A COMPARISON OF PROCEDURES TO DETERMINE ADSORPTION CAPACITY
OF VOLATILE ORGANIC COMPOUNDS ON ACTIVATED CARBON
Richard J. Miltner
Drinking Water Research Division
Municipal Environmental Research Laboratory
and
0. Thomas Love, Jr.
Water Supply Branch, Region VI
Dallas, Texas 77570
WATER ENGINEERING RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
Cincinnati, Ohio 45268
-------
A COMPARISON OF PROCEDURES TO DETERMINE ADSORPTION CAPACITY
OF VOLATILE ORGANIC COMPOUNDS ON ACTIVATED CARBON
Richard J. Miltner
Environmental Engineer
Drinking Water Research Division
Water Engineering Research Laboratory
Cincinnati, Ohio 45268
0. Thomas Love, Jr.
Sanitary Engineer
Water Supply Branch, Region VI
Dallas, Texas 77570
United States Environmental Protection Agency
A number of volatile organic compounds (VOCs) are regulated or
are under consideration for regulation in drinking water. Maximum
contaminant levels for trihalomethanes were promulgated in 1979.*
Recommended maximum contaminant levels for nine VOCs were proposed
in 1984,2 and numerous compounds, volatile in nature, were cited in an
Advanced Notice of Proposed Rulemaking in 1983.^ In addition to aeration,
adsorption on-to granular activated carbon (GAC) is a cost-effective
treatment process for many of these compounds.^ As a result, labor-
atories will seek to develop GAC adsorption capacity data as one step
in.the process of comparing adsorption to other processes, in order to
make comparisons among carbons, or as the first step in the design of
the adsorption process.
A review of the literature found little discussion on stan-
dard procedures for adsorption capacity tests, but much discussion and
presentation of adsorption capacity data. A significant contribution in
this area was made by Randtke and Snoeyink, wherein design and interpre-
tation of GAC adsorption capacity tests were examined.5
This paper presents GAC adsorption capacity data, expressed as
equilibrium isotherms, for a number of VOCs under regulatory considera-
tion. Several procedures were employed; each is described and resulting
data and procedural ease are compared.
MATERIALS
Adsorbent Filtrasorb 400* was employed. It was stored in
manufacturer's 50 pound bags or in glassware under Teflon seal. After
sieving, samples were stored in glassware under Teflon seal.
Adsorbates All VOCs were reagent-grade quality. VOCs were stored
in their manufacturer's containers, or aliquots were taken and stored in
Teflon-sealed glassware.
Solution Waters Distilled water and two field waters were studied.
Field waters were filtered though a bed of sand or mixed-media bed of
sand and anthracite coal. One, Ohio River water, provided a moderate
total organic carbon (TOC) background. The other, a gravel pit water,
provided a relatively high TOC background. A capillary column, gas-
*Mention of commercial products does not constitute approval or endorse-
ment by U.S. EPA.
-------
chromatographic profile found the latter to be heavily laden with low-
to mid-molecular weight organic compounds relative to typical surface
water profiles.^ These waters are described in Table 1. All waters
were stored in glass under Teflon seal.
Reaction Vessels All contact surfaces were glass, Teflon, or
stainless steel. Glassware was distilled water washed and fired to
O
400 C; stainless steel and Teflon were distilled water rinsed several
times and air dried. Other materials used in pumps or meters were not
observed to produce interfering chromatographic peaks or to exhibit VOC
demand.
ANALYTICAL METHODS
A purge-and-trap gas chromatographic procedure was used to determine
VOC concentrations.^ TOC concentrations were determined by using a
Dohrmann/Envirotech DC-54 low^level analyzer.
EXPERIMENTAL PROCEDURES
VOC Preparation Microliter volumes of VOC were injected into
water during mixing. Mixing was briefly stopped, water was added to
eliminate a headspace, the vessel was Teflon sealed, and mixing continued
until steady-state, mg/L concentration stock solutions were achieved.
Volume dilutions were mixed in a similar manner until ug/L concentration
working solutions were achieved. Selected VOCs were studied to determine
the required mixing time' for stock solutions of mg/L concentrations;
VOC stock solutions were typically mixed two days.
