Equilibrium Adsorption of Polycyclic Aromatic Hydrocarbons from Water onto Activated Carbon Richard W. Waiters*+ and Rlchard G. Luthy' Department of Civil Engineering, University of Maryland, College Park, Maryland 20742, and Department of Civil Engineering, Carnegie-Mellon University, Pittsburgh, Pennsylvania 15213
rn Adsorption isotherm data for 11 polycyclic aromatic hydrocarbons (PAH) from water onto Filtrasorb 400 activated carbon a t 25 OC have been generated by batch shake testing. An evaluation of Henry's law, Langmuir, BET, Freundlich, and Redlich-Peterson equations indicates that the Langmuir equation is most useful for representing the data. Fitting the data to two linearized forms of the Langmuir equation provides estimates of the capacity term qo and the energy term b which apply over different equilibrium concentration (C,) ranges. The products of the parameters bqo obtained by using the low C, linear form of the Langmuir equation approximate Henry's law adsorption constants. These values range from 2390 (mg/g)/(mg/L) for naphthalene to 326000 (mg/ g)/(mg/L) for chrysene. The parameter qo obtained by using the high C, linear form of the Langmuir equation reflects limiting adsorption capacity. These values range from 580 mg/g for naphthalene to 14.7 mg/g for benz[alanthracene. Both log (qo)and log (bqo)correlate with making it possible to predict relative log (CJ and log (KO,,,), adsorption of PAH from these latter properties. The adsorption of PAH onto activated carbon is (1)much stronger than is adsorption of PAH onto soils, sediments, and suspended organic matter and (2) comparable to or greater than the adsorption of other hydrophobic organics onto activated carbon.
Introduction Polycyclic aromatic hydrocarbons (PAH) are a ubiquitous class of organic compounds which have been identified in a variety of waters and wastewaters (I,2). In recognition that PAH are generally toxic and that several PAH exhibit carcinogenic behavior, the World Health Organization (WHO) has recommended a limit for PAH in drinking water (3), and the United States Environmental Protection Agency (EPA) Effluent Guidelines Division has included PAH in its list of priority pollutants to be monitored in industrial effluents ( 4 ) . Several mechanisms can be identified that account for removal of PAH from water and wastewater during or following treatment. These are reviewed by Walters (5) and include chemical transformations (e.g., autoxidation, ozonation, chlorination, and photooxidation), biological uptake, volatilization, and adsorption. The removal of PAH by association with wastewater Suspended solids appears to be quite important Department of Civil Engineering, University of Maryland. of Civil Engineering, Carnegie-Mellon University.
t Department
0013-936X/84/09 18-0395$01.50/0
on the basis of results of a recent study (2) which indicate that the distribution of PAH between dissolved and suspended solid phases often favors the suspended solid phase, with suspended solid-phase partitioning for a given PAH occurring to a greater extent with decreasing aqueous solubility (C,) and increasing octanol-water partition coefficient (I&). Thus, partitioning aids PAH removal by adsorption and by removal of suspended solids containing adsorbed PAH. Consequently, an understanding of the adsorptive behavior of PAH is critical in unders h d i n g the environmental control and environmental fate of these compounds. At least five different adsorption isotherm equations have been employed by researchers in previous studies. El-Dib and Badawy (6),McGuire and Suffet (7), and Arbuckle (8)have used the Freundlich equation, Karickhoff et al. (9) and Means et al. (10) have used a Henry's lawtype equation, Belfort (11)and Martin and Al-Bahrani (12) used the Langmuir equation, Radke and Prausnitz (13) used the Redlich-Peterson equation, and Weber (14) discusses the use of the BET equation. This paper presents data and discussion regarding the adsorption of 11PAH from water onto activated carbon. The isotherm equations mentioned above are evaluated by using the PAH data, and relationships between isotherm equation parameters and aqueous solubility and octanol-water partition coefficient are considered in a semiquantitative fashion. Also, the adsorptive behavior of PAH onto activated carbon observed in this study is compared to PAH adsorptive behavior observed in previous studies. Adsorption Isotherm Equations. Five adsorption isotherm equations were considered for fitting the data. These equations are summarized in Table I. Representative plots of each equation are presented in Figure 1. Although each equation contains a term that provides some indication of relative adsorption capacity, each equation is limited in applicability to specific systems or to specific concentration ranges, and it is often misleading to compare adsorption capacity values obtained from the different methods on both an absolute and a relative basis. A review of adsorption isotherms and adsorption isotherm equations is given by Walters ( 5 ) .
Experimental Methodology Materials. PAH standards were used as received from Aldrich and Eastman Kodak. PAH standards were assayed by the manufacturer as having a purity of 98% or
0 1984 American Chemical Society
Environ. Sci. Technot., Vol. 18, No. 6, 1984 395
Isotherm A -Henry's Law Region -Low Surface coverage
qe
Y
Ce
E
Isotherm B
-Type I B E T - Langmuir isotherm - Monolayer a d s o r p t i o n
qe
ce
CS
Isotherm C -Type I I B E T - B E T isotherm - M u l t i l a y e r a d s o r p t i o n or m o n o l a y e r with enhanced c r y s t a l l i z a t i o n near c s
e
h
/Isotherm D
qe
8
Freundlich isotherm
Y
1 h
M M , , M , M ; E" a
I
Ce
Y
Figure 1. Typical shapes of adsorption isotherm curves.
