                                   STOCK = 52 ppb
                                                         PH5
                                                        STOCK= 50
  30 -
  20 -
  10 -
          RUN 3
             _«--  PH7
       ^.^-*     STO(

.^"•-"
                                         STOCK =49 ppb
                                             PH5
                                            STOCK = 50 ppb
                                           .l       I
            200     400      600    800     1000     1200

                              BED VOLUMES
                                    1400
 I

1600
              Figure 32.  Run  1-3, Capacity Tests
                                 95

-------
 volumes,  and pH 9 - 400 bed volumes.  This equates to the following capacities for
 Se(IV) in terms of milligrams of Se(IV) per liter of activated alumina (mg/1): 67.5 mg/J,
 40.5 mg/1, and 18 mg/1.

      Based  on  the results from the  3-inch  column studies, the regeneration after
 runs 1 and 2 was:

      5 bed volumes of 0.5% NaOH at 3 gpm/ft2, up
                                           2
      10 bed  volumes of D.I. water at 3 gpm/ft , up
                                             2
      5 bed volumes of 0.05N H-SCX  at 3 gpm/ft , up
                                          2
      5 bed volumes of D.I. water at  3 gpm/ft , up

      It is evident that the capacity for Se(IV) at pH 5 and  7 decreased in both cases
 and leakage  of  Se(IV)  was 5 ppb for pH 7 during run 3.  At pH 7, the breakthrough
 capacity dropped to 300 bed volumes (13.5 mg/1).  Since the pH 5 run was terminated
 prior to 10 ppb breakthrough, its capacity can only be estimated as 1300 bed volumes
 (58.5 mg/1).

      Table  13 includes a list of the removals and recoveries of Se(lV) during all the
 9-inch column tests.   This table will be referred to often in  the  remainder of  this
 section.  For runs 1  through 3, the amount of Se(IV) recovered during  regeneration
 stays the  same;  approximately 1.9 mg for pH 5 and 2.9 mg for pH 7.  Notice that the
 total amount of Se(IV) removed during each run steadily decreased.   The recoveries
 ranged from  20 to 40 percent; obviously not as much was recovered as was removed.

 Run No. 4; Capacity Test, pH 5,  6, and 7

      Figure  33  shows  run 4.  Following run 3, the media used for the pH 9 runs  was
 discarded and replaced with virgin  media, which was  rinsed  with acid  only.   The
 breakthrough curve for pH 6 is longer than  pH 5 or pH 7, approximately  1400 bed
 volumes (77  mg/1  capacity for  Se(IV) with  C.  = 60 ppb).   Capacity for pH 5  was
*                                           ^
 1100 bed  volumes (60.5 mg/1) and 900 bed volumes  for  pH 7  (49.5 mg/1). The  larger
 breakthrough capacity for pH 6 is' attributed to the initial high capacity of the virgin
 media.
                                       96

-------
                               TABLE 13

            SUMMARY OF REMOVALS AND RECOVERIES OF Se(lV)
                DURING ALL RUNS WITH 9-INCH COLUMNS
                    (Regeneration with 5 bed volumes of
                      0.5% NaOH at noted flow rate)
Run
  3

  4


Se(IV)
Stock
Con. (ppb)
68
67
67
50
52
51
50
49
61
62
60
112
113
115
110
110
111
210
212
205
211
209
206
190
195



Stock
pH
5
7
9
5
7
9
5
7
5
6
7
5
6
7
5
6
7
5
6
7
5
6
7
5
6
Se(IV)
Removed
During
Run
(mg)
9.3
11.8
6.9
8.7
6.7
4.4
6.3
4.5
7.3
7.5
7.0
9.3
9.3
9.1
12.3
11.8
11.9
17.7
17.0
16.7
22.6
22.2
21.8
21.3
21.0
Se(IV)
Recovered
During
Subsequent
Regeneration
(mg)
1.8
2.9
2.4
2.1
2.9
3.5
1.8
2.8
2.7
2.3
2.7
10.9*
9.1
8.9
8.8
9.0
5.9
7.7
7.5
8.0
6.8
8.8
8.7
8.1
7.1


NaOH
Flow Rale
(gpm/ft2)

3


3


3

3


1


1


1


1


1
Recovery


   19
   25
   34

   24
   44
   79
   29
   62

   37
   31
   39

   117
   99
   97

   72
   76
   50

   43
   44
   48

   30
   40
   40
   38
   34
                                  97

-------
                                  TABLE 13
                                (CONTINUED)
                      (Regeneration with 5 bed volumes of
                         0.5% NaOH at noted flow rate)

Run

10

11

12

13


14


15


16


17


18


Se(lV)
Stock
Con. (ppb)
215
204
205
201
195
202
205
205
200
201
198
205
18200
18000
18000
200
204
210
185
195
192
200
204
197

Stock
PH
5
6
5
6
5
6
5
6
7
5
6
7
5
6
7
5
6
7
5
6
7
5
6
7
Se(IV)
Removed
During
Run
(mg)
26.3
24.8
25.1
24.0
23.1
23.7
23.8
23.5
14.9
29.5
28.6
28.6
1391.0
1195.0
891.0
12.4
4.9
-14.5
16.4
16.2
13.3
19.1
19.2
16.9
Se(IV)
Recovered
During
Subsequent
Regeneration
(mg)
7.9
8.0
9.2
10.2
9.0
9.1
19.7
19.3
13.1
20.5
19.9
21.3
1154.0
1086.0
892.0
128.9
112.6
91.3
48.6
42.1
36.9
35.2
31.4
27.8

NaOH
Flow Rale
(gpm/ftT

1

1

1

.5


.5


.5


.5


.5


.5


%
Recovery
30
32
37
43
37
43
83
82
88
70
70
74
82
91
100
— „
—
~
— —
__
—
— —
—
__
*Run #5 was completed as usual, but the columns were allowed to sit for 1 week prior
to regeneration.  The high figures would appear to be due to diffusion from the media
into the surrounding water during this "rest period.
                                       98

-------
    20-


    15-


    10-
RUN 4
        pH7
       STOCK = 60 ppb
                                  pH5
                                 STOCK=61 ppb
                                                   »..p—*"

                                                 /  pH 6 - VIRGIN MEDIA
                                                /   STOCK = 62 ppb

3
    20-i
    15 -\
u.
Ill
 u
(A
           RUN 5
                                            • pK 5, STOCK =112 ppb

                                            • PH7, STOCK = 115 ppb

                                            A pH6, STOCK = 113ppb
   20-


   15 -


    10-
RUN 6
            PH 7
           STOCK=IIIppb
              V-T
                                                    pH 6
                                                    STOCK = MO ppb
                                                                pH 5
                                                                STOCK = IIOppb
                 200
400        600

      BED VOLUMES
                                       800
1000
 I
1200
                   Figure  33.  Runs 4-6, Capacity Tests.
                                      99

-------
                                                     2
Run Nos. 5 and 6; Capacity Tests. C^ = 100 ppb, 1 gpm/ft- NaQH

      From runs  1 through 4, we observed two important points:

      1)    With the  NaOH being applied at 3 gpm/ft2,  approximately 30% of the
           selenium  was  being recovered.   If left  in  this cycle, the breakthrough
           curves  would  have continued  to  get  worse, and the leakage  would have
           increased  until some equilibrium was  established between the amount
           adsorbed  during the run  and the  amount eluted during regeneration.  To
           increase the amount  of  Se(IV) eluted,  the flow  rate was decreased to
           1 gpm/ft  .  If  diffusion  plays a large role in the adsorption process, then
           lowering the flow rate threefold  should recover at least three times the
           amount of Se(IV).  For more  discussion on the diffusion problem, reference
           is made to the  section following runs 13-14.

      2)    In an effort to increase  the  slope of the  breakthrough  curve from shallow
           to  steeper,  the stock  concentration  was  increased  to  about  100 ppb
           (=0.10 mg/1).  Expectedly, the amount of Se(IV) adsorbed should increase
           due to the  increased "driving  force."    But  by increasing  the  stock
           concentration, perhaps once  the alumina had adsorbed as much as it could,
           a rapid breakthrough up to the influent concentration would be noticed.

      Based on  the  above discussion,  runs  5 and 6 were conducted with  a stock
concentration  of Se(IV) of roughly  100 ppb  and the subsequent  regenerations with
1.5 ///ft3 of NaOH at a flow  rate of 1 gpm/ft .  This lowered flow rate increased the
flow-through time of the caustic in the bed from  10 minutes to 30 minutes.
     Figure 33 shows runs 5 and 6. Because of the increased stock concentration and
                                                                2
the  fact  that  the  previous  regeneration  was  done  at  3 gpm/ft ,  the  resultant
breakthrough pH 5 drifted slowly up, with no breakthrough to 10 ppb. After run 5, the
columns were not regenerated immediately.  The columns  sat unregenerated for one
week before  elution. As seen  in  Table 13, the amounts recovered from the columns
were at least equal to the amounts removed during run.5.  The long rest period helped
to diffuse some 5e(IV) out of the alumina into the surrounding fluid.
                                      100

-------
      The following run (Figure 33) shows that the leakage dropped to zero for all the
pH's and that breakthrough bed volumes for pH 6 and 7 were 900-1000 (94.5-105 mg/1).
Since the run was stopped prior to breakthrough with pH 5, breakthrough can only be
estimated to occur at 1300 bed volumes (136.5 mg/1). Note that the amount removed
during  treatment increased  significantly from run 5.   The  following regeneration
                                               2
recovered at least three times what the 3 gpm/ft   regeneration recovered in  runs 1
through 4.