Care was taken to dose below solubility limits and to avoid the use
of intermediate co-solvents to aid in solution. A co-solvent was used,
however, in the case of lighter-than-water VOCs. An ethanol/ water,
V/V ratio of approximately 1/8000 was used in preparing mg/L stock
solutions. Ethanol has been reported to be non-adsorbable®, thus it
was not believed to compete with VOCs for adsorption sites.
GAC Preparation The adsorbent was grab sampled from 50-pound
bags, placed in a ball mill, pulverized for 30 days and graded using
US-standard sieves and a sieve shaker. 50x200 mesh adsorbent was
stored in Teflon-sealed glassware. Distilled water was added to a
weighed amount of 200x400 mesh adsorbent until desired volumes were
reached. The slurries were mixed for three days and stored in glass-
stoppered bottles. Non-graded, grab-sampled adsorbent was termed
12x40 mesh as per manufacturer's specifications.
Adsorption Capacity Isotherm Procedures Two procedures were
employed. One was a bottle (250 mL) isotherm, batch procedure using
three mesh sizes: 200x400, 50x200 or 12x40. The other was a large
volume (100 gallon), recirculating procedure, termed macroisotherra,
using only 12x40 mesh GAC. The macroisotherra procedure attempted
to evaluate whether capacities developed from large-volume isotherms
better correlated with field capacities than those from bottle-size
isotherms.
Bottle Isotherms Required volumes of slurry containing 200x400
mesh adsorbent were added to 250 mL centrifuge tubes via dispensing
pipets, providing a range of adsorbent doses. As required, make-up
water was added to each dosed tube or undosed control tube to keep VOC
dilution constant. Ug/L working solutions of VOC were pumped into each
-------
tube. The tubes were Teflon-sealed without headspace. Mixing was
provided by a device that tumbled the tubes end-over-end at room
temperature at a rate of 10-12 rpm. At the termination of tumbling,
tubes were centrifuged at 2000 rpm for 60 minutes at constant temperature
(20-21 ). Following centrifugation, samples were withdrawn by pipet
and transferred to Teflon-sealed vials. For selected VOCs, control
tubes were withdrawn during tumbling, centrifuged and sampled to verify
no loss of VOC over the reaction period. For selected VOCs, identically
GAC-dosed tubes were withdrawn during tumbling, centrifuged and sampled
to establish the required time to equilibrium. Equilibrium was defined
as the time beyond which changes in subsequent concentrations were
within expected analytical precision for the procedures employed,
generally +6%. ^
Control of volatilization during adsorbent separation was necessary.
Paper filtration, with or without a vacuum, was not attempted, assuming
excessive losses. Filter-disk separation was abandoned after significant
volatile loss of trichloroethylene (TCE) as observed. No attempt was
made to determine whether the loss was attributable to an uptake of TCE
by the filter material or whether volatilization of TCE was enhanced by
passing the filters. Centrifugation was found to produce no loss of
VOC that could be distinguished from expected analytic precision.
Required" weights of 50x200 mesh adsorbent were added to 250 tnL screw-
cap bottles, providing a range of adsorbent doses. Ug/L working solutions
of VOC were pumped into each dosed or control bottle. The bottles were
Teflon-sealed headspace-free and tumbled as described. Following
tumbling, the bottles stood for 16 to 20 hours to allow for gravity
settling of the adsorbent. Following settling, samples were collected
by pipet, as described. Both VOC stability over the reaction period
and time to equilibrium were determined, as described. Bottle isotherms
utilizing 12x40 mesh adsorbent were handled in an identical manner,
with the exception of gravity settling time, which was rapid.
Macroisotherms In this procedure, VOC-spiked water was pumped
from a 100-gallon stainless steel tank, passed through a meter, up
through a column containing 12x40 mesh adsorbent, and returned to the
tank (see Figure 1). The vessel was filled with water in an upflow
manner, with the headspace being displaced along the walls of the
conical top through the funnel. During recirculation, a mg/L concen-
tration VOC stock solution was introduced at the apex while an identical
volume of water was withdrawn at a sample tap to maintain a headspace-
free condition. Recirculation continued until a ug/L concentration
working solution was uniformly diluted throughout the vessel. Control
tests with selected VOCs showed uniform dilution to occur in two hours
or less. 12x40 mesh adsorbent was weighed, added to the column, flushed
for several minutes to fill void spaces, affixed to the vessel, and
recirculating water was directed from the bypass line to the column.