greater. A listing of the 11 PAH used in this study along with various physical and chemical data pertaining to each is summarized in Table 11. Abbreviations used for the PAH are also listed in Table 11. Water used for preparation of aqueous solutions was purified by an ion-exchange and activated carbon bed system from Continental Water Co. Water from this system typically had a pH in the range 6.8-7.2. The acetonitrile, methanol, and methylene chloride were certified UV grade solvents appropriate for use in HPLC analysis and were obtained from Fisher Scientific. The water and solvents were filtered and deaerated under vacuum prior to use. The activated carbon used in this study was Filtrasorb 400 and was obtained from Calgon Corp. To reduce equilibration time, the activated carbon was mechanically ground and classified to 200 mesh (0.074 mm). The powdered carbon was boiled in and washed with organic-free, deionized water, decanted, and oven-dried a t 105 OC overnight prior to use in adsorption studies. Carbon prepared in this manner had the following properties (15): surface area 998 m2/g; apparent density 0.48 g/mL; iodine number 1030; molasses number 312; total pore volume 0.825 mL/g. Analyses. Analyses were performed with a high-pressure liquid chromatographic (HPLC) system obtained from Perkin-Elmer. This system consists of a Series 3 liquid chromatograph and a Model S-210 fluorescence detector. Analyses were performed by injecting an equal-volume mixture of the aqueous sample in methanol onto a PAH-10 analytical column (Perkin-Elmer). Acetonitrile and water were used for elution. Detection was performed by using excitation and emission settings which were optimized for each compound to give the greatest response relative to the base line. Glassware. All glassware used in this study was specially cleaned to minimize contamination. Glassware was washed and rinsed with organic-free, deionized water and was soaked overnight in a 1:l HNO, acid bath. Glassware soaked in this manner was rinsed successively with organic-free deionized water, methanol, and methylene 396 Environ. Sci. Technol., Vol. 18, No. 6, 1984
h
h
h
u" u" u"
-a g -a .e e .3
i
w
0
0
P
E
a
g
s
$ I1
d
P
.c
0
P
Table 11. Data Pertaining to the 11 PAH Evaluated in the Adsorption Study
compound naphthalene
abbreviation
M,
molecular formula
molecular structure
aqueous
solubility" 31.7
log 3.36
NA
128
acenaphthylene
ACY
152
acenaphthene
ACE
154
3.93
4.03
fluorene
FLE
166
1.98
4.47
phenanthrene
PH
178
1.29
4.57
anthracene
AN
178
0.073
4.54
fluoranthene
FLA
202
0.260
5.22
pyrene
PY
202
0.135
5.18
16.1
@ /
3.94
/
triphenylene
TRP
228
0.043
5.45
benz[a]anthracene
BaA
228
0.014
5.91
chrysene
CHR
228
0.002
5.91
"Data reported by Mackay and Shiu (22). Units are milligrams per liter at 25 "C. Value for ACY was determined in the present study. bK,,wis 1-octanol-water partition coefficient. Data reported by Karickhoff et al. (9)for NA, PH, AN, and PY. Values for ACE, ACY, FLE, FLA, TRP, BaA, and CHR were determined by the method described by Yalkowsky and Valvani (29). chloride and oven-dried a t 105 "C. Precautions. Because PAH have low aqueous solubilities (2.0 pg/L for CHR), experiments performed with aqueous solutions of these compounds are particularly vulnerable to contamination, analyte losses, and degradation effects. Contamination was minimized by the methods noted above for preparing the water, carbon, and glassware used in this study. Losses by photodegradation were minimized by covering all containers used to hold PAH solutions with aluminum foil. Losses by oxidation were minimized by deaerating the water prior to use in making PAH solutions, by filling any space above solutions in containers with nitrogen, and by sealing containers tightly with Teflon-lined screw caps. Sealing sample containers in this fashion also minimizes volatile losses. Contact of aqueous solutions with extraneous glassware was avoided to minimize adsorptive losses. Centrifuge tubes were also silanized to minimize adsorptive losses (16). The use of centrifuge tubes as the shake test sample containers allowed shake tests and carbon particle removal (by centrifugation) to be performed in one container. Prior to analysis, aliquots of the aqueous solutions were diluted in an equal volume of methanol to help minimize all losses noted above. Tests performed on blanks indicated that these degradation effects were as high as 10% for solute
concentrations near solubility but were typically below 2 % for solute concentrations in the range of the equilibrium concentrations for which data are reported in this study. Saturated Solution Generation. Saturated solutions were generated by using the technique described by May et al. (17)in which water is pumped through a generation column containing glass beads precoated with the compound of interest. Batch adsorption shake tests were employed to obtain isotherm data in this study. An alternative to this method, the coupled-column system described by May et al. (17))was evaluated for use in the present study. This technique involves pumping water first through a generation column and routing the column effluent directly to a carbon adsorption column. By this method, a breakthrough curve is obtained from which a single isotherm point can be plotted. Performing several such breakthrough tests at different influent (equilibrium) concentrations provides points necessary to obtain an adsorption isotherm curve. Despite the analytical advantage of this system (5),several problems were identified which prohibited its application to the study of adsorption of PAH from water onto activated carbon. These problems generally related to kinetic and equilibrium factors, the low solubilities of PAH, and the relatively high adsorption capacities of activated carbon for these compounds. Due Environ. Scl. Technol., Vol. 18, No. 6, 1984
397
Henry's Low
L 103
io2 \ cn
1
LOW
ce Langmuir
F
//
G
m
High ce Langmuir
I
/ l o o k Hlgh c, Langmuir NA DATA L
10-'1 '
IO'
ACY DATA
I
""""
'"""'1
'
'"''''~ '
'"""1
8
io-'
'',''''l
'
" ' 1 ' " "
100
10'
'""J
102
Ce, m s l l
Figure 2. log-log plot of naphthalene isotherm data.