Run Nos. 7-12; Capacity Tests. C. a 200 ppb. 1 gpm/ft- NaOH

      Because the  100 ppb stock concentration didn't appear that  it would  rapidly
produce a saturation of the alumina with Se(IV), and  therefore not produce the desired
rapid breakthrough, it  was  decided to jump to a  200 ppb stock  solution.   It  was
expected that the  capacity of the alumina for Se(IV) would increase again, perhaps
double  from  the  removals seen with C^ = 100 ppb and  four  times as much  as with
C. = 50 ppb.  We  were still trying to quickly  fill up the sites on and in the alumina to
achieve rapid breakthrough.

      The initial  run (#7) in  this series with  C. = 200 ppb showed poor breakthrough
characteristics, similar  to what happened after we changed the stock from 50 ppb to
100 ppb  in run 5.   Figure 34  shows that during run  7, the pH 7 column exhibited no
capacity to achieve an effluent concentration of less than 10 ppb.  The breakthrough
capacities for  pH 5 and  pH 6 were, respectively,  600 bed  volumes (123 mg/1)  and
500 bed volumes (102.5 mg/1).

      Run 8 shows a greatly improved breakthrough curves, with leakage being  5 ppb,
2 ppb and 0 ppb  for pH 7, 6, and 5, respectively.   The  breakthrough bed volumes
increased to  500 (102.5 mg/1),  900 (184.5 mg/1), and 1100  (225.5 mg/1), respectively.
The total amount  of-SeilYl_removed increased  from run 7 to about 22 mg.

      After run 8, one of the motors broke down and we decided to continue testing
pH 5  and 6, as they seemed  to produce the  optimum removals.  Subsequent  runs 9
through 12 were  run under the same condition.  See also Figure 35.  For these four
runs,  the average breakthrough capacity for  pH 5 was 900 bed volumes (175.5  mg/1),
and for pH 6 was 500 (97.5 mg/1).

                                      101

-------
   20 -

   15

   10


    5
     RUN 7
                                          pH6
                                         STOCK = 2l2ppb
          pH 7
          STOCK=205ppb
                                    pH 5
                                   STOCK =210 ppb
   20 -
I
ft
^  15
§
3  I0
LL
U.
    5
 u
10
     RUN 8
                                  PH 7
                                 S TOOK = 206 ppb   .f
--jr.---'
                                               PH 6
                                              STOCK = 209 ppb
                                                     pH 5
                                                    STOCK =211 ppb
   20 -


   15 -


   10 -


   5 -
     RUN 9
                                 PH 6             .
                                 STOCK = 195 ppb   /
                                                pH 5
                                               STOCK = 190 ppb
               200
                             600
                        BED VOLUMES
800
1000
1200
                Figure 34.  Runs 7-9,  Capacity Tests.
                                102

-------
                                                          pHS,  STOCK = 2l5ppb
                                                        OpH6, STOCK = 204ppb
I
Dj
h

1
u.
U.
U
 III
U)
20
15 •
10-
           RUN II
                                 pH6, STOCK = 201 ppb
,
5 J


pH5, STOCK = 205 ppb

^


     20
     15
           RUN 12
                           PH6, STOCK=202ppb
                                pH5, STOCK=l95ppb
                 200
                        400         600
                           BED VOLUMES
800
1000
                                                                           1200
                 Figure 35.   Runs 10 - 12,  Capacity Tests.
                                      103

-------
      Notice  that for runs 9-12, there is a general  trend of increased leakage and
lower breakthrough capacity.  If you look at  the  recoveries during  these runs, the
percent recovered  averages  approximately 37%.   The  amount  recovered  has not
increased  from the  100 ppb stock tests by the  same factor that the amount removed
has.   The decreasing breakthrough  capacities attest  to  the fact  that the  poor
recoveries are affecting how the alumina performs in the following run.  The slower
                         2
regeneration rate (1 gpm/ft ) was not slow enough.

      During run 12, composite samples were taken of  the first 500  bed volumes of the
effluent for  the  three columns.   These  composites  were analyzed  for bicarbonate
alkalinity, chloride, sulfate, nitrate, fluoride,  and  hardness.  These  were compared
with the stock concentrations of each column.  Results are listed in  Table 14.
                                  TABLE 14
                COMPARISON OF REMOVALS OF INTERFERING
                    ANIONS AND CATIONS DURING RUN 12
                        pH5
                           pH6
                                     pH7
   Species
Alkalinity
(mg/1) (as
CaCO.,
Chloride
(mg/1)
Sulfate
(mg/1)
Nitrate (as N)
(mg/1)
Fluoride
(mg/1)
Calcium
(mg/1)
Magnesium
(mg/1)
Influent   Effluent    Influent    Effluent   Influent   Effluent
 22.8
402
 98
  4.5
  0.85
 46.3
 26.0
420
 97.3
  3.3
  0.15
            82.4
 42.3
 90.6
359
 96.6
  4.8
  0.90
           91.4
 42.5
 92.0
361
109.3
  2.9
  0.20
 40.5
200.0
312
102.6
  4.5
  0.89
                      86.2
 43.6
190.4
311
102.6
  3.2
  0.40
                      80.1
 11.9
                                      104

-------
      As seen, the levels of chloride remain fairly constant, except there is a slight
 rise at pH 5.  Alkalinity and sulfate vary, with some levels increasing through the bed,
 others decreasing and some staying  the same.  This variation is  not  thought to  be
 significant. However, decreases in nitrate  and fluoride are seen in each run.  Because
 of  its low  rank in the selectivity  series, nitrate wasn't expected to be removed.
 However, because each column shows some removal and nitrate's detection limit very
 low (0.01 mg/1), the results show that activated alumina is removing some NO"! . The
 removal of fluoride is expected and greater removals were achieved at lower  pH. The
 hardness  ions, calcium  and magnesium,  decrease in each  of  the columns.   This
 decrease, noted  by other authors using activated  alumina  to  remove  fluoride,  is
 thought  to be due to secondary  adsorption.   The extremely low magnesium concen-
 tration at pH 7 is unexpected and may be due to poor analytical technique.

      During runs 9-12, the acid rinse was  analyzed more closely.  Its objective is  to
 reduce the pH of the bed to an operating level that will not reduce the capacity of the
 media for  Se(IV).  Rubel discovered that removal of  fluoride begins  to occur at pH's
 less than 10. His neutralization mode involved adjusting the pH of the raw water to be
 treated in steps to bring the pH of the treated water down to the pH of the raw water.
 In our system, we had no  easy way  to  continuously adjust the pH  of the stock feed
 solution.  We  tried to neutralize the bed with 0.05N H2SO^ until the pH of the bed
 dropped to around 7.  In determining when this occurs,  some very interesting facts
 came to light.  Figure 36 shows plots of averages of the acid rinse waters' pH just out
 of the top of the bed versus the bed volumes of acid applied at 1 gpm/ft2 (= 20 ml/min)
 for these four runs.  Actually, pH was  measured out of  the teflon tube that fed the
 2000 ml graduated cylinders to collect the  regenerant. The pH out of the top of the
 bed  was then back-calculated  by subtracting the volume of the column and  tubing
 above the top of the media. As can be seen, the pH decreased considerably near 5 bed
 volumes.  While slowly decreasing for the first few bed volumes, the  pH suddenly drops
 6 or 7 pH units in 1 bed volume or less.  At this point, the acid had neutralized most of
the hydroxide  ions from the caustic rinse. The pH seemed to  level off at 3.5, and since
the pH of the acid was 1.8,  some neutralization of the acid must still have been  taking
place.  To ensure that the  pH of the bed dropped to at least  7, all ensuring runs were
conducted  with 6 bed volumes  (700 ml)  of  0.05N  (0.25%) H2SO,  at  1 gpm/ft2.
This was also  done  in  the  upflow  mode.    It  would  not  be  necessary   to  do

                                       105

-------
  Q
  Ul
  m

  u.
  o
  H
  U
  O
  U
  K

  U.
  O


  s
   10-
      8-
      6-
                                     PH 6 COLUMN
                          pH 5 COLUMN
         NOTE; PH OF 0.05 N  H2SO4 =1.8

               1 BED VOLUME = 116 ml
                   2468


                     BED VOLUMES OF ACID APPLIED
                                                             10

  LJ
    7-




    6-




    5-
S   4-
H
       3-
       2-
                                  PH 6 COLUMN
                     PH 5 COLUMN
                   200
                           400        600


                            BED VOLUMES
800
1000
Figure 36.  Neutralization of activated alumina by 0.05N H2SO4  and

            effluent pH of subsequent treatment run.
                                106

-------
 this in a large-scale removal facility.  As a matter  of fact, co-current acid rinsing
 would probably make the neutralization step more efficient by decreasing the chances
 for  channelization.   Up flow  acid  rinsing  in these tests was  done to keep  the
 regeneration steps  as simple as possible, without a lot of changes in tubing and pump
 arrangements.

      Figure 36 also  shows  a  plot of average  effluent pH versus treated water  bed
 volumes during the four runs, 9-12.  The pH started to approach that of the stock
 solution immediately and levelled off at 300 bed volumes for pH 6 and 400 bed volumes
 for pH 5.

      For a more economical use of the acid, this  approach  is suggested.  Run 0.05N
 H-SCX through the bed until the pH of the water at the  exit of the bed is 10. Then
 start the treatment run with the pH adjusted raw water and let it bring the treated
 water pH down the  rest of the way. The amount of time to bring the pH down will be
 shprt compared to  the  removal  runs' length,  A short operating time at a pH higher
 than the optimum,  but below 10 will probably not affect  the capacity of the alumina
 for Se(IV) to any great degree.

