b

REMOVING

RADIONUCLIDES

Exhaust

and
regenerating
resin

foruranium
remova
As expected, lowering the pH

adversely affected uranium

removal by anion resin-but at a
lower than anticipated pH.
Zhihe Zhang and
Dennis A. Clifford

Agency
he US Environmental
Protection
1) recently proposed a maximum cant .aminant
ACL) ot 20 pg/L and a maximum
con t aminant
)a1 (MCLG) of zero pCi/L for uranium 1 because
of concerns about its
association with kidney
disease and cancer.
Based on the results of a
national survey, it was
estimated
that 1,500
community water supplies in the United States
could be out of compliance and would have to
reduce uranium
concentrations to the proposed level. l Although
not well documented,
several small community water supplies have
uranium concentrations
in the high micrograms-

228

JOURNAL

AWWA

Houston-US
Environmental
Pm
mobile laboratory
was used at
to study possible
nium removal.

rbonate anions to form stable,

(or both) uranyl carmost significant species
in drinking water supplies.4 Under mildly
acidic conditions of pH 5.0-6.5, the principal
uranyl carbonate species is UO,CO,a, whereas
under neutral and alkaline conditions the
principal
species are U02(C03),2and
UO, (CO,) a4-. Uranyl hydroxy complexes such
as U020H+
and (UO2)3(OH)s+
are also
formed, but generally in small percentages
unless at high temperature or in carbonatedepleted water at pH >lO.s Figure 1 shows
the distribution of typical uranium complexes
as a function of pH in aerobic groundwater
at a carbon dioxide partial pressure of 10-2
atm.4 The reported stability constants of various complexes of uranium are given in Table
1.6-l Because of their high solubility and consequent dissolution from minerals and mine
tailings, uranium compounds are present in
both groundwaters and surface waters.
The three naturally occurring uranium isotopes,
with their corresponding percentage occurrences and
half-lives, are as follows: 238U (99.2745 percent, t, =
4.468 x lo9 years), 235U (0.7200 percent, t, = 7.037
x lo8 years), and 234U (0.0055 percent, t, = 2.454 x
10s years).ll When these isotopes are present in a
water at these percentages, the relationship between
uranium radioactivity
and concentration
is 0.69
pCi/pg.3 A slightly different specific activity-O.67
pCi/pg-has
been reported by others.12,13 Caution
must be exercised when using an average specific
activity because uranium isotopes in groundwater are
not always found in their reported natural abundance
ratios because of varying geochemical conditions.12

per-litre range-even
up to the milligrams-per-litre
range.2 The abandoned community water supply well
used for this research in the Chimney Hill subdivision, 25 miles northwest of downtown Houston, Texas,
was considered typical of a groundwater highly contaminated with uranium. Its uranium concentration
was normally in the range 1 lo-125 pg/L.
This article is one of several generated from the
results of the USEPA-funded 20-month bench- and
pilot-scale field study in the Chimney Hill subdivision on possible techniques for uranium removal. As
detailed in the article on page 2 14,3 the University of
Houston-USEPA
mobile drinking water treatment
research facility was used for the field study.
Uranium
chemistry
and radioactivity
Uranium is not found in the elemental form because of its strong reactivity.
It has four oxidation states-III, IV, V, and
VI-with the IV and VI states being most
stable. The high affinity of uranium for
oxygen results in the formation of uranyl
ion, UO,2+, which is stable under acidic
conditions (pH ~5.0). The uranyl ion readily combines with ligands such as Cl-,
NO,-, SOd2-, HPOa2-, and C032- to form
stable complexes depending on the prevailing water chemistry. In the near-neutral pH range, UO, 2+ combines with bicar-

TABLE

Stability
constants
groundwater

of uranium

compounds

occurring

APRIL 1994

in

229

t-b@

Summary

of runs

used

to remove

uranium

1,360
mL for
41,000
EMS, 25
mL for 2,940 BVs
61,CJOO

26,300
42,700

q Fractions
l,23,,:a;.&$
c.,-, -L*$%+v of pH in
.ti~~;-.p~~~,;p&*J

of carbonate
species
aqueous
system*

&&&&
I,
/ #.Q2f@ : 8 *
0.974
r

present

as a function

Uranium
removal
Several techniques for removing uranium from
contaminated water have been investigated either in
the laboratory or in pilot plants. Among these are
conventional coagulation-filtration
with iron or alum,
lime-softening
using varying amounts of Ca(OH),
and MgCO,, cation or anion exchange, activated alumina and activated carbon adsorption, and reverse
osmosis hyperfiltration.
The highlights of these laboratory and pilot-plant studies have been summarized
in a recent review article.2
Because of its high capacity, simplicity, and cost
effectiveness, ion exchange with anion resin has
received considerable attention
for removing uranium from small community water supplies. Bondiettir4 studied uranium removal by anion resin and
achieved 70 percent removal at 4,000 bed volumes
(BVs) throughput. Bench-scale studies conducted by
USEPA demonstrated that when actual drinking water
contaminated
with 300 pg/L uranium was used,
>9,000 BVs could be treated before uranium was
detected in the effluent.t5,16 Hathaway16 reported
two column runs using anion resins operated to
17,400 BVs and 31,300 BVs still removed 95 percent

230

JOURNAL

AWWA

txlp
0.00456
6.47 x IO-'
7.95 x 10-g

1,360 mL for
41.000
BVs. 264
m<for 19,600
BVs
II

(total)

1,360 mL for
41,000
BVs, 25
ml for 3 9Ar-l RVs

and 90 percent, respectively, of the total

uranium in the feed. These results and
others summarized
in the companion
article3 demonstrated that anion resins
have exceptionally
high selectivity for
charged uranyl carbonate complexes.
In the typical anion exchange process,
uranium-contaminated
water is passed
through a bed of strong base anion (SBA)
resin in the chloride form. The simple
exchange reactions between uranium
compounds and resin can be expressed as

2 RCl + U02(C03)22-

-RzUO,(CO3)*

+ 2 Cl- (1)