to these problems, obtaining sufficient data for an adsorption isotherm by the coupled-column technique was cumbersome, time consuming, and not sufficiently precise. For these reasons, the coupled-column technique was not used in this study. Adsorption Shake Tests. Batch adsorption isotherm shake tests were performed with 50-mL glass centrifuge tubes fitted with Teflon-lined screw caps. One set of shake tests with a particular compound entailed the use of 20-32 such tubes, four of which were blanks and contained no carbon. Two or more tests with this number of tubes were typically performed for each compound. One data point was determined for each centrifuge tube sample for which the final equilibrium concentration was within the limits of detection. Varying amounts of saturated solution, dilution water, and carbon were added to each centrifuge tube to give a total water volume of 40 mL. Solution in the centrifuge tubes was topped with a nitrogen atmosphere prior to sealing with Teflon-lined screw caps. Sealed tubes were then shaken with a wrist-action shaker a t 25 f 2 "C for periods of 1-2.5 days as necessary to reach equilibrium. Each tube was then centrifuged for approximately 5 min a t 8000 rpm, and 5-mL aliquots were withdrawn by pipet and added to an equal volume of methanol in 15-mL centrifuge tubes. Injection volumes of 5-175 p L of the water-methanol mixture were used for analysis by HPLC. Uncertainty in Experimental Measurements. Uncertainty analyses are presented by Walters (5). These analyses provide an indication of the accuracy associated with determining C,, C,, and qe in the present study for NA and CHR and other PAH of intermediate solubility. Aqueous solubility determinations were made with an uncertainty ranging from f4.1% for NA to f7.6% for CHR. Uncertainty in determining equilibrium concentration was from f4.6 to f4.8% for NA and from f7.6 to f10.5% for CHR. Uncertainty in equilibrium capacities were from f12.3 to f13.5% for NA and from f13.7 to *17.6% for CHR. The major source of uncertainty in these determinations was determining the area of the fluorescence detector response.
Results and Discussion The data for each of the 11 compounds were used to obtain best-fit estimates of the parameters in the five isotherm equations mentioned previously. These estimates, along with sample coefficient of determination (r2) values, are summarized in Table 111. Curve fitting for the Henry's law equation was performed by linear regression for a line passing through the origin (i.e., zero intercept). Curve fitting for the Langmuir, BET, and Freundlich 398
Envlron. Scl. Technol., Vol. 18, No. 6,1984
104
c IO 3
cr"
I
E
IO' ILangrnuir L O W ce
i 0"10-5
10-4
10-3
10-2
io-'
100
IO'
io2
C e , mg/l
Flgure 4. log-log plot of fluorene isotherm data. 104
High c, Langmuir
equations was performed by computer using linear regression analysis of manipulated data and the linearized equations in Table I. Nonlinear regression analysis was performed by using a time series processing computer program a t Carnegie-Mellon University to fit the Redlich-Peterson equation to the data. Summarized 1.2 values indicate that, on a statistical basis, the five equations studied generally provide a comparable fit of the data. Comparison of plotted data and the isotherm equations is necessary to evaluate the capability of each equation to reflect relative adsorption. log-log plots of the adsorption isotherm data for NA, ACY, FLE and P H are presented in Figures 2-5. The Langmuir equation corresponding to the best-fit param-
Table 111. Summary of Isotherm Equation Parameters
Henry's law compd
Kh
NA 2390 ACY 5270 ACE 16800 FLE 5640 213000" PH AN 6140 FLA 159000" PY 30500 TRP 15900" BaA 220400" CHR 2136000"
r2
Lanamuir high C, linear form low C. linear form qo
580 500 465 440 231 23.1 c 104 0.29 82.8 36.8 c c 14.7 38.4 c 0.59 0.22 0.72 0.28 c 0.54
b
r2
2.26 3.41 21.7 13.8 12.4 244 181 522 121 933 3220
0.96 0.97 0.98 0.93 0.75 0.72 0.88 0.97 0.16 0.56 0.57
qo
b
475 5.67 25.4 275 52.1 332 248 52.1 101 115 21.4 290 60.2 983 802 64.0 17.0 3860 8.77 2660 16.7 19500
r2 0.85 0.70 0.79 0.66 0.62 0.91 0.70 0.68 0.38 0.58 0.28
qo
Freundlich
BET B
283 -516 100 -54 286 1220 64.2 256 40.7 129 17.8 24.5 87.2 17.0 48.8 1340 8.80 22.8 19.4 10.8 14.4 20.9
r2
Kf
l/n
0.93 0.87 0.98 0.97 0.84 0.80 0.93 0.96 0.26 0.64 0.77
277 266 624 674 273 330 242 389 532 216 716
0.430 0.302 0.457 0.604 0.406 0.620 0.314 0.386 0.685 0.500 0.458
Redlich-Petersonb Khr
r2
Kf,
r2
m
0.96 0.74 0.80 0.90 0.84 0.81 0.77 0.78 0.55 0.67 0.62
0.87 1310 580 0.0 0.88 70 000 359 0.392 0.87 17 600 390 0.0643 674 0.604 0.90 C C 273 0.406 0.84 C 330 0.620 0.81 0.81 249 000 215 0.285 C 389 0.386 0.78 C 532 0.685 0.55 C 216 0.500 0.67 C 716 0.458 0.62
