United States
Environmental Protection
Agency
Environmental Research
Laboratory
NarraganMtt Rl 02882
EPA-600/3-80-011
January 1980
Adsorption of Trace
Metals by  Hydrous
Ferric Oxide in
Seawater

-------
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                        EPA-600/3-80-011
                                        January 1980
    ADSORPTION OF TRACE METALS BY
   HYDROUS FERRIC OXIDE IN SEAWATER
  K. C. Swallow and Francois Morel
     Ralph M. Parsons Laboratory
For Water Resources and Hydrodynamics
   Department of Civil Engineering
Massachusetts Institute of Technology
   Cambridge, Massachusetts 02139
           Grant R-803738
           Project Officer

            Earl W. Davey
  Environmental Research Laboratory
  Narragansett, Rhode Island  02882
   ENVIRONMENTAL RESEARCH LABORATORY
  OFFICE OF RESEARCH AND DEVELOPMENT
 U.S. ENVIRONMENTAL PROTECTION AGENCY
  NARRAGANSETT, RHODE ISLAND  02882

-------
                                 DISCLAIMER
       This report has been reviewed by the Environmental Research Laboratory,
Narragansett, 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 conmerclal products constitute endorsement or
recommendation for use.
                                     ii

-------
                                  FOREWORD
      The Environmental Research Laboratory of the U.S. Environmental
Protection Agency is located on the shore of Narragansett Bay, Rhode Island.
In order to assure the protection of marine resources, the laboratory is
charged with providing a scientifically sound basis for Agency decisions on
the environmental safety of various uses of marine systems.  To a great
extent, this requires research on the tolerance of marine organisms and their
life stages as well as of ecosystems to many forms of pollution stress.
In addition, a knowledge of pollutant transport and fate is needed.

     This report describes a three-year project aimed at improving modelling
capabilities for the fate of metallic waste constituents in coastal waters.
The particular focus of the report is the study of metal adsorption onto
hydrous ferric oxide.


                                            Donald  K.  Phelps
                                            Acting  Director, ERLN
                                     iii

-------
                                   PREFACE

     Assessment of the environmental impact of activities which alter the
chemical characteristics of a body of water is made especially difficult by
the great number and complexity of the possible reactions through which a
chemical can be mobilized or precipitated and can take biologically active—
sometimes toxic—or inactive forms.  One approach to such a problem is to
build reasonable models of the systems under study by assembling all relevant
thermodynamic, kinetic and analytical information.  In this fashion at
least "chemical boundary conditions" can be established that permit the
isolation of the essential processes and point the way to relevant research
on smaller and more manageable experimental systems.

     This report describes a three year project aimed at improving our
modelling capabilities for the fate of metallic waste constituents in
coastal waters.  The particular focus of the report is the study of metal
adsorption onto hydrous ferric oxide.  This process was singled out as the
one whose ignorance was most responsible for the Inadequacies of our present
thermodynamic modelling of the metal chemistry in waste fields.

     As part of the same project, new computer programs for chemical
calculations have been developed.  These (particularly the M1NEQL series)
have been widely distributed among universities, industry and government
research centers and are used routinely for environmental studies.  A
particular focus of the modelling effort has focussed on the question of
alternative models for adsorption of solutes at solid water interfaces.

     The following reports and publications should be consulted for specifics
regarding these various models:

     MINEQL, A Computer Program for the Calculation of Chemical Equilibrium
      Composition of Aqueous Systems.  J.C. Westall, J.L. Zachary, F.M.M.
      Morel.  1976.  Water Quality Laboratory, Ralph M. Parsons Laboratory
      for Water Resources and Environmental Engineering, Department of
      Civil Engineering, Massachusetts Institute of Technology, Cambridge,
      MA 02138.  Technical Note No. 18.

     The Use and Abuse of MINEQL-II.  J.L. Zachary.  1977.  Ralph M. Parsons
      Laboratory for Water Resources and Environmental Engineering,
      Department of Civil Engineering, Massachusetts Institute of Technology,
      Cambridge, MA 02138.

     Morel, F.M.M., J.G. Yeasted, J.C. Westall.  "Adsorption Models:  A
      Mathematical Analysis in the Framework of General Equilibrium
      Calculations", In:  M.A. Anderson and A. Rubin [eds.], Adsorption of
      Inorganics at the Solid-Liquid Interface.  Ann Arbor Science, Ann Arbor,

                                     iv

-------
 Michigan.  1979 (in press).

Chemical Equilibrium Including Adsorption on Charged Surfaces.  J.C.
 Westall.  1979 (in press).  Swiss Federal Institute of Technology,
 EAWAG, CH-860Q, Duebendorf, Switzerland.

-------
                                  ABSTRACT

    The adsorption of trace metals by amorphous hydrous ferric oxide in sea-
water is studied with reference to simple model systems designed to isolate
the factors which may have an effect on the isotherms.  Results show that
the complex system behaves in a remarkably simple way and that the data ob-
tained under various conditions of total metal concentration and total oxide
concentration can be reduced to an apparent reaction constant, K, which is a
function of pH only.  The high capacity of the oxide for trace metals renders
the concept of a surface reaction useless to explain the uptake of metals.  A
physical picture of the oxide as a swollen hydrous gel permeable to hydrated
ions is presented.

    This report was submitted in fulfillment of Grant Number R-803738 under
the sponsorship of the Environmental Protection Agency.  The work cited
in this report was completed as of April 1978 and is included in the Ph.D
thesis of Kathleen C. Swallow (Massachusetts Institute of Technology, 1978).
This work has also been submitted (July 1979) for publication to the journal,
Environmental Science and Technology in a paper entitled, "Sorption of Copper
and Lead by Hydrous Ferric Oxide" by K.C. Swallow, D.N. Hume and F.M.M. MoreL
                                     vi

-------
                               CONTENTS






Foreword ----------------------------------------------------------  ill




Preface  ----------------------------------------------------------   iv
         — • — •• — •• — —•. ,.„ — •_•!,- — — — —• — — • — ---- --- -- — — --- -- -i -- ——————— — ----- ^T-  TT^ 4 ^
                     BI™™—!.      ^      ^  — — «   ^ ^     ^  ^ —   — _  VJ.J.X




Tables   —————————— ____ — _ — __ — ___ ————————    xl




   1*  BacK^rouiiu  — — — — — — — _ — — _ — — _ — « — — — — — — — — — — — — —     ^




             Introduction ----------------------------------------     1




   2.  Experimental Materials and Methods ------------------------    13




             Materials ——————————————————————    13




             Methods  —————— ——————— —————————    15




   3.  Experimental Results -------------------------------------    19




             Potentiometric Acid-Base Titrations ------------------    19




             Effect of Ionic Strength ---------------------------    22




             Effect of Composition of the Background Electrolyte  ---    22




             Effect of variation of metal and oxide concentrations —    22




             Competition between metals for the hydrous  ferric oxide    39




   4.  Discussion  -------------------------------------------------    44




References —————————————————————————————————    50
                                  vii

-------
                                   FIGURES



Number                                                                    Page



  1    Adsorption in the electrical double layer 	  8



  2    Experimental apparatus	16

                                 _3
  3    Titration curve for 1 x 10  M amorphous hydrous ferric oxide



       in 0.5 M NaClO^	20


                       +                                      -3
  4    Surface excess H  or OH  as a function of pH for 1 x 10  M



       hydrous ferric oxide 	 21


                                                           -4
  5    Changes with aging in the titration curve for 1 x 10  M



       hydrous ferric oxide in S.O.W.	 23


                                2+                    -5    2+
  6    Effect of aging on the Cu   isotherm for 1 x 10  M Cu   on



       1 x 10~4M hydrous ferric oxide in S.O.W, 	 24


                                         2+                    -5
  7    Effect of ionic strength on the Cu   isotherm for 1 x 10  M


         2+          -3
       Cu   on 1 x 10  M hydrous ferric oxide prepared in the



       different background electrolytes 	  25

                                             2+                    -5
  8    Effect of the ionic strength on the Cu   isotherm for 1 x 10  M


         2+          -4
       Cu   on 1 x 10   hydrous ferric oxide prepared in one large



       batch and diluted with the appropriate background electrolyte 	  26


                                         2+                     S
  9    Effect of ionic strength on the Pb   isotherm for 1 x 10  M

                  c    f\ i           I

       for 1 x 10~ M Pb   on 1 x 10~ M hydrous ferric oxide prepared



       in one large batch and diluted with the appropriate background



       electrolyte	27


                                                                 2+
 10    Effect of the background electrolyte composition on the Cu




                                      viii

-------
                         -5    2+          -3
      isotherm for 1 x 10  M Cu   on 1 x 10  M hydrous ferric



      oxide prepared in different background electrolytes 	 22


                                                            2+
11    Effect of background electrolyte composition on the Cu


                         -5    2+          -4
      isotherm for 1 x 10  M Cu   on 1 x 10  M hydrous ferric



      oxide prepared in one large batch and diluted with the



      appropriate background electrolyte •	 29


                                                            2+
12    Effect of background electrolyte composition on the Pb


                         -5    2+          -4
      isotherm for 1 x 10  M Pb   on 1 x 10  M hydrous ferric oxide



      prepared in one large batch and diluted with the appropriate



      background electrolyte 	 30


                          -5    2+          -3
13    Isotherms for 1 x 10  M Cu   on 1 x 10  M hydrous ferric oxide



      under various conditions of ionic strength and background



      electrolyte composition obtained with different batches of



      hydrous ferric oxide prepared in the different background



      electrolyte	 32



14    The ratio of Cuad"sorbed/Cufree as a functlon of PH for variable

                   -4
      CuT on 1 x 10  M hydrous ferric oxide	 33


                 2+
15    Moles of Cu   adsorbed per mole hydrous ferric oxide as a function



      of pH for variable Cu  on 1 x 10~ M hydrous ferric oxide	34


                  2+         2+
16    The ratio Cuadsorbed/Cufree fls a function of pH for variable



      CuT on 5.0 x 10~ M hydrous ferric oxide	 35


                 2+
17    Moles of Cu   adsorbed per mole hydrous ferric oxide as a function



      of pH for variable CuT on 5.0 x 10~ M hydrous ferric oxide 	 36


                     2+         2+
18    The ratio of Pbadsorbed/Pbfree as a function of pH for variable



      PbT on 5.0 x 10~ M hydrous ferric oxide	 37
                                      ix

-------
                                                                        Page
                 2+
19    Moles of Pb   adsorbed per mole hydrous ferric oxide as a



      function of pH for variable PbT on 5.0 x 10  M hydrous



      ferric oxide 	  38


                 24*
20    Per cent Cu   adsorbed as a function of pH for various



      concentrations of hydrous ferric oxide in S.O.W. 	  40


                 2+
21    Per cent Pb   adsorbed as a function of pH for various



      concentrations of hydrous ferric oxide in S.O.W. 	   41


                                                   -5    2+
22    Reduction of data from isotherms for 1.0 x 10  M Cu   on



      various concentrations of hydrous ferric oxide in S.O.W. to



      Cu2*   , j/Cu2+  Fe,,, as a function of pH	   42
        adsorbed   free  T                  *