 Run Nos. 13 and 14; Capacity Tests, C. = 200 ppb, 0.5 gpm/ft-

      Run 13, shown in Figure 37 was conducted the same as runs 9-12, except that  the
 broken motor had been replaced and testing of three pH's  was resumed.  However,  the
 regeneration after  run 13 was  done  at a 0.5% NaOH  flow rate  0.5 gpm/ft2.  As
 explained previously, a slower regeneration was expected  to significantly increase  the
 amount of Se(IV) recoverable during regeneration.  This regeneration took 60 minutes
 to accomplish, compared to 30 minutes for 1 gpm/ft  and 10 minutes for 3 gpm/ft2.

      Unexplainedly, the breakthrough curves for run 13 improved from run 12.  5 ppb,
 2 ppb, and 0 ppb leakage were present in the pH 7, 6 and 5 columns respectively.  The
 breakthrough volumes concurrently increased  to 600 bed volumes  (120 mg/1), 800 bed
 volumes (160 mg/1),  and  1100 bed volumes (220 mg/1).

      The amount  of Se(IV) recovered in the regeneration  after run 13 was more than
twice  as  much as  what was recovered during  any of the 1 gpm/ft  regenerations.
                                      107

-------
           RUN 13
I
2
u.
bl
 u
(fl
                                                              pH 7

                                                             STOCK = 200 ppb
                                                               pH6
                                                              STOCK *205ppb
                                                              STOCK* 205 ppb
20




15




10 •




 5
      RUN 14
StTO
                                                        TOCK-205ppb
                                                     •'STOCK «i98Ppb
                                                       pH  5
                                                       STOCK = 201 ppb
           RUN IS
LL
UJ
 ui
tn
                                                 • pH 5, STOCK - 18.2 ppb

                                                 • pH 6, STOCK = 18.0 ppb

                                                 A PH 7, STOCK - 18.0 ppb
                  200
                       400        600        800

                              BED VOLUMES
      1000
1200
        Figure 37.  Runs 13 - 15,  Capacity Tests and Saturation Test.
                                      108

-------
Obviously, the slower rate had quite an effect.  Subsequent run l(f showed a vastly
improved  breakthrough  curve,  with  no  leakage  for  any of  the  columns  and
breakthrough bed volumes of 700 (136.5 mg/1), 1000 (195 mg/1), and 1*00 (273 mg/1) for
pH 7, 6, and 5, respectively. The regeneration after run 1* collected more Se(IV) than
after run 13, but the percent recovery decreased due to the increased amount of Se(IV)
removed in run 1* as compared to that of run 13.

      The regenerant after run 13  was collected  differently than after the previous
runs.   100 ml  aliquots of  the regenerant  after  the  NaOH  rinse were collected  to
develop an elution curve.  This curve, plotted as a percent of Se(IV)  recovered during
NaOH regeneration versus bed volumes, will tell  a lot  about whether kinetics play a
large role in adsorption.  If  the  curve has a very sharp peak, with very little "tail,"
then diffusion may not play a key role in regeneration.  However, a curve with a long
"tail" usually  indicates that the regeneration  is  not as  rapid and  is  controlled  by
diffusion.

      Figure 38 plots  data  for the regeneration after run 13 for the three columns.
The curves are corrected for the actual bed volumes of  NaOH rinse and the deionized
water rinse afterwards.  The initial 200 ml of the regenerant collected was that water
which was left over in the column  when the treatment  run was stopped.  It had very
little Se(IV) present.   That volume was  discarded  and, as  explained before, 100 ml
aliquots were collected up to 1100 ml (9.5 bed volumes).

      Note that the highest  peak occurs earlier with increasing pH.  This is probably
due to the lower pH columns requiring  more NaOH to initially bring the pH up to a
level  where OH" ions are readily adsorbed by the  media.  As can be seen, the curves
have long, drawn-out tails, suggesting that Se(IV) adsorption is diffusion-limited.

      Based on  these results  so far,  a discussion of the  diffusion phenomena  is
presented in order to more clearly define the resistances to efficient mass-transfer in
the adsorption process.

                            20
      As explained  by Weber  ,  there are essentially three consecutive steps  in the
adsorption of  materials from  solution by  porous  adsorbents.    Listed in order of
occurrence, they are:
                                        109

-------
    30n
    20-
     10-
 Z
 o
UJ

UJ
o
ui
K
X

9

O
z
£

a
a
UJ
(E
Id
    30 -i
    20
     10
 o

 £   o
 u
 (A

 U.
 O
    30 i
     20
                        REGENERATION CONDITIONS FOR ALL COLUMNS:

                        5 BED VOLUMES 0.5% NAOH AT 1/2 gpm/ft2

                        S BED VOLUMES D. I. RINSE AT I gpm/ft2	
                                             PH 7 COLUMN
                                             pH 6 COLUMN
                                             PH 5 COLUMN
       0!     23456789    10


BED VOLUMES OF REGENERANT DURING N*OH AND SUBSEQUENT D. I. RINSES






     Figure 38.  Elution curves  for Regeneration after  run  13.
                                 110

-------
      1)    Transport of the adsorbate through a  surface film to the exterior of the
           adsorbent ("film diffusion")

      2)    Diffusion of  the adsorbate  within the  pores of  the adsorbent ("pore
           diffusion")

      3)    Adsorption of the solute on  the  interior  surfaces bounding the pore  and
           capillary spaces of the adsorbent.

      After step 1, there will  be a small amount of adsorption that occurs on  the
exterior surface of the adsorbent after transport across the surface film.

      Investigations have suggested  that the adsorption process itself (in  activated
alumina's  case, the exchange of anions discussed  in Chapter 1) is probably  not rate-
determining, and that a much slower process must control the overall rate of  uptake.
                                            •v
   \
      Film diffusion describes the resistance to mass transfer in the region separating
a turbulent bulk  solution and a  solid surface.  Many theories  have been postulated to
explain this phenomena, but boundary layer theory seems to be the most plausible.

      Since boundary layer theory accounts  for  a velocity distribution  and is  more
realistic than theories which assume a laminar film surrounding the particle, it seems
to be the best answer within today's knowledge.

      Based on experimental evidence outlined by Weber, the rate of uptake of a solute
by  many  porous adsorbents is  governed by "intraparticle  transport",  the rate of
transfer of adsorbed materials  from the exterior sites of  an adsorbent  to surfaces
bounding inner pore spaces. This  was outlined in 2 as  pore  diffusion.  The rate  for
diffusion processes  can be expressed by Pick's first  law,
                                    ° - _ D  _
                                      -  ui  ax '
                                         111

-------
where     Dj is the diffusion coefficient
           F° is the mass transport through a unit cross-section in unit
           time (f Jux),
           C is the mass concentration of diffusing substance, and
           x is the space coordinate in the direction of diffusion, so
            *\ >-»
            -£-TT  becomes the concentration gradient.

      When this diffusion is accompanied by a surface process such as adsorption, then
Pick's law must be modified to include a term for adsorption.

      In relating the above discussion to our results, the effective removal of Se(IV) by
activated alumina and its  subsequent elution by NaOH  seem to be controlled  by the
pore diffusion process.  Figure 24 shows "plateaus" in the breakthrough curves and also
increased removal for a short time after the columns had "rested."  This would imply
that  significant  film  diffusion is occurring, with  its co-current small amount of
adsorption on the exterior surfaces.  But when the concentration gradient between the
surface and the  interior of  the alumina becomes sufficient, rapid particle  diffusion
depletes the surface layer, promoting a period of time during the run when the rate of
increase in effluent  concentration versus time decreases, i.e., the curve flattens out.
If this column is  allowed to  rest overnight, the pore diffusion occurs deeper into the
particles of alumina until a "semi-equilibrium" state is reached between the  selenium
molecules and  the activated alumina.  Restarting the  column in the  morning will
repeat the above described process,  but initially the alumina will be able to remove
more Se(IV) due to the availability  of sites within the pores closer to the surface of the
alumina.

      During elution  of the Se(IV)  saturated column with NaOH, a great difference in
recovery was noted in the three different flow rates through the bed. By slowing down
the NaOH flow within the  bed  while  keeping the dosage of NaOH the same, the OH"
ions have more time to diffuse into the pores and exchange with the Se(IV) molecules.
                                        112

-------
                             2               2
    In comparing the 1 gpm/ft and 1/2 gpm/ft  rates, more than twice the Se(IV) was
 recovered in  the  slower  regeneration.  This implies  that an  even slower rate may
                                       2
 recover even more Se(IV)  than 1/2 gpm/ft .  An optimum has not yet been found.

      Based on  the  results of the batch kinetic tests, a smaller mesh (48-100) of
 activated alumina should appreciably improve the diffusion kinetics of a continuous
 flow column.  The smaller particles have a larger total  surface area per unit weight,
 therefore increased  adsorption at the surface should  occur.  By adsorbing the Se(IV)
 more rapidly  at the surface,  the  alumina should exhibit higher capacities. - It  would
 also make available  more inner sites per unit weight.  Unfortunately, 48-100 mesh is
 very fine and dissolution  during regeneration would be  greater.  In addition  to this,
 headless during  a  treatment run would increase and the particles would be subject to
 washing out of the bed during  backwash, if it is employed.

      To effectively  test the optimum dose and flow-through contact time of NaOH to
 regenerate Se(IV)-saturated alumina, one must continue decreasing the flow rate until
 no  appreciable  change in recovery is noted.   Then,  using this flow rate, vary the
 amount of NaOH  applied  (///ft ) until the  maximum amount of Se(IV)  is  recovered.
 Both  variables,  of course, affect each other. It may be, that  by using a slower flow
 rate (increasing  the flow-through contact time of the NaOH), a much smaller amount
 of NaOH can eventually be used as the optimum regenerant dose.  After diffusion was
 understood to  be more significant than batch experiments indicated, time restrictions
 on this project forced us to  delay this type of testing.  We recommend that  these tests
 be done in the future to completely investigate the diffusion phenomena and to further
 optimize NaOH flow rate  and  dose.  An attempt was made to determine  the operating
 characteristics of the columns in a "worst case" condition; as described in run 15.