"Tabulated value was estimated from inspection of log-log isotherm plots and represents a minimum value. Nonlinear regression analysis indicated that the Redlich-Peterson equation reduced to the Langmuir equation for NA and to the Freundlich equation for FLE, PH, AN, PY, TRP, BaA,and CHR. Tabulated Redlich-Peterson parameters for these compounds correspond to those tabulated for the Langmuir or Freundlich equations. Nonrealistic values obtained. eters are also plotted in Figures 2-5. Representative plots of the Henry law (Figure 3), Freundlich (Figure 2), and BET (Figure 5) equations are also shown. The RedlichPeterson equation is not plotted because curve fitting for the data of only three compounds (ACY, ACE, and FLE) resulted in values for the three parameters of the Redlich-Peterson equation. For the other compounds, the best fit of this equation was achieved with only two parameters; in these cases the equation reduced to the Langmuir equation for NA and the Freundlich equation for FLE, PH, AN, PY, TRP, B d ,and CHR. Apparently, data covering a wider concentration range or data that more precisely define the adsorption isotherm curve are necessary to accurately evaluate the parameters in the Redlich-Peterson equation. However, it is noted that this equation could potentially be very useful for describing the data for PAH and other compounds which show a curved isotherm over a wide concentration range. Visual inspection of the log-log plots in Figures 2 and 3 reveals the inapplicability of the Freundlich and Henry's law equations for describing the entire range of data. These equations entail a linear isotherm relationship, whereas the isotherm data suggest a curved isotherm, particularly for the lower molecular weight compounds for which data span a t least a 3 order of magnitude range in concentration (e.g., NA, ACY, FLE, and PH). As shown in Figure 2, the Freundlich equation overestimates adsorption at low C,. As shown in Figure 3, the Henry's law equation overestimates adsorption a t high C,. It is constructive also to consider additional limitations of the Freundlich equation. This equation is generally assumed to be appropriate for characterizing relative adsorption in which the constant Kf reflects adsorption capacity a t a concentration value of unity. Hence, Kf values are sensitive to the concentration units employed (e.g., mg/L or mmol/L) (8), and as discussed by Walters (5),Kf values may have no meaningful relationship to the range of experimental data. For a curved (type I BET) isotherm, the extrapolated value for Kf can be quite sensitive to the region of the curve in which the data lie; a t higher concentrations the extrapolated value of Kf decreases when extrapolated above the C, range of the data. One way to overcome this problem is to normalize equilibrium concentrations using aqueous solubility. Figures 6 and 7 show the plotted Freundlich isotherm equation for each of the 11PAH evaluated in the present study by using log (C,) and log (C,/C,) as abscissas, respectively. The plots in
F
NA
t 10-tl ' 10-5
'"""1
' ' 10-4
"""'
' '"""1 ' 10-3 10-2
'
''"'''l
10-1
' '
''('"'1
11''''1
100
' '
"'U
102
10'
Ce,mg/L
Flgure 6. log-log plot of Freundlich isotherms for
all
PAH studled.
NA,ACY,ACE
(High Soiubility)
CHR,AN,TRP (Low S o l u b i l i t y 1
10-1' '
'"c"'i
' '
"""'
' '
"""'
' '
"""l
' '
i'''a'l
'
'"U
'
Environ. Sci. Technol., Vol. 18, No. 6, 1984 399
When data are plotted in this manner for a homologous series of compounds (of which the PAH studied here clearly represent a homologous series), the isotherms would be expected to collapse to a single curve (28). It was observed that the isotherm data did not generally exhibit an increase in qe as C, approached C, (as is typical of the BET isotherm) but rather appeared to reach a limiting capacity as shown for ACY in Figure 3. Although the data for P H (Figure 5) and to a lesser extent FLE (Figure 4) show a slight increase in qe as C, approaches C,, the remaining compounds did not exhibit this trend, and it was concluded that the BET isotherm was not appropriate. It was also observed that the remaining equation, the Langmuir equation, was not capable of describing the adsorption isotherm data over the entire concentration range. Hence, as indicated in Table 111, two linear forms of the Langmuir equation were employed to determine adsorption parameters. This resulted in curves which fit different regions of the data. The Langmuir equation determined from the first linear form is applicable at values of C, greater than about 2% of C, but underestimates capacity at lower C, by up to a factor of 2 (Figure 2). The Langmuir equation determined from the second linear form fits the data well at low C, but underestimated capacity a t higher C, by up to a factor of 8 (Figure 4). As is shown later, these observations may be due to different trends in relative adsorption a t high and low relative concentration regions of the curved isotherms. Therefore, it is instructive to consider relative adsorption in each of these regions. Adsorption at Low C,. Adsorption at low C, is described by Henry’s law relationship between q, and C,. Values for Kh in Table I11 were estimated from isotherm data which corresponded to the lowest C, values for each compound and which appeared to exhibit a slope of 1.0 on the log-log plots, and thus a linear relationship between qe and C,. Henry’s law region was apparently attained for the lower molecular weight PAH with relatively higher aqueous solubilities and with low analytical detection limits with respect to solubility. Compounds in this group include NA, ACY, ACE, FLE, AN, and PY. These compounds exhibited 10 or more data points within