SEPA
United States
Environmental Protection
Agency
EPA/540/SR-93/515
September 1993
SUPERFUND INNOVATIVE
TECHNOLOGY EVALUATION
Emerging Technology
Summary
Metals Treatment at Superfund
Sites by Adsorptive Filtration
This project evaluated an innovative
approach for removing inorganic con-
taminants from the liquid phase at Su-
perfund sites. In the process, called
adsorptive filtration, metals are col-
lected by attachment to a thin layer of
ferrihydrite (iron oxide) that has been
immobilized on the surface of sand
grains. The modification of the sand
surface allows the grains to simulta-
neously adsorb soluble heavy metals
and remove particulate metals by filtra-
tion from a wastewater.
The metals studied were Cd, Cu, and
Pb, present at concentrations of 0.5 or
5 mg/L each, in synthetic solutions. A
few preliminary tests were also con-
ducted to evaluate removal of the toxic
oxyanions of As and Se. The effects on
process performance of solution pH,
the empty bed detention time (EBDT),
and the presence of complexing agents,
oil, surfactant, and biodegradable sub-
stances were evaluated. In addition, the
potential to regenerate the media after
a run was investigated, including both
the kinetics and overall efficiency of
the regeneration process. Finally, a
model waste solution from a Super-
fund site was treated by adsorptive fil-
tration in a small-scale test.
In general, adsorptive filtration
proved to be an efficient and effective
treatment process. Soluble and particu-
late forms of all the metals tested could
be removed from the water stream at
both concentrations tested. The con-
tact time required for treatment was
minimal (<5 min), and treatment was
successful at moderate pH values (near
9). Removal efficiencies ranged from
about 70% to >99%, depending on treat-
ment conditions. Regeneration was also
fairly rapid and efficient, metal concen-
trations in the regenerant solutions
reached several hundred times those
that were in the influent. Over the dura-
tion of the tests, there was no indica-
tion that the media were deteriorating.
Finally, the process was shown to be
applicable for removing Zn from a waste
stream generated at a Superfund site.
Overall, the project adequately demon-
strated the potential applicability of the
process, and it appears that the pro-
cess is appropriate for. testing on a
larger scale.
This Project Summary was developed
by EPA's Risk Reduction Engineering
Laboratory, Cincinnati, OH, to announce
key findings of the research project
that is fully documented in a separate
Printed on Recycled Paper
-------
report of tho samo title (see Project
Report ordering Information at back).
Introduction
Adsorption is the binding of chemical
species on the surface of suspended par-
ticles, if the adsorbent (solid surface) is
chosen carefully and the solution chemis-
try is adjusted appropriately, adsorption-
based processes are capable of removing
metals over a wider pH range and to much
tower levels than can processes based on
precipitation. Additionally, sorption may
remove some metals in systems where
precipitation will simply not work at all. For
instance, adsorption can often remove in-
organically- and organically-complexed
motals that would not be removed by con-
ventional treatment methodology. Adsorp-
tion can also be very effective for remov-
ing antonte metals such as oxyanions of
Se, Cr, and As. These capabilities make
adsorption an effective process for reme-
diation at sites where a wide variety of
catfonte, anionic, and complexed metals
co-exist.
One adsorbent that is commonly present
in metaltreatment processes is amorphous
iron oxide, or ferrihydrite, which forms
when iron salts precipitate in neutral to
slightly alkaline solutions. Since adsorp-
tion processes are most effective when
large quantities of adsorbent are present,
a higher coagulant dose will improve treat-
ment efficiency in most cases. However,
increasing the ferrihydrite concentration
leads to a corresponding increase in
sludge mass. Thus, while optimization of
the metal removal efficiency requires that
large amounts of ferrihydrite be used, op-
timization of the sludge processing step
requires that minimal amounts of this ma-
terial be disposed. This trade-off can be
dealt with successfully by first adsorbing
contaminant metals onto ferrihydrite and,
then separating the metals from the ad-
sorbent, so that the adsorbent can be
retained and reused to treat subsequent
batches of waste.