The vessel was sampled below the column until the VOC concentration was
shown to have reached equilibrium. As a sample was collected, unspiked
make-up water was introduced from the funnel to maintain a headspace-free
condition; the resulting dilution (approximately 1/13000) was insig-
nificant. Equilibrium was defined as described. For chlorobenzene,
samples were collected above and below the column to confirm exhaustion
of the adsorbent. Several vessels were operated in this manner providing
a range of carbon doses. Vessels operated without adsorbent were
sampled over time to verify no loss of VOC during the time required to
reach equilibrium.
-------
RESULTS AND DISCUSSION
Time-to-Equilibrium Determination The time required to reach
equilibrium was determined for several VOCs. Generally, for each VOC
studied in bottle tests, the time required for the smallest adsorbent
weight to equilibrate was determined, as it requires a longer time than
larger adsorbent weights.^ In some cases, a minimum time to equilibrium
was not determined; rather, tests revealed that equilibrium had occurred
before the first, time-sequenced sample was collected. Table 2 presents
mean values of time-to-equilibrium conditions for bottle isotherm
tests. In support of the literature, the required time was dependent
on particle size.^ While a mean three hours or less were required to
reach equilibrium using 200x400 mesh adsorbent, a mean reaction time of
near three days was utilized to ensure equilibrium conditions. Similarly,
longer-than-required times were utilized for larger-mesh-sized adsorbent
to ensure that equilibrium had been reached. A typical time-to-equilibriura
test is seen in Figure 2 where eleven days were required to reach
equilibrium under the cited conditions.
Generally, each macroisothenn vessel was sampled to determine time
to equilibrium for each VOC studied. Using this procedure, the time
required for the smallest adsorbent weight used for each VOC was not
necessarily the longest, as recirculation rate varied Inversely with
headloss through the column which, in turn, varied with adsorbent
weight. Figure 3 shows a typical time-to-equilibrium test requiring
approximately 800 gallons throughput under the cited conditions.
Comparison of Procedures Data were fitted to the Freundlich
equilibrum expression:
1/n
qe = KCe
where qe = adsorbent equilibrium in mg VOC adsorbed/gram adsorbent;
Ce = adsorbate equilibrium in mg VOC/L solution water; and K and 1/n are
experimentally determined constants. Data were fited to the Freundlich
expression using the method of least squares. For comparative purposes,
a prediction band was calculated for the line. The prediction interval
estimates the range of a new qe, for a given Ce, on the basis of the
fitted line. The family of q= intervals over the experimental range of Ce
provides a prediction band.®»* A 95 percent confidence level was chosen.
Procedures were compared by testing whether the Freundlich line fitted to
data from one procedure, i.e., the family of new qe's, fell within the
prediction band fitted to data from another procedure. One such compari-
son is seen in Figure 4 where the data from the bottle isotherm procedure
utilizing 200x400 adsorbent fell within the prediction band about data
from the macroisotherm procedure. Conversely, the latter fell within
the prediction band of the former. Other representative comparisons
are seen in Figures 5 and 6. Generally, data from macroiostherm and
bottle isotherm procedures were comparable using 95 percent prediction
bands, as were data from bottle isotherm procedures utilizing different
mesh-size adsorbent. When procedures were not comparable using 9 5%
confidence levels, they were if 90 percent confidence levels were used.
Because procedures were considered comparable, the macroisotherm
procedure was abandoned in preference to the bottle procedures, as the
latter are less labor-, time-, space- and sample-intensive. These studies
-------
showed scale to be of little importance in developing adsorption capacity
data. Laboratories seeking to develop such data can do so with generally
available laboratory equipment and supplies.
