Attainment of Equilibrium in Activated Carbon Isotherm Studies

water and phosphate buffer and found no discernible differ- ences. Similarly, Ying .... 7.0. 0.002 M PO4. 4. 6 phenol. 192.8. P. 20. -7 distilled wate...
0 downloads 0 Views 592KB Size
Attainment of Equilibrium in Activated Carbon Isotherm Studies Russell G. Peel and Andrew Benedek" Water Research Group, Department of Chemical Engineering, McMaster University, Hamilton, Ontario, Canada L8S 4L7

This paper examines the adsorption of phenol and 0 chlorophenol from aqueous solution onto activated carbon. The results show that granular activated carbon took up to 3 weeks to reach equilibrium with phenol and up to 5 weeks to reach equilibrium with o-chlorophenol. Powdered carbon isotherms took from 3 to 5 days to reach equilibrium. Up to 80%of the adsorptive equilibrium was reached in the first few hours, but the remaining capacity was utilized very slowly. This type of behavior can be described by a dual rate mechanism macropore-micropore adsorption model, which can be used to explain variations in isotherm behavior previously reported in the literature. The study clearly shows that extended contacting periods should be allowed during isotherm evaluations to ensure that equilibrium is obtained. Wherever possible, powdered carbon isotherms should be used instead of granular carbon isotherms. In any study involving activated carbon, an accurate isotherm is essential, whether for system design or research purposes. Without knowledge of the isotherm, bed life and location of the breakthrough curve cannot be predicted and kinetic data cannot be modeled meaningfully. Research on adsorption kinetics by the authors indicated several problems related to isotherm evaluation and led to the present study. A review of published isotherms for adsorption of phenol on activated carbon (Fitrasorb F400) yielded the information shown in Figure 1 and detailed in Table I. All isotherm data have been converted to consistent units (mg/g, mg/L), and temperature, pH, and other relevant data noted. As discussed below, while some differences in equilibrium capacity could be expected due to variations in carbon properties and environmental conditions, the observed differences are greater in magnitude than might reasonably be attributed to such causes. There is a substantial volume of evidence in the literature showing that the effects of environmental conditions such as temperature, pH, and buffer strength are usually quite small. The effect of temperature on isotherm capacity was investigated by Snoeyink ( I ) and was found to be slight. Over the narrow range reported or assumed in Table I this influence can be regarded as insignificant. Zogorski (2) studied uptake rates of phenol from distilled water and phosphate buffer and found no discernible differences. Similarly, Ying ( 3 ) found no difference between isotherms measured in tap water, distilled water, or weak phosphate buffers under otherwise identical conditions. Thus the effect of differing solvent conditions among the reported studies could not account for the large observed variations. The influence of pH is complex and potentially could have had a significant effect on the results of studies conducted in the vicinity of the pK,. However, the phenol pKa value of 9.86 is well above the pH range reported in most of the studies. At p H 7 phenol exists predominantly in the neutral form, and changes in capacity appear to be due largely to competition with hydronium ions ( I ) . Myers and Zolandz ( 4 ) found that in the p H range of 2-6 the adsorption of phenol was little affected, and they showed significant changes in capacity to be due to a shift from the neutral to the anionic species a t high PH. Biodegradation could result in an incorrect isotherm measurement if suitable conditions for the growth of bacteria were present. In this work, however, the possibility of biodeg66

Environmental Science & Technology

radation was unlikely due to the absence of nutrients and seed organisms and, as discussed in a later section, the results indicated that biodegradation did not occur. Therefore, it seemed possible that part if not all of the observed variations in reported isotherms were due to problems in determining attainment of the real equilibrium and the present study was carried out to test this hypothesis.