4 RCl + U0,(C03)34-

-R,UO,(CO,),

+ 4 Cl- (2)

in which R represents an anion exchange site on the

resin. When exhausted, the SBA resin is regenerated
with 14 N sodium chloride (NaCl), and the exhaustion reactions (Eqs 1 and 2) are reversed.
Anion exchange has been used for many years to
recover uranium in the mining industry;17 nevertheless, prior to this research, no systematic study of
uranium removal by anion exchange had been published for water supply applications. The objectives of
this research were
l
to determine the anion resins removal efficiency and capacity for uranium,
l
to investigate the effect of pH on uranium
removal and regeneration,
l
to determine the optimum regenerant for the
partially uranium-exhausted
resin,
l
to establish the effect of NaCl concentration
and mass loading on regeneration,
l
to determine the effect of the degree of ura-

nium exhaustion on regeneration, and

to predict the effects of
sulfate and feed uranium
concentrations on run length.
l

sElG&##?t
;;!
: ,;

Distribution
of typical
= 1W2 atm, 25W,
ho,

and methods
In addition to
1 lo-125 ug/L uranium, the
Chimney
Hill water contained 20-25 pCi/L radium.
Details of the feedwater composition, pretreatment
facility, and the flow schematic
for the mobile laboratory
have been described in the
companion article.3
Because the well was out
of service, it was pumped occasionally to fill a 2 1 O,OOO-gal
storage tank, which provided
water for the experiments. The
observed variations in feedwater uranium concentration
were attributed to its adsorption onto the traces of hydrous
iron and manganese oxide precipitates that formed or were
stirred up each time the tank
was filled. Although the uranium concentration was normally in the range of 11O-l 2 5
pg/L, it once reached a low 67
ug/L. When low-pH water was
tested, the feedwater was first
acidified to the required pH
range in the feed tank with 0.2
N hydrochloric
acid (HCl)
before being pumped through
the resin beds.
Resins. The type of resin
used was not a variable in this
research because previous
studies had shown the typical
polystyrene quatemary amine
SBA resins are adequate for
uranium remova12,17 Among
the many possible anion
resins, the one* chosen had
successfully been used in fullscale uranium removal. la This macroporous resin consists of a polystyrene-DVB matrix, with fixed quaternary ammonium-N(CH3)3+-exchange
groups in the
Cl- form. Its advertised capacity is 1 meq/mL.
Column studies. Generally, the downflow fixedbed column runs were conducted at ambient temperature (20-24C) in l- and 2-in.-diameter
glass
columns with graded-gravel support. Eight columns,
consisting of pure SBA resin, were tested in oncethrough mode. Cyclic runs to predetermined
BVs
were aimed at studying (1) the removal of uranium

uranyl
complexes
as a function
and Chr = 2.4 pg/L4

of PH for

Materials

Feedwater.

and (2) the regeneration of the partially uraniumexhausted resin with a variety of regenerants. Oncethrough exhaustions were performed in an attempt
to characterize the capacity of the columns for removing uranium under different pH conditions. The partially uranium-exhausted
resins were used later during the regeneration studies, which were generally
carried out in 50-mL burettes. The details of the total
BVs and run conditions, including the empty bed
*lonac

A-642,

Sybron

Chemical

Co., Birmingham,

N.J.

APRIL

1994

231

Concurrent
downflow
regeneration
was employed for all the regeneration runs,
for which the regenerant EBCT was always
10 min. A regeneration was typically followed by slow and fast rinses to displace
desorbed uranium. EBCTs for the slow and
fast rinses were 10 and 5 min, respectively.
During regeneration, the water level was
maintained at 6 in. above the resin.
For the so-called complete regenerations, six BVs of 1.7 N NaCl (37 lb/cu ft)
were employed to give a regeneration
stoichiometry of 10.2 eq Cl-/eq resin. For
other regenerations, the appropriate number of BVs of 0.5, 0.8, 1.0, 1.3, 1.7, 2.0,
3.0, or 4.0 N NaCl was used to give a mass
loading of salt corresponding
to 4.0 or
6.0 eq Cl-/eq resin.
Some experiments
determined
the
influence of high pH on uranium recovery efficiency. In these experiments,
NaCl was mixed in a
.
. . . .._.
n of virgin
2:l molar ratio (NaCl:base)
with NaOH, NaHCO,, or
Na,CO,. Each of these mixtures was tested for regeneration efficiency at various concentration levels with the total
amount of regenerant un.:i
changed. Finally, 1 N HCl was
_. ,
j
.,;, tested as a regenerant and
::.,.A .--,
1
compared with the NaCl and
:. : ,:
$.,
;
,,
3
NaCl plus base regenerations.
v.
Standards and reagents.
s
Uranium
standards were
6
obtained
from
three sources,
.
the USEPA Environmental
Monitoring Systems Laboratory (EMSL) in Las Vegas,
Nev., and two chemical coma
-9
c
5
I ,%
panics,* and were used suc2&o
30,ioo
cessfully to cross-check one
another. All the chemicals
used for the pilot runs and
analyses were analytical
reagent grade chemicals.
contact time (EBCT), for all the exhaustion runs are
Uranium analyses. Uranium concentrations were
determined using the analyzer+ on loan from USEPA,
summarized in Table 2.
For purposes of tracking data, samples,and analywhich had used the instrument and the associated
ses,the runs were numbered asfollows: E designates laser fluorescent technique for a national radionuexhaustion, R designates regeneration, and the
elidessurvey.12 The principle of the method is described
number after the decimal point designates the cycle
in the accompanying article.3 This uranium analyzer
number. For example, run E 11.1 is the first exhauswas standardized at least twice per analysis run using
tion during cyclic run 11, and R 11.1 designates the
2-, 20-, 120-, or 200-pg/L standards, depending on
first regeneration during cyclic run 11.
the expected range of sample concentrations. The preRegeneration. After the termination of exhauscision of uranium analyses was improved by always
running duplicates or triplicates and taking the avertion runs E 4.1, E 11.1, and E27.1 of the virgin resin,
age. The relative standard deviation for the uranium
the resin was removed from the column and well
mixed. The mixed resin was then divided into several
*Mallmckrodt
Chemical,
Inc., Chestcrflcld,
MO.; and Aldrich
Chemcal
25-mL portions, which were packed into 50-mL
Inc., Milwaukee,
WE.
tUA-3,
Scmtrex Ltd , Concord,
Ont , Canada
burettes for the regeneration study.
BVs of