One efficient way to retain ferrihydrite in
a system is to use it as the media in a
packed-bed treatment process. Unfortu-
nately, this approach is not practical, since
ferrihydrite is a bulky, flocculant material
with extremely low hydraulic conductivity.
To overcome this problem, a means was
developed by which the ferrihydrite could
be coated onto the surface of sand. A
photograph of some coated sand grains
used in the current study is shown in
Figure 1.
After treatment has proceeded for a pe-
riod of time, either the coating reaches its
maximum capacity to remove metals or
the fitter requires backwashing. At this time,
the column can be backwashed to re-
cover particulate metals from the column,
and an acidic solution can be used to
recover the adsorbed metals, thereby re-
generating the column. Because the
ferrihydrite is trapped on the sand par-
ticles, only the contaminant metals and
not the ferrihydrite are released. Thus, the
need to dispose of large amounts of iron
oxide with the metal sludge, one of the
main drawbacks of a conventional treat-
ment process, is eliminated.
In sum, there are six essential aspects
of adsorptive filtration that combine to
make it a potentially valuable and widely
applicable technology:
• Ferrihydrite is a strong metal
adsorbent that can be regenerated
by changing pH.
• Ferrihydrite can be coated onto sand,
retaining much of its adsorbent
activity.
• Ferrihydrite can adsorb some metal
complexes that are not removed from
solution by conventional precipitation.
• Many metal oxyanions that cannot be
treated by conventional precipitation
can adsorb onto ferrihydrite.
• A column of coated sand acts as a
filter as well as an adsorbent.
• The technology appears to be
applicable over a wide range of
contaminant concentrations.
Methodology
The project consisted primarily of pilot-
scale testing of the ferrihydrite-coated
sand. The experiments addressed the ki-
netics of adsorption and regeneration, the
adsorption capacity of the media for both
soluble and particulate metals, the effects
of various organic contaminants on met-
als removal by the media, the long-term
stability of the adsorbent, and the metal
concentrations achievable in the
regenerant solution. In addition, although
model influents were studied during most
of the project, a few tests were conducted
with a metal-containing solution from a
Superfund site.
The coating was applied by heating a
solution containing an iron salt to dryness,
under controlled conditions and in the pres-
ence of Ottawa sand with an average
diameter of 400u.m. Some relevant fea-
tures of the plain sand and the two batches
of coated sand used in the project are
presented in Table 1.
The runs were conducted using a
packed bed containing 250 mL (bulk vol-
Figure 1. Electron micrograph of the coated sand magnified 25x. Large areas of the grains are
completelycoated with a skin-like layer of iron oxide, which appears to be about ten \im thick.
-------
Table 1.
Characteristics of Plain Ottawa Sand and Iron-oxide-coated Sand Used in Adsorptive
Filtration Study (All Sands Were 20-30 Mesh Size)
Plain Sand
Media I
Media II
Fe salt used for coating
% iron by weight
surface area by BET, rrf/g
pH of the PZC
none
0
0.04
0.7
FefNOJ,
2.4
9.2
Fed,
3.2
9.1
9.8
ume) of the coated sand. The rest of the
setup consisted of automated instruments
for maintaining pH of the influent water,
pumping water through the column, and
regenerating the column at fixed intervals.
Samples were collected automatically and
were analyzed for metal content. Headloss
across the bed was also monitored. Once
a predetermined criterion was met (re-
lated to either the duration of the run or
the headloss), the bed was cleaned by
backwashing and/or acid regeneration, and
the cleaning solutions were analyzed. A
solution adjusted to and maintained at pH
near 2.0 was generally used for regenera-
tion.
Model influent solutions contained 0.5
or 5.0 mg/L of Cu, Cd, and Pb, some-
times individually and sometimes in com-
bination. The pH of the test solutions
ranged from 7.0 to 9.5. with most tests
conducted at pH 9.0. Test solutions also
contained 0.01 M NaNO3. Also, several
tests were conducted in which an addi-
tional substance was added to the influent
solution to assess its effect on metal be-
havior in the column. The substances
tested in this way included ammonia (as a
complexing agent), EDTA (as a chelating
agent), sodium dodecyl sulfonate (a sur-
factant), motor oil, and antifreeze. In addi-
tion, some tests were conducted using a
column containing biogrowth. Finally, as
noted above, a few tests were run with a
solution collected from a Superfund site
where conventional treatment is currently
being applied.