Experimental The activated carbon used in this study was Filtrasorb F400 (Calgon Corp., Pittsburgh, Pa.) supplied as 12 X 40 mesh. The carbon was prepared by sieving to 16 x 30 mesh to produce a more uniform size range with a mean particle diameter of 0.094 cm. After sieving, the carbon was boiled in distilled deionized water and then dried a t 103 "C for 24 h and stored in a desiccator until used. A quantity of the wet granular carbon was ground with a mortar and pestle until the entire sample passed through a No. 200 U.S. Standard Sieve. The resulting powdered carbon was dried a t 103 OC for 24 h and stored until use. The solutes, phenol and o-chlorophenol (OCP),were chosen as they are easily analyzed using UV spectrophotometry and are common industrial waste and water supply contaminants. Both reagents were obtained in liquid form, phenol as Liquefied Phenol (Baker Analyzed Reagent) and OCP (BDH Laboratory Reagent, minimum assay 99.5%).Stock solutions of 10 OOO mg/L were made up in distilled, deionized, activated carbon filtered water and stored a t 4 OC in the dark. These solutions were used to make up all working solutions and standards. Either distilled, deionized, carbon treated water alone, or with a 0.002 M phosphate buffer giving a pH of 7, was used in making up the working solutions. Concentration analyses were performed on a Beckman Model 25 spectrophotometer, using a 1-cm cell. Wavelengths of 270 nm for phenol and 274 nm for OCP were used, and standards covering the experimental range indicated a linear relationship between absorbance and concentration. Duplicate and multiple determinations indicated a standard deviation of 0.065 mg/L. Wavelength scans of both components in the neutral and anionic forms were obtained, and periodic scans of isotherm samples both before and after adsorption were in all cases identical with those of the neutral species. Isotherm studies were performed by preparing 5-L batches of working solution and adding either 500-mL quantities for granular isotherms, or 300-mL quantities for powdered carbon isotherms, to 500-mL bottles containing preweighed quantities of carbon. The bottles were sealed with Teflon cap liners and placed on a rotator in a constant temperature room maintained a t 20 f 1 "C. The rotator was designed to gently tip the bottles end over end so the carbon was kept fully in suspension and attrition was prevented. For each isotherm two blanks without carbon were run and all samples were filtered through washed 0.45-pm Metricel filters prior to UV measurement. Tests showed neither phenol nor OCP significantly adsorbed on the filters. Adsorption Equilibria Rather than rely on kinetic studies to determine the time required to reach equilibrium, concentration progressions were measured during the actual isotherm experiment. In the granular carbon studies the isotherm bottles were temporarily removed from the rotator and 10-mL samples withdrawn for

0013-936X/80/09 14-66$01 .OO/O

@ 1980 American Chemical Society

Table 1. Published Isotherm Data time period

reference

present study Ying ( 3 )

Crittenden (8) Wedin (9) Huang ( 75) Mathews ( 76) Usinowicz ( 7 7)

23-40 days 6 days-6 weeks

several weeks 2-8 days

particle size

16/30, 200 30/35 50/60

granular 10130 30135, 60/100 30140, 80/100

24 h N/A N/A

EQUIL

isotherm, mglg of C; mglL

PH

7.0 0,= 78.1 Ceo.212 lo-* M PO4 0,= 56Ce0 288 7.0 0,= 16160Ce/(1 183.9Ce086'7) 0,= 68Ce0 0,E 56Ce024 7.0 0, = 43.29Ce/(1 0.7138Ce0794*) 0,= 15.64Ce/(1 O.133Ce)

+

+ +

LIQUID CONC

T , OC

20

room (?) room (?) 25 23.5

room (?) room (?)

MG/L 1

Figure 1. Published isotherms for phenol on F400 (legend in Table I )

h

c3 v

0 Z

0

0

u

# U

33

200 4oOl

r a n

H

L I Q U I D PHRSE CONC

(

MG/L 1

Figure 2. Apparent OCP loadings as a function of time

filtration and analysis. The samples were subsequently returned to the bottles and rotation was recommenced. In the powdered carbon studies the bottles were removed from the rotator and 20-mL samples of the completely mixed solution were rapidly withdrawn for filtration and analysis. In these studies the withdrawn samples were not returned to the bottles. The progress with time of apparent carbon loadings for one

such OCP granular carbon isotherm is shown in Figure 2. As carbon loadings did not change after 33 days, the final loadings must be in true equilibrium with the measured liquid concentrations. Although about 60% of the final capacity was obtained after only 4 h, the remaining capacity was utilized very slowly. This finding is in agreement with that of Zogorski ( 2 ) and Snoeyink ( 1 ) . In all, six isotherms were conducted under the conditions detailed in Table 11. Excess time was Volume 14, Number 1, January 1980 67

Table II. Experimental Isotherm Conditions isotherm no.

1 2

3 4

5 6

adsorbate

Co, rnglL

granlpowd.

echlorophenol o-chlorophenol o-chlorophenol phenol phenol phenol

100.7

G G P

20

203.5 101.8 99.0 104.5 192.8

G

20 20 20

250 226

T, OC

PH

-7

7.0

20 20

P P

buffer

7.0 7.0 7.0 -7

l i m e period, days

distilled water 0.002 M PO4 0.002 M PO4

33 40 14

0.002 M PO4 0.002 M PO4

23

distilled water

7

4

.

200.

v5 u

0 150

z

0 125 0

u 100 (13 a 75

r n

. .