Henry’s law region. Fewer data were available to define Henry’s law region for PH, FLA, TRP, BaA, and CHR, so an estimate of Kh was made for these compounds from low concentration data. Since these data may possibly correspond to the intermediate region of the curved isotherm at C, greater than that corresponding to Henry’s law, this estimate represents a minimum value of Kh for these compounds. For the group of PAH in which Henry’s law region was clearly evident, Kh values ranged from 2390 (mg/g)/ (mg/L) for NA to 30500 (mg/g)/(mg/L) for PY. Estimated Kh values for the remaining five compounds for which Henry’s law region was not confirmed ranged from 6000 (mg/g)/(mg/L) for TRP to 140000 (mg/g)/(mg/L) for CHR. As will be shown later, comparison of these estimates of Kh with a comparable term calculated from parameters of the Langmuir equation suggests that reasonable estimates of Kh were obtained for eight PAH including NA, ACY, ACE, FLE, PH, AN, FLA, and PY. The applicable concentration range for Henry’s law was generally less than 1% of aqueous solubility for the six compounds exhibiting ten or more data points in this region. The estimated Henry’s law region for P H and FLA also was below 170 of aqueous solubility. Data for TRP, BaA, and CHR were obtained at C, values above 1-570 of aqueous solubility, and these data may lie beyond the 400
Environ. Sci. Technol., Vol. 18, No. 6, 1984
Table IV. Estimates of Henry’s Law Constants from the Langmuir Equation compound NA ACY ACE FLE PH AN FLA PY TRP BaA CHR
Kh
2 390 5 270 16 800 5 640 113000 6 140 1 5 9 000 30 500 2 5 990 1 2 0 400 5136 000
bqO (low C , Langmuir) 2 690 6 990 17 300 12 900 11600 6 210 59 200 51 300 65 600 23 300 326 000
bqo/Kh 1.13 1.33 1.03 2.29
1.01 1.68
range of applicability of Henry’s law. It was not possible to obtain adsorption data a t low equilibrium concentrations relative to solubility for compounds such as TRP, BaA, and CHR by the analytical method used in this study. An accurate evaluation of Henry’s law adsorption constants for these compounds would require a different analytical method, perhaps involving extraction and concentration. Adsorption a t low C, is also described by the low concentration limit of the Langmuir equation. Parameters obtained from the low concentration linear form of the Langmuir equation can be used to evaluate Henry’s law constant, which corresponds to the product bqo. Henry’s law constants derived from the low C, Langmuir equation are shown in Table IV. By inspection of the ratios of the values from the Langmuir and Henry’s law equations, it appears that good agreement is obtained between both equations. The relative adsorption data for Kh obtained from Henry’s law equation were correlated with log (C,) (r2= 0.41) and log (KO,)(r2 = 0.63), respectively. Although neither parameter correlated well with Kh,the data show that the value of Kh (or equivalently relative adsorption a t low concentration) increases with decreasing C, and increasing KO,. These results were derived from the data for NA, ACY, ACE, FLE, PH, AN, FLA, and PY; data for TRP, BaA, and CHR were not used in the regression analysis. Adsorption in the Range from Low C, to C , . The capacity term qo obtained from the high C, linear form of the Langmuir equation is best suited for describing adsorption data beyond the low C, region. Best-fit parameters for the Langmuir equation are shown in Table 111. Values of qo determined from the high concentration linear form of the Langmuir equation range from 580 mg/g for NA to 14.7 mg/g for B d , while values of b range from 2.26 (mg/L)-’ for NA to 3220 (mg/L)-l for CHR. Regression analysis indicated that values of log (qo) and log ( b ) obtained from this form of the Langmuir equation correlate well with log (C,) (r2values of 0.84 and 0.92, respectively) and not as well with log (KO,)(r2values of 0.65 and 0.86, respectively). The qo values generally decrease with decreasing solubility, while the energy of interaction term, b, shows the reverse trend in that it increases with decreasing solubility. The qo values of the Langmuir equation have further significancein that, in cases where a monolayer adsorption assumption is valid, the apparent surface coverage at monolayer capacity can be calculated. A discussion of surface area calculations is presented elsewhere (5), which leads to the hypothesis that PAH adsorption from water onto activated carbon occurs in steps in which adsorption
Table V. Comparison of Henry's Law Coefficients (Kh)for Adsorption of PAH onto Different Adsorbents
Kh, (mg/g)/(mg/L)
compoundu
suspended organic matter
9-mAN
FLA PY TRP BaA 7,12-dmBaA
hypothetical soil and sediments (K0J
26'
3b
activated carbone
0.083' 1.3' 8.5'
0.0025' 0.039' 0.257'
B NA 2-mNA ACY
ACE FLE PH AN
soils and sediments
clay
2 690 6 990 17 300 12 900 11600 6 210
0.695' 0.785' 1.960'
23.0' 26.0' 65.0'
2.54' (0.071-1.16)d
84' (43.8-85.3)d
(0.56-6.78)d
(117-407)d
59 200 51 300 65 600 23 300
(1250-3130)d (2.26-37.4)d 3-mCHO 326 000 CHR 650' 19.6' NAC BbF BkF BaP (1.76-55.7)' DBahA Bghip "See text for compound names. *Datareported by Herbes (19). CDatareported by Karickhoff et al. (9). dData reported by Means et al. (10).eData reported in Table IV.