232

JOURNAL

AWWA

method was 5.4 percent as determined from 55 pairs

of water samples that exhibited an average concentration of 5.6 pg/L.
To further ensure accuracy, uranium standard solutions and some actual test samples ranging from a few
micrograms per litre to lo6 pg/L uranium were also
cross-checked using the inductively coupled plasma
(ICP)-mass spectrometry (MS) method at the USEPA
Risk Reduction Engineering Laboratory in Cincinnati,
Ohio. Good agreement was obtained between the laser
fluorescent technique and the ICP-MS method.
Other analyses.
Total
hardness, total alkalinity, total
dissolved solids, and pH measurements were performed
as specified in the 16th edition of Standard Mefhods.i9
Results

and discussion

Resin capacity and ura

nium leakage. Two columns

containing 12- and 27-m. bed

depths of virgin SBA resin
operated
to 30,000 and
41,000 BV throughputs,
respectively. These natural-pa
runs exhibited excellent uranium removal; at pH 7.6-8.2
the effluent uranium concentration was always CO.1 pg/L,
i.e., >99 percent removal.
The effluent history for
run E4.1 is shown in Figure
2. This 2-in.-diameter, 20-in.bed-depth column exhibited
excellent uranium removal. It
operated continuously for 478
days for a total throughput of
302,000 BVs at pH 7.6-8.2.
Even after 300,000 BVs, the
effluent uranium concentration was still <6 pg/L, while
the feed concentration
was
125 pg/L, i.e., a removal efficiency of 95 percent. Before
260,000 BVs, the uranium
leakage was <l pg/L, i.e., ~99
percent removal. After that,
the uranium leakage in the
effluent slowly increased at a
rate of about 1 pg/L per
10,000 BVs of throughput.
Compared with granular activated carbon columns, which
exhibited only a few thousand
bed volumes of capacity during a parallel test, the observed
capacity of anion resin for
uranyl carbonate complexes
was enormous in this sulfatefree water.

Although the uranium was surely more concentrated near the bed inlet, the uranium concentration
averaged over the length of the bed was 30 g/L resin,
which corresponds to approximately
7.8 x lo4 pCi/g
dry resin. This loading is equivalent to 2.2 1 lb U,Os/cu
ft anion resin, which is approaching the 5 lb U,O,/cu
ft anion resin reported for uranium mining operations that utilized feedwater typically containing-l,000
times as much uranium as in the Chimney Hill water.
As expected, no measurable radium was removed
from the raw water by the pure anion resin.

The 1Pff wide x 13-ft

lo
/aboratory
(rignt)
uranium
and radium
Houston-area
supply.
Feed p4
top) delivered
raw
exchange
columns,
and
rignt, bottom)
mo

Effect of pH on uraniud

Although the natural pH I

water was 7.6-8.2, several runs were
conducted
at pH 5.6-6.0 and pH
4.2-4.5 to investigate the effects of
lower-pa
feedwater
on uranium
removal. A substantial decrease in the
resins capacity for uranium removal
was expected at lower pH because the
charge on the uranyl carbonate complex decreases with decreasing pH as
does the carbonate concentration
in
the feedwater. However, when pH was
decreased-by
HCI addition-from
7.6-8.2 to 5.6-6.0, uranium removal
efficiency was almost unchanged. If
the pH was further decreased to 4.2-4.5, however,
serious uranium leakage occurred from the very
beginning
of the run. The overall relationship
between uranium leakage and pH is illustrated in
Figure 3.
pH 5.8. Figure 4 shows the effluent history of
run 31 at pH 5.8, which lasted for nearly 20 days
for a total of 26,300 BVs of throughput. The highest
uranium leakage observed during the entire run was
~0.5 ug/L. As observed during the pH 8.0 runs, excellent uranium removal (~99 percent) was achieved at
pH 5.8.
Runs 28 and 29 were conducted
in two pH
stages. First the resin was fed pH 8.0 groundwater
for the first 41,000 BVs, then the pH was lowered
to 5.8 as the runs continued. Figure 5 demonstrates
the results from run 29.1, which were identical to
those of run 28.1. Lowering the pH of a partially
exhausted resin to 5.8 produced an insignificant
increase in uranium leakage to approximately
0.5
from CO. 1 pg/L.
The excellent pH 5.8 uranium removal results
indicate that, as far as leakage is concerned, uranium
removal is nearly as efficient at pH 5.8 as it is at pH
8.0. This unexpected performance at an acidic pH
requires some explanation. At pH 5.8, the dominant
uranium species in the aqueous system in the absence
of anion resin is the zero-charged UO,( CO,)O complex
(Figure 1). Nevertheless, as the following discussion
demonstrates, uranium removal can still be explained
as the uptake of a tetravalent UO,( C03)a4- complex
in exchange for four Cl- ions.
It is theorized that the direct formation of charged
U0,(C03)z4- from uncharged IJO,CO,a occurs within
the ion exchanger because of the resins high affinity

224

JOURNAL

AWA

for polyvalent anions and the high stability constant

of the tricarbonate uranium complex. The suggested
exchange reactions are
4 RCl + 4 HCO,- -

4 RHCO, + 4 Cl-

(3)

4 RHCO,+UO,CO,a-R4U0,(C03),
+2

H2C03

(ad

t4)

Eq 3 is the conversion of four resin sites to the

bicarbonate form by simple ion exchange for chloride.
In Eq 4, these sites are converted to the much preferred uranyl tricarbonate (R,U02(C03)3)
form with
the production of carbonic acid [H,CO, (aq)]. The
resulting lowering of pH by the HzCO3 (aq) produced
is insignificant because of the very low concentration (0.0005 mM) of U02COso in the feedwater compared with the total carbonate concentration ( CT,co3)
of the carbonate buffer system-3.00
mM.
Adding Eqs 3 and 4 yields the net ion exchange
reaction:
4 RCl + UO,CO,a