                                                   -5    2+
23    Reduction of data from isotherms for 1.0 x 10  M Pb   on



      various concentrations of hydrous ferric oxide in S.O.W. to



      Pt»2j   i. j/Pb£+  FeT as a function of pH	   43
        adsorbed   free  T                  K

                                             -5    2+            -4
24    Depression of the isotherm for 1.0 x 10  M Cu   on 4.0 x 10  M



      hydrous ferric oxide in S.O.W. in the presence of 6.0 x 10~ M



      Fe2+	   45


                                          2+
25    Log K as a function of pH for all Cu   data	   47
                                   TABLES



Number                                                                    Page



 1   The Composition of S.O.W.	  14

-------
                                  SECTION 1

                                 BACKGROUND
INTRODUCTION

     A comparison of the measured concentrations of trace metals in seawater
with calculations of the solubilities of the least soluble compounds which
can be formed from trace metals and the major components of seawater has led
to the conclusion that seawater is undersaturated with respect to many trace
elements (1,2,3).  Various mechanisms for removal of trace metals to the sedi-
ments have been proposed as the controlling factor for their concentrations in
seawater.  Krauskopf (1) performed a series of simple experiments to test the
efficiency of a number of processes for removing trace metals to a solid phase:
precipitation as the least soluble species, precipitation by sulfides and ad-
sorption by ferrous sulfide in reducing environments, and adsorption on
hydrous ferric oxide, hydrous manganese oxide, apatite, clay and organic
matter.  He concluded that adsorption on one or more of these substrates could
account for the low aqueous concentrations of eight of thirteen metals studied.
In particular, hydrous ferric oxide removed 95% of the Zn, 98% of the Cu, 92%
of the V, 86% of the Pb, 89% of the Co, and 80% of the W from seawater solu-
tions with initial concentrations of 0.1 to 10 ppm metal.

     Although the adsorption of metals on hydrous ferric oxides has been in-
vestigated (4,5), the experimental data exist mainly for simple, carefully
controlled systems designed to allow for maximum insight into the process of
adsorption at the microscopic level.  Isotherms were generated for one metal
adsorbing on a crystalline hydrous ferric oxide surface in the presence of an
inert background electrolyte, usually at low ionic strength.  Adsorption in
seawater, on the other hand, involves many adsorbing metals adsorbing onto an
ill-defined, possibly largely amorphous surface in the presence of a complex
background electrolyte at high ionic strength.  Any or all of these factors
may make it unfeasible to extrapolate data from the simpler systems to sea-
water.

Factors Which May Affect Adsorption of Trace Metals by Hydrous Ferric Oxide in
Seawater

The Nature of the Hydrous Ferric Oxide—

      In a study of the effects of ionic strength, temperature and Fe(III) con-
centration on the hydrolysis and precipitation of ferric oxides from ferric
nitrate solution, Dousma and deBruyn  (6) found that 6.25 x 10~3M Fe(III) at
24°C  produced an amorphous ferric oxide.  Higher temperature and Fe(III) con-
centration were necessary to produce  geothite  (a FeOOH).  This effect of

-------
temperature had earlier been reported by Kolthoff and Moscovitz (7) who also
noted that the surface area of the crystalline oxide was smaller than that of
the amorphous oxide and that the former adsorbed three to four times less
copper than the latter.

     In general, slow hydrolysis and/or elevated temperature are required to
produce the crystalline ferric oxides.  Hematite (Fe203) is usually prepared
by boiling solutions of ferric nitrate at low pH under reflux for 18 days
(8).  Goethite is obtained by aging a solution of ferric nitrate at pH 12 for
24 hours at 60° (9).  These methods produce particles of considerable uni-
formity in size and shape whose composition depend most strongly on the pH
and the anion present in the solution.

     Gadde and Laitinen (10) prepared amorphous hydrous ferric oxide by ad-
justing the pH of a 0.1M ferric nitrate solution to 6 and allowing the pre-
cipitate to settle for 2 hours at room temperature.  After filtering, washing
and aging overnight in distilled water at pH 6, the suspension was used for
the experiments.  X-ray powder diffraction confirmed that the oxide was
amorphous.

     Matijevic reports that the background anions, although responsible for
the particle characteristics, are not found in the solid phase.  Ellis et al.
(11), on the other hand, found that Cl~ ion is specifically adsorbed by
3 FeOOH and cannot be removed completely even with extensive washing with
distilled water or by ion exchange.

     An extensive investigation of the slow hydrolysis of partially neutral-
ized Fe(III) solutions in the nitrate, chloride and perchlorate solutions
was carried out by Murphy, Posner and Quirk (12).  They found that the initial
polymerization process yields spherical polycations independent of the anion
present.  The anion does affect the subsequent aging and crystalline structure
of the precipitate as it progresses from the spherical polycations to rods to
rafts.  Rods are formed from linear coalescence of spherical polycations and
addition of unpolymerized ferric species in solution.  Both rods and spheri-
cal polycations are X-ray amorphous.  Rafts result from lateral coalescence
of rods and show the X-ray pattern of geothite or 3 FeOOH.

     In nitrate and perchlorate solutions a FeOOH (goethite) predominates,
while in chloride solutions 3 FeOOH is formed.  The formation of 3 FeOOH in
chloride solution is thought to arise from the penetration of the Cl~ ion
into the polycations resulting in a different internal structure when the
polycations coalesce.  The Fe(III) concentration does not affect the crystal-
line structure1 in chloride solutions and increased ionic strength only in-
creases the rate of precipitate formation.  In more dilute solutions both
nitrate and perchlorate preparations also contained traces of y FeOOH (lepido-
crocite).  High ionic strength or anion affinity for the Fe3+ ion inhibit the
formation of this crystalline form.

     In a study of Fe(III) speciation in seawater, Byrne and Kester (13) note
that the X-ray amorphous solid phase formed initially in solutions supersat-
urated with respect to Fe(OH)-j is slowly transformed to a more stable crystal-
line solid phase.  This transformation is extremely slow, however, and after

-------
several months of aging a substantial portion of the solid phase still exists
as the amorphous oxide.  Dousma and deBruyn had also noted this slow trans-
formation, but found that when the X-ray amorphous solid initially formed in
nitrate solution was studied by infrared techniques, the presence of a FeOOH
(geothite) was suggested.  It has been proposed that the iron oxide found in
seawater, while X-ray amorphous, actually has a largely geothite nature (14).

     The ferric oxide in seawater may have originated from terrestrial sources
or may have been precipitated in situ.  Since an in situ precipitation would
occur in the presence of the potentially adsorbing trace metals, the question
is raised whether coprecipitation will be the same as adsorption on a pre-
viously formed surface.  Kolthoff and Moskovitz (7) found that if the oxide
was formed at room temperature (amorphous), there was only a slight increase
in the amount of copper coprecipitated versus copper adsorbed.  With pre-
cipitate formed at 98°C, however (crystalline), the differences were more
pronounced.  Kurbatov, Kulp and Mack  (15) found that a higher percentage of
strontium and barium was coprecipitated than was adsorbed on an amorphous
oxide, but that after several days of standing at pH 8 the amounts of adsorbed
and coprecipitated metal were identical.  Gadde and Laitinen (10) found that
more lead was adsorbed on hydrous ferric oxide if it was present in the solu-
tion during the precipitation than if it was added immediately after the pre-
cipitation.  Aged hydrous ferric oxide adsorbed less lead than fresh, and
lead adsorbed on the aged oxide was more efficiently recovered, suggesting
that some occlusion occurs in the coprecipitation process.  In a later paper,
however (16), under the same conditions they report that the effect of aging
of the hydrous ferric oxide, either in the presence of lead or before it was
added, on the amount of lead adsorbed was not pronounced.

The Complexity of the Background Electrolyte—
     The inert background electrolyte used in adsorption experiments usually
consists of a dilute solution of a simple salt whose ions have no specific
chemical interaction with the surface.  Seawater is a complex solution which
contains high concentrations of the divalent ions Ca^+ and Mg^+ which may
specifically adsorb on the hydrous ferric oxide and out-compete a less con-
centrated trace metal for the available surface sites.  Alternatively,
ligands such as Cl~ that form stable  complexes with the adsorbable metal may
outcompete the surface for the metal.  In either case, a reduction in adsorp-
tion of trace metal would be expected.

     McNaughton and James (17) reported no exchange or competition between
Na+ or Mg2* and the hydrolyzed Hg(II) species adsorbed on aQuartz.  On the
other hand, O'Connor and Kester (18)  found that Mg in a concentration equal to
that in seawater suppresses Co2+ adsorption onto illite, but not Cu   adsorp-
tion.  It is possible that adsorption on the clay, illite, and the oxide
aQuartz occur through different mechanisms and that results from one surface
are not relevant to the other.

     The effect of competing ligands has been interpreted in several different
ways.  Forbes, Posner and Quirk (4) interpreted the failure of HgCl2 to adsorb
on goethite as an indication that coordinated hydroxyl groups play a vital
role in adsorption.  The same reasoning was used by McNaughton and James to
explain the same result for HgCl2 on aQuartz.  They had found that the abrupt

-------
authors have expressed doubt as to the applicability of this method to ad-
sorption on hydrous oxides.

     There is no reason to expect that adsorption of a gas occurs via the
same mechanism as specific chemical adsorption of ions from solution.  In
addition, removing the solid from suspension and drying it in order to per-
form a BET determination may drastically affect the physical configuration
of the solid and alter the surface.

     Another technique which is sometimes used involves the adsorption of a
dye, such as wooly violet, followed by removal of the solid from the sus-
pension and colorimetric measurement of the residual dye in solution.  Again
the adsorption of large dye molecules is not necessarily the same as adsorp-
tion of metal cations, but this technique can be used on an aqueous suspension
of the solid.

     When the surface charge is attributed to specific sites on the surface,
the available reaction sites may be determined by acid-base titration of the
surface.  This assumes that all sites which react with acid and base will
also react with adsorbing metals.

     Regardless of the physical picture of surface charge distribution, the
determination of the available surface on an amorphous hydrous ferric oxide
suspended in the medium in which it was precipitated is difficult to carry
out.  B.E.T. determinations cannot be made on solution; dye methods, while
feasible, are not likely to be appropriate; and acid-base titrations are
difficult to interpret because of the unknown concentration of strong acid
or base already present in the suspension after precipitation of the oxide.

     If the available surface is limited, it should be possible to demonstrate
that the amount of metal adsorbed on a given amount of oxide reaches a limit-
ing or plateau level above which an increase in the equilibrium concentration
of the metal does not result in further adsorption.  James and Healy (21)
found that for Co^+ adsorption on Si02, isotherms plotted as adsorbed Co2+
versus free Co^+ did not reach a plateau as such.  With increasing Co^+ the
isotherms converged toward the region where free Co^+ concentration was
limited by the solubility of Co(OH)2-  They estimated a maximum adsorption
density corresponding to bare Co^+ ions being separated by two to four water
molecules.