onto the more active sites occurs first, followed by additional adsorption onto lower energy surfaces. This hypothesized, multistep adsorption process may explain why two different forms of the Langmuir equation are obtained when each of the linear forms are fitted to the isotherm data. Comparison of PAH Adsorptive Properties for Different Adsorbents. The adsorptive behavior of PAH has been reported in the literature for several systems including microorganisms, soils and sediments, and activated carbon. Results from these studies are summarized in Table V and are discussed below with emphasis on comparison of relative adsorption capacities. Herbes (19) studied the adsorption of AN onto yeast cells; the yeast cells were used as a model material to represent naturally occurring suspended organic matter. By use of a suspended solid concentration of 35 mg/L, adsorption was found to follow Henry's law for equilibrium concentrations in the range 0.02-31 pg/L (Le., up to 42% of CJ. A Kh value of approximately 26 (mg/g)/(mg/L) may be calculated from the data of Herbes for a suspended solids concentration of 35 mg/L. This is more than a factor of 200 less than the value of 6140-6210 mg/g per mg/L reported in this study for Henry's law constant for adsorption of AN onto activated carbon. Karickhoff et al. (9) and Means et al. (10) have studied the adsorptive behavior of several PAH from water onto soils and sediments. These adsorbents were made up of a variety of materials, including sand, clay, minerals, and organic matter. Karickhoff et al. (9) studied seven PAH: NA, 2-methylnaphthalene (2-mNA), PH, AN, 9-methylanthracene (9-mAN), PY, and tetracene (naphthacene, NAC). Means et al. (10) studied five PAH: PY, 7,12-dimethylbenz [a]anthracene (7,12-dmBaA), 3-methylcholanthrene (3-mCHO), and dibenz[a,h]anthracene (DBahA). The adsorption capacity for the compounds studied on various size fractions of soil and sediment samples generally followed Henry's law relationship which applied for concentrations up to 60-70% of C,. Henry's law constants from these studies are summarized in Table
-
E. N A , ACY, ACE ( H i g h Solubilily )
F L E , PH, F L A
I Inlermediate Solubility
1
AN (Low Solubility)
BbF, BkF, BaF, B,ghiP, DBahA (Very Low Solubility
10-21 10-6
''j'"'l
10-5
'
""""
I
' "'''''1 ' 10-4 10-3
/ ' 1 ' ' ' 1 1 '
' ' 1 ' ' ' 1 1
10-2
'
' ' U
IO-!
100
Ce I C s
Figure 8. Normalized log-log Dobbs and Cohen (20).
plot of
PAH
Freundlich isotherms of
V. Henry's law coefficients were also normalized by dividing by the fraction organic carbon content of the sediment (typically in the range 0.001-0.03) to obtain a value designated as K,, and log (K0Jwas shown to be directly proportional to log (Kow).Table V shows that the average values of KO,which were reported for coarse silt size fractions for NA, PH, AN, and PY, range from 1.3 (mg/ g)/(mg/L) for NA to 84 (mg/g)/(mg/L) for PY. Dobbs and Cohen (20) have evaluated the adsorptive behavior of 12 PAH from water onto Filtrasorb 300. Compounds which were studied were benzene (B), NA, ACY, ACE, FLE, PH, AN, FLA, benzo[b]fluoranthene (BbF), benzo[k]fluoranthene (BkF), benzo[a]pyrene (BaP), DBahA, and benzo[g,h,i]perylene(BghiP). The Freundlich isotherms for these compounds are shown on a log-log plot of capacity vs. relative concentration in Figure 8. The data of Dobbs and Cohen typically span no more than 2 orders of magnitude in concentration, and unlike the previously cited studies involving suspended organic matter and soils and sediments, Henry's law region of adsorption was apEnviron. Sci. Technol., Vol.
18, No. 6, 1984
401
I o3
Table VI. Comparison of Adsorption Capacities at 50% Solubility for PAH onto Different Adsorbents
102
compd
adsorption capacities a t 50% solubility, mg/g suspended organic soils and activated material clay sediments carbon
B NA 2-mNA ACY ACE FLE PH AN 9-mAN FLA
PY TRP BaA 7,12-dmBaA 3-mCHO CHR NAC BbF BkF BaP DBahA BghiP
2230b 0.622b 3.26b
0.95"
I
53000d 420: 900e
HALOGENATED HYDROCARBONS
250: 500" 240: 850e 330: 670e 180: 230e 37,d 42e
0.488b 0.1" 0.0287' 0.256b
I
io6
10'
I
io4
1
10)
io2
I
I
ioo
10'
I
io1
1
io2
1 io3
Ce mg/,l
Figure 9. log-log plot of isotherms for several hydrophobic solutes.
190: 130e 0.171b (0.0048-0.078)' 14Y 38" 18" (0.0068-0.087)c (0.0036-0.060)' 30e 0.0049' 3.7d 1.7d 0.77d 0.46d 0.3ge
(0.0022-0.0069)'
"Data reported by Herbes (19). bData reported by Karickhoff et al. (9). 'Data reported by Means et al. (10). dData reported bv Dobbs and Cohen (20). e Data reported in the present studv.