 Run 15: Saturation Test, C. = 18 ppm

     The remaining  tests  for Se(IV) were done to  approximate the  "worst  case"
condition, when  the  alumina  has filled all  its sites with  Se(IV).  The  alumina was
saturated with Se(IV) in run 15 with a stock concentration of approximately 18 mg/1.
 Figure 37 shows a fairly rapid breakthrough from 0 to 18 ppm for all the columns, with
pH 7  breaking through faster than  pH 6  and  pH 5.    Table 13 shows  that  almost
                                      113

-------
•1,400 mg of Se(IV) were removed at pH 5, with 1,200 mg removed at pH 6, and 900 mg
at pH 7. The following regeneration recovered 82, 91,  and  100 percent of the Se(IV)
removed in the pH 5, 6, and 7 columns, respectively.  From the regeneration data, the
          o
1/2 gpm/ft  NaOH regeneration would not elute all of the Se(lV) adsorbed during the
previous saturation run.  This would lead one to believe  that the steady-state capacity
for Se(IV)  for  the pH 5 and pH 6 columns was  less than shown for run 14,  with
1/2 gpm/ft2 NaOH elution.

Run Nos. 16-18; Capacity Tests, GI = 200 ppb. 1/2 gpm/ft2 NaOH

     After  the saturation  test, runs  16-18  show  an approach to a  steady-state
condition where all of the Se(IV) would be recovered in regeneration that was adsorbed
during the  treatment run.   The data in Table 13 show that the amount of Se(IV)
removed during the treatment run increases as the alumina approaches an equilibrium
between the Se(IV) removed and the Se(IV) recovered.   Notice  that  for  the 3 runs
(Figure  39) following the saturation run,  more Se(IV) is  recovered in the regeneration
than was removed in the previous treatment run.  The differences in the  two figures
decreases as  the cycles are  continued,  with run 18  suggesting  that  the   amount
removable and recoverable  in a steady-state condition is probably 20 to 25 mg Se(IV).
The breakthrough curves for  the  3 different  pH's  are very different,  with  pH 5
producing more treated water with  Se(IV) < 0.010 mg/1.  The breakthrough for runs 16,
17, and  18 suggest that they will get progressively  more bed volumes of treated water
from each pH condition until steady-state is observed.  Quite possibly, an even slower
            2
(< 1/2 gpm/ft )  regeneration rate  may   be  needed  to  remove enough  Se(IV) to
effectively  prepare  the  alumina for  a  long  enough  removal  run  with sufficient
breakthrough capacity.

     Due to the  time constraints  on the project,  it was  necessary to initiate Se(VI)
removal testing to get a good  grip on the parameters of interest concerning Se(VI).
Therefore, no further testing was done on Se(lV).

     In conclusion, the  ultimate capacity of  activated alumina  for Se(IV) will be a
function of  many variables: pH, dosage of NaOH, and NaOH  regeneration flow rate.
Depending on factors such as the cost of pH adjustment, the cost of chemicals,  and the
                                        114

-------
500-


400-


300-


200-


100-
RUN 16
                                               pB 7
                                               STOCK =210 ppb
                                               pH 6
                                              STOCK = 204 ppb
                                               PH5
                                              STOCK =200 ppb
        RUN 17
                                                      3TOCK =
                                                      192 ppb
                                                      J?H 6
                                                      STOCK = 195 ppb
                                          — — ••-*-
                                                      pH5
                                                      STOCK= 185 ppb
 50-


 40-


 30-


 20-


 10
RUN 18
            ,..-••**
*****
                                                   TOCK= 197 ppb
                                                    ••*
                                              PH6
                                              STOCKs 204 ppb
                                              oH 5
                                             STOCK = 200 ppb
             200
                400        600
                      BED VOLUMES
                                         800
1000
1200
          Figure  39. Runs 16 -  18,  Capacity Tests.
                               115

-------
influent  Se(IV) concentration, an optimum  operating plan may be different for two
different sources of water.  The kinetics of regeneration play the biggest role of any
of the above-mentioned parameters in determining the ultimate capacity of activated
alumina for Se(IV).

      Based on data gathered during this phase of work, the following breakthrough
capacities  are estimated for an influent Se(IV) concentration of 200 ppb, with 0.5%
                               2                       3
NaOH regeneration at 1/2 gpm/ft  (a dose of 1.5 //NaOH/ft ) and similar water quality
to that tested in this study.
              pH 5 - 1,200 bed volumes =    235 liter ot
              pH 6 - 900 bed volumes = 175 mg/1
              pH 7 - 500 bed volumes = 100 mg/1

      Work at lower concentrations suggested that similar breakthrough bed volumes
for each pH could be achieved.  This implies a linear relationship between adsorptive
capacity and  concentration.   Some  isotherm  models  predict  a linear  relationship
between capacity and concentration, qe a c, for very low amounts of adsorption.  It is
assumed that the levels of adsorption discussed  in this report are low compared with
adsorption  of organics by  activated carbon.  We estimate that with lower concen-
trations  of Se(lV)  in the influent, capacities will be reduced.   The following table
predicts capacities for influent concentrations of Se(IV) of 50 and 100 ppb at the three
pH's tested.

                             50 ppb Se(IV)              100 ppb Se(IV)
             pH5              60 mg/1                    120 mg/1
             pH 6              4 5 mg/1                     90 mg/1
             pH 7              25 mg/1                     50 mg/1
     These capacities are all based on a 9-inch bed.  Increasing the bed depth will
probably increase the breakthrough capacity  in  bed volumes  and should minimize
leakage. Remember from Figure 24 that initial testing with 3", 6", 9" depths produced
                                       116

-------
 respectively,  600,  700,  and  1100 bed  volumes  of  treated  water  within  Se(IV)
 concentration less than 0.01 mg/1 at pH 6.5.  This trend indicates that increasing the
 depth will increase the number of bed volumes produced prior to breakthrough. There
 will be some optimum level  when increasing the depth provides  no  better removal.
 Pilot testing should verify this.

 CHRONOLOGICAL SUMMARY OF Se(VI) REMOVAL TESTS
 USING 9-INCH COLUMNS

 Run No. 1; Initial Test. 3-Inch Column

      Prior to testing  Se(IV) removal with 9-inch columns, and after  testing Se(IV)
 removal with 3-inch columns, preliminary studies looked at  rough estimates of Se(VI)
 removal.  (Runs 1 through 5). Figure 40 (run 1) shows that the breakthrough was very
 rapid with a 3-inch column and the slope  of the breakthrough curve was much steeper
 than with Se(IV).  Since the run lasted  only one-half an hour  until breakthrough, it was
 immediately decided to try a deeper column of 9-inches.

 Run Nos. 2 and 3; 9-lnch Columns, Repeatability Tests

      Runs 2 and 3 (Figure 40) were done at  pH 6.1 with a feed concentration of about
 50 ppb Se(VI).  The regeneration was as follows:

                                       2
      10 bed volumes D.I. water at 9 gpm/ft  , upflow
      27 bed volumes of 1% NaOH at 3  gpm/ft2, upflow
      10 bed volumes of D.I. water at 6 gpm/ft , upflow
                                          2
      16 bed volumes of 0.05N HC1 at 3 gpm/ft , upflow
      10 bed volumes of D.I. water at 6 gpm/ft , upflow

      Remember that we had done no  Se(IV) testing with 9-inch columns and had not
optimized regeneration yet.  However, HC1 was used as the acid  rinse  because  data
from  the  batch  tests  showed HC1  worked better for Se(VI) removal  than H-SO..
Table 15 shows removals and recoveries for runs 2  and 3.
                                     117

-------











a
5
gj
a
u.
LI
70-
60-
50-

40-

30-
20-
10-
0
70-
60-

50-
40 -


• - _ _ 	 *
. 	 ^^^
/
J RUN 1
j • COL. I 1 VIRGIN MEDIA (3 INCHES)
I • COL. 2 j
I STOCK= 67 ppb
/
-/



to*«-£»<4
A^^^
n RUN 2
// «COL. l] VIRGIN MEDIA (9 INCHES)
// .COL. 2J
' 1 STOCK= 50 ppb
20-

10-

 0
70-

60-

50-

40-

30-

20-

10-

 0
RUN 3
• COL. I
• COL. 2
  STOCK =65 ppb
                         ISO     200      250
                           BED VOLUMES
                     300
350
400
         Figure 40. Runs 1 -3,  Initial  Tests.
                            118

-------
     It  was apparent that the  liberal  amount of  NaOH used  to  regenerate  was
sufficient to recover all of the Se(VI) removed during the previous run.

                                  TABLE 15
          REMOVALS AND RECOVERIES OF Se(VI) FOR RUNS 2 AND 3

                             Se(VI)         Se(VI Recovered
                           Removed       During Subsequent
                             During          Regeneration           %
       Run      pH         Run (mg)             (mg)           Recovery
                6.1            0.28              0.35              100
                6.1            0.29              0.31              100
                6.2            0.68
                6.2            0.65
Run Nos. fr and 5; Regeneration Tests, Varied Amounts of NaOH

     The regenerations after runs 3, 4, and 5 were done to delimit the amount of
NaOH necessary to efficiently regenerate the column.  0.5, 5.3, and 53.0 bed volumes
of 1 percent NaOH were tested to see  if there were differences in recoveries.  The
                      2
flow rate was 3 gpm/ft  and  the other  regeneration  steps were kept the same as
before.  Table 16 shows the removals and recoveries for runs 3, 4, and 5.  Runs 4 and 5
are shown in Figure 41.