     It was pointed out by James (25), however, that the adsorption and pre-
cipitation processes are apparently independent since a smaller percent of
the total (k>2+ is adsorbed with increasing added metal, which is opposite to
the concentration dependence of precipitation and polymerization.

     Further evidence that adsorption and precipitation are independent pro-
cesses is found in very early work by Kurbatov (26) on the preparation of
"highly emanating preparates".  These are preparations of hydrous ferric
oxide containing adsorbed Ra used to obtain nuclear emanation in exactly
measured quantities.  He found that exceeding the Ksp of Ra or Ba salts re-
sulted in preparates of diminished radioactivity.  This was attributed to
the formation of microscopic crystals of the salts which resulted in less Ra

-------
rise in the Isotherm for Hg(II) adsorption on aQuartz corresponded to the pH
at which Hg(OH)2 predominates in the speciation diagram.  The correlation
between adsorption and hydrolysis led Matijevic (20) to the conclusion that
the hydroxyl group is responsible for the adsorption of ions on colloid
particles.  He postulated that hydrogen bonding may play a decisive role.

     James and Healy (21) rejected the idea that adsorption and hydrolysis
were related because of the presence of the -OH group.  Instead, they suggest-
ed that in lowering the charge on the central cation, hydrolysis made it
easier to disrupt the secondary hydration sheath of the metal cation because
the charge dipole interactions were weakened.  The disruption of the secondary
hydration sheath is necessary if an adsorbing cation is to move to within one
water molecule of the surface, their definition of specific adsorption.  To
test James and Healy1 s interpretation of the correlation between hydrolysis
and adsorption, Stanton and Burger (22) used acetate and phosphate ions to
reduce the charge of Zn2+.  Acetate had little effect, but phosphate enhanced
the adsorption of Zn^+ on amorphous iron and aluminum oxides.  They interpret-
ed the different results for the two anions as indicative of bridging between
the zinc ions and the oxide surface by the phosphate ions, similar to the
bridging role played by hydroxyl ions.  Acetate could not form oxo bridges,
so had no effect.
     In ammoniacal media, Kolthoff and Moskovitz  (7) found that increased
concentration decreased Cu^+ adsorption on hydrous ferric oxide and attributed
it to lowering of the OH~ concentration by the Nlty"1".  Increased NH3 concentra-
tion, while increasing OH~ concentration, also reduced Cu?+ adsorption, how-
ever.  This effect was due to the formation of the stable Cu2+ -NH^ complexes
which apparently did not adsorb.

     With illite, O'Connor and Kester (18) found  less copper adsorbed at a
given pH from seawater or 0.7M NaCl than from an  artificial river water
solution.  There was no difference if the system  was C02  free or contained
a carbonate buffering system, however.  Vuceta (23) also found that a carbon-
ate buffering system neither enhanced nor depressed adsorption of Cu   and
     on aQuartz.  In both cases the HC03~ concentrations were in the range
of the concentration in natural waters and the C02 was atmospheric.  Under
these conditions, the carbonate complexes of the metals are not the predomin-
ant species.

The Presence of Many Potentially Adsorbing Trace Metals and Limited Surface
for Adsorption
     Much of the existing data for adsorption of metals on hydrous oxides has
been obtained with an excess of adsorbing surface and one metal in solution.
The meaning of available surface is not clear and has been measured or defined
by various methods.

     If the charge on the surface is pictured as evenly distributed over the
entire surface area, it is common to assume that the limit of adsorption
density is surface monolayer coverage.  This value is usually obtained by
measuring available surface area per amount of solid with the B.E.T. gas
adsorption method (24) or a similar technique.  Then from calculations based
on hydrated ionic radii, a maximum adsorption density is obtained.  Various

-------
or Ba being uniformly distributed throughout the hydroxide.

     Gadde and Laitinen (16) were able to demonstrate a plateau value for
Pb^+ adsorption on amorphous hydrous ferric oxide at pH 6.  Their value was
0.28 moles Pb^+ per mole of Fe(III).  They could not demonstrate a plateau
at pH 5 or for Pb^+ on hydrous manganese oxide.  Cd  , Zn^+ and Th^" reached
a limiting plateau on amorphous hydrous manganese oxide at about 0.2 mole/
mole Mn at pH 6.  The values for the plateau levels were found to be pH-
dependent, decreasing with decreasing pH.

                                                                         2+
     With a hematite suspension, Breeuwsma and Lyklema (27) found that Ca  ,
Mg2+, Sr^+, and Ba^+ were specifically adsorbed and shifted the pH of the
suspension toward lower pH until it reached pH 6.5 in 10~% salt solution.
No further shift was observed up to 10~^-N salt solution, indicating no further
adsorption.  The surface charge computed from maximum adsorption density was
lower than 1 micro coulomb per cm^.  Comparison of this value with the com-
puted maximum surface charge obtained from potentiometric tltration and B.E.T.
surface area data of 75 yC cm~^, indicates the presence of special groups or
adsorption sites on the surface.

     The fact that a hydrous oxide surface can be saturated opens the possi-
bility of competition between adsorbing metals as the total metal concentra-
tion exceeds the concentration of available sites.  Gadde and Laitinen (16)
were able to demonstrate that Pb^+ outcompetes Cd^+, Zn^+, and Thr+ for
hydrous manganese oxide sites regardless of which metal is added first.  This
demonstrates not only competition but the existence of a reversible adsorption
mechanism through which an adsorbed metal ion can be replaced by other com-
peting metal ions.  On hydrous ferric oxide, Zn^+, Ca^+, and 1C1" all in large
excess over Pb^+ had no effect on the Pb^+ adsorption.

     Kurbatov, Kulp and Mack (28) in a study of Ba and Sr adsorption on
hydrous ferric oxide found plateaux for both metals at about 10~2 mole of
metal per mole of ferric oxide, but could demonstrate competition only if one
metal was present in great excess over the other.

Theoretical Models of Adsorption

     Predicting the possible effects of each of these factors on the adsorp-
tion isotherms on hydrous ferric oxide in seawater depends to a great extent
on the physical and chemical picture of adsorption that is considered.  The
basic concepts on which most adsorption models have been built were expounded
in the Gouy-Chapman theory of the electrical double layer as modified by
Stern -Grahame.  This theory, known as the electrical double layer or EDL
theory, was derived with reference to either mercury drops which become
charged as the result of the external application of a known potential differ-
ence, or solids such as Agl which become charged as the result of assymetrical
dissolution of the lattice ions and for which these ions are potential
determining.  The double layer potential for these solids is gotten from the
Nernst Equation:

                               ^o - ™ ln %.                             (1>

-------
where V  = the potential difference across the electrical double layer formed
           by the reversible transfer of potential determining ions

      R  = the gas constant

      T  ™ the temperature in  K

      F  « the Faraday constant

      a. - the activity of the positively charged potential determining ion in
           solution

     a   = the activity of the positively charged potential determining ion
           at the point of zero charge

For either surface, rigorous thermodynamic calculations relating the surface
charge and the potential of the electrical double layer are possible.  The
adsorption of ions at either of these surfaces is a function of the coulombic
attraction or repulsion of the charged surface for the ions and a specific
chemical interaction energy.
                              A G  .  = ZWX - *                             (2)
                                 ads      o

where Z - the charge on the adsorbing ion

      F = the Faraday constant

     ¥~ = the potential at the distance of closest approach of the adsorbing
          ions  (6)

      $ « the specific chemical interaction energy

The physical picture of adsorption that results from this treatment is repre-
sented in Figure 1.

     Application of EDL theory to hydrous oxides is not completely straight-
forward.  With hydrous oxides, much higher surface charges are developed than
can be accounted for by the electrokinetically measured double layer poten-
tials (zeta potentials).  The surface charge arises from reaction of the sur-
face with H+ and OH~ ions, which are therefore the potential-determining ions
for oxides.

                             EM-OH + H+  •* 31-OH-+
                                                L                           (3)
                             EM-OH + OH~ -»• 31-0" + H20

The Nernst equation for oxide surfaces takes the form:
where pHp7r = the pH at which the surface contains equal numbers of adsorbed
              H+ and OH" ions

-------
00
               Distribution  of
                Charge In the
              Electrical  Double
                  Layer
Solid    -
                                                    olid surface
                                                     Stern Plane
                Potential Relative to
                   iulk Solution
Solution
                                                   Surface,  Charga = o
                                             !lo  Specific Adsorption

                     Tha Electrical Double Layer

                Figure 1.   Adsorption  in the electrical  double layer.
                                                         .— solid surface
                                                        /—Inner Helnholtz
                                                       ///—Outer Helrcholtz Plane
                                                       frr              —
                                                       ©
                                                       9!   *
                                                                                         ©
                                                                                         ?!
                                                                                         ©!
                                                      V\A-Outer  Helmholtz Plane, Charge * o.
                                                       \  Inner llelinholtz Plane, Charge » a.
                                                        ^-Surface, Charge = a               °
                                                    •r    I

                                                      With S-iecific Absorption

-------
      pH  = any other pH

The use of the Nernst equation to calculate double layer potentials for oxides
is only considered valid near pHpzc where the number of adsorbed potential
determining ions is small compared to the number of surface lattice ions.
Lyklema (29) points out that in order for equation (4) to hold, the chemical
potential of H* on the surface, ^IT" must be assumed to be independent of the
activity of H"1" on the surface.  He maintains that this is unlikely since IT1"
is not a constituent of the unhydrated solid, and thus adsorption of H  will
cause the surface layer composition to be different from that of the bulk
solid.

     This problem is also discussed by Berube and deBruyn (30) who observed
that the differential capacity of the double layer on hydrous oxides, given
by:
                                       9a
                                   c -
where a  = surface charge

      4»
       o = surface potential relative to the bulk solution

is much larger than the differential capacities for mercury or silver iodide
surfaces.

     Several models of the double layer on hydrous oxides which could account
for the existence of high surface charge despite low double layer potential
have been proposed.  The porous double layer model of Breeuwsma and Lyklema
(31) is based on the ability of the potential determining ions and some
counter ions to penetrate into the surface layers of the solid.  Thus the
specificity of a charged surfaced in adsorbing ions of the opposite charge
depends on the geometric penetrability of the ions as well as on their affin-
ity for the solid.  Their data suggests that of the two, porosity is the more
Important factor.  Since much of the surface charge would be neutralized
within the pores of the solid, the potential across the electrical double
layer would be lower than that expected from consideration of the total sur-
face charge.

     Berube and deBruyn in their structured water model place the locus of the
potential determining H*" and OH~ ions not on the surface itself, but in a
layer at least one molecular layer of water away.  This highly structured
water layer between the surface and the adsorbed potential determining ions
is dissociated, possibly to a greater extent than bulk water, because of its
chemisorption at the surface.  Since H"1" and OH~ are now lattice ions, the
constraints on the Nernst equation are no longer a problem.  Adsorbed counter-
ions, especially strongly hydrated ones, can now approach the region occupied
by the surface charge very closely, again allowing high surface charges to
exist despite low potentials some distance from the surface.