Results
Systems with 0.5 mg/L Cu, Cd,
and Pb In the Influent
The effluent concentrations of the met-
als for a run with 0.5 mg/L each of Cu,
Cd, and Pb in the influent at pH 9.0, using
a 2-min EBDT and a hydraulic loading
rate of about 11 gal/min-ft2, are shown in
Figure 2. Initially, concentrations of all three
metals in the effluent were less than 0.1
mg/L and were gradually increased to
around 0.1 to 0.2 mg/L after 7000 bed
volumes of influent had been treated.
Headloss was usually under 5 psi at the
beginning of a run and increased gradu-
ally thereafter. When the headloss reached
around 10 to 13 psi, the column was
backwashed to remove particulate matter
that had been trapped. This process was
successful in that the pressure drop
through the column was reduced after
backwashing and the metal removal effi-
ciency was at least as good, and often
better, after backwashing.
Systems with 5 mg/L Cu, Cd,
and Pb In the Influent
In the runs with 5 mg/L of each metal in
the influent, most of the metal load was
particulate; soluble influent concentrations
were typically around 1.5 mg Cd/L, 0.8
mg Pb/L, and 0.2 mg Cu/L Under these
conditions, for the process to be success-
ful, filtration must be at least as significant
a mechanism of metal removal in the col-
umns as adsorption.
During each run, effluent was sampled
until the pressure drop across the column
reached a predetermined value, usually
either 20 or 25 psi. At that point, an auto-
matic shut-off switch was activated, and
flow to the column was terminated. The
column was then backwashed with pH 9.0
water, and flow was reinitiated. Influent to
the column at that point was identified as
"Batch 2." A similar sequence occurred
when the pressure reached the target
value again, and Batch 3 was treated.
When the pressure reached the maximum
allowable value after treatment of Batch
3, the column was backwashed and then
regenerated with water adjusted to pH
2.0. Data for a typical run under these
conditions are shown in Figure 3. Batch 2
began at Bed Volume 260, and Batch 3 at
Bed Volume 500.
The total concentrations of all the met-
als in the effluent were well below 0.1 mg/
L until a few hundred bed volumes had
been treated (a 6- to 12-hr period), at
which point particulate metals began
breaking through the column. Backwashing
of the media allowed additional influent to
be treated effectively. Removal of soluble
metal was substantial throughout these
runs, with typical removal efficiencies of
80% for Cu, 90% for Pb, and 98% for Cd,
and typical overall removal efficiencies
(comparing total effluent and total influ-
ent) of 99% or greater for all three metals.
Backwashing
The metal concentration in the back-
wash water was on the order of a few
hundred mg/L for each metal. Interest-
ingly, the amount of metal recovered by
backwashing was consistently and sub-
stantially less than the computed particu-
late load that had been applied to the
column. Most of the difference between
the amount of particulate metal removed
and the amount recovered by backwashing
was recovered during the acid regenera-
tion step: overall recovery efficiencies
(backwash plus regeneration) were almost
always greater than 80% and were often
100% ±10%.
Regeneration
The regeneration protocol was typically
to circulate two bed volumes of water ad-
justed to pH 2.0 through the column for 2
hr, although in some cases a larger vol-
ume of solution was used. After 2 hr,
another, equal volume of acidified water
was passed through the column and was
not recirculated. The metal concentration
in the recirculation fluid increased rapidly
at first and then only slowly thereafter
(Figure 4). Based on these results, it ap-
pears that a recirculating period as short
as 10 min. would release a large fraction
of the available metal. Metal concentra-
tions in the first and second regenerant
solutions were as high as 3000 and 500
mg/L after the 5 mg/L runs.