     The data from these runs indicate that small amounts of NaOH (.5 bed volumes
of 1 percent NaOH = 0.3 ••    3   can recover as much Se(VI) as the amount recovered
using  100 times the applied  dose  of NaOH.  Se(VI)  is not as preferred by activated
alumina as Se(IV) and,  therefore, is much easier to recover during regeneration.  Also,
note that the treatment runs lasted on the order of one-tenth as long as the Se(IV)
runs.  This shortened run length limited the role that diffusion could play  in adsorbing
more selenium.
                                     119

-------
ft
V_X
I
t
UJ
 III
in
70

60

50

40

30

ZO-

IC-

 CD
             RUN 4
70

60

50-|

40

30

20

10-

 0
             RUN 5
                                              • 0.5 B. V.  I %
                                              • 5.3 B. V.  I %
                                              A 53.4 B. V. 1 %
                                               STOCK-52 ppb
                                              •  0.5 B. V. I  %
                                              •  5.3 B. V. I  %
                                              A  53.4  B. V.  I %
                                                 STOCK=50 ppb
                      50             100
                               BED VOLUMES
                                              150
200
          Figure 41. Runs  4 and 5, Regeneration Tests.
                                 120

-------
    Run

     3

     4
                                  TABLE 16
                    REMOVALS AND RECOVERIES OF Se(IV)
                            FOR RUNS 3, 4, AND 5




PH
6.2
6.2
6.2
6.2
6.2
6.2
6.2
6.2

Se(VI)
Removed
During Run
(mg)
.68
.65
.53
.49
.51
.47
.47
.49
Se( VI) Recovered
During
Subsequent
Regeneration
(mg)
.72
.70
.50
.52
.48
.50
.45
.46


Bed Volumes
of
1% NaOH
5.3
53.0
0.5
5.3
53.0
0.5
5.3
53.0



%
Recovery
100
100
94
100
94
100
96
94
Run Nos. 6-8; Regeneration Tests.
                                      Versus HC1
     Following run 5, Se(VI) testing was delayed until after the completion of Se(IV)
testing. With  the knowledge gained about Se(IV), we were able to quickly  determine
the important parameters concerning Se(VI) removal and regeneration.  Based on Se(IV)
data, the following regeneration scheme was  adhered to throughout  the  remaining
Se(VI) testing:

     5 bed volumes of 0.5% NaOH, flow rate varied, up
     5 bed volumes of D.I. water at 1 gpm/ft , up
     6 bed volumes of 0.05N H2SO^ or HC1 at 1 gpm/ft2, up
     5 bed volumes of D.I. water at 1 gpm/ft , up

     Runs 6 through 8 (Figure 42) were regenerated prior to the treatment runs with
H9SO..  as the  acid rinse.   As we noticed in  the batch tests, a sulfuric  acid rinse
  £  *r                                                                  ^
interferred with activated alumina's ability to adsorb Se(VI).  Because SO.    and
SeO,~2 are  relatively close  in  the selectivity series for adsorption  by  activated
                                               29
alumina, the much greater concentrations of SO^~  (compared to SeO^) increased
                                       121

-------
                                            RUN 6
                                          • PH 5, STOCK = 46 ppb
                                          • pH 6, STOCK = 49ppb
                                          * pH 7, STOCK =48 ppb
                                            VIRGIN MEDIA
u
tf)
    SO-i
    40-
S|    30-

fc    20
     10
                                            RUN 7
                                          • pH 6, STOCK=50ppb
                                          • pH 7, STOCK= 50 ppb
    1001
                                            RUNS
                                          •  pH 5, STOCK= 60 ppb
                                          •  pH7, STOCK =59 ppb
                                100           ISO
                           BED VOLUMES
                                                         200
Figure 42.  Runs 6-8, H2SO. versus HCL  regeneration.
                            122

-------
its adsorption capacity in relation to Se(\~ . Therefore, the competition presented by
     *\                                "
SOj.    ions increases the difficulty of Se(VI) adsorption.  The synthesized well water
quality remained the same as in the Se(lV) removal runs.

      The graphs  indicate  that  pretreatment  with  HjSO^  will  not  produce any
breakthrough capacity for  Se(VI)  at  pH 5,  6, or  7.  The alumina does adsorb some
Se(VI), as shown in Table 17, but it didn't remove enough to bring the concentration
lower than  10 ppb.  Therefore, the regeneration following run 8 and all subsequent
regenerations were done with a 0.05N HC1 rinse.  Chloride should not interfere with
Se(VI) adsorption.  During the regeneration after run 6, the motor for the pH 5 column
broke down.  No replacement was readily available and a new motor was not obtained
until run 15 was initiated.  pH 5 and pH 7 were done on runs 8 through  1*, while the
data for pH 6 was estimated by interpolation.

Run  No. 9; Capacity Test. pH 5 and 7

      The regeneration following runs 8 and 9 was done with the NaOH surface loading
               2                                    2
rate  of 1 gpm/ft .  It appeared as though the 1 gpm/ft rate wasn't recovering all of
the Se(VI).  As with Se(IV), the amounts removed were calculated by integrating the
area above the breakthrough curve and below the stock concentration level.  Figure 43
shows run 9 and it is apparent  that the concentration of Se(VI) increases  above the
stock concentration for a short period of time after breakthrough.  This  means that a
more selective  species is being adsorbed on the alumina causing the Se(VI)  previously
adsorbed to be desorbed,  thus increasing the concentration  of Se(VI) in the effluent.  It
is thought that SO£  is causing this.  In this case, after  breakthrough has occurred,
more Se(VI) is coming off than is being put on the column. This requires subtracting
the amount  desorbed after  breakthrough  from the amount adsorbed prior  to  break-
through.  The amounts of Se(VI) removed during the run shown in Table 17 reflect this
calculation.   From  the  graph, 100 bed volumes of  treated  water  with  an Se(VI)
concentration less than  0.01 mg/1 are produced at  pH 5.   This  is  equivalent  to
4.5 milligrams of Se(VI) per liter of activated alumina.  For pH 7, 35 bed volumes, or
1.6 mg/1 Se(VI) were adsorbed.
                                       123

-------
Run Nos. 10 and 11: Regeneration Tests. 1/2 gpm/ft2 NaOH
                                        o
     Because the regeneration at 1 gpm/ft  didn't appear to recover all of the Se(VI)
removed during runs 6 through 9, the  regenerations after runs 10 and 11 (Figure 43)
                      2
were done at 1/2 gpm/ft . Table 17 shows that the amount of Se(VI) in the regenerant
did not increase from the  previous runs.  Testing of Se(IV) regeneration showed that a
decreased regeneration rate increased  the amount recovered in the eluant.  This was
not found in runs 10 and 11.
                                              o
Run Nos. 12 and 13; Regeneration Tests. 2 gpm/ft   NaOH

     To verify  that the kinetics of regeneration did not play as big a role as they did
                                                                           2
with Se(IV), regenerations  after runs 12 and 13 (Figure 44) were done at 2 gpm/ft .  As
seen in Table 17, no change in the amount of Se(VI) recovered was noticed, but the
apparent percentage of recovery still indicated that only 50 to 60 percent  of the Se(VI)
removed during a run was being recovered in the regeneration.

Run No. 14; Capacity Test, ph 5 and 7

     We decided  to collect all the  effluent  from  both  columns during  run 14
(Figure 44). This represented 160 bed volumes or 18.6 liters of water for each column.
The Se(VI)  concentration  in the  composite samples for each column were:  pH 5 -
45 ppb, pH  7 - 55 ppb.  These amounts, subtracted from the initial  concentration of
Se(VI) in the stock solution and multiplied by the volume of effluent, should give the
total amount of Se(VI) remaining on the alumina after termination of the run.  The
resultant amounts  were: pH 5 - 0.26 mg, pH 7 - 0.09 mg.   Table 17 shows that these
amounts correspond to the amounts recovered in the regeneration after run 14.

     The previous method for calculating the amounts of Se(VI)  removed during the
run was subject to some  error,  due to the small number  of sample points  on the
breakthrough curve.  Apparently, right  after breakthrough, the curve goes much higher
than the sample points indicate.  There must be a very sharp peak immediately after
breakthrough which made the integration by averaging the values of consecutive points
inaccurate.  Therefore, the amounts listed as Se(VI) removed during run for runs 6-13
                                      124

-------
     100-


      80-


      60-


      40-


      20-


       0


     100-

2T
'8   8
s^x
H
U   60 H
 u   40H
 Ul
 (A
        RUN 9
A
            r
                                                 pHS
                                PH7
                                STOCKS: 58 ppb
20-


 0


[00


80


60-


40-


20-
        RUN 10
                             PH7
                             STOCK =57 ppb
                     pKS
                    STOCK = 55 ppb
            RUN II
                       / PH7
                      / STOCK = 60ppb
                                       PH5
                                      STOCK = 60 ppb
               -I	1	—
                50            100
                        BED VOLUMES
                                               ISO
                                        —1
                                         200
        Figure  43.  Runs  9  - 11, Regeneration Tests.
                             125

-------
too-i
                                      PH5
                                    STOCK = 60ppb
[00-i
 80-
               50
       100
BED VOLUMES
ISO
200
  Figure 44. Runs  12  - 14, Regeneration Tests.
                       126