     In the site-binding model of Yates, Levine and Healy (32), the charged

                                       9

-------
surface sites arising from reaction of the surface with potential determining
ions, react further with ions of the indifferent (non-specif ically adsorbed)
supporting electrolyte, e.g. NaCl.
                          EM-OH* + Cl~  J  EM-OH-Cl
                                    Na+  J  =M-0-Na
                                                                          (6)
All sites involving the potential determining H  and OH  ions are in one plane
at potential ¥o.  The supporting electrolyte ions are situated in a different
common plane, the Inner Helmholtz Plane.  The Outer Helmholtz Plane is the
plane of closest approach of ions in the diffuse part of the double layer.
Again high surface charge and differential capacity of the double layer can
exist despite low potential at the Outer Helmholtz Plane.

     These models have been proposed to explain characteristics of the elec-
trical double layer peculiar to hydrous oxide surfaces.  They produce refine-
ments in the isotherms for electrostatically bound adsorbants, but do nothing
to elucidate the nature of specific chemical adsorption.  The fact that
charge-reversal due to super-equivalent adsorption of cations on a negatively
charged surface and the adsorption of cations on a positively charged surfaces
exist, indicates that in many cases the specific chemical interaction is more
important than the electrostatic interaction.

     James and Healy modified the EDL treatment of adsorption by adding a
third energy term to equation (2).  They observe that for each metal specifi-
cally adsorbing on a given surface there is a critical pH range, usually less
than two units wide, over which the fractional amount of metal adsorbed
increases from zero to almost unity.  The onset of the range is related to
the onset of hydrolysis and the subsequent lowering of the charge of the
adsorbing ion.  The added term opposes adsorption of highly charged free aquo
metal ions and decreases in magnitude as the charge on the metal decreases
with hydrolysis.  Thus:

                     AG°   - AG°  .  +  AG°    + AG° .                     (7)
                       ads     coul       chem     sol

AG  solvation represents the energy required to disrupt the hydration sheath
of an adsorbing metal ion so that it can move to within one water molecule
of the surface.  It depends on the dielectric constant of the adsorbant and
the charge and radius of the adsorbing metal ion.  The more hydrolyzed the
adsorbing metal ion is, the less charged it is and the less unfavorable the
energy of solvation is because the water of hydration is less tightly bound.
This model represents a departure from the classical EDL treatment of adsorp-
tion in that it includes more information about the nature of the specific
chemical adsorption process.

     A different approach to adsorption on oxides is taken in the surface com-
plexation model (33).  Adsorption is treated as a chemical reaction whose
equilibrium constants are affected by the presence of the charged surface.
First, since the development of the surface charge is dependent on the acid-
base properties of the oxide, one can write the constants for the surface


                                      10

-------
reactions:

                          +            +
                    EM-OH   + EM-OH + H   ;  Ka. -
                         2                     l   {EM-OH+}
                                                         2  .               (8)
                    EM-OH   •> EM-0  + H


where {  } indicates surface groups

      t  ]  indicates species in solution

     These constants can be evaluated experimentally from data for the poten-
tiometric acid-base titration of the surface.  The constants obtained in this
manner are "microscopic" constants, however, which are pH dependent.  The
"microscopic" constants are analogous to the series of acidity constants ob-
tained as protons are removed from proteins or other polymers.  As each
successive proton is removed, the remaining protons become more tightly bound
because of the increasing number of negatively charged groups.  Likewise,
for a charged surface, the acidity of a proton depends on the relative charge
on the rest of the surface which in turn depends on the pH.  These microscopic
or apparent acidity constants can be converted to intrinsic constants, inde-
pendent of pH, by extrapolating the apparent constant versus surface charge
curve to the zero charge condition.  From EDL theory it follows that the
change in the acidity constants with surface charge is given by:

                           K  - K      exp (F ¥/RT}                         (9)
                            a    a.
                                  int

where K     ** intrinsic acidity constant
       aint

      K     = microscopic or apparent acidity constant
       SL
                                                         2+
     Similarly, for a specifically adsorbed metal ion (Me  ), the surface
reactions and stability constants are given by:
                                                         _i_   i
          EM-OH + Me2+ -> EM-0-Me+ + H+  ;  K
                                            1    {EM-OH} [Me2+]             (1Q)
          2EM-OH + Me2+ -> (EM-0)2Me + 2H+
                                                    (EM-OH}2[Me2+]
for mono and bidentate complexes.  Again, these are microscopic constants and
can be extrapolated to the pH of zero charge to obtain the intrinsic constants.

     This model presents adsorption as a chemical reactions which is influenc-
ed by electrostatic interactions.  It does not give any direct insight into
the electrical properties of the surface or address itself to purely coulombic
interactions.

                                      11

-------
     In order to determine what are the effects of the various factors dis-
cussed above on the adsorption of trace metals by hydrous ferric oxide in sea-
water, it is first necessary to isolate the factors.  This can be done by
starting with a simple system of one metal, for example Cu  , adsorbing on
hydrous ferric oxide in a simple, non-specifically adsorbed background electro-
lyte, for example NaC104.  Once a baseline result is established for the
simple system, the system can gradually be changed and/or made more complex
and any changes in the adsorption isotherms can be ascribed to the appropriate
factor.  Finally, the isotherm for Cu   on hydrous ferric oxide in the
presence of the other adsorbable metals in seawater can be interpreted on the
basis of the known influence of each of the factors.  Using the experimental
results, it should be possible to determine which, if any, of the theoretical
models of adsorption provides a satisfactory explanation for the process of
adsorption in seawater.
                                      12

-------
                                  SECTION 2

                     EXPERIMENTAL MATERIALS AND METHODS
MATERIALS

Chemicals

     All reagents were prepared from Analytical Reagent grade chemicals using
distilled, deionized water.

     Sodium hydroxide solutions were prepared from 50% NaOH solution and
standardized against potassium acid phthalate with phenolphthalein indicator.
They were stored in polyethylene dispenser bottles and kept C02 free.

     6 N hydrochloric acid was distilled from a quartz still.  Solutions were
standardized against standard base with phenolphthalein indicator.

     Synthetic Ocean Water (S.O.W.) was prepared according to the formulation
of the FWPCA (34).  This is a complex solution of eleven salts containing the
major constituents of seawater.  The composition is given in Table I.  The
S.O.W. was passed through a Chelex 100 chelating resin column at pH 8.5 to
remove trace metals present as impurities in the reagent grade major salts
(35).

     Copper and lead stock solutions were 0.01 and 0.1M Cu(NO_)2 and Pb(NO-) .
50 yl of the appropriate 0.01 M solution was added to 50 ml of sample gave
a final trace metal concentration of 1 x 10  M.  This concentration was used
for most experiments with 1 x 10~^M (as Fe) hydrous ferric oxide.  In the
experiments where total trace metal concentration was varied, differing
amounts of the stock solutions in both concentrations were used to give the
desired final concentration.  Stock solutions were naturally acidic as a
result of the hydrolysis of the Cu^+ and Pb  .  Periodically fresh stock
solutions were prepared and both old and fresh solutions were analyzed by
Atomic Absorption Spectrophotometry.  Results showed that the stock solutions
remained stable indefinitely.

     The ferric iron stock solution was 1.0 M Fe(N03)3.  It had a pH of 2.7
immediately after preparation, so no acid was added to stabilize it.  The
amber color of this solution increased in intensity with time, indicating
increasing polymerization of the iron hydrolysis species, but no cloudiness
or precipitate appeared.  Periodic iron analysis by the orthophenanthroline
procedure indicated that the concentration was stable indefinitely.

     Other trace metal solutions were prepared fresh as needed from reagent

                                      13

-------
                    Table 1



           The Composition of S.O.W.
salt



NaCl



CaCl,
    i



KBr



NaF



KC1
NaHCO,
SrCl,
final concentration



  4.20 x lO'Hl



  1.05 x l
  8.40 x 10~4M



  7.14 x 10~5M



  9.39 x 10~3M



  4.85 x 10~4M
                                     2.88 x 10~2M
  2.38 x 10~3M
  6.38 x 10~5M
MgCl,
  5.46 x 10
                       14

-------
grade chemicals.

Apparatus

     Before use, all glassware was cleaned, rinsed with distilled, deionized
water, soaked in 3N HCL for at least 12 hours, and rinsed with distilled,
deionized water.  All tubing and stirrers which contacted the sample solutions
were Teflon.  Some experiments were done in acid-soaked glass beakers, others
in Teflon beakers.  The apparatus used for most of the experiments is shown
in Figure 2.  The sample was placed in a 100 ml beaker, either glass or Teflon
with a Teflon coated  stirring bar.  This beaker was placed inside a 250 ml
jacketed glass beaker which was connected to a Haake circulating constant
temperature bath.  The temperature was maintained at 25°C for all experiments.
A #10 silicone rubber stopper bored to accommodate two electrodes, a N2 inlet
tube, a sample withdrawal tube and a reagent introduction inlet fit snugly
into the top of the beaker.  The N2 was passed first through an Ascarite-
filled drying tube to remove C02 and then through distilled, deionized water
to saturate it before it entered the reaction vessel through Teflon tubing.
The sample withdrawal tube was a Teflon tube ending in a Luer-loc fitting.
This end was kept immersed in a test tube of water to provide a pressure
outlet without allowing air to enter the system.  The other end was only
immersed in the sample when an aliquot was being withdrawn.  Reagents were
introduced into the system with Eppendorf pipettes.  The reagent introduction
inlet was an Eppendorf pipette tip (5-100 yl size), with about 2 cm cut off
from the tip inserted in the stopper until the cut off end was flush with the
stopper bottom.  Except when in use, this inlet was stopped with another
Eppendorf pipette tip either filled with Ascarite or closed off with a medi-
cine dropper bulb.  When completely assembled this apparatus was a (X^-free
system.  The pH was measured with an Orion Model 91-01 glass electrode versus
an Orion Model 90-02 double junction reference electrode.  The electrodes were
calibrated daily using Fisher Certified Buffer Solutions at pH 7 and 4.
Readout was obtained on either an Orion Model 701 digital pH meter of an Orion
Model 801A digital ionalyzer.  The output was recorded on a Cole-Farmer dual
channel strip chart recorder.
METHODS

Preparation of Hydrous Ferric Oxide

     The hydrous ferric oxide sols used for these experiments were prepared
by filling a volumetric flask with the appropriate background electrolyte
solution, stirring with a magnetic stirring bar, and adding stock 1M Fe(NO-j)3
with an Eppendorf pipette.  In NaC104 and NaCl solutions a stoichiometric
amount of IN NaOH, calculated by assuming the formula Fe(OH)o for the pre-
cipitate, was added to raise the pH and ensure complete, fast precipitation.
In synthetic ocean water  (S.O.W.), the natural buffering made this unnecessary,
as the pH stayed around 7.5.  All preparations were done open to the atmos-
phere.  For experiments using freshly precipitated hydrous ferric oxide  (HFO),
only enough suspension was prepared for one day's experiments, usually 100 or
250 ml.  For experiments with aged HFO, 2 liters were prepared, which yielded
forty 50 ml samples.