+ 4 HCO,- = R,UO,(CO,),
+ 2 H2C03 W
+ 4 Cl(5)

Eq 5 is, effectively, the exchange of a neutral

U02COso molecule for chloride ions.
The production of U02(C03)24- in the resin can
also be attributed to the production of C032- in the
resin from the abundant bicarbonate ions in the
water. In this mechanism, the proposed reactions
are as follows:
2 RCl + HCO,--

R,CO, + 2 Cl- + H+

(6)

,*, .,,.
I_ ..-

In Eq 6, two resin sites, which much prefer one

divalent ion to two monovalent ions, convert bicarbonate to carbonate with the production of a proton. Like Eq 4, Eq 6 disturbs the normal uranyl carbonate equilibrium described in Figure 1. Again, the
system is buffered against significant pH reduction
because H+ reacts with the abundant HCO,- ions to
produce H,CO, (aq) as indicated in Eq 7.
HCO,- + H+ -

H,CO, (aq)

(7)

The carbonate on the resin from Eq 6 can be

incorporated into an exchangeable and highly
preferred uranyl tricarbonate anion starting with the
neutral uranyl carbonate complex, according to Eq 8.
directly

2 R2CO3 + UO,(CO3)-R,UO,(CO3)3

(8)

Eqs 338 explain how an apparently neutral

species-UO,CO,a-can
be taken up by anion resins
that strongly prefer the charged uranyl carbonate
complexes or can convert bicarbonate to carbonate
with the expulsion of a proton. (Essentially all strong
base resins are capable of performing this latter
process20) As seen in Eqs 3-8, HCO,- drives the systern because it is the source of the COa2- ions that
further complex the uranyl ion and make it progressively more negatively charged. Table 321 illustrates
that sufficient bicarbonate was available irrthe Chimney Hill water to promote U02(C03)o uptake at pH 5.8.
Based on the waters total alkalinity of 150 mg/L as
CaCO,, its C,,,, was 3.00 mmol/L. Thus, at the nat-

ma1 pH of 8.0, the HCO,- concentration

was 2.92 mmol/L; at the adjusted pH value
of 5.8 and 4.2, HCO,- was still significant
at 0.66 and 0.026 mmol/L, respectively.
Considering that a uranium concentration
of 120 pg/L translates to 0.0005 mmol/L
UO, (CO,)O or UOz2+, much more than the
required stoichiometric
amount of HCO,was available in the water to react with
the resin to produce the additional CO,zneeded to progressively
complex the uranium according to Eq 5 or a combination
of Eqs 6 and 8.
pH 4.3. Run 33.1, as presented in Figure 6, was conducted at an adjusted feed
pH of 4.3 f 0.1 for 340 h with a total
throughput of 6,800 BVs. Lowering the
feed pH to 4.3 with HCI drastically
impaired the uranium removal performance of the anion bed. Only about 50
percent uranium removal was achieved
right from the start of the run. Significant
uranium breakthrough began :at about
4,000 BVs and was complete at about
6,000 BVs when the column was apparently in equilibrium with the feedwater
and no uranium was being removed. This
early breakthrough is in sharp contrast to
the excellent performance at pH 8.0,
when the macroporous resin was operated to 302,000
BVs, at which point the uranium leakage was still
only 5 percent of the feed uranium concentration.
Although poor, this 50 percent uranium removal
was better than expected for an anion resin, basedon
the cationic nature of the UOz2+ present in the feedwater at pH 4.3.
Here again, the explanation for uranium uptake is
basedon the known high affinity of strong baseanion
resinsfor divalent and polyvalent anions, which, under
appropriate conditions, can be produced within the
resin.2aJ2The suggestednet reaction for the uptake of
cationic U022+ by chloride form anion resin is
2 RCl + 2 U022+ + 4 HCO,- = R,UO,(CO,),

+ H2C03
taq)
+2C1-+2H+
t U02C03
(9)
In Eq 9, the divalent-selective resin sites in conjunction with HCO,- ions in the water promote the
formation and uptake of divaIent UO2(CO3)2- from
cationic U022+, with the corresponding production of
H,CO, and HCl. In contrast to Eq 5, Eq 9 produces a
neutral UO2CO3 molecule not retained by the resin.
This helps to explain the 50 percent uranium leakage
observed at pH 4.3 which was not observed at the
higher, but still acidic, pH of 5.8.
The protons produced by Eq 9 will either add to
the H+ concentration or be consumed by HCO,- to
form H,CO, (aq). To convert the U022+ to a 50-50
mixture of UO~CO~O-IJO~(CO~)~- would require 2

APRIL

1994

235

mmol HCO,-/mmol
UO12+. In the Chimney Hill water
this conversion would produce about 0.001 mmol/L
H+, which
might easily be consumed
by what
remained available of the original 0.026 mmol/L
HCO,-. In summary, the carbonate species fraction
changes and the pH changes produced by Eqs 5 and
9 would not be measurable
in the bicarbonatebuffered Chimney Hill water because of the stoichiometrically
small amount of uranium present.
Figure 7 shows the uranium-dumping
effect
caused by reducing the feedwater pH to 4.4 after partially exhausting the anion resin at pH 8.0. This column was first operated at pH 7.6-8.2 for 915 h with
a total of 41,100 BVs of throughput, during which the
maximum
uranium leakage was ~0.1 pg/L. Then,
the pH of the feedwater was reduced and controlled
at 4.2-4.5 by adding 0.2 N HCl. Serious uranium
leakage (60 pg/L) was observed immediately
after
the pH was reduced. The uranium concentration
in
the effluent rose sharply to 500 from 0.1 pg/L within
400 BVs after lowering the pH. At 1,600 BVs following the pH reduction,
the uranium concentration
reached 2,000 pg/L, i.e., 17 times the feed concentration. Clearly, lowering the pH caused uranium to
be dumped from the partially exhausted SBA resin.
The proposed mechanism
of dumping is shown in
Eq 10.
ZR,UO,(CO,),