-------
               TABLE 17

   REMOVALS AND RECOVERIES OF SE(VI)
IN TESTS USING 9-INCH COLUMNS, RUNS 6-18


Run

6*

7*
8*
9
10

11
12

13
14
15


16



pH
5
6
7
6
7
5
7
5
7
5
7
5
7
5
7
5
7
5
7
6(S04"2 =5)
6(S04'2 = 50)
6(S0^2 = 500)
6(50 -2 = 5)
6(S04'2 = 50)
6(SO ,'2 = 500)


Se(VI;
Removed
During
Run (mg)
0.22
0.09
0.01
0.30
0.21
0.33
0.17
0.40
0.16
0.45
0.15
0.43
0.10
O.*15
0.45
0.11
0.26
0.09
2.50
0.73
0.06
2.45
0.69
0.05
Se(VI)
Recovered
During
Subsequent
Regeneration
(mg)
0.17
0.11
0.02
0.13
0.09
0.19
,0.20
0.25
0.12
0.24
0.11
0.24
0.12
0.23
0.12
0.23
0.10
0.24
0.09
2.15
0.68
0.04
2.13
0.63
0.04


NaOH
Flow Rale
(gpm/ft^)

1

1
1
1
.5

.5
2
-
2
.5
.5


.5



Recovery
77
100
100
43
43
58
100
63
75
53
73
56
100
52
80
51
91
92
100
85
93
67
87
91
80
                  127

-------
Run
 17
 18
      pH
6(Alk = 5)
6(Alk = 50)
6(Alk = 500)

6(Alk = 5)
6(Alk = 50)
6(Alk = 500)
                                   TABLE 17

                                 (CONTINUED)
  Se(VI)
Removed
 During
Run (mg)
  0.52
  0.32
  0.18

  0.57
  0.27
  0.14
Se(VI)
Recovered
During
Subsequent
Regeneration
(mg)
0.48
0.28
0.15
0.53
0.27
0.14



NaOH
Flow Rale

.5


.5





Recovery
92
88
83
93
89
100
*Prior to these runs, 0.05N H-SCX was used as the acid rinse.  The removals of Se(VI)

were resultantly low.
are probably in error to some degree, and it is likely that 100 percent recovery by the
regenerations  may have been accomplished.   The remaining four runs (15-18) had the
effluent samples analyzed every hour as before, but all of the effluent was collected
and analyzed, as per run 14.


     Notice for runs 9 through 24 that leakage is present in all the runs.  This leakage
didnt affect the breakthrough volumes of any run since the effluent concentration of

Se(VI) never got above 10 ppb until the rapid breakthrough.   We assumed that all the
Se(VI) was being removed during regeneration, so there should not be any present as-
leakage.   Leakage was not present in  runs 1  through  5 when the HC1 acid rinse was
initially used to treat the virgin media. It is possible that the  FUSO^. rinses  prior to
runs 6 through 8 affected the  performance of the following runs.
     Based  on  the  data acquired  in  runs 9 through 14, the following  breakthrough
capacities are  estimated  at  pH 5,  6, and  7:  100 bed  volumes,  65 bed  volumes, and
                                      128

-------
30 bed  volumes,  given  the  water  quality  explained in  Table 4  and  an  initial
concentration of 50 ppb Se(VI).  These volumes correspond to the following capacities:
pH 5 - 4.5 mg/1, pH 6 - 3.0 mg/1, and pH 7 - 1.5 mg/1.

     Similar to the data  presented in  Figure 38  for  Se(IV),  an elution curve was
prepared for the regeneration after run 14. Because  the kinetics of the regeneration
did not play an important role in the recovery  of Se(VI),  it was assumed  that the
elution curve would show a sharp peak with very little tail.  Compared with  Se(IV),
Se(VI) is not adsorbed nearly as well, therefore it makes sense that it should be easier
to desorb.

     Figure 45 shows  that the elution curves for the pH 5  and 7  columns are almost
identical in shape,  with no relevant elution occurring after 2.5 bed volumes.  The sharp
peak occurred almost instantaneously with  the first amount of NaOH that was applied.
From this graph, it is apparent  that OH"  ions are much  preferred over Se(VI) ions and
that the rate of application of  the NaOH is not important.   Probably much less NaOH
could be  used  to  effectively regenerate Se(VI), based on  these  results and earlier
                                                                  2
testing done with 0.5,  5.3, and 53.0 bed volumes of NaOH at 3 gpm/ft .  To ascertain
what minimum  dose of NaOH would be appropriate  to regenerate these columns, side-
by-side tests with decreasing  amounts of NaOH applied should be done.  When  less than
100% recovery  is noted, it will indicate that the minimum dose has been used.

     Because HC1 would be  the required acid for  neutralization  of the alumina in a
full-scale removal facility removing a combination of  Se(IV)  and Se(VI), in  order to
             »
optimize removals, its use had to be compared to that of H2SO^.  A pH  breakthrough
diagram was constructed, similar to Figure 36 for 0.05N HC1 during the regeneration,
after run  14. Figure 46 shows the results.  The amounts of HC1 required to reduce the
                                     O
pH of the alumina  after the 1/2 gpm/ft   NaOH regeneration are virtually identical to
those for H2SO^.   We expect  no  decrease in the  capacity of activated alumina for
Se(IV) due to the use of HC1.  Cl" is lower on the selectivity series than SO^  and using
this as the acid rinse may enhance Se(IV) adsorption.
                                      129

-------
      70
      60
      50-
Z   40
O

P


S   30


§

S   20

a


9   10
   o
   1   »
          73
                  REGENERATION CONDITIONS FOR BOTH COLUMNS:

                  5 BED VOLUMES 0.5% NAOH AT 1/2 gpm/ft2

                  5 BED VOLUMES D. I. RINSE AT I gpm/ft2
                                          pH 7 COLUMN
   a  70
   ui
   8  60
   HI
   K
   ^.  50
    40-





    30





    20





    10
                                          pH 5 COLUMN
         01     234     5     6    78     9    10


BED VOLUMES OF REGENERANT DURING N*OH AND SUBSEQUENT D. I. RINSES
      Figure 45.  Elution curves  for  regeneration after run 14,
                                 130

-------
      a
      m
      u.
      o
      h
      Z
      el
      s
      £
      li.
      O
12


10
4-
2-
                      PH 5 COLUMN
                                      PH 7 COLUMN
     NOTE: pH OF 0,05 N HCu =1.8
            I BED VOLUME =116ml
                           -»     I     I	T	1	,	
                            345678
                            BED VOLUMES OF ACID APPLIED
                                                  T
                                                   9
 T"
  10

     a
                                             PH 7 COLUMN
                                 T
                                400        600
                                BED VOLUMES
                                            800
1000
Figure 46.  Neutralization of activated  alumina by 0.05 HCL and subsequent
            treatment run.
                                    131

-------
 Run Nos. 15 and 16; Sulfate Interferences

      Because  it  was  noted that  sulfate  and bicarbonate  interfered  with  Si  VI)
 adsorption in batch tests and that rinsing  with PUSCX prior to a treatment run causes
 poor removal of Se(VI), it was decided to test varied levels of SCL~  and HCO^" for a
 few runs.
      Runs 15  and 16 were  done with sulfate concentrations of 5, 50, and  500 ppm
     2
S
-------
80-

60-

40 •


20-

0-
RUN 15 -^
\ .^".*. ' ""X^

V** / STOCK
SOOppm SOj /
STOCK- SOppb 1
1
••* 1 =
' 5 ppm SO4
MfcJ— *• " STOCK" 55 ppb

m
nS04
- 59 ppb




— 	 — 	 '

80


60


40 -
            RUN 16
       I  \
             X
                       '•*.«  50ppmS04
                           STOCKs 57 ppb
                    J..A  500 PP» S04
                    J  STOCK- 60ppb
                               •  5 ppm ALK, STOCK -54 ppb

                               •  50 ppm ALK, STOCK -49 ppb

                                  500  ppm ALK, STOCK - 49 ppb
80


60


40-


20-


 0
    RUN
nr-f^i
      • /
  + mm**mJ
                               •  5 ppm  ALK, STOCK-52 ppb

                               •  50 ppm ALK, STOCK-48 ppb

                               A  500  ppm ALK, STOCK - 49 ppb
             100
                         200          300
                           BED VOLUMES
                                                400
 Figure  47.  Runs 15 - 18,  SO4 and HC03 interferences.
                                                            500
                           133

-------
     100
8
      10
  8  i.o
  ui
 U)  0.5
    O.I
                                                   —-so4-
                                                   (ALK = 100 ppm

                               ALKALINITY
                               (AS
                                        ppm )
       10
50
100
500
1,000
5,000   10,000
                      RATIO OFSO~4  OR ALKALINITY TO Se(VI)
                   SO4-     ALK
                         OR	  ALL CONCENTRATIONS IN ppm
                  St(VI)
  Figure 48.  Se(IV)  removal versus SOj and alkalinity concentrations,
                                   134

-------
all  of  the  Se(VI) adsorbed.   Diffusion allowed some of the Se(VI) to penetrate far
enough into the bed to escape desorption by the NaOH. The shorter runs did not allow
enough  time  for  significant  diffusion  to occur, and all of the adsorbed  Se(VI) was
recovered.