                                      15

-------
Figure 2.  Experimental apparatus.

-------
     In experiments done on "fresh" precipitates, the suspension was stirred
for one hour before sampling.  During stirring the oxide remained uniformly
suspended and no clumping of particles or accumulation of solid on the walls
of the flask occurred.  Samples were withdrawn with 50 ml volumetric pipettes
while the suspension was stirred vigorously.  The sample was transferred to
the beaker and placed in the apparatus shown in Figure 2.  The experiment
was begun after pH equilibrium was reached.  A signal drift below 0.002 pH
units/min was used as the criterion for equilibrium.  The total time elapsed
between preparation and the start of an experiment on fresh hydrous iron
oxide was around 3 hours.

     In the first experiment done on "aged" precipitates, the preparation was
stirred continuously and sampled at appropriate times measured as time after
preparation.  After 144 hours it was noticed that small shavings of Teflon
had flaked off the magnetic stirring bar and had become nuclei for agglomera-
tions of the ferric hydroxide sol.  This was extensive enough to significant-
ly alter the concentration of iron in the samples withdrawn, since these
agglomerates were not taken into the sampling pipettes.  Later experiments
on aged precipitates were done with sols that were quiescent between sampl-
ings and only stirred during sampling.  The settling out of the precipitate
did not appear to affect its later dispersability or chemical  characteristics.

Potentiometric Titrations

     The titrations were performed in the C02~free system by adding a known
excess of acid—either HC104 for NaC104 solutions or HC1 for NaCl or S.O.W.
solutions—to the sample in the beaker, to lower the pH below 4.  The N2 was
turned on and the CC>2 purged from the sample.  After the pH equilibrated, but
not before 10 minutes, the pH was recorded and a known volume of standard
0.05N NaOH (C02~free) was added with an Eppendorf pipette.  At the low and
high ends of the pH range,   50 yl additions were made.  At the break,  addi-
tions were reduced to 10 yl.  The pH was allowed to re-equilibrate after
each addition and was recorded.  A drift of less than 0.002 pH units/min was
used as a measure of equilibrium.  This was continued until the sample reach-
ed pH 10.  All titrations were kept free of (X^.

Generation of Isotherms

     To generate an isotherm a 50 ml portion of the hydrous ferric oxide sol
was placed in the beaker in the C02~free apparatus shown in Figure 2.  HC10A
or HC1 was added to lower the sample pH below 4 and the sample was purged of
C02.  After pH equilibration a known volume of a trace metal solution was
added with an Eppendorf pipette.  An addition of 0.05N NaOH (C02~free) was
made to raise the pH.  At low pH, 0.05N NaOH was added in 100 yl portions.
Near the break of the titration the size of each addition was lowered to
50, 10, and finally 5 yl.  This made it possible to obtain samples distribu-
ted over the entire range from pH 4 to pH 7.  In the low and high pH regions,
pH equilibrations occurred within one minute.  Near the break, the time
necessary for equilibration became longer, reaching a maximum of about one
hour.  Equilibration times of longer than one hour were indicative of a
clogged reference electrode, a pressure build-up in the system or inadequate
stirring.  A pH drift back to lower pH indicated C02 contamination which

                                      17

-------
necessitated termination of the experiment.

     After pH equilibration the sample withdrawal tube was lowered into the
suspension and a 5 cc disposable syringe was attached to the Luer-loc fitting.
About 3 ml of the suspension and another addition of base was made to the
sample in the beaker.  A Swinnex Millipore filter holder with a 0.025 ym
Cellulose acetate filter was attached to the syringe containing the with-
drawn sample.  The sample was expressed through the filter into a 25 ml
Erlermeyer flask with a standard taper glass stopper.  10 yl of 6N HC1 were
added to the flask, it was stoppered, marked for identification and set
aside until all samples were collected in the same way.

     When all the samples for an isotherm were collected they were analyzed
for the metal or metals of interest by flame atomic absorption spectrophoto-
metry.

     The efficiency of the 0.025 ym Millipore filters was tested by analyzing
the filtrate for iron by the orthophenanthroline procedure.  1.2 x 10~"M Fe^+
was left in the filtrate from a 1 x 10~^M  (as Fe) suspension of hydrous
ferric oxide.  This corresponds to 99.8% removal of the iron.

Flame Atomic Absorption Spectrophotometry

     Flame Atomic Absorption Spectrophotometry was done on a Perkin-Elmer
Model 360 double beam atomic absorption spectrophotometer.  All of the metals
analyzed required an air-acetylene flame.  Metal concentrations were always
within the detectable range of the instrument so no pre-concentration steps
were necessary.  Standards were prepared in concentrations to bracket the
sample concentrations using the same background electrolyte as the sample.
The high salt content of standards and samples caused some variability in
absorption values obtained for different aspirations of the same sample.
Because of this, all standards and samples were aspirated in random order at
least three times.  The average of the three values was taken as the absor-
bance for a sample.  The relative standard deviation of the three readings
ranged from 1 to 3%.  Standard curves prepared from the samples were linear
through all points, but had non-zero intercepts on the absorbance (y) axis
corresponding to the absorbance reading for background electrolyte with no
added metal.  It was determined that this was due to the high salt content
and its interference with the flame and not to ambient metal concentrations
in the reagents by using the Cd lamp to analyze a solution of Cu.  Cd has a
spectral line at 326.1 nm, very close to the Cu line at 324.8 run.  A dis-
tilled water solution of Cu has zero absorbance at this wavelength, but the
0.5 M NaC104 0.5M NaCl and S.O.W. solutions all gave absorbance readings
corresponding to the non-zero intercept of their standard curves.  The non-
zero intercept was reproducible and stable from day to day.
                                      18

-------
                                  SECTION 3

                           EXPERIMENTAL RESULTS
POTENTIOMETRIC ACID-BASE TITRATIONS

     Existing models of adsorption of trace metals by hydrous oxides rely
heavily on data from acid-base titrations of the solid.  These data are used
to compute the pHpzc and the number of reactive sites on the solid surface.
In this work, to avoid possible changes in the nature of the precipitate due
to filtration, the titrations were performed in the same medium from which
the hydrous ferric oxide had been precipitated.  Reproducible titrations of
the amorphous hydrous ferric oxide used in this work were very difficult to
obtain.  Similar difficulties were encountered by Davis (36).  The experi-
mental errors in the addition of FeCNO^)^ and NaOH to the medium resulted in
variability in the position of the titration curves.  A representative titra
tion curve is shown in Figure 3 for 1 x 10~-*M amorphous hydrous ferric oxide
in 0.5 M NaClO^.  Similar curves were obtained in 0.1 and 0.25 M NaCIO,.
     The proton excess or deficiency in the solid phase (Fjjf - FOH~) is cal-
culated from the titration curves as the difference in the amount of acid or
base needed to reach a given pH in the sample and the blank.  It is related
to surface charge by
where a  = surface charge

     F + - r „- = surface excess H  over OH~
      H     On

      A  = surface area

      F  = Faraday constant

     Because the methods available for determining surface area are unsatis-
factory for use with suspensions in the medium in which they have been pre-
cipitated, surface excess in eq/mole has been substituted for surface charge
in this work.  If a constant specific surface area is assumed the two differ
by only a constant.

     Plots of the surface excess shown in Figure 4 for the three ionic
strengths are the same within experimental error.  They do not exhibit the
sigmoid characteristics obtained for other oxides (5, 30, 31).  Therefore a


                                      19

-------
N3
o
                PH
10




9




8




7




6




5




4




3
                               0.5 M  NaCI04
                            acid
                                  eq x I04
                                                            base
                             3.  Titration curve for 1 x 10  M amorphous hydrous ferric oxide
                                 in 0.5 II NaClO..
                                             4

-------
                                                   .IM   NaCI04

                                                 * .25M NaCI04

                                                 o .5M   NaCI04
                                    7

                                   PH
8
10
                      +     —                          —3
Figure 4.  Surface excess H  or OH  as a function of pH for 1 x 10 M hydrous ferric oxide.

-------
value for maximum reactive sites cannot be obtained, although it must be in
excess of 0.15 site/Fe.  The pH_7f, appears to be in the range of 7.9 - 8.5.
This is in agreement with data reported for various crystalline and amorphous
iron oxides (4, 5, 8, 29).

     Titration curves obtained as soon as possible after preparation of the
oxide (ca. 2 hours due to pH equilibration time) required more acid than
curves obtained for the same oxide after 24 hours aging.  No further change
occurred after 24 hours.  This shift in the titration curves, shown in Figure
5, can be interpreted as a release of acid or a consumption,of base by the
aging oxide.  The two curves are roughly parallel 1.3 x 10   eq apart for
1 x 10~^M Fe.   Copper isotherms obtained simultaneously with the same batch
of oxide showed no effect of aging.  This is shown in Figure 6 which also
illustrates that the same isotherm was obtained when copper was coprecipitated
with the iron.

EFFECT OF IONIC STRENGTH
                                           2+
     The effect of ionic strength on the Cu   isotherm was first studied us-
ing hydrous ferric oxide precipitated in media of varying ionic strength.
The results shown in Figure 7, exhibit a non-systematic variability presum-
ably due to differences among the batches of ferric oxide.  To eliminate
batch to batch variations in the oxide a large batch of 2 x 10~^M hydrous
ferric oxide was prepared in 0.01 M NaClO^.  By appropriate dilution (1:1)
with 0.99 M NaC104, 0.09 M NaC104, or deionized, distilled water, a back-
ground electrolyte concentration of 0.5 M NaClO^ 0.05 M NaC104 or 0.005 M
NaClO^ could be had with the same batch of oxide.  As expected, the results
were less variable and demonstrate no effect of ionic strength either on the
Cu2+ isotherm  (Figure 8) or on the Pb2+ isotherm (Figure 9).

EFFECT OF COMPOSITION OF THE BACKGROUND ELECTROLYTE

     The effect of the composition of the background electrolyte was studied
in the same way.  The Cu2+ isotherms obtained with hydrous ferric oxide pre-
cipitated in 0.5 M NaC104, 0.5 M NaCl and S.O.W. (Figure 10) again show a
small non-systematic variability.  Diluting portions of the large batch of
2 x 10~4M hydrous ferric oxide in 0.01 M NaC104 with 2 x S.O.W. and 1.0 M NaCl
gave background electrolyte compositions of S.O.W. and 0.5 M NaCl for the
same batch of iron.  While the Cu2+ isotherms were identical to the one
obtained in 0.5 M NaC104 (Figure 11), the Pb2+ isotherms were equally depress-
ed relative to the Pb2+ isotherm in 0.5 M NaC104.  The difference follows
from the relative strengths of the Cl~ complexes of the two metals.  While
the chloride complexes of Cu(II) are relatively unimportant, Pb(II) is known
to form important di and trichloro complexes at (Cl~) - 0.5 M.  Apparently
the Cl~ complexes are not adsorbed and the result of Cl~ complex formation
is a decrease in available Pb2+ which depresses the isotherm.  Neither the
Cu2+ nor the Pb2+ isotherm is affected by high concentrations of Mg2+ and
Ca2+, indicating either than Mg2+ and Ca2"1" are not adsorbed or that they can
be displaced by Cu2+ and Pb2+.