+ 4H,CO,

-UOz2+
+ 2HCO,+ R,UO,(CO,),
+ 6 RHCO,
(10)

In Eq 10, after partial exhaustion at pH 8.0, eight

resin sites (2 x R4+) are occupied by two uranyl tricarbonate complexes [2 x U02(C0,)a4-1.
When the
pH 4.4 feedwater containing H,CO, contacts the sites

236

JOURNAL

AWA

occupied by the uranyl tricarbonate complexes,

U022+ is
released
to the aqueous
phase, while the resin retains
some uranium as the divalent
dicarbonate
complex-UO,
(COs)22-. During this reaction,
the pH increases slig,htly as
the H,CO,
is neutralized
(converted to HCO,-) by the
uranyl carbonate complexes.
It is known
that not all the
uranium
is eluted from the
resin at pH 4.3 because of the
results from run 3 3.1 (Figure
6) just discussed. During run
33.1, virgin resin achieved 50
percent uranium removal at
pH 4.3.
In summary, the uptake of
uranium
from
carbonate
buffered solutions containing
predominantly
cationic and
neutral uranium complexes at
pH 4.2-6.0 can be explained
by ( 1) the very high affinity of strong base resins for
polyvalent anions in general and uranyl carbonate
complexes in particular, (2) the ability of strong base
resins to produce divalent CO,2- from HCO,-, and (3)
the relatively high buffer capacity of the carbonate
species present. Resins containing uranyl carbonate
complexes exhibit significant buffer capacity. Such
uranium-loaded
resins can release uranium into the
aqueous phase when the feed pH is lowered significantly, i.e., to ~4.6 in these experiments.
Effects of uranium and sulfate concentrations
on uranium breakthrough.
Uranium feed conccntration was expected to have a big influence on column capacity; however,
because of the extremely
long runs and the potential regulatory problems associated with uranium spiking of the raw water, it was
not practical to vary the uranium concentration
in
the Chimney Hill water to determine its influence
on bed volumes to uranium
breakthrough.
Thus,
computer predictions of uranium breakthrough
were
made using the Equilibrium Multicomponent
Chromatography Program with Constant Separation Factors (EMCT-CSF).23
This program is a user-friendly
implementation
of the multicomponent
chromatography theory of Helfferich and I<lein,2* which has
been used by one of the authors,
Clifford,25
for
selected ion separations.
Using this program, the bed volume throughputs
to uranium breakthrough
were calculated at various
uranium concentrations
in the Chimney Hill water.
The results from the computer prediction are given in
the Figure 8 curve labeled Chloride
water. This
curve predicts that for a water containing chloride
and bicarbonate but no sulfate, like the Chimney Hill
feedwater, an increase in the uranium concentration
to 240 pg/L would result in a decreased run length of

203,000 BVs. Decreasing the uranium concentration

in groundwaters containing the usual carbonate and
to 20 pg/L would increase the run length to 815,000
bicarbonate concentrations. This is the justification
BVs. This high sensitivity of run length to uranium
for excluding uranyl sulfate complexes from the
concentration is due to the fact that uranium is an
computer predictions.
exceptionally highly preferred species and occupies a
Regeneration.
Effect of NuCZ concentration.
The
significant fraction of the resin sites at exhaustion.
results of the pure NaCl regenerations on columns
The EMCT-CSF program was also used to pre11 and 27 are summarized in Figures 9-12. NaCl
diet the influence of sulfate on uranium run length
concentrations ranging from 0.8 to 4.0 N were applied
at variable uranium concentrations. In this simulation,
sulfate replaced chloride in the
Chimney Hill water, which is
bough not well documented,
several
naturally almost free of sulfate. The line labeled Sulfate
rail community
water supplies have
water in Figure 8 shows that
uranium concentrations
in the high
replacing chloride with sulfate
greatly affects the SBA resins
micrograms-per-litre
range-even
up
capacity for uranium removal.
At the natural uranium conto the milligrams-pelclitre
range.
centration of 122 ug/L, if the
sulfate concentration increased
from 0 to 1.32 meq/L (64 mg/L), run length drops
to columns partially exhausted to 30,000 or 41,000
from the experimental
340,000 BVs to a predicted
BVs to establish the effect of regenerant concentration
135,000 BVs, a decrease of 60.3 percent. This
on uranium recovery efficiency. Although the NaCl
decrease at higher sulfate concentration
is attribconcentration varied, the total amount of NaCl applied
uted to the fact that compared with chloride, sulfate
was maintained constant at 4.0, 6.0, or 10 eq NaCl/eq
ion has a strong affinity for the SBA resin sites.
resin (14.4, 2 1.6, or 36 lb NaCl/cu ft resin) during
Therefore, water containing significant concentrathree sets of experiments.
tions of sulfate will produce significantly
shorter
Figures 9-12 demonstrate that uranium recovery
run lengths than low-sulfate waters. As shown in
efficiency strongly depends on the regenerant NaCl
Table 1, the stability constants of the uranium sulconcentration. Increasing NaCl concentration resulted
fate complexes [U0,(S0,),4and U0,(S0,),2-]
are
in increased uranium recovery at a fixed regeneration
much lower than those of uranium carbonate comlevel. As shown in Figure 9, eluted uranium exhibplexes. Thus, it is expected that uranium sulfate
ited a narrow peak at high NaCl concentration and a
complexes will not occur in any significant degree
relatively broader peak at low NaCl concentration. At
a regeneration level of 6.0 eq
NaCl/eq resin (2 1.6 lb NaCl/cu
ft resin), the recovery effiBed volume
throughput
calculated
for a resin column
before
uranium
,)I&&
8
cienciesfor4.0, 3.0,2.0, 1.33,
breakthrough
at different
concentrations
of uranium,
chloride,
1
1.0, and 0.8 N NaCl regenerand sulfate
ant were 91, 86, 75, 67, 54,
and 47 percent, respectively,
as shown in Figure 10. The
uranium
recovery did not
appear to level off at 4.0 N,
which suggests that even
higher NaCl concentration
might more closely approach
100 percent uranium recovery. Thus, the use of saturated
NaCl (6.15 N at 2OoC) to
regenerate uranium-spent
columns seems appropriate.
Effect of NaCZ regeneration level. The results of using

two NaCl regeneration levels (4.0 and 6.0 eq Cl-/eq

resin) are also shown in Figure 10, which indicates that
uranium recovery depends
less on regeneration
level