Run Nos. 17 and 18; Alkalinity Interferences

     The final series of tests were done to test alkalinity's effect on Se(VI) adsorption.
The other parameters in the synthetic water were kept as before (SO^ = 100 ppm) and
only the alkalinity was varied to achieve 5, 50, and 500 ppm alkalinity (as CaCOJ at
pH  6. This  required adding varied amounts of NaHCO3 to each solution, as the amount
of buffering capacity each system had would vary, and different amounts of HC1 must
be  added to  achieve  the  desired alkalinity  at pH 6.   For 100 liters  of water, the
following amounts of NaHCO, and HC1 were added:

    5  ppm -    .33   meq/1 HCO" =  2.8 grams NaHCO-

                .23   meq/1 H+ = 3.9 ml of 6N HC1

    50 ppm -    2.72 meq/1 HCO^  = 22.9 grams of NaHCO3


                1.72 meq/i H+  =  28.7 ml of 6N HC1

    500 ppm - *  33 meq/1 HCO^ = 277.2 grams of NaHCO3

                23 meq/1 H"1"  =  383.3 ml of 6N HC1

     The results are shown in runs 17 and 18 (Figure 47). The effect of varied HC
-------
100 bed volumes are produced.  The effect of reducing the alkalinity from 100 ppm to
50 ppm  was  barely noticeable,  comparing  results  obtained  in the  previous  runs.
Figure 48 shows the relationship of alkalinity to Se(VI) capacity.  Compared with  the
plot for sulfate, its effects are not as great.

      In summary, Se(VI), being  lower in the selectivity series, is more susceptible to
interferences with adsorption. For the same operating conditions and water composi-
tion, Se(VI) was removed about  one-tenth as much as Se(IV). Based on data gathered
during Se(VI)  testing, the following  breakthrough  capacities are  estimated for an
influent concentration of 50 ppb Se(VI), with 0.5% NaOH applied as the regenerant at
2 gpm/ft  (a dose of 1.5 #NaOH/ft ), a neutralization step using HC1 as the acid, and
similar water quality to that tested in this study.
       PH 5 - 100 bed volumes =    *.J  liter      to    alumina
       pH 6 -  70 bed volumes =     3.2 mg/1
       pH 7 -  35 bed volumes =     1.6 mg/1

     These capacities are  13  times less than Se(IV)  adsorption under similar  condi-
tions.  If sulfate  concentrations in the water are less than those used in this testing,
(<100 ppm), increased capacity can be expected. If sulfates were present at 50 ppm,
then the above capacities could be doubled.  Alkalinity does not play as important a
role in interferring with Se(VI) adsorption.
     It is also predicted that for increased Se(VI) initial concentrations, the capacity
for Se(VI) will  linearly  increase, as explained  in the summary of the Se(IV) testing.
However, the actual number of bed volumes of  treated water with Se(VI) concentration
less than 0.01 mg/1 would remain the same.

     Based on the  results of  these  tests, Se(VI)  would be  the limiting  factor  in
determining the capacity of activated alumina for a mixture of Se(IV) and Se(VI).  If
concentrations of Se(VI) exceeded 0.01 mg/1 in the influent to a full-scale removal
facility, then the run length would be limited to the breakthrough capacity  of Se(VI).
                                       136

-------
Regeneration would therefore be required more often, with resultant increased costs
due to chemicals and replacement of bed media.

     One last test was performed on the activated alumina after the  runs had been
completed for approximately 1 week.  The remaining media in the columns was mixed
together, then 9-inches was put into the three columns.  Another test  was done with
different concentrations of  NaOH to see if the reduced regeneration rate and the
different concentrations effected the amount of alumina dissolved. Since the alumina
had been previously filled with Se(VI), the differences in this test of desorption were
not considered.

     The three columns were dosed equally with 1.5 #/ft  NaOH at  a flow  rate of
         2
1/2 gpm/ft , the only difference being the concentrations of NaOH, 2%,  1%, and 0.5%.
They were  then rinsed with  5 bed volumes of  D.I. water.  Following this was a
D.05N HC1 rinse (6 bed volumes at 1 gpm/ft2) and subsequent 5 bed volume rinse with
D.I. water.  The results are listed in Table 18.

                                 TABLE 18

             DEGRADATION OF ACTIVATED ALUMINA BY VARIED
                      CONCENTRATIONS OF NaOH AND
                          BY 0.05N HC1 ACID RINSE
                     Al recovered during Caustic
                           and Subsequent
                           D.I. Rinse (mg)
                    (% by weight of 9 -inch column
                    _ in parentheses) _
       0.5% NaOH
       1.0% NaOH
       2.0% NaOH
    (0.86%)
480 (0.87%)
489 (0.89%)
Al Recovered During
     0.05N HC1
   Rinse and Sub-
    sequent D.I.
     Rinse (mg)
     48 (0.09%)
     41 (0.07%)
     38 (0.07%)
                                    137

-------
      The table  shows that the decreased  regeneration rate had  a large  effect  on
dissolution of alumina when compared with Table 10.  The differences noted between
the 3 different concentrations of NaOH are not as great  as  with  the  previous  test,
however there is a slight increase in amount dissolved as  concentration is  increased.
The amount of alumina dissolved  during the acid rinse increased a little also,  from
0.03% with 3 gpm/ft2 rate to 0.08% with 1 gpm/ft2.

      Based on these  results, the following numbers of cycles could be completed  at
each  concentration of NaOH before all  of  the activated alumina would be dissolved
(included in loss due to acid rinse, also):

      0.5% NaOH - 106 cycles
      1.0% NaOH - 105 cycles
      2.0% NaOH - 103 cycles

      If this regeneration were done once a day, the media would have to be replaced
3.5 times a year. If it were possible to regenerate less often, then the resultant loss
of media would be less.

      This information would have to be evaluated more closely in pilot studies.  This
work  indicates that increasing the amount of flow through contact time of NaOH  in
                                 2                          2
the bed fr.om 10 minutes (3 gpm/ft ) to 60 minutes (1/2 gpm/ft ) will greatly increase
the amount of alumina dissolved.  The associated replacement costs for this could  be
very expensive.
                                      138

-------
                                  SECTION 7
                        PRELIMINARY COST ESTIMATE

      Based on the capacities and regeneration techniques developed during the column
studies, separate cost estimates have been prepared for the removal of either Se(IV) or
Se(VI) from a ground water similar in chemical composition to the water used during
these tests.  Main design criteria used to develop the costs include:

      Average Flow - 1 mgd
      Se(IV) or Se(VI) concentration - 0.10 mg/1
      Capacity of Activated Alumina at pH6:
          Se(IV)     -    90 mg/1
          Se(VI)     -    7 mg/1
      Minimum Duration of Treatment Run - 2k hours
      Regeneration:
          NaOH - minimum flow  through time with 0.5% NaOH of 1.5 hours
             @ 1 gpm/ft2
           H-SO^ or HC1 - same volume of NaOH with a concentration
             of 0.25% at 2 gpm/ft2

     The system considered would be very basic, with no automatic control systems.
Regeneration would be done  manually.   Capital  costs  include  all manufactured
equipment, activated alumina,  piping and valves,  electrical  and instrumentation, a
small building for chemical storage and other operations, and contingencies.  A small
clearweil was provided to supply  sufficient  regeneration water.   Land acquisition is
included.

     Annual operation and mantenance costs  include:
                                      139

-------
           chemicals for regeneration and pH adjustment
           labor @ $15,000/yr
           media replacement (9 0.5% per regeneration
           electrical energy (d 7£/kw-hr

     Current costs used were:
           50% NaOH = $175/ton
           28% HC1 = $65/ton
           100% H2SOlt = $60/ton
           Activated alumina = $0.60/lb

     To develop costs on a cents per 1000 gallons produced (£/1000 gal) or dollars per
acre-foot ($/ac-ft.) basis, estimated capital costs were amortized over the life of the
equipment, estimated to be 20 years.  An annual interest rate was assumed to be 10%.
Table 19 lists these estimated costs.

      As can be seen, ground water contaminated with only Se(VI) will cost much more
to treat than will water with Se(IV) present.  If  there is an appreciable amount of
Se(VI) (>10 ppb) present in a water with Se(IV), then the removal of  selenium will be
dependent on how well Se(Vl) can be removed during treatment.
                                   TABLE 19
          COST ESTIMATE FOR Se(lV) OR Se(VI) REMOVAL FACILITIES
                                                           Cost
        	    Item	          Se(IV)          Se(VI)
        Amortized Capital Costs ($/yr.)            $27,000        $115,000
        Operation and Maintenance ($/yr.)          53,000         1^5,000
        Total Annual Costs ($/yr.)                 $80,000        $ 260,000

        Cost to Treat 100% Se(IV) =  23
-------
      Because the reduction of Se(VI) to Se(IV) requires very vigorous reduction, sulfur
dioxide or other mild reducing agents could not be used for this purpose.  It would be
virtually impossible to chemically pretreat a drinking water supply to reduce Se(VI) to
Se(IV).

      The estimated costs, prepared for this report, are based solely on nine-inch deep
columns of alumina. It is quite possible that deeper beds may show a greater capacity
for both species of inorganic selenium.  To develop truly accurate  cost  estimates,
capacities would have to be developed from pilot scale testing.  This testing  would also
develop other engineering parameters that would allow the rapid determination of the
feasibility  for using activated  alumina  to  remove selenium  from  drinking  water
supplies.
                                       141

-------
                                 REFERENCES
1.   Committee  on  Medical  and Biologic  Effects of  Environmental  Pollutants,
     Selenium, National Academy of Sciences, Washington, D.C. (1976) 203 pp.

2.   Ball,  R.  "Removal  of  Selenium from Drinking Water Supply Using  Activated
     Alumina." Senior Thesis, University of Cincinnati, Ohio, 1977. 62 pp.

3.   Anderson,  M.S.,  Lakin,  H.W.,  Beeson,  K.C., Smith,  F.F., and Thacker,  E.
     Selenium in  Agriculture, Agriculture Handbook #200, U.S. Dept. of  Agriculture,
     1961,65pp.

4.   Klayman, D.L.  and  Gunther, W.H.H.   Organic  Selenium Compounds;  Their
     Chemistry and Biology.  Wiley-Interscience, New York, New York,  1973, 1188 pp.