parently not achieved. Consequently, the Freundlich capacities at 50% C, (referred to as KW)for each compound which was studied will be used for comparison here. The K,, adsorption capacities for data of Dobbs and Cohen (20) are summarized in Table VI. These data show that the capacities determined in the present study for lower molecular weight PAH are generally higher by a factor of 2-4 than those determined by Dobbs and Cohen. However, capacities are comparable for intermediate molecular weight PAH (viz., PH, AN, and FLA). Differences in capacities may be a result of differences in the activated carbon used in the respective studies or in the preparation of the powdered carbon. Dobbs and Cohen used Filtrasrb 300 while Filtrasorb 400 was used in the present study. Also, Dobbs and Cohen used carbon classified as 200 X 400 mesh, while carbon particles smaller than 400 mesh were included in the present study. Despite these differences,
the data are in general agreement. Table VI also summarizes KW values of PAH for various adsorbents as determined in the studies mentioned earlier. AN is the only solute common to studies on the various adsorbents. Data for this compound indicate that adsorption is highest on activated carbon; adsorption on suspended organic matter is roughly a factor of 40 lower, while adsorption on soils and sediments can be lower by more than a factor of 1300. It is noted that the comparison here is made for relative capacities a t 50% C, and not a comparison of Kh values. The distinction is that Kh values for activated carbon are comparatively much greater than those for the other adsorbents but do not apply over the entire concentration range. Comparison of Carbon Capacities for Different Solutes. The adsorption of various hydrophobic solutes onto activated carbon has been studied by several researchers (12, 20, 21). Although differences in relative solubilities of the various solutes studied and in the C, range for which data are available complicate a comparison of relative adsorption of these solutes, it is instructive to consider the data available regarding their relative adsorptive characteristics. Data presented by Dobbs and Cohen (20) enable comparison of relative adsorptive properties of PAH to those of other classes of compounds for one type of carbon (Filtrasorb 300). Data presented in Figure 9 show some general trends in relative adsorption between PAH (shaded) and pesticides, PCBs, halogenated hydrocarbons, monocyclic aromatics, and phenolics, all of which are important classes of compounds on the priority pollutant list. The indicated adsorption region for compounds other than PAH reflects the data of only one to three representative compounds within each class chosen
Table VII. Comparison of Aqueous Solubilities Determined in the Present Study to Literature Values
compounds
present study (determinations)
NA ACY ACE FLE PH AN FLA PY TRP BaA CHR
32.9 f 1.2 (9) 16.1 f 0.45 (4) 4.16 f 0.57 (3) 1.90 f 0.057 (3) 1.29 f 0.14 (4) 0.0698 i 0.0075 (4) 0.199 f 0.011 (4) 0.133 f 0.033 (4) 0.0307 f 0.0043 (4) 0.0168 f 0.0011 (3) 0.00327 f 0.00043 (4)
aqueous solubility, mg/L at 25 "C May et al. Mackay and Shiu (17) (22) 31.69 f 0.23 1.685 i 0.005 1.002 f 0.011 0.0446 f 0.0002 0.206 f 0.002 0.132 i 0.001 0.0094 f 0.0001 0.0018 f 0.0001
31.7
jz
0.2
3.93 f 0.014 1.98 f 0.04 1.29 i 0.070 0.073 i 0.0005 0.260 f 0.020 0.135 i 0.005 0.043 i 0.00013 0.014 i 0.0002 0.002 i 0.0002
other studies 34.4," 30,b 30.3; 31.3,d 31.2", 34.d 3.93" 3.47," 3.88," 3.47d 1.98," 1.90e 1.29," 1.6,b 1.151: 1.070,d 1.180" 0.073," 0.075,' 0.041,' 0.075" 0.26," 0.26P 0.14," 0.129 0.148" 0.014," O.OIOe 0.002,' 0.006e
"Versar (23) as reported by Dobbs and Cohen (20). 'ISHOW (25) as reported by Dobbs and Cohen (20). 'Schwarz (25). dEganhouse and Calder (26) as reported by Neff ( I ) . e Wauchope and Getzen (27). f Arnold et al. (28) as reported by May et al. (17). 402
Environ. Sci. Technol., Vol. 18, No. 6,1984
on the basis of reflecting high and low relative solubilities. The compounds for which data are plotted include the following: dieldrin and chlordane (pesticides); PCB-1221; carbon tetrachloride, chlorobenzene, and hexachlorobenzene (halogenated hydrocarbons); benzene and toluene (monocyclic aromatics); phenol and 2,4,6-trichlorophenol (phenolics). As a result, trends shown in Figure 9 should not be considered as comprehensive. The general observations are that, for the class of compounds, PAH adsorption is comparable or greater than that of many other types of organics. Relative to PAH adsorption a t a given C,, the adsorption region for pesticides is up to an order of magnitude greater, the adsorption region for PCBs is comparable, the adsorption regions for halogenated hydrocarbons and phenolics are comparable or up to nearly 2 orders of magnitude lower, and the adsorption region of monocyclic aromatics is lower by 1-2 orders of magnitude. These observations are limited to the compounds noted over the range of C, values as indicated. Aqueous Solubility Determination. Table VI1 s u m marizes aqueous solubilities which were determined for all compounds in this study. These data are in good agreement with values reported recently by May et al. (17)using the coupled-column technique and by Mackay and Shiu (22). The aqueous solubilities for ACY, ACE, and T R P appear to be the first published values using the coupled-column technique. While the solubilities for ACE and T R P agree closely to literature values, the determined solubility for ACY of 16.1 mg/L is greater than that reported by Versar (23) by more than a factor of 4.