EFFECT OF VARIATION OF METAL AND OXIDE CONCENTRATIONS

     An important aspect of the amorphous hydrous ferric oxide system is the

                                     22

-------
    PH
10
CO
10


9


8


7


6


5


4
        21
                                            I
                    19
17               15
meq acid x 100
13
                                                                  _A
         Figure 5.  Changes with aging in the titration curve for 1 x 10  M hydrous ferric oxide in S.O.W.

-------
   100
   90
   80
   70
I 60
 k.
 o
 V)
XJ
< 50
+
CM
   30
   2O
    10
                          FeT = lxlO-4M
                          CUT = I x IO~5M
                                           2 +
o
x
HFO generated in presence of Cu'
HFO prepared, Cu2+added  immediately
    after pH equilibration
HFO prepared .aged 24hrs
HFO prepared .aged 48 hrs
HFO prepared,aged 72 hrs
                           PH
                                     2+                 -5
       Figure 6.  Effect of aging on the Cu   isotherm for 1 x 10 M
                  2+        -4
                Cu   on 1 x 10  M hydrous ferric oxide in S.O.W.
                              24

-------
  100
   90
   80
   70
O
£60
a:
O
en
o
< 50
CM
O

^40
   30
   20
   10
   0.5 M

   0.05 M

   0.005M
CuT=IO-5M
                                                                    8
                                   pH
                                                    2+
       Figure 7.   Effect of ionic  strength  on  the Cu    isotherm for
                        -5    2+         -3
                  1 x 10  M Cu  on 1  x  10  M  hydrous ferric oxide prepared

                  in the different background  electrolytes.
                                    25

-------
100


90


80


70


60
I 50
CM
 <3 40
 55

   30


   20


     10
       4


       Figure 8.
                   FeT =  I x IO"4M
                   CuT =  I x IO"5M
              •  0.5 M  NaC!04
              A 0.05 M  NaC!04
              oQ.005 M  NaCI04
                             PH
                                              2+
             Effect of the ionic strength on the Cu
                  -5    2+         -4
                                    hydrous ferric oxide prepared
                                   isotherm for

1 x 10~5M Cu2+ on 1 x l(f''

in one large batch and diluted with the appropriate

background electrolyte.
                               26

-------
100


90


80


70
         I  60
         <  50
        CM
         .Q
         Q_
            40


            30


            20


             10
                                         -4
                              FeT  = I x 10"* M

                              PbT  = Ix IO"5M
          •     0.5 M NaCl04
          A    0.05 M NaCI04

          o  0.005 M NaCI04
                                                               o_i_          y
Figure 9.   Effect of ionic strength on the Pb2+ isotherm for 1 x 10~ M Pb  on 1 x 10~ M

          hydrous ferric oxide prepared in one large batch and diluted with the

          appropriate background electrolyte.
                                     27

-------
       too
       90
       80
       70
     S60

     cc
     o
     CO
     o

     < 50
     -t-
     PJ
      3
     O


     35 40
       30
       20
        10
«  0.5MNaCl04


•  O.SMNaCl


•  SOW


FeT = IO'3M
                                                                         8
                                        pH
Figure 10.  Effect of the background electrolyte composition on the Cu

                               -5    2+          -3
            isotherm for 1 x 10  M Cu   on 1 x 10  M hydrous ferric


            prepared in the different background electrolytes.
                                                                      2+
                                    28

-------
     
-------
    100




    90



    80



    70
CM
 .a
 a.
 o
 CO
 TJ
 <  50
    40
    30
    20
     10
                         0.5 M  NaCI04

                         0.5 M  NaCI

                         SOW
                                                              7
                                  PH
                                                       2+
                                                         isotherm
Figure 12.  Effect of background  electrolyte composition on the Ph

                    -5    2+         -4
           for 1 x 10 M Pb   on 1 x 10 M hydrous ferric oxide prepared in

           one large batch and diluted with the appropriate background

           electrolyte.
                             30

-------
batch to batch variability under controlled laboratory conditions.    Figure
13 which consolidates the Cu^"1" data from seven different batches of iron
under various conditions of ionic strength and background electrolyte compo-
sition, gives a measure of this variability.  Some of the low data at high
pH may reflect contamination of the filtrate from new apparatus and not true
batch to batch variations.  Nevertheless there are significant differences
among the isotherms obtained with different batches which must be considered
when comparing the data of different researchers or applying the results to
natural systems.

     To elucidate the nature of the reaction, metal and oxide concentrations
were systematically varied.  The precipitation of the Cu^+ and Pb^+ hydrox-
ides limited the metal concentrations that could be used to ca. 1 x 10~^M
Cu?+ and 4 x 10~^M Pb  .  The minimum oxide concentration yielding measurable
uptake of metals was 5 x 10~^M.

     In the first set of experiments, the hydrous ferric oxide concentration
was kept at 1.0 x 10~^M while the Cu^+ concentration was varied.  From
consideration of a simple mass balance

                 EM-0- + Me2+ :  =M-0-Me+  ;  K = i=*±
                                                  (EM-O }[Me  ]

it was expected that below saturation at a given pH the ratio of adsorbed to
free metal would remain constant while the total moles of metal adsorbed
increased as long as the available reaction sites were in large excess.  As
saturation was approached, the ratio of adsorbed to free metal would begin
to decrease as the available reaction sites decreased.  According to this
interpretation, the results shown in Figures 14 and 15 imply that saturation
is not approached up to a Cu2+/Fe ratio of 0.2.  The ratio cu|Jsorbed/^ufree
remained constant while the moles of Cu   adsorbed increased with increasing
Cu~+.  Increasing the Cu^+ concentration above 1.0 x 10~^M would have led to
precipitation of Cu(OH)2 in the range in which the isotherms showed signifi-
cant adsorption of Cu^"*".  The ferric oxide concentration was therefore de-
creased to 5.0 x 10~^1 and the experiment repeated with the same Cu   con-
centrations.  It appears that even under these conditions, an approach to
saturation cannot be reached before the Cu?+ precipitates as Cu(OH)2
(Figures 16-17).  The distinction between adsorption and precipitation of
Cu^+ in these experiments was very clear.  Before the onset of precipitation,
a gradual decrease in free Cu   accompanied a gradual increase in pH as
base was added.  When precipitation of Cu(OH)2 began, a rapid decrease in
free Cu~+ occurred while the pH remained constant with the addition of base.
As previously mentioned, further reduction of the hydrous ferric oxide con-
centration to 1.0 x 10"% resulted in negligible Cu^+ adsorption in the pH
range 4-7.
                 2+
     Since the Cu   hydrolysis and precipitation made it impossible to demon-
strate saturation of the hydrous ferric oxide with Cu?+, the experiments
were repeated with Pb+ which precipitates as Pb(OH)2 at a slightly higher
pH than Cu^+ precipitates as Cu(OH)2, but gives a similar isotherm on hydrous
ferric oxide.  The results shown in Figures 18 and 19 are not as straight-
forward as those for Cu2+.  During the generation of the isotherms, the
                                      31

-------
     100
     90
     80
     70
     60
   cc
   o
   to
   O
     50
  M


   O


   8*
     40
     30
     20
      10
        Hr
                                               o    o
                                   o p
                                     pH
T = IO'5M
                                                                       8
                                -5    2+          -1
Figure 13.   Isotherms for 1 x 10  M Cu   on  1  x  10 M hydrous ferric oxide


            under various conditions of  ionic  strength and background electrolyte


            composition obtained with different  batches of hydrous ferric oxide


            prepared in the different background electrolyte.
                                       32

-------
       1.0

       .9

       .8

       .7
     : -6
    +w
     V)
   CVJO
    3  .4


        .3


        .2
CUT
CUT
CuT
CUT
FeT
• 5 x !(T6  M
= Ix IO"5 M
= 5x !0"5  M
- Ix IO"4M
= Ix IO"4M
                                 PH
                    2+       2+
Figure 14.  The ratio Cu ,  ,  d/cuf   as a function of pH for variable CIL,
          on 1 x 10~% hydrous ferric oxide.
                               33

-------
           .20
         O
         u.
            .15
         0)

         o
         0)
         Q.
         0)

         £  .10
         o
         I/)
         •a
        M


         O

         o)
           .05
                       2+
Figure 15.  Moles of Cu   adsorbed per mole hydrous ferric oxide as a function

                                            -4
            of pH for variable GIL on 1 x 10  M hydrous ferric oxide.

-------
   1.0
    .9
    .8
    .7
    -6
-I -5
IM O
 o
    .4
                         ,-5,
                  • CuT=5xlO"5M


                  a CuT*lxlCT4M


                    FeTs5xlO'5M
                        2+          2+  pH
Figure 16.  The ratio Cu     . d^Cuf     as  a  funct^-on

            on 5.0 x 10~^M hydrous  ferric oxide.
                                                         for variable GI
                                  35

-------
           .8
        o
        UL
        o
        o
           .6
0)
Q.

TJ
0)
.D

O
O)
       CM

        O

        (O
        O)
          .4
              D


              A


              O
CuT

CuT

CUT

CuT
I x !0~5  M

4 x  IO'5M

5 x  IO'5M

Ix IO"4M


5xlO"5M
                     2+
Figure 17.  Moles of Cu  adsorbed per mole hydrous ferric oxide as a function

           of pH for variable Cu on 5.0 x 10 "M hydrous ferric oxide.
                                 36

-------
        1.0
         .9
         .8
         .7
     +   5
     NO  • 3
      .0

      0_
•  PbT=lxlO~4M


x  PbT = 2xlO~4M



•  PbT = 3xlO~4M



a  PbT=4xlO'4M



   FeT=5xlO'5M
                                        pH



                           2+         2+
Figure 18.   The ratio  of Pb      j/Pb.    as a function of pH for variable
                           adsorbed   free


            Pb  on 5.0 x 10~ M hydrous ferric oxide.
                                    37

-------
       1.0
        .8
    O
    u.
    I
     0)
     O
     0)
     Q.
     O
     (A
     TJ
    •f
    CM
     0>
     o
        .6
.4
       .2
                                              o  I x I0~4 M Pb2*
                                              A  2x IO"4 M Pb2*
                                              •  3x IO"4 M Pb2*
                                              n  4x IO'4M Pb2*
                                   PH
                     2+
Figure 19.  Moles of Pb"' adsorbed per mole hydrous ferric oxide as a function of

           pH for variable Pb  on 5.0 x 10 M hydrous ferric oxide.
                            T       38

-------
Pb2+ solutions apparently became supersaturated with respect to Pb(OH)2 and
an addition of base actually lowered the pH if precipitation commenced.  The
complex and variable hydrolysis of Pb2"*" in these solutions reflects the known
polynuclear behavior of Pb2+ hydrolysis products.  Britton (37) in a study
of the precipitation of trace metal hydroxides, found that Kgp for Pb(OH>2
increased with the amount of alkali added, ranging from 3 x 10~-*-  to
1.35 x 10~^5-  A phenomenon of this kind is indicated in the data presented
here.  At high Pb2+ concentration (4.0 x 10~^M) the precipitation of Pb(OH>2
caused the hydrous ferric oxide to coagulate and a white solid was carried
down with the brick red ferric oxide.  Before the appearance of the white
solid, however, the distinction between adsorption and precipitation of the
hydroxide was not as clear for Pb2+ as it was for Cu2+.  Despite these
difficulties, an approach to saturation is suggested with the highest Pb
concentration at a 0.8Pb/Fe ratio.