5.   Sorg, T.3.,  and  Logsdon, G.S.  "Treatment Technology  to  Meet  the  Interim
     Primary Drinking Water Regulations for Inorganics:  Part 2." 3.  AWWA.  3uly,
     1978, pp. 379-396.

6.   Gupta,  S.K. and Chen,  K.Y.  "Arsenic Removal by Adsorption."  3. WPCF.
     March,  1978, pp. 493-506.

7.   Wu, Y.C.  "Activated Alumina Removes Fluoride Ions From Water."  Water and
     Sewage Works 3.  3une, 1978,  pp.  76-82.

8.   Ames, L.L.  and  Dean,  R.D.  "Phosphorus Removal From Effluents in Alumina
     Columns." 3. WPCF. May, 1970, pp. R161-R172.

9.   Bellack, E. "Arsenic Removal from Potable Water." 3. AWWA, 3uly 1971, pp 454-
     458.

10.   Choi, W.W. and Chen, 3.Y. "The Removal of Fluoride from Waters by Adsorption"
     3. AWWA October, 1979, pp. 562-570.

11.   Clifford, Dennis, Assoc.  Prof, of Civil Engineering,  Univ. of  Houston, Texas
     Personal Communication, 1980.

12. . Clifford, D., Matson, 3., and Kennedy, R.  "Activated  Alumina: Rediscovered
     'Adsorbent1 for Fluoride, Humic Acids, and Silica" University of Houston, Texas,
     1979.
                                      142

-------
13.  Rubel, F. and Woosley, R.D.  "Removal of Excess Fluoride from Drinking Water."
     EPA  570/9-78-001, U.S. Environmental  Protection  Agency, Washington, D.C.
     1978.  16pp.

14.  Perry,  R.H. and  Chilton,  C.H.   "Adsorption and  Ion Exchange"  Chemical
     Engineers'  Handbook,  5th  Edition, McGraw-Hill  Book  Co., New  York, N.Y.
     pp 16-1 to 16-10.

15.  Bellar, T.A. and Lichtenberg, 3.3. "Determining Volatile Organics at Microgram-
     Per-Litre Levels by Gas Chromatography."  J.AWWA, Dec., 1974 p 739.

16.  Raihle, J.A.  "Fluorometric Determination of Selenium in Effluent Streams with
     2,3 - Diaminonapthelene."   Environmental Science and Technology.   Vol.  6:7,
     pp. 621-622, 1972.

17.  Martin, T.D.,  Kopp, J.F., and  Ediger, R.D.  "Determining  Selenium  in  Water,
     Wastewater,  Sediment,  and   Sludge   by   Flameless  Atomic   Absorption
     Spectroscopy."  Atomic Absorption Newsletter. Vol. 14:5 pp. 861-868, 1975.

18.  APHA-AWWA-WPCF.   Standard Methods for  the Examination of  Water and
     Wastewater, 14th  Edition.    American Public Health  Association, Washington,
     D.C., 1975, pp.  237-242.


19.  Zingaro, R.A. and Cooper,  W.C.  Selenium.  Van  Nostrand Reinhold Co., New
     York, New York, 1974  835 pp.

20.  Weber,  W.3., Physicochemical  Processes for Water  Quality Control.   Wiley-
     Interscience, New York, New York, 1972.  640 pp.

21.  National Research  Council,  Drinking Water  and Health, National Academy  of
     Sciences, Washington, D.C.  (1977) 939 pp.

22.  U.S. Environmental Protection Agency, Office of Water Supply, National Interim
     Primary Drinking Water Regulations (1977), EPA-570/9-76-003.

23.  Hadjimarkos, t».M. "Selenium in Relation  to Dental Cavies."  Food and Cosmetics
     Toxicology 11(6), 1083-1095 (1973).

24.  Lakin, H.W. "Selenium in Our Environment." Trace  Elements in the Environment
     (1972), pp. 96-111.

25.  Kubli, H. "On the Separation of Anions by Adsorption on  Alumina"  Helvetica
     Chemica Acta. Vol. 30, No.  2 , 1947, pp. 453-463.

26.  Geering, H.R., Gary, E.E., Jones L.H.P., and Allaway, W.H.  Solubility and Redox
     Criteria for  the Possible Forms  of Selenium in Soils.  Soil Sci. Soc.  Amer. Proc.
     32: 35-40, 1968.
                                      143

-------
27.   Stumm, W. and Morgan,  3.3.  Aquatic Chemistry; An Introduction Emphasizing
     Chemical Equilibria in Natural Waters, Wiley-Interscience, Ne,w York, 1970  593
     pp.

28.   EPA "Estimating Water  Treatment Costs: Volume 2, Cost Curves Applicble for
     1 to 200  mgd Treatment Plants."   EPA-600/2-79-1626,  U.S.   Environmental
     Protection Agency, Cincinnati, Ohio 1979, 506 pp.

29.   Trussell, R.R.,  "A Summary of  Fluoride Removal Technlogy" presented at the
     National Defluoridation Meeting, November 8, 1977, Dallas, Texas.
                                      144

-------
                                 APPENDIX A
                        MATERIALS AND INSTRUMENTS

     All of  the analytical  instruments  that were  used have been  described  in
Chapter 4, except for an Orion pH meter. A list of materials includes:

     1.   Three Cole-Palmer peristaltic pumps, model #7015.

     2.   Three variable-speed motors with speed controls,  1/30 hp  with an output
          gear ratio of 12:1 and a maximum speed of  290 rpm, distributed by Minarik
          Electric Company, Los Angeles.

     3.   Three automatic samplers, Wastewatcher II,  with  24-200 ml conventional
          polyethylene sample bottles each, manufactured  by  Raymond  Jensen
          Company, Los Angeles.

     it.   One  six paddle stirrer, manufactured by Phipps and Bird, Inc., Richmond,
          Virginia.

     5.   Three 15-inch glass columns of 1-inch inside diameter.  Fitted with ground
          glass  joints with ball and socket clamps and  stopcocks.

     6.   Three conventional, nonlinear polyethylene tanks, 55 each.

     7.   Tygon tubing, 3/16 inch I.D., 3/8 inch O.D.

     8.   Teflon tubing, 3/16 inch I.D.,  I/* inch O.D.

     9.   Swagelock fittings, 1/4-inch.
                                      145

-------
10.   Polyethylene   sample  containers  (1-  and  4-liter),  "Qubetainers"
     manufactured by Hedwin Corporation, Laporte, Indiana

11.   ALCOA activated alumina, type F-l, 14-28, 28-48, and 48-100 mesh.

12.   Reagent grade  concentrated HC1,  H-SO^,, and HNO, and 50 percent
     NaOH.
                                  146

-------
                                   TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
1. REPORT NO
  EPA-600/2-80-153
                                                           3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
  SELENIUM REMOVAL FROM GROUND WATER USING
  ACTIVATED ALUMINA
                                                           5. REPORT DATE
                                                           August 1980 (Issuing Date)
                                                           6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
  R.  Rhodes Trussell,  Albert Trussell, Peter Kreft
                                                           8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
  James M. Montgomery Consulting Engineers,  Inc.
  555 East Walnut Street
  Pasadena, California  91101
                                                           10. PROGRAM ELEMENT NO.

                                                            C110 1CC824 Task 50
                                                            11. CONTRACT/GRANT NO.

                                                            68-03-1515
12. SPONSORING AGENCY NAME AND ADDRESS
  Municipal Environmental Research Laboratory—Cin.,OH
  Office of Research and Development
  U.S. Environmental Protection Agency
  Cincinnati, Ohio  45268
                                                            13. TYPE OF REPORT AND PERIOD COVERED
                                                            Research 8/78 -  6/80
                                                            14. SPONSORING AGENCY CODE
                                                            EPA/600/14
15. SUPPLEMENTARY NOTES
  Project Officer:  Richard P.  Lauch  (513) 684-7467
         Laboratory  studies were performed  to  determine optimum conditions for
    using activated  alumina to remove selenium from drinking water supplies.   Column
    tests showed that  the capacity of alumina  for Se IV decreased as  influent pH
    increased.  Best removal of Se IV occurred at pH 5 where 1200 bed volumes of water,
    with influent concentration of 0.2 mg/L, were treated before breakthrough (Se
    cone. >0.01 mg/L)  occurred.  Optimum regenerant was 1.5 lbs/ft3 of 0.5% NaOH at
    1/2 gpra/ft* upflow and 0.7 lb/ft3 of 0.25% H,S04 at 1 gpm/ft^ downflow.  The
    capacity of alumina for Se VI decreased as either pH or sulfate concentration
    increased.  Removal costs would be three to four times higher if  Se VI is the
    predominant form of selenium present in water supply.

         Tests showed  that the flourometric technique can be used to  determine Se IV
    concentrations.  Atomic Absorption Spectroscopy remains the best  method for total
    selenium and Se  VI plus organic selenium concentratiions can be determined by
    difference.

         Information from this project should  be useful to research and design
    engineers, and state personnel concerned with the removal of selenium from ground
    water.
17.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                              b.lDENTIFIERS/OPEN ENDED TERMS  C.  COSATI Field/Group
 Potable Water
 Water  Treatment
 Adsorption
 Selenium Removal
 Activated Alumina
                                                                             13B
                                                                          21. NO. OF PAGES
                                                                             159
18. DISTRIBUTION STATEMENT

 Release  to Public
19. SECURITY CLASS (ThisReport)
  Unclassified
                                              20. SECURITY CLASS (This page)
                                                Unclassified
                                                                         22. PRICE
EPA Form 2220-1 (R.v. 4-77)
                                            147
             U.S. GOVERNMENT PRINTING OFFICE: 1980--657-165/0104

-------