Results Table 3 summarizes relevant adsorption capacity data from
isotherm studies of VOCs in distilled water using Filtrasorb 400. In
addition to the Freundlich constants, the least squares correlation
coefficient and range of experimental data are given. In the application
of these, or any Freundlich data, the user must note that extrapolation
beyond the experimental range of Ce leads to wider intervals about qe than
would be the case if confined within the range. By definition, the
interval about K is relatively wide as K = qe at Ce = 1.0 mg/L, i.e.,
beyond the experimental range of Ce in these studies. Where several pro-
cedures were used, the least squares fit given In Table 3 was made
using the summation of data points, as listed.
Similar data are cited in Table 4 for VOCs in solution waters of
variable TOC concentration. An examination of Freundlich K constants
for tetrachloroethylene (PCE) and TCE suggests that adsorption capacity
decreases with increasing TOC, i.e., with increasing competition for
adsorption sites. Figure 7 supports this, wherein the 95 percent
prediction bands about TCE in distilled water (TOC <0.1 mg/L) and in
filtered Ohio River water (TOC = 1.97 mg/L) demonstrate no coincidence
over the range of experimental data. Both VOCs in filtered gravel pit
water (TOC = 8.39 mg/L) show similar separation of prediction bands
over the experimental range. For 1,1,1-trichloroethane (TCA) and
cis-1,2-dichloroethylene (CIS), however, no significant differences in
capacity appear to exist over the experimental range. An example of
this may be seen in Figure 8 for cis-1,2-dichloroethylene in high- and
low-TOC waters. Why compounds of higher adsorption capacity (PCE or
TCE) are more affected by competative matrices than compounds of lower
capacity (TCA or CIS) is not well understood and needs further investi-
gat ion.
ACKNOWLEDGEMENTS
The authors appreciate the efforts of Kenneth Kropp, James Horton,
Christian Nyberg, Michelle Hamlin, Ihor Melnyk and John Lauch for
conducting isotherm studies, Bradford Smith for performing organic
analyses, Herbert Braxton and Kim Fox for their assistance with graphics,
Leown Moore, James Westrick and Alan Stevens for technical review of
the paper, and Pat Pierson and Maura Lilly for typing the paper. At
the time of these studies, 0. Thomas Love was with the Drinking Water
Research Division.
REFERENCES
1. Federal Register, 44:231:68624 (November 29, 1979); 45:49:15542
(March 11, 1980).
2. Federal Register, 49:114:24330 (June 12, 1984).
3. Federal Register, 48:194:45502 (October 5, 1983).
4. Love, O.T. and Eilers, R. , Treatment of Drinking Water Containing
Trichloroethylene and Related Industrial Solvents, Journal AWWA,
74:8:413 (August, 1982).
5. Randtke, S.J. and Snoeyink, V.L., Evaluating GAC Adsorptive Capacity,
Journal AWWA, 75:8:406 (August, 1983)
-------
6. Stevens, A.A., et al., Gas Chromatographic Techniques for Controlling
Organic Removal Processes, Journal AWWA, 73:10:548 (October, 1981).
7. Methods for Organic Chemical Analysis of Municipal and Industrial
Wastewater: Purgeable Halocarbons -Method 601; Purgeable Aromatics -
Method 602, EPA 600/4-82-057, Environmental Monitoring and Support
Laboratory, U.S. EPA, Cincinnati, Ohio (July 1982).
8. Dobbs, R.A. and Cohen, J.M., Carbon Adsorption Isotherms for Toxic
Organics, EPA 600/8-80-023, Municipal Environmental Research
Laboratory, U.S. EPA, Cincinnati, "Ohio (April 1980).
9. Remington, R.D. and Schork, M. A., Statistics with Applications to
the Biological and Health Sciences, Prentice-Hall, Inc., LC 71-100588
(1970).
10. Draper, N.R. and Smith, M., Applied Regression Analysis, John Wiley
and Sons, Inc., LC 66-17641 (1967).
-------
TABLE 1
Properties of solution waters used in isotherm studies
Water
Mean pH
Mean TOC
mg/L
Turbidity
ntu
Dist illed
Filtered
Ohio River
Filtered
Gravel Pit
6.68
7.98
8.04
0.09
1.97
8.39
< 0.1
<1.0
< 1.0
TABLE 2
Bottle isotherm reaction times, days
Mesh Size
Mean
Required Time
Mean
Test Time
200 x 400
50 x 200
12 x 40
0.125
4.5
15.7
2.8
7.1
23.9
FUNNEL
—SAMPLE
100 GAL TANK
_ GAC COLUMN
UJ
-®-I SAMPLE
Q METER
fill/Drain
~Q~i
PUMP
® VALVE
Figure 1. Ilacroisotherm apparatus.