APRIL 1994

237

FlOURa-4)

Uranium
elution
by OS, 1.0, 2.0, or 3.0 N NaCl during first cocurrent
*a 30,000
BVs; regeneration
regeneration
of a resin bed erh=l=*ad
level-10
eq Cl-/eq
resin

applied was less than 5 percent of the total uranium

recovered. The data in Figure
9 suggest that no more than
about 2 BVs of concentrated
regenerant
are required to
efficiently elute uranium. For
2.0 and 3.0 N NaCl, 2 BVs of
regenerant
translate to regeneration levels of 4 and 6
eq NaCl/eq resin (14.4 and
21.6 lb NaCl/cu
ft resin),
respectively.
It is clear from Figure 9
that a dilute regenerant, e.g.,
0.5 N NaCl, is quite inefficient;
after about 5 BVs, very little
additional uranium is eluted
by the final 15 BVs of regenerant applied. The inefficiency
of dilute regenerants is further verified in Figure 10, in
which 0.8 N NaCl is shown to
recover only about 46 percent
of the uranium
even at a
regeneration
level of 6 eq
Cl-/eq resin.
Base and acid regeneration. Samples of resins from

than on the NaCl concentration.

Nevertheless,
increasing the regeneration level from 4.0 to 6.0 eq
Cl-/eq resin did tend to improve uranium recovery,
with bigger improvements noted at 1.3 N and higher
NaCl concentrations.
In Figure 9, it is clear that uranium was mostly desorbed in the first few bed volumes during regeneration, especially at higher NaCl concentrations.
For
example, examination of the elution curve during 2
N NaCl regeneration shows that the area under the
portion of the curve from 2 to 5 BVs is <5 percent of
the total area under the curve. Thus, the uranium
eluted by the final 60 percent of the regenerant

238

JOURNAL

AWWA

columns 11 and 27 were also

regenerated by mixtures of
NaCl with NaOH, NaHCO,, or
Na,C03 to investigate
the
effects of alkaline regenerants
on regeneration
efficiency.
The results from the NaCl +
NaOH regenerations are summarized in Figures 11 and 12.
They clearly show that the
addition of NaOH to the NaCl
regenerant greatly decreases
the uranium
recovery efficiency. Given that the total
amount of regenerant used
equals 6.0 eqleq resin and
NaCl:NaOH
at a 2:l molar
ratio, the recovery efficiency
by 4.0 N NaCl dropped from
9 1 percent to 15 percent as a result of NaOH addition.
For the other five NaCl concentrations ranging from
0.8 to 3.0 N NaCl, the recovery efficiency dropped
below 10 percent with a decrease ranging from 80
to 90 percent upon the addition of NaOH. The addition of NaHCO, or Na,CO, to NaCl regenerant also
reduced uranium recovery percentage but to a lesser
extent than NaOH.
Overall, for all the NaCl concentration levels tested,
addition of NaOH, NaHCO,, or Na,CO, resulted in
decreased uranium recovery percentages from the partially uranium-exhausted
SBA resin. This is attributed
to the formation of zero-charged uranium precipitates

at high pH values after the addition of strong base. The formed

precipitates were then retained
in the resin pores or stuck to
the resin structure and were
not eluted by the regenerant,
which caused the substantially
lower uranium concentration
in the regeneration elute. As
reported, in uranium mining
practice, strong base is added
to the regenerant elute to precipitate uranium. 17,26-29During regeneration by NaCI, the
uranyl tricarbonate complex is
first replaced from resin sites
by chloride. Upon the addition
of strong base, the regenerated
uranyl tricarbonate ions react
with NaOH and produce
sodium pyrouranate or uranium yellow, Na2U20,,26J7
which, like all the other
uranates, is insoluble. The suggested reactions during this process are as follows:
R,UO,(CO,),

2 U02(C0,),4-

+ 4 NaCl-4

RCl
+ uo2(co3)34+4Na+

R,UO,(CO,),

(11)

+ 6 OH- + 2 Na+ = Na,U,O, (s)

+ 6 C0,2+ 3 H,O
(12)

In Eq 11 uranyl tricarbonate is eluted from the

resin by excess chloride. Then, in Eq 12 the eluted tricarbonate combines with hydroxide to form the insoluble pyrouranate. Adding Eqs 11 and 12 yields:
2 R,UO,(CO,),

+ 8 NaCl+ 6 NaOH= 8 RCI

+ Na2U207 (~1
+ 6 C032+ 3 H,O
+ 12 Na+ (13)

Along with these reactions, the production of uranium oxides is also possible. 17,28-30According to this
mechanism, uranyl tricarbonate generated during
regeneration may react with NaOH and produce insoluble uranium oxides in the presence of strong base.
The suggested reaction is
U02(C03)34-

+ 2 NaOH = UO, (s)

+ 3 CO,2- + 2 Na+
+ H,O
(14)

For Eq 14, there may be some uranyl hydroxide

complexes formed as intermediate
products, which
hydrolyze to UO, at very high pH values.izJs-30
Adding Eq 11, the elution of uranium tricarbonate, to
Eq 14 gives the net reaction

+ 4 NaCl + 2 NaOH= 4
t
t
t

RCl
UO, (s) + H,O
3 co,26 Na+
(15)