Anlon Removal
Five runs were conducted to evaluate
the removal of As and Se from model
solutions by the coated sand. In this case,
the influent was adjusted to pH 3.5, but
conditions were otherwise similar to those
for removal of cationic metals. Significant
amounts of Se or As began appearing in
the effluent after about 200 to 300 bed
volumes of solution had been treated (Fig-
ure 5). The removal pattern was remark-
ably consistent, regardless of the metal
(As or Se) being treated or its oxidation
state (+3 or +5 for As; +4 or +6 for Se).
The latter result was particularly surpris-
ing, since selenate (SeO42-) is generally
much more difficult to remove from solu-
tion than is selenite (SeO32-).
-------
0.8
0.6 '
0.4 • •
0.2 ..
1000
2000
3000
4000
Bed volumes
5000
6000
7000
8000
Breakthrough curves for Pb, Cd, and Cu for an influent containing 0.5 mg/L of each metal. EBDT'= 2 mm, pH = 9.0.
7 T
-•-- Lead
-*— Cadmium
•*••• Copper
—I-
200
roo
300
500
600
400
Bed volumes
Figure 3. Breakthrough curve for Run 14. Influent contained Cd, Cu, and Pb at 5 mg/L each. EBDT=2 mm, pH=9.0.
4
700
800
-------
I
2000 T
7600 --
1200 --
800 --
400 --
8
Time, minutes
10
12
14
16
Flgure4.
J±3^^
Effects of Other Contaminants
on Removal of Cu, Pb, and Cd
Tests were also conducted in which the
metals in the influent solution were
complexed by ammonia or EDTA. Only
Cd and Cu were tested with ammonia
present as a complexing agent, since the
ammonia did not maintain Pb in a soluble
form. All three metals were present in the
influent when EDTA was used as the
chelating agent.
Substantial amounts of ammonia-
complexed metal were sorbed by the iron-
coated sand: about 1500 mg of each metal
sorbed per liter of media before the efflu-
ent concentration exceeded a few tenths
of a mg/L, and about 4000 mg of each
metal sorbed per liter of bed before the
effluent concentration reached 4mg/L Re-
generation of this column at pH 2.0 recov-
ered 93% of the sorbed Cd and 100% of
the Cu.
When the metals were complexed with
EDTA, they broke through the column al-
most immediately. The capacity of the
media to remove metals under these con-
ditions is not significant either at pH 10.0
or at pH 4.5, and the adsorptive filtration
process would not be applicable for treat-
ment of waters containing EDTA-
complexed metals.
Sodium lauryl sulfonate is a surfactant
that might interfere with the adsorptive
filtration process by interacting either with
the metals or the surface of the media.
The presence of 0,15, or 30 mg/L of this
surfactant had no noticeable effect on
metal sorption.
An attempt was made to investigate the
behavior of media on which biogrowth had
occurred. To induce the biogrowth, a col-
umn was operated for 25 days with a feed
containing 50mg/L acetate and 20 mg/L
yeast extract. At the end of this period,
substantial biogrowth was visible above
and within the media. The column was
then backwashed to remove the easily
dislodged particles, potentially leaving be-
hind a biofilm attached to the media. The
biofilm apparently reduced the capacity of
media for the metals by about 50%. It is
expected that this interference could be
partially reversed by exposing the column
to a high pH solution, which would prob-
ably solubilize a substantial amount of the
biofilm. In any case, the interference is
relatively small considering that the biofilm
was grown under very favorable condi-
tions.
One test was conducted in which the
media were exposed to motor oil, to simu-
late a situation where, by accident, a large
amount of some oily substance entered
an adsorptive filtration column. Normally,
such substances would be removed up-
stream of the column. To investigate a
worst-case scenario, a damp sample of
the coated sand was exposed to a 10%
by volume mixture of SAE 30 motor oil in
water and was then packed into a col-
umn. Oil remained attached to the media,
and air bubbles were trapped in the col-
umn. When the influent was applied, se-
vere channeling was observed, and break-
through occurred almost immediately. The
coated sand was then removed from the
column and cleaned by rinsing it twice in
isopropanol. The cleaned sand was re-
turned to the column, and the standard
ammonia-complexed influent was applied.