-------
TABLE 3
Freundlich isotherm values
distilled water, Filtrasorb 400
K
Ce Range
Number
Compound
mg/g
1/n
r
ug/L
Points
Procedures
tetracliloroethylene
143
0.52
0.983
3."6-4 21
29
M,BS
trichloroethylene
56.0
0.48
0.991
7.7-442
17
M,BL
1,1-dichloroethylene
16.5
0.52
0.998
6.9-392
6
M
cis-1,2-dichloroethylene
11.7
0.59
0.981
5.1-615
38
M,BS,BL
trans,1-2-dichloroethylene
17.0
0.51
0.983
14-415
10
BL
1,1,1-trichloroethane
13.2
0.53
0.976
15-860
10
M
1,2-dichloroethane
5.1
0.53
0.984
42-727
9
M
1,1-dichloroethane
7.9
0.64
0.996
26-495
11
BL
carbon tetrachloride
23.6
0.59
0.968
9.1-429
18
M
benzene
50.0
0.53
0.998
3.2-462
10
BL
ethyl benzene
120
0.44
0.958
5.6-158
12
BA.BS
chlorobenzene
101
0.34
0.991
15-353
6
M
o-dichlorobenzene
407
0.50
0.986
14-208
8
M
p-dichlorobenzene
472
0.65
0.891
5.7-284
21
BA, liS
1,2-dibromomethane (EDB)
25.3
0.51
0.924
13-123
12
BA.BS
M = macroisotherra; B = bottle isotherm
S = 200x400 mesh; A = 50x200 mesh; L = 12x40 mesh
-------
TABLE 4
Freundlich isotherm values
variable TOC, Filtrasorb 400
TOC
K
Ce range
Number
Compound
mg/L
mg/g
1/n
r
ug/L
Points
Procedures
tetrachloroethylene
<0.1
143
0.52
0.983
-3.6-421
29
M, BS
8.39
59.2
0.39
0.993
4.6-330
10
BL
trichloroethylene
<0.1
56.0
0.48
0.991
7.7-442
17
M, BL
1.97
31.6
0.44
0.979
13-380
16
BL.BS
8.39
24.4
0.44
0.979
13-399
19
BL, BS
1,1,1-trichloroethane
<0.1
13.2
0.53
0.976
15-860
10
M
1.97
12.4
0.52
0.986
7.6-601
22
M,BA
ci8-l,2-dichloroethylene
<0.1
11.7
0.59
0.981
5.1-615
38
M,BS,BL
1.97
12.7
0.64
0.989
19-230
8
BL
8.39
9.1
0.52
0.993
25-527
9
BL
M = macroisotherra; B «= bottle isotherm
S = 200x400 mesh; A = 50 x 200 mesh; L ° 12x40 mesh
-------
100-]
Co = 805 ug/L
—r-
IS
—I—
ia
—i—
20
—i
22
10
TIME, DAYS
Figure 2. Time-to-equilibrium study using 12x40 mesh adsorbent and
bottle isotherm procedure for cis-l,2-dichlor'oethylene in
distilled water.
Co = 921 ug/L
200
400 aoo »oo iooo uoa v«oo
THROUGHPUT, GALLONS
1100
1100
2000
2200
Figure 3.
Time-to-equilibrium study using 12x40 mesh adsorbent and
macroisotherm procedure for 1,1,1-trichloroethane in
distilled water.
-------
100-,
ca
s~.
s£0
to
a
cd
a.
co
u
10-
tetrachloroethylene
in
distilled water
macrobothe rm
K = 14-3, Vh
95% band
-200 x 400 mesh battle isotherm
K = 146, 1/n = 031
= 034
i i—r i i i i |
1
0.001
-1 1 I I I I I I I
0.01
I I ' I—
0.1
equilibrium concentration, mg/L
Figure 4.
Freundlich isotherms for tetrachloroethylene in distilled
water using different procedures.