. .

0 50. H J 25 a

0

cn

0.

0.

L I Q U I D PHRSE CONC I M G / L 1 Figure 3.

Granular carbon phenol isotherm as a function of time: (0)14 days: ( 0 )23 and 30 days

I n

l

l

1

I

l

l

I

I

I

L I Q U I D PHRSE CONC Figure 4.

I

I

Environmental Science & Technology

I

I

I

I

I

-

MG/L 1

Powdered carbon phenol isotherm as a function of time: (0)2 days, Qe

allowed to ensure equilibrium in the remaining OCP isotherms (no’s. 2 and 3). Similar studies were carried out on the phenol isotherms. In Figure 3 the data for isotherm no. 4 a t 14 and 23 days are shown. Because of the small change, equilibrium appeared to have almost been obtained after 14 days. Thus, phenol reaches equilibrium more quickly than o-chlorophenol. This result was expected as diffusion coefficients for both compounds should be similar and the capacity for phenol is less than the 68

I

2251-

N

72Ce0.217; (0) 5 and 7 days, 0, = 78.1Ceo2’*

capacity for o-chlorophenol. The studies on the powdered carbon isotherms (no’s.5 and 6) indicated that equilibrium was achieved after about 3 days. Data for isotherm no. 6 are plotted in Figure 4 a t 2, 5 , and 7 days. This figure clearly shows that 2 days contacting is insufficient for powdered carbon isotherm studies with phenol. The final equilibrium isotherms for experiments 1 to 6 are plotted in Figure 5. All the data for each adsorbate were re-

-

400 m

\

380.

c3

c3 E 320. v 0 280. Z H Q

240.

5

200.

Q

160.

a

H

-1 0 120. (D

1 83. H 3 40. (3

W

0.

20.

10.

30.

EQUIL

40.

601

50.

LIQUID CONC

80.

70.

90.

100.

MG/L 1

Figure 5. o-Chlorophenol and phenol isotherms. OCP: (A)no. 1; ( 0 )no. 2; (M)no. 3. Phenol: (0)no. 4; (A)no. 5; (0) no. 6

1 . b

I

TIME

I

I

I

I

I

HRS 1

( I

1

I

I

1

I

I

I

1

I

I

I

1

I

T

I 2.

Om

I

I

I

4.

I 6.

I

8.

10

12.

I 14

.

1

TIME (DRYS) Figure 6. Batch kinetic data plotted over 15 h and 15 days

gressed using a single Freundlich isotherm by a nonlinear least-squares procedure. The parameters for the equation Qe = MCen are given in Table I11 with the corresponding 95% confidence limits. For each substrate all data points were well fitted by a single line, no difference being observed between powdered and granular isotherms. On the basis of these results, it is apparent that neither particle size nor the presence or absence of a weak buffer has any effect on the equilibrium relationship. Additionally, changes in the initial concentration were observed to have no effect over the limited concentration range studied. The possibility that biological degradation was responsible

for part of the observed removal was precluded for the following reasons: (a) all phenol concentrations were eventually constant; (b) different initial concentrations of phenol gave the same end points; (c) TOC (Beckman Model 915) and direct UV phenol analyses were identical within experimental error; (d) wavelength scans at the beginning and end of the adsorption period showed no changes in the spectra. Since the above results showed powdered and granular carbon isotherms to be identical, the use of powdered carbon in evaluation of most isotherms seems justified. I t has previously been shown ( 5 )that crushing carbon has a slight effect on the largest pore volume only, and negligible effect on the surface area. Thus, there is no logical reason why the isotherms should not be identical.

Table 111. Freundich Isotherm Parameters

Effects of Slow Capacity Utilization The slow approach to equilibrium as well as the existence of a unique isotherm were well established by the above reported experimental results. This slow approach to equilibrium is illustrated in Figure 6 in which data from a batch ki-