     The expected increase in Cu2+ and Pb2+ adsorption with increasing FeT
at a given pH, is shown in Figures 20 and 21.  The reduction of the data to
                                        KH
                                   adsorbed                                , .
                                                                           *• '
                                   2+
                                 CufreeFeT
and


                                   adsorbed
                                   2+
                                 PbfreeFeT

presented in Figures 22 and 23, respectively shows that the increase is
approximately proportional to the concentration of the oxide.  The high points
in the Cu2+ plot at low pH and the low points in the Pb2+ plot at high pH
reflect large indeterminant errors in the isotherms.

COMPETITION BETWEEN METALS FOR THE HYDROUS FERRIC OXIDE

                                                            2+       2+
     It was impossible to demonstrate competition between Pb   and Cu   for
the iron.  This was consistent with the difficulties encountered in demon-
strating saturation for either of the metals.  An apparent competition was
demonstrated between Cu2+ and Fe2+ when 1.0 x 10~5M Cu2"*" and 5.0 x 10~^M Fe2+
were present with 4.0 x 10~^M hydrous ferric oxide in an 02~free system.*
This result is shown in Figure 24 as a large depression in the isotherm com-
pared to the same system in the absence of Fe(II).  This demonstration is of
academic interest only, as an 02~free system is not applicable to seawater
but it does indicate the possibility of competition if combined trace metal
concentrations are sufficiently high.
* The N2 used to exclude C02 from the system was also passed through a trap
containing Cu metal in NH^OH solution to remove 02, a second trap containing
HC1 to remove NH~ and a water trap to remove NH,C1 and HC1.


                                      39

-------
                      CuT = Ix IO"5M
                   FeT =
                      2+
Figure 20.  Per cent Cu   adsorbed as a function of pH  for various concentrations
           of hydrous ferric oxide in S.O.U.
                                      40

-------
      100
       90
       80
       70
    I  60
    w
    O
    V)
    TJ
    <  50
   +
   CVl
       40



       30



       20



        10
•   FeT = I  x IO"3 M

A   FeT = I x IO'4 M

o   FeT =5  x IO"5M
                                PH
                                         6
                     2+
Flgure 21.  Per cent Pb" adsorbed as a function of pH for various

           concentrations of hydrous ferric oxide in S.O.W.
                                41

-------
                                                         -5    2+
Figure 22.  Reduction of data from isotherms for 1.0 x 10  M Cu   on various

                                                                  2+
            concentrations of hydrous ferric oxide in S.O.W. to Cu ,   ,  ,/
                                                                  adsorbed

            Cu
,   Fe^ as a function of pH.
                      42

-------
      50
      40
      30
   £
   (M O

    a

    0.
      20
       10
•   FeT=  5x IO"5 M



0   FeT =  I x IO"4 M



x   FeT =  I x IO'3 M
                  PbT = I  x  ICT5 M
                                      PH


                                                       -5    2+
Figure 23.   Reduction of data from isotherms for  1.0 x 10  M Pb   on various


            concentrations of hydrous ferric oxide  in S.O.W. to Pl»-"t   ,  ,/
                              J                                 adsorbed

            Pb|+  FeT as a function of pH.




                                    43

-------
                                  SECTION 4

                                 DISCUSSION
     The mechanism by which hydrous metal oxides remove trace metals from
solution has been postulated by various workers to be adsorption, chemisorp-
tion, ion exchange, surface complexation and coprecipitation.  Several simple,
well-defined experimental systems yielded data which could be fitted to
equations derived for models based on these mechanisms.

     The electrical double layer theory and numerous modifications treat the
phenomenon of electrostatic adsorption of ions on a charged surface.  James
and Healy added a term to the established double layer treatment to explain
the relationship between hydrolysis of metal ions and their specific adsorp-
tion (chemisorption).  The surface complexation model proposes a chemical
reaction between surface sites and metal ions in solution for which the
equilibrium constants are modified by the surface charge.  All of the mathe-
matical models require that the data can be reduced to an apparent constant
for a given pH and a given background electrolyte according to a mass law
expression


                             K - t51"""\                             (1)
                                 (EM-O }[Me  ]

     As the reduction in Figs. 14, 16, 18, 22 and 23 demonstrates, the reac-
tion of Cu2+ and Pb^+ with amorphous ferric oxide is, to a first approximation
proportional to the free metal concentration in solution and to the amount
of oxide over the range 5 x 10~°M to 4 x 10~^M metal and 5 x 10~5M to
1 x 10"% oxide.

     To reduce the data further to test equation 1, requires an estimate of
the total number of reactive sites on the oxide in order to calculate
{=M-0~}.  The acid-base titration data are of little help for this purpose as
not all of the acid-base sites appear to be titrated in the pH range of
interest.  An estimate of total reactive sites based on such data would result
in orders of magnitude underestimation of the removal of Cu^+ and Pb^"*" by
the iron.  Upper and lower limits for the number of reactive sites can be
obtained directly from the adsorption data.  The Pb^+ data (Figure 19)
provides the lower limit of 0.8 site/Fe; while an upper limit of 1.5 site/Fe
is implied from the Fe(II)-Cu competition experiment (Figure 24).  This is
consistent with the estimate obtained by Davis (36) of 0.87 site/Fe on the
basis of Yates (38) data obtained from rapid tritium exchange experiments.
Figure 25 shows the reduction of the data using a value of 1 site/Fe for the
Cu2+ isotherms in seawater for all values of Cu^ and FeT.  It should be noted

                                      44

-------
    4-

    CVJ
100




90




80




60




50




40




30
        10
                     I x IO"5 M Cu2*only


                     I x KT5 M Cu2*  + 6xlO'4M  Fe(ll)


                     FeT= 4xlO'4M                 ^^"*
                                         -5   2+         -4
Figure 24.  Depression of the isotherm for 1.0 x 10  II Cu   on 4.0 x 10  M

                                                         -4    2+
          hydrous ferric oxide ir. S.O.W. in the presence of 6.0 x 10 M Fe
                              45

-------
both that the data are consistent with the mass law expression and that since
all the isotherms were below saturation, this is not a sensitive test of the
value chosen for total reactive sites.  Although the Pb2+ data would provide
a more sensitive test, their scatter renders such reduction meaningless.
The curve in Figure 25 provides a convenient means to predict Cu2+ removal by
the iron oxide for any combination of concentrations and pH.

     The very complex system of amorphous ferric oxide in seawater behaves in
a remarkably simple way with respect to Cu2+ and Pb   removal.  There is no
effect of ionic strength, no effect of background electrolyte (except for the
expected Cl~ effect on Pb2+), no effect of aging, no difference whether the
iron is precipitated in the presence or absence of the metal.  Such simplicity
was not expected on the basis of the various mechanisms that have been pro-
posed.

     The expected effect of ionic strength depends to some extent on the model
being considered.  The ionic strength enters into electrical double layer
calculations and affects the coulombic interactions between the charged
surface and the adsorbing ions by changing the thickness of the double layer.
The parameter 1/K, which can be conceived of as the distance between the
plates of a hypothetical electrical condenser representing the electrical
double layer is 10.0 nm in 1.0 x 10"% NaCl and reduces to 0.36 nm in 0.7M
NaCl (39).  The collapse of the electrical double layer at high ionic strength
decreases the diffuse layer potential.  What effect this has on adsorption
of trace metal cations, which are specifically adsorbed against the electro-
static repulsion of a positively charged surface, is unclear.

     In all cases, however, the coulombic interactions in the adsorption pro-
cess can only play a role for reactions resulting in a net change of charge.
The remarkable consistency of the isotherms for various ionic strengths
                                                             O 1       o I
invites a speculation that the reaction of iron oxide with Cuz  and Pb^"1"
results in no net change in charge in the pH range of interest.  For example,
two protons may exchange for one metal ion.  This is reinforced by the aging
study which demonstrates a clear change in charge (evidenced by a change in
uptake of acid by the solid) with no concomitant change in the isotherm.

     Britton (37) in extensive studies of the precipitation of hydroxides
found that in general the amount of base needed to precipitate a metal
hydroxide did not correspond to the computed stoichiometric amount.  He con-
cluded that other anions were incorporated into the solid forming what he
calls "basic salts".  Such non-stoichiometric precipitation would lead to the
observed batch to batch variations, but it also implies a possible effect of
the background electrolyte on the nature of the iron oxide.  The effect of
the background electrolyte on the formation of the crystalline ferric oxides
has been reported (12).  Whatever effect the background electrolyte may have
on the nature of the amorphous iron oxide, however, it is not reflected in
the isotherms for Cu2+ and Pb2+.

     It is puzzling that the release of 1.3 eq/mole of H /Fe over 24 hours of
aging is not accompanied by any change in the Cu2+ isotherms.  There is a
strong implication that the protons lost in aging are not the exchangeable
protons on the reactive sites.  This may be a consequence of different

                                      46

-------
      0>
      o
                                        pH


Figure 25.  hog K as a function of pll for all Cu"  data where:

                           2+
                         Cu
                     K -
adsorbed

     2+
                         (Ci^-Cu
                                                2+
     adsorbed^ (r7eT~Cuadsorbed'
                                      47

-------
coordination of the oxygen atoms or of geometrical inaccessability of some
of the oxygen atoms.  Dousma and de Bruyn (6) discuss the polymerization of
Fe(III) oxide and note that in addition to the olation processes involving
-OH addition to iron oxide polymers, oxolation also occurs.  This process
is represented by
    H

Fe/'Se
                   H
                          — >
              H
                                + H
(1)
The oxo bridges thus formed react very slowly with acid.  Oxolation would
explain the release of acid over the first 24 hours.  The oxygen atoms
involved in oxolation are already bound to two iron atoms and are not likely
to be the reactive sites.  That the acid-base sites with rapidly exchangeable
protons and the coordination sites for Cu^+ are indeed the same is supported
by the dependence of the apparent adsorption constant (K) on pH.  Note that
from pH 6 to 7 the apparent constant increases by an order of magnitude
(Figure 15).