10 cis—1.2— dichloroethylene
in
E
e
CO
O)
E V
o
D
CL
0.1-
distilled water
macroisotherm
K = 11.0,1/n = 038
95% band
12 X 40 mesh bottle isotherm
K = 11.5, Vh = 033
-i—i—i—i i 11—
0.01
~t 1 1—I I I |
0.1
'III
1
0.001
Figure 5.
equilibrium concentration, mg/L
Freundlich isotherms for cis-l,2-dichloroethylene in
distilled water using different procedures.
-------
100-1
trichloroethvlene in
filtered Ohio River water
T") v/ / H fcnfharm
200 x 400 mesh bottle isotherm
K = 31.8, Vh = 0.45
95 % band
-i—i—
0.1
-I 1 I I—I
0.01
Figure 6.
equilibrium concentration, mg/L
Freundlich isotherms for trichloroethylene in Ohio River
water using different mesh-size adsorbent.
-trichloroethylene
95 % bands
distilled water
V
>- filtered Ohio River water
K = 31.6, Vh = 0.44
TOC = 197 mg/L
^—i—i—i i i 11
0.001
I I—I—I—I I I I—
0.01
1—I—I—I I I I—
0.1
equilibrium concentration, mg/L
Figure 7. Freundlich isotherms for trichloroethylene.
-------
10-, cis—1.2—dichloroethylene
95% bands
E
E
jy>
CD
E
o
o
Cl
8
1-
Oc
Is
% 4
p
fc88?c
«aiPi
IP5
?5H
0.1-
0.001
filtered gnovel pit water
K = 9.1, Vh = 0.52
TOC = 8.39 mg/L
distilled water
K = 11.6, Vh = 059
TOC < 0.1 mg/L
t 1—r
¦t—n—
0.01
« ' » ¦ I
0.1
I I I I I
1
equilibrium concentration, mg/L
Figure 8. Freundlich isotherms for cis-l,2-dichloroethylene.
-------
TECHNICAL REPORT DATA
fPlease read Irturucnons on the reverse bejore complenng)
1 REPORT NO 2. /. 1
(JUOOS~£j
3. RECIPIENT'S ACCESSION NO
4 TITLE ANO SUSTITLE
A Comparison of Procedures to Determine Adsorption
Capacity of Volatile Organic Compounds on Activated
Carbon
S. REPORT OATE
6 performing organization cooe
7 AUTHORISI
Richard J. Miltner and 0. Thomas Love, Jr.
8 performing ORGANIZATION REPORT NO
9 PER FORMING ORGANIZATION NAME AN 0 AOORESS
Drinking Water Research Division
Water Engineering Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
10 PROGRAM ELEMENT NO.
T! COW TP ACT/C P ANT NC.
12. SPONSORING AGENCY NAME ANO ADDRESS
Same as 9.
13. TYPE OF REPORT ANO PERIOD COVEREC
14 SPONSORING AGENCY COOE
15 SUPPLEMENTARY NOTES
Presented at: American Water Works Association, Uater Quality Technology Conference
Denver, Colorado. Der.pmhpr 3-S 1Qfli Fnr. fnnfprpnro PrnrppHincrc
16 AaSTHACT
, Numerous volatile organic compounds {VOCs) are under regulatory consideration
for inclusion in the National Primary Drinking Water Standards. Adsorption is a
cost-effective treatment technology for control of VOCs. Adsorption capacities
were determined for fifteen VOCs in distilled and field waters using different
procedures. A raacroisotherm procedure, testing 100 gallons of water, was compared
with a bottle isotherm procedure, testing 250 mL of water. While procedures were
comparable, conventional bottle procedures were better suited to water utility
laboratories. Scale was found to have no significant effect on adsorption capacity.
17 KEY WOROS AN0 DOCUMENT ANALYSIS
1 DESCRIPTORS
b I06NT1FIERSVOPEN ENOEO TE^MS
c COSATI Field/Group
13 DISTRIBUTION STATEMENT
19 SECURITY CLASS (Thti Report)
21 NO Of fiAGBS
20 SECURITY CLASS (This pa$e)
22 PRICE
EPA Form 2220-1 (9-73)
------- |