phenol o-chlorophenol

M

n

78.1 3.3 240.2 f 2.1

0.212 f 0.011 0.098 f 0.003

Volume 14, Number 1, January 1980

69

netic experiment with o-chlorophenol are shown. The upper curve represents data from 0 to 15 h and the lower curve from 0 to 15 days. (The solid lines are the predictions of a kinetic adsorption model discussed briefly in the next section.) Although the data in the upper curve give the appearance of having reached equilibrium if concentration measurements are made over a short time span, the system is clearly a long way from the true equilibrium at this time. The lower curve shows that about 70% of the capacity is utilized in the first few hours, and the subsequent approach to equilibrium is very gradual. For this system the change in concentration between the third and fourth days was less than 1mg/L and thus within the experimental error of many measurement techniques. The phenomena noted above demonstrate some underlying problems in typical techniques for evaluating isotherms. Usually a series of supposedly identical carbon masses is contacted with the adsorbate solution and a t various times flasks are withdrawn for analysis. When no further changes in concentration are observed, equilibrium is assumed to have been reached. As the above analysis shows, slight errors in concentration measurement, in weighing out carbon masses, or just in allowing insufficient time between evaluations could lead to the incorrect assumption that equilibrium had been reached. Isotherms which would then be evaluated after this determined time period may not be at true equilibrium and would give a lower apparent capacity than that which would be obtained in a full term experiment. Existing adsorption kinetic models (6, 7) are not consistent with the observed behavior noted above. All, in effect, assume a single rate-determining parameter which is operative throughout the adsorption period and cannot cope with rapid initial uptake followed by a much slower approach to equilibrium. One way of circumventing the basic inconsistency between a single rate model and the true equilibrium isotherm has been to introduce “pseudo”-isotherms which only account for that fraction of the capacity utilized in the initial rapid adsorption period (8, 9). These would correspond to an apparent end point implied by the upper curve in Figure 6. If the aim of the experimental work is only to model the adsorption kinetics in the initial region, this approach is valid. If, however, the information is used to predict performance of systems operated for longer time periods, and in which some or all of the slow uptake capacity is uti1ized;then the approach is incorrect. A similar problem lies in using short rapid flow columns to evaluate adsorption capacities. If the time span of operation is not long enough to utilize the slow uptake capacity, the results will not be applicable to larger systems operated for much longer time spans in which the slow uptake capacity is more fully realized. Mechanisms of Slow Adsorption It was apparent that only a dual rate mechanism model for carbon adsorption could satisfactorily explain the observed data. In the course of comprehensive studies of equilibrium, kinetics, and column performance of activated carbon such a model has been developed (10). The proposed model assumes that two diffusion modes are operable throughout the adsorption period and divides the carbon particle into two regions. The first is a homogeneously distributed network of pores that are much larger than the molecular radius of the diffusing species. In such pores a relatively rapid surface diffusion mechanism is assumed to be predominantly responsible for transport, and a specified fraction of the total adsorption capacity is utilized within this region. This capacity corresponds to that described by the “pseudo”-isotherms noted above. The second region consists of those narrow pores that are available for adsorption but within which diffusion is restricted by the close proximity of the walls. These pores are assumed to be 70

Environmental Science & Technology

-

100

I\

I 0.

l

2.

l

l

I.

l

l

l

6.

TIME

(

l

8.

l

l 10.

l

l 12.

l

l

I+.

DRYS I

Figure 7. Predicted rate of equilibrium attainment for varying initial concentration

homogeneously distributed throughout the particle and do not contribute to radial transport. They are instead regarded as branch pores into which adsorbate enters via ends open to the rapid diffusion pore network. The model is described in detail elsewhere (10) and is introduced here only to explain some potential problems related to isotherm studies. The solid lines in Figure 6 are the predictions of the model fitted to the data by a nonlinear leastsquares procedure, and the ability of the model to describe the observed behavior is excellent. The dashed line is the equilibrium predicted by the isotherm of Figure 5. The model was used to look at the effect of varying initial concentrations in systems having the same final equilibrium concentration. In Figure 7 , model predictions for systems having initial o-chlorophenol concentrations of 500 and 100 mg/L are drawn. Even though both systems evefitually approach the same final equilibrium concentration, the 500 mg/L system is significantly slower. Thus if the same time period were allowed for isotherms a t both 100 and 500 mg/L initial concentrations, the higher initial concentration isotherm would always appear to have a slightly lower capacity. If enough time were allowed for both systems to reach true equilibrium, no difference in capacities would be observed. This behavior could potentially be the cause of the “concentration-history” effect noted by Crittenden (8)and others ( 3 , l l ) .Isotherms evaluated from high initial concentrations were found to give lower capacities than those measured from lower initial concentrations. The results over the narrow concentration range of the study reported herein tend to support the hypothesis that observed capacity differences may be due to incomplete attainment of equilibrium. Disregard of slow uptake effects could also be the cause of disagreement in the literature regarding the effect of particle size on uptake capacity. Weber and Morris (12)found that equilibrium capacity was strongly affected by particle size; however, Dedrick and Beckmann (13)and Martin and Al-Bahrani (14) found no such effect. Since diffusion into large particles is slower, equilibrium takes longer to achieve. If equilibrium is not fully attained when data for the isotherm are measured, the results would show a decrease in capacity with an increase in particle size. It is also possible, however, that differences in the activation process for various size ranges of activated carbon could affect adsorption capacity. Conclusions and Recommendations A slow, long term activated carbon adsorption of phenolic compounds has been demonstrated experimentally. Ap-