When reducing agents are present, the production

of U,Os, a mixture of UO, and UO,, is possible.
In summary, Eqs 11-15 explain how insoluble
uranium compounds including sodium pyrouranate
or uranium oxides are produced from uranyl tricarbonate complex during regeneration
by NaCl
with the addition of NaOH. Similar reactions can
be written
for uranyl
dicarbonate
complex,
2-,
when
it
is
released
from
the resin
u2(co3)2
sites. It appears that the resulting uranium precipitates are mostly retained in the resin, which substantially
reduces the uranium
recovery during
regeneration.
In contrast to the negative effect of NaOH on
regeneration by NaCl, pure HCl exhibited a slightly
higher uranium recovery than pure NaCl at 1 N concentration. This is probably because under acidic conditions the uranyl tricarbonate complex is converted
to cationic U022+, which is easily eluted from the
positively charged anion resin. These results agree
with the prediction that the addition of HCI to NaCl
regenerant will enhance uranium recovery.al However, it is questionable whether the small increase in
uranium recovery by HCI is worth the increased
chemical and corrosion prevention costs.
Effect of percentage resin exhaustion.
Because
resin beds are seldom if ever operated to 300,000
BVs, SBA resins were exhausted to 500, 31,000, or
41,000 BVs and then regenerated with 6 eq 2.0 N
NaCl to study the effects of the degree of uranium
exhaustion on the regeneration efficiency. Compared
with run E4.1 which was terminated at 302,000 BVs,

APRIL

1994

239

FMMJRE

f2

Influence
recovery

runs 11 and 27, which were operated only to 3 1,000

and 41,000 BVs, respectively, were far from complete uranium exhaustion.
When the beds were
regenerated with 6 eq of 2.0 N NaCl, uranium recovery efficiency was constant in the range from 75 to 78
percent, irrespective of the run length of 50041,000
BVs. The uranium leakage after cocurrent resin regeneration was insignificant and similar to that of the
virgin column for all three runs.
Summary
and conclusions
This uranium removal study was conducted on a
natural, well-buffered,
and almost sulfate-free
groundwater containing 11 O-l 2 5 pg/L uranium and
25 pCi/L radium. The macroporous SBA resin used
exhibited enormous capacity for removing uranium.
A bed of this resin was operated continuously
for
478 days at pH 7.6-8.2 for a total throughput
of
302,000 BVs, at which time its effluent uranium
concentration
was still <6 pg/L, i.e., more than 95
percent removal. The total uranium loading on the
column was 30 g /L resin, which corresponds to 7.8
x lo4 pCi/g dry resin.
Predictably, lowering pH reduced the amount of
uranium
removal by anion resin, but the effect
occurred at lower-than-anticipated
pH. At a feed pH
of 4.3, 55 percent uranium leakage was observed
from the start of the run, and a virgin column was
completely exhausted at only 6,000 BVs. For resin
that had been partially exhausted at pH 8, a reduction in the feedwater pH to 4.3 caused massive uranium desorption such that the effluent uranium after
1,000 BVs reached 1,400 pg/L, i.e., 12 times higher
than the feed concentration. However, lowering feedwater pH to 5.8, at which the predominant
species
was the neutral complex, U0,C030, had no significant effect on uranium removal performance of a
column partially exhausted for 41,000 BVs at pH

240

JOURNAL

AWWA

8.0. Apparently,
the resins
strong affinity for polyvalent
u02(C0,)22- and UO, (CO,),+
promotes the formation and
sorption of these species in
the resin using the HCO,ions in the feedwater.
During the regeneration
study, it was determined that
the recovery of uranium from
exhausted strong base anion
resin depended more on the
concentration
of NaCl than
on the regeneration level, i.e.,
than on the mass of salt
applied. At a regeneration
concentration
of 6 Cl- per
exchange site, the recovery
efficiencies for 4.0, 3.0, 2.0,
1.33, 1.0, and 0.8 N NaCl
were 91, 86, 75, 67, 54, and
47 percent, respectively. At 1
N concentration,
HCl exhibited a slightly higher uranium recovery than NaCl.
Uranium was mostly desorbed in the first few bed
volumes of regenerant. When 2.0 N NaCl was used,
95 percent of the uranium elution occurred in the
first 40 percent of the regenerant applied, i.e., 4.0 eq
Cl-/eq resin. The same was true for 1.0 and 0.8 N
NaCl. Although not tested in this research, the results
suggest that an efficient regenerant for uranium
removal would be saturated NaCl (6.15 N) at a level
of 4 eq Cl-/eq resin (14.4 lb NaCl/cu ft resin).
Addition
of base to the NaCl solution greatly
reduced the recovery of uranium. When NaOH represented one third of the regenerant equivalents at 4.0
N regenerant concentration, uranium recovery efficiency dropped from 91 to 15 percent. Reductions
in regeneration efficiency were also observed when
Na,CO, and NaHCO, were added to NaCl.
Uranium recovery efficiency was not affected by
the degree of exhaustion when the anion resin was
far away from complete exhaustion. Columns that
were exhausted to 500 BVs exhibited the same uranium recovery percentage as columns exhausted to
30,000 and 40,000 BVs.
Sulfate, although almost absent from the test water,
is expected to have-major influence on uranium run
length. A computer calculation using the EMCT-CFS
program predicts that if sulfate is substituted for the
1.32 meq/L chloride in the Chimney Hill water, the run
length to uranium breakthrough
will drop 60 percent, i.e., from 340,000 to 136,000 BVs.

of regenerant
stoichiometry
and NaOH addition
on uranium
during
regeneration
with 2.0 PI NaCl of resin exhausted
to

Acknowledgment
This research was made possible through a cooperative research agreement (CR-81 3 148) between
the University of Houston and the US Environmental Protection Agency, Drinking Water Research Division, Cincinnati, Ohio. The authors are grateful to
the project officer, Tom Sorg, for his advice and to

the members of Chimney Hill MUD Water Board,

especially Jay Sinzei, president, for their cooperation
and support.