The media performed reasonably well, but
a little more than half of the sorption ca-
pacity was lost. It is not known whether
this loss was due to residual oil on the
media, which might be removed by more
strenuous cleaning efforts, or whether it
reflects a permanent loss of capacity.
-------
4 "
3 ••
2 • •
1 ..
-Arsenate, Run 18
-Arsenate, Run 19
-Selenite, Run 18
-Selenite, Run 19
-H-
200
—I-
400
600 SOO
Bed volumes
Figure 5. Breakthrough curve for arsenate and selenite In Runs 18 and 19. EBDT = 2 min, pH = 3.5.
1000
1200
1400
Adsorbent Longevity
During the course of these runs, the
media were backwashed over 20 times
and regenerated about 10 times over a
period of a few months, with no apparent
deterioration in performance. Although this
resutt is fairly qualitative, it is possible
that, over the course of several months of
testing, any significant changes in column
behavior resulting from repeated regen-
eration would have been observed.
Treatment of A Real Superfund
Solution
Once reasonable operating parameters
for the technology were established, a real
waste from a Superfund site was collected
and treated. The untreated water at this
site contains tens of mg/L of ferrous iron
in addition to a few mg/L Zn and less than
1 mg/L of several other metals. It is treated
at the site by conventional precipitation/
coagulation at pH around 8.0, and, since
the ferrous iron is oxidized and precipi-
tated in the process, the metals are ex-
posed to a large amount of iron oxide in
tha process. As a result, a significant frac-
tion of the metals that can adsorb onto
iron oxide do so in the treatment process.
We chose to treat the effluent from the
precipitation/coagulation process being
used on site. The water was treated, as
received, at pH near 8.0, which is well
below the pH that would have been opti-
mal for metal removal. This test gives an
indication of the additional metal removal
that can be obtained by using adsorptive
filtration as a polishing step after a con-
ventional treatment process.
Zn was the only metal present in signifi-
cant quantities. The total and soluble Zn
concentrations in the samples collected
were in the ranges 0.6 to 4.0 and 0.3 to
0.6 mg/L, respectively. The corresponding
Zn concentrations in the effluent were
around 0.2 and <0.1 mg/L, respectively
(Figure 6). Thus, even though the test
might have been run under nonoptimal
pH conditions and exposure to relatively
high concentrations of iron oxide solids
for adsorptive filtration, the process re-
moved Zn to concentrations considerably
lower than those achieved by known con-
ventional processing.
Conclusions
Simultaneous sorption and filtration of
Cu, Cd, and Pb are feasible with the use
of iron-oxide-coated sand under reason-
able engineering conditions. Total and
soluble effluent concentrations of less than
100 u.g/L, and sometimes considerably
less, are achievable. The media can be
backwashed to recover most of the par-
ticulate metals and be regenerated by ex-
posure to an acid solution to recover the
remaining participates and most of the
adsorbed metals. The regenerant solution
typically contains metal concentrations a
few hundred times as concentrated as the
influent. In tests with 5 mg/L of each of
three metals in the influent, filtration lim-
ited process performance more than sorp-
tion did. It should be recognized that this
outcome is not generalizable: the limiting
factor in any application depends on the
specific operating conditions and chemi-
cal composition of the influent solution.
Modifications to the influent, such as ad-
justing solution pH or adding a polymeric
filter aid, and modifications to the opera-
tion, such as adjusting the hydraulic load-
ing rate, would certainly affect the relative
importance of headloss and effluent con-
centration limits. There was no indication
that such factors could prevent adsorptive
filtration from serving as a viable treat-
ment technology.
•frU.S. GOVERNMENT PRINTING OFFICE: 1993 - 750471/80070
-------
-------
Mark Benjamin and Ronald Sletten are with University of Washington, Seattle,
Washington, 98195.
Norma Lewis is the EPA Project Officer (see below).
The complete report, entitled "Emerging Technology Report: Metals Treatment
at SuperfundSites byAdsorptive Filtration,"(OrderNo. PB93-231165; Cost:
$19.50, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
EPA/540/SR-93/515
BULK RATE
POSTAGES FEES PAID
EPA
PERMIT No. G-35
------- |