Summary and Conclusions The adsorptive behavior of 11PAH from organic-free, deionized water onto activated carbon has been studied. Adsorption isotherm data show a curved relationship between equilibrium capacity (q,) and equilibrium concentration (C,) when plotted on log-log axes; this is indicative of two distinct adsorption regions. The first region occurs a t low C, in the range of C, less than approximately 1% of solubility, for which adsorption follows Henry’s law relationship, and the second occurs a t high C,, in which a limiting capacity appears to be achieved. The Langmuir adsorption isotherm equation has been determined to be more useful in describing the isotherm data than either Henry’s law, Freundlich, BET, or Redlich-Peterson equations. Both adjustable parameters in the Langmuir equation are correlated with either the logarithm of aqueous solubility (C,) or octanol-water partition coefficient (K,,). This is true when isotherm data are fitted to either of two linearized forms of the Langmuir equation. However, parameters determined from the linear form which weights low C, data differ from those obtained from the linear form which weights high C, data; the former are applicable a t low C,, and the latter are applicable a t high C,. Henry’s law constants can be estimated from the product of the two parameters of the Langmuir equation that have been determined by using the low C, linear form of the equation. The logarithms of the monolayer capacity parameters of the Langmuir equation determined by using the high C, linear form appear to be better correlated with log (C,) than with log (K&. These observations suggest the possibility of two different adsorption mechanisms in which adsorption a t low C, is governed by factors which influence partitioning while adsorption a t high C, is governed by factors which influence solubility. PAH adsorption from clean water onto activated carbon observed in the present study was much greater than PAH adsorption onto other adsorbents observed in previous studies. For activated carbon, Henry’s law constants range
from 2390 (mg/g)/(mg/L) for naphthalene to approximately 326 000 (mg/g)/(mg/L) chrysene, and limiting capacities range from 580 mg/g for naphthalene to 14.7 mg/g for benz[a]anthracene. These capacities are applicable to clean-water, single-solute systems; different capacities may be observed in more complex aqueous systems. The adsorption of PAH onto activated carbon is (1) greater than adsorption of PAH onto soils, sediments, and suspended organic matter and (2) comparable to or greater than the adsorption of other classes of hydrophobic organics onto activated carbon. Registry No. C, 7440-44-0; NA, 91-20-3; ACY, 208-96-8; ACE, 83-32-9; FLE, 86-73-7; PH, 85-01-8; AN, 120-12-7;FLA, 206-44-0; PY, 129-00-0; TRP, 217-59-4; BaA, 56-55-3; CHR, 218-01-9.
Literature Cited Neff, J. M. ”Polycyclic Aromatic Hydrocarbons in the Aquatic Environment”; Applied Science Publishers Ltd: Essex, England, 1979. Walters, R. W.; Luthy, R. G. Water Res., in press. World Health Organization “International Standards for Drinking Water”, 3rd ed.; Geneva, Switzerland, 1971. Keith, L. H.; Telliard, W. A. Environ. Sci. Technol. 1979, 13, 416. Walters, R. W. Ph.D. Thesis, Carnegie-Mellon University, Pittsburgh, PA, 1981. El-Dib, M. A.; Badaway, M. I. Water Res. 1979, 13, 255. McGuire, M. J.; Suffet, I. H. In “Activated Carbon Adsorption of Organics from the Aqueous Phase”; Suffet, I. H.; McGuire, M. J., Eds.; Ann Arbor Science: Ann Arbor, MI, 1980; Vol. I, Chapter 4. Arbuckle, W. B. Environ. Sci. Technol. 1981, 15, 812. Karickhoff, S. W.; Brown, D. S.; Scott, T. A. Water Res. 1979, 13, 241. Means, J. C.; Wood, S. G.; Hassett, J. J.; Banwart, W. L. Environ. Sci. Technol. 1980, 14, 1524. Belfort, G. Environ. Sci. Technol. 1979, 13, 939. Martin, R. J.; Al-Bahrani, K. S. Water Res. 1979,13,1301. Radke, C. J.; Prausnitz, J. M. Ind. Eng. Chem. Fundam. 1972, 11, 445. Weber, W. J., Jr. “Physicochemical Processes for Water Quality Control”; Wiley: New York, 1972. Wagner, J. J. Calgon Corp., Pittsburgh, PA, personal communication, 1981. Watson, J. T. “Introduction to Mass Spectrometry: Biomedical, Environmental, and Forensic Applications”; Raven Press: New York, 1976. May, W. E.; Wasik, S. P.; Freeman, D. H. Anal. Chem. 1978, 50, 997. Adamson, A. W. “Physical Chemistry of Surfaces”; Wiley: New York, 1976. Herbes, S. E. Water Res. 1977, 13, 241. Dobbs, R. A.; Cohen, J. M. Cincinnati, OH, 1980, U.S. EPA Report EPA-600/8-80-023. Weber, W. J., Jr.; Pirbazari, M. J-Am. Water Works Assoc. 1982, 74, 203. Mackay, D.; Shiu, W. Y. J. Chem. Eng. Data 1977,22,399. Versar, unpublished report to U.S. EPA under Contract 68-01-3832, 1979. ISHOW, University of Minnesota, unpublished report, 1979. Schwarz, F. P. J. Chem. Eng. Data 1977, 22, 273. Eganhouse, R. P.; Calder, J. A. Geochim. Cosmochim. Acta 1976, 40, 555. Wauchope, R. D.; Getzen, F. W. J. Chem. Eng. Data 1972, 17, 39. Arnold, D.; Plank, C.; Erickson, E. Chem. Eng. Data Ser. 1958, 3, 253. Yalkowsky, S. H.; Valvani, S. C. J . Chem. Eng. Data 1979, 24, 127.
Received for review August 3,1982. Revised manuscript received September 19,1983. Accepted January 3,1984. This work was sponsored by the U.S. Department of Energy under Contract DE-FG22-80PC30246-84 to Carnegie-Mellon University. Environ. Sci. Technol., Vol. 18,No. 6, 1984
403