     The capacity of amorphous hydrous ferric oxide for metal ions makes the
meaning of the terms "surface" and "adsorption" somewhat arbitrary.  It
becomes necessary to provide a more satisfactory image of the system than
that of an interface separating two semi- infinite phases.  One is clearly
not concerned with the interface between the visible precipitate and the bulk
solution, but rather the interface between some microstructure and its
immediately surrounding water.  Amorphous hydrous ferric oxide has been
described as "amorphous, randomly crosslinked aggregates containing large
and indefinite amounts of water" (40).  This description calls to mind the
structure of a swollen ion exchange resin which is permeable to hydrated ions.
The ions are free to diffuse throughout the structure and are not restricted
to external "surface" sites.
     The formation of hydrous ferric oxide proceeds from the hydrolysis of
the hexahydrated aquo ion
Fer
                                                     H
to formation of the dimer
                        2Fe(H20)5OH
           2+  ->  
-------
linking and crystallinity.  Byrne and Kester (13) found evidence for increas-
ing crystallinity after very long aging periods for amorphous ferric oxide.

     The loose, highly hydrated structure of the ferric oxide used in this
work readily accommodates foreign ions which become incorporated into the
solid as they hydrolyze.  A loss of H  accompanies the formation of -OH and
-0- bridges.  The observed effect of chloride is a consequence of the forma-
tion of complexes with the metal ions which are not incorporated into the
solid.  Rudnev et al. (41) found a similar effect of NH3 on copper coprecipi-
tation with ferric hydrous oxide.

     In addition to accounting for the high capacity of the iron for metals,
this Image also provides an explanation for the similarity between isotherms
obtained by precipitation of the iron in the presence and absence of Cu^ .
It is also consistent with what is known regarding the formation of metal
ferrites (42).  The amorphous complex oxides, such as those obtained in this
work, are precursors of the crystalline ferrites.  Adamovich et al. (43) exa-
mined the solid resulting from the coprecipitation of Fe and Cu by X-ray
diffraction and thermogravimetry and found evidence for a single solid phase
rather than a mixture of iron and copper oxides.  Heating the solid produced
a compound whose X-ray diffraction pattern was suggestive of copper ferrite.
Gmelin (44) documents the preparation of copper ferrite by heating a
coprecipitated iron and copper oxides.  When the oxides are precipitated
separately, mixed, and heated only CuO and Fe20o are formed.
     The results of this study show that despite its complexity, the amorphous
iron oxide in seawater system behaves rather simply.  The removal of metal
can be approximately predicted under all concentration and electrolyte
conditions by a constant which is only a function of pH.  This pH dependence
can presumably be accounted for in a numerical model by adjusting the acid-
base chemistry of the solid which is difficult to quantitate experimentally.
Although the mathematics appears to be similar to the surface complexation
model for adsorption on crystalline metal oxides, the reaction between
amorphous iron oxide and metals should not be viewed as a macroscopic surface
phenomenon.  In many ways, amphorous ferric oxide can be viewed as more
closely analogous to polyelectrolytes than to crystalline solids such as
quartz.
                                      49

-------
                                 REFERENCES

1.  Krauskopf, Konrad B.  Geochim. et Cosmochim. Acta, 9:1, 1956.

2.  Goldberg, E.D.  J. Geology, 62:249, 1954.

3.  Goldschmidt, V.M.  J. Chem. Soc., 1937:655, 1937.

4.  Forbes, E.A., A. Posner, and J.P. Quirk.  J. Colloid Interface Sci.,
    49(3), 1974.

5.  Yates, D.E., and T.W. Healy.  J. Colloid Interface Sci., 52(2):222, 1975.

6.  Dousma, J., and P.L. deBruyn.  J. Colloid Interface Sci., 56(3):527, 1976.

7.  Kolthoff, I.M., and B. Moskovitz.  J. Phys. Chem., 41:629, 1937.

8.  Parks, G.A., and P.L. deBruyn.  J. Phys. Chem., 66:967, 1962.

9.  Atkinson, R.J., Posner, A.M., and J.P. Quirk.  J. Phys. Chem., 71(3),
    1967.

10. Gadde, R.R., and H.A. Laitinen.  Environ. Lett., 5(4):223, 1973.

11. Ellis, J., R. Giovanoli, and W. Stumm.  Chimia, 30:194, 1976.

12. Murphy, P.J., A.M. Posner, and J.P. Quirk.  J. Colloid Interface Sci.,
    56(2):270, 1976.

13. Byrne, R.H., and D.R. Kester.  Marine Chem., 4:255, 1976.

14. Stumm, W.  Personal communication.

15. Kurbatov, J.D., J.L. Kulp, and E. Mack, Jr.  J. Am. Chem. Soc., 67:1923,
    1945.

16. Gadde, R.R., and H.A. Laitinen.  Anal. Chem., 46(13):2022, 1974.

17. McNaughton, M.G., and R.O. James.  J. Colloid Interface Sci., 47(2):431,
    1974.

18. O'Connor, T.P., and D.R. Kester.  Geochim. et Cosmochim Acta, 39:1531,
    1975.

19. Forbes, E.A., A. Posner, and J.P. Quirk.  J. Colloid Interface Sci.,


                                      50

-------
    49(3), 1974.

20. Matijevic, E.  In:  Principles and Applications of Water Chemistry,
    Faust and Hunter.

21. James, R.O., and T.W. Healy.  J. Colloid Interface Sci., 40(1):42, 1972.

22. Stanton, D.A., and R. duT. Burger.  Geoderma, 1(1):13, 1967.

23. Vuceta, J.  Ph.D Thesis, California Institute of Technology, 1976.

24. Brunauer, S., P.H. Emmett, and E. Teller.  J. Am. Chem. Soc., 60:309,
    1938.

25. James, R.O., P.J. Stiglich, and T.W. Healy.  Faraday Disc. Chem. Soc.,
    59:142, 1976.

26. Kurbatov, J.D.  J. Phys. Chem., 36:1241, 1932.

27. Breeuwsma, A., and J. Lyklema.  J. Colloid Interface Sci., 43(2), 1973.

28. Kurbatov, J.D., J.L. Kulp, and E. Mack, Jr.  J. Am. Chem. Soc., 67:1923,
    1945.

29. Lyklema, J.  Croatica Chem. Acta, 43:249, 1971.

30. Berube, Y.G., and P.L. deBruyn.  J. Colloid Interface Sci., 27(2):305,
    1968.

31. Breeuwsma, A., and J. Lyklema.  Disc, of Faraday Soc., 52:324, 1971.

32. Yates, D.E., S. Levine, and T.W. Healy.  J. Chem. Soc., 70:1807, 1974.

33. Stumm, W., H. Holh, and F. Dalang.  Croatica Chem. Acta, 1976.

34. FWPCA Methods for Chemical Analysis of Water and Wastes, U.S. Dept. of
    Interior, 1969.  p. 240.

35. Morel, F.M.M., J.C. Westall, J.G. Rueter, and J.P. Chaplick.  Technical
    Note #16, Ralph M. Parsons Laboratory, M.I.T., Cambridge, Massachusetts,
    1975.

36. Davis, J.A.  Ph.D Thesis, Stanford University, 1977.

37. Britton, H.T.S.  Hydrogen Ions, Vol. II.  D. Van Nostrand and Co., Inc.,
    Princeton, New Jersey, 1956.

38. Yates, D.E.  Ph.D Thesis, University of Melbourne, 1975.

39. Parks, G.A.  Adsorption in the Marine Environment.  In:  Chemical
    Oceanography, J.P. Riley and G. Skirrow, eds. Academic Press, London,
    1975.

                                      51

-------
40. Walton, H.F.  Principles and Methods of Chemical Analysis.  Prentice-
    Hall, Inc., Englewood Cliffs, New Jersey, 1964.

41. Rudnev, N.A., G.I. Malofeeva, N.P. Andreeva, and T.V. Tikhonova.  Zhurnal
    Anal. Khimmii, 26(4):697, 1971.

42. Hilpert, S., and A. Wille.  Z. Physik. Chem., B-18:291, 1932.

43. Adamovich, T.P., V.V. Sviridov, and A.D. Lobanok.  Dokl. Akad. Nauk.
    Belorussk. SSR, 8(5):312, 1964.

44. Gmelin's Handbuch der Anorganischen Chemie.  Kupfer Tell B., Lief. 3,
    System Num. 60:1276, Verlag Chem., 1965.
                                      52

-------
                                    TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
1. REPORT NO.
  EPA-600/3-80-011
                              2.
                                                            3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
   ADSORPTION OF TRACE METALS BY HYDROUS FERRIC
   OXIDE  IN SEAWATER
              5. REPORT DATE
               January 1980 issuing date
             6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
    Francois Morel  and K. C. Swallow
                                                            8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
  Ralph M.  Parsons Laboratory  for Water Resources  and
  Hydrodynamics,  Department of Civil Engineering
  Massachusetts Institute of Technology
  Cambridge,  Massachusetts  02139
              10. PROGRAM ELEMENT NO.
               1BA819
              11. CONTRACT/GRANT NO.
              R803738
12. SPONSORING AGENCY NAME AND ADDRESS
  Environmental  Research Laboratory - Narragansett,  R.I.
  Office of Research and Development
  U.S. Environmental Protection Agency
  Narragansett,  Rhode Island  02882
              13. TYPE OF REPORT AND PERIOD COVERED
              Final      	.
              14. SPONSORING AGENCY CODE
              EPA/600/05
15. SUPPLEMENTARY NOTES
16. ABSTRACT

       The adsorption of trace metals  by amorphous hydrous ferric oxide  in seawater
  is studied with  reference to simple  model systems designed to isolate  the factors
  which may have an  effect on the  isotherms.   Results  show that the complex system
  behaves in a remarkably simple way and that the data obtained under various
  conditions of total metal concentration and total oxide  concentration  can be reduced
  to an apparent reaction constant, K,  which  is a function of pH only.   The high
  capacity of the  oxide for trace metals renders the concept of a surface  reaction
  useless to explain the uptake of metals.  A physical picture of the oxide as a
  swollen hydrous  gel permeable to hydrated ions is presented.
17.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS  C.  COSATI Field/Group
    Trace elements
 Modelling
 Metalic Wastes
    06/F
18. DISTRIBUTION STATEMENT
 RELEASE TO  PUBLIC
19. SECURITY CLASS (This Report)
 UNCLASSIFIED
21. NO. OF PAGES
      63
                                               20. SECURITY CLASS (This page)
                                               UNCLASSIFIED
                                                                          22. PRICE
EPA Form 2220-1 (9-73)
                                             53
                                                                    s US GOVEHNHEN1 MINTING OfflCt I9K> -657-146/5546

-------

-------
TD427.T7S93
  Swallow, K.C. (Kathleen C.)-
    Adsorption of trace materials
by hydrous ferric oxide in
seawater.

                          14437687

-------
Una*** StMOT
Agency
Cirwinnati OH 45268
and ^""""— ""• «^
Environmental ^^Jlf
Protection mfSbm
EPAJ35 V. ' ^

Official Business
Penalty far Private Use, $300

                                                                                                                                                                                      Rate

                                                                      Ptesse make 3ll necessary changM on the ab«w» label.
                                                                      detach of copy, and return to the addrss* in the upp«i
                                                                      led-hand corner

                                                                      H you do no! wish to receive ihese reports CHECK HEH6 U.
                                                                      detach, or cqav this cover, and return to the addre&s in th®
                                                                      upper left-hand corner
                                                                               EPA-600/3-8Q-011

-------