References
1. US Environmental
Protection Agency, National
Primary Drinking Water Regulations; Radionuelides; Advanced Notice of Proposed Rulemaking.
Fed. Reg., 56:138 (40) CFR Parts 141 and 142
(1991).
2. SORG, T.J. Methods for Removing Uranium From
Drinking
Water. Jour. AWWA, 80:7:105 (July
1988).
3. CLIFFORD, D.A. 6 ZHANG,
Z. Modifying
Ion
Exchange for Combined Removal of Uranium
and Radium. Jour. AWWA, 86:4:214 (Apr. 1994).
4. LANGMUIR, D. Uranium Solution-Mineral
Equilibria at Low Temperatures With Applications to
Sedimentary
Ore Deposits.
Geochemica et Cosmochimica ( 1978).
5. MO, T.J. Chemistry of Uranium in Aqueous Environments. USEPA Ofce. of Radiation Programs,
Criteria 6 Standards Div., Radioactive
Waste
Standard Br. ( 1980).
6. BRUSILOVSKII, S.A. Tr. IGEM Akad. Nauk SSSR
(1960).
7. KOMAR, N.P. Osnovy Kachestvennogo Khimicheskogo
Analiza, 1 (Fundamentals of Qualitative Chemical
Analysis, I). Kharkov (1955).
8. SCANLAN, J.P. Equilibria in Uranyl Carbonate Systems-11, The Overall Stability Constant of
U0,(C03)22- and the Third Formation Constant
of U02(C0,),4-. Jour. Inorg. Nuclear Chem.,39:635
(1977).
9. NELSON, F. 6 KRAUS, K.A. Chemistry of Aqueous
Uranium (v) Solutions. III. Jour. Amer. Chem.,
73:5:2157 (1951).
10. YATSTMIRSKII,
K.B. 6 VASILYEV,
V.P. Konseanty
Neotoikosti Kompleksnykh Soyedinenii (Instability Constants of Complex Compounds). Izd.
Akad. Nauk. (1959).
11. BROWNE, E.; FIRESTONE, R.B.; &- SHIRLEY, V.S. Table
of RadioactiveIsotopes.John Wiley & Sons, New
York (1986).
12. LONGTIN,
J.P. Occurrence of Radon, Radium,
and Uranium in Groundwater. Jour. AWWA,
80:7:84 (July 1988).
13. LOWRY,
J.D. S LOWRY,
S.B. Radionuclides in
Drinking Water. Jour. AWWA, 80:7:51 (July 1988).
14. BONDIETTI,
E.A.; LEE, S.Y.; &- HALL, S.K. Methods of Removing Uranium From Drinking
Water: II-Present
Municipal Treatment and
Potential Removal Methods. EPA 570/9-82003 (1982).
15. LASSOVSZKY, P. 6 HATHAWAY, S.W. Treatment Technologies to Remove Radionuclides From Drinking Water. Presented at the USEPA Natl. Workshop on Radioactivity in Drinking Water, May
23-26, 1983.
16. HATHAWAY,
S.W. Uranium Removal Processes.
Proc. 1983 AWWA Ann. Conf., Las Vegas, Nev.

17. MILLER,W.S. 6 MINDLER,A.B. Ion Exchange Separation of Metal Ions From Water and Waste
Waters. RecentDevelopmentin Separation Science.
CRC Press, Cleveland, Ohio (1977).
18. JELINEK, R.T. S SORG, T.J. Operating a Small FullScale Ion Exchange System for Uranium
Removal. Jour. AWWA, 80:7:79 (July 1988).
19. Standard Methodsfor the Examination of Water and
Wastewater. APHA, AWWA, and WPCF, Washington, D.C. (16th ed., 1985).
20. HORNG,
L.L. Reaction Mechanisms and Chromatographic Behavior of Polyprotic Acid Anions
in Multicomponent Ion Exchange. Doctoral dissertation, Univ. of Houston, Texas ( 1983).
2 1. SNOEYINK,
V.L. 6 JENKINS, D. Water Chemistry.John
Wiley S Sons, New York ( 1980).
22. SENGUPTA,
A.K. Chromate Ion-exchange Study
for Cooling Water. Doctoral dissertation, Univ.
of Houston, Texas (1984).
23. CLIFFORD,
D.A. 6 MAJANO,
R.E. Computer Prediction of Ion Exchange. Jour. AWWA, 85:4:20
(Apr. 1993).
24.
HELFFERICH,
F. 6 KLEIN, G. Multicomponent Chromatography. Marcel Dekker, New York (1970).
25.
CLIFFORD,
D.A. Multicomponent Ion-Exchange
Calculations for Selected Ion Separations. Jour.
Indus. Engrg. Chem.Fundamentals, 2 1 (1982).
26. BAILAR,
J.C. JR. ET AL. ComprehensiveInorganic
Chemistry. Pergamon Press, London (1973).
27. THORNE, I?C.L. 6 ROBERTS, E.R. Inorganic Chemistry.
Interscience Publishers,New York (6th ed., 1954).
28. SMITH,S.E. S GARRE?T,
K.E. Some Recent Developments in the Extraction of Uranium From Its
Ores. Chem.Engrg. Jour. (Dec. 1972).
29. The Uranium Institute. Ufanium and Nuclear
Energy. Proc. 4th Intl. Symp., London, Sept.
10-12, 1979.
30. HANSON,
S.W.; W.JILSON,
D.B.; 6 GUNA.JI,
N.N.
Removal of Uranium From Drinking Water by
Ion Exchange and Chemical Clarification. EPA
Contractor CR-810453-01-0,
EPA 6OO/S287/076. Cincinnati, Ohio ( 1987).
31. MINDLER,
A.B. & TERMINJ, J.P.K. The Vital Role
of Ion Exchange in Uranium Production. Engrg.
&Mining Jour., 157:9 (Sept. 1956).
About the authors: Zhihe Zhang
isa postdoctoralresearchassociatein
the Department of Environmental
Engineer,University of Houston,4800
Calhoun St., Houston, TX 772044791. A graduate of the ChineseUniversity of Scienceand Technology,
Peking, with a bachelork degreeand
of the University of Houstonwith a doctoral degree,Zhang
has more than 15 years of experience in the water and
wastewater treatment, power, and petroleum-chemical
industries. Zhang is a member of AWWA and ACS, and
his work has beenpublishedpreviously by Journal AWWA.
DennisClifford isprofessorof environmental engineeringat
the University of Houston.

APRIL1994

241