proximately 60-80% of the ultimate capacity was utilized within a few hours in the case of granular carbon. The remaining capacity, however, took several weeks to be exhausted. The powdered carbon was shown to reach equilibrium in 3 to 5 days of contact. The experimental isotherms were shown to be independent of the type of buffer used, the carbon particle size, and the initial adsorbate concentration. Both phenol and o-chlorophenol isotherms were well fitted by the Freundlich equation over the range studied. The reason for the apparent lack of consistency among isotherms previously reported in the literature is attributed to difficulty in defining the attainment of true equilibrium. The slow uptake can account for a significant proportion of the carbon capacity and can easily be overlooked when performing an isotherm study. Disregard of this capacity or equivalently nonallowance of sufficient time for equilibrium can be the cause of previously unexplained observations such as the variation of isotherm capacity with the initial concentration of adsorbate. A model describing adsorption in terms of rapid radial diffusion in series with slower diffusion into branch pores fits experimental data well and can be used to assess potential effects of varying particle size, initial concentration, and so forth. I t is suggested that granular carbon isotherms be evaluated over periods of a t least 30 days unless rigorous studies are carried out to ensure equilibrium is obtained. Where possible, powdered carbon obt.ained from the granular carbon of interest should be used in the isotherm evaluations and these

should be conducted for a t least 3 days.

Literature Cited (1) Snoeyink, V. L., Weber, W. J., Mark, H. B., Enuiron. Sci. Technol., 3, 918 (1969). (2) Zogorski, J. S., Faust, S. D., Haas, J. H., J . Colloid Interface Sci., 55,329 (1976). (3) Ying, W., Ph.D. Thesis, University of Michigan, Ann Arbor, Mich., 1978. (4) Myers, A. L., Zolandz, R. R., 171st National Meeting of the American Chemical Society, Miami, Fla., Sept 1978. (5) Rankin. P. R.. M.Enz. Thesis. McMaster Universitv. Hamilton. 'Ontario, 1975. ' (6) Vermeulen, T., Adu. Chem. Eng., 2, 147 (1958). (7) Fleck, R. D., Kirwan, D. J., Hall, K. R., Ind. Eng. C'hem.Fund., 12(1),95 (1973). (8) Crittenden, J. C., Weber, W. J., J . Enuiron. Eng. Diu., Am. Soc. Civ. Eng., 104, 185 (1978). (9) Wedin, J., M.Eng. Thesis, Royal Institute of Technology, Stockholm, Swedenr1976. (10) Peel, R. G., Ph.D. Thesis, McMaster University, Hamilton, Ontario, 1979. (11) Usinowicz, P., Ph.D. Thesis, University of Michigan, Ann Arbor, 1972. (12) Weber, W. J., Morris, J. C., J . San. Eng. Diu., Am. Soc. Ciu. Eng., 90 (SA3),79 (1964). (13) Dedrick, R. L., Beckmann, R. B., CEP Symp. Ser., 63(74), 68

-

(1967). \-ll.,.

(14) Martin, R. J., Al-Bahrani, K. S., Water Res., 12,879 (1978).

(15) Huang, J., Garrett, J. T., Water Sewage Works, 124, 64 (1977). (16) Mathews, A. P., Ph.D. Thesis, University of Michigan, Ann Arbor, 1974. Received for review July 9, 1979. Accepted October 4 , 1979.

Effects of Fuel, Lubricant, and Engine Operating Parameters on the Emission of Polycyclic Aromatic Hydrocarbons Peter S. Pedersen* and Jan lngwersen Laboratory for Energetics, Technical University of Denmark, Building 403, DK-2800 Lyngby, Denmark

Torben Nielsen and Elfinn Larsen Chemistry Department, Ris0 National Laboratory, DK-4000 Roskilde. Denmark

The effects of fuel, lubricant, and engine operating parameters on the emission of polycyclic aromatic hydrocarbons (PAHs) associated with particles were studied by means of an automobile engine operated with a number of special fuels under steady-state conditions. More than 90% of the mass of the heavier PAHs (e.g., benzo[a]pyrene) was found on particles with a diameter below 1pm, while for the light PAHs, a substantial part was in the vapor phase. The aromatic content, the type of aromatic fraction, and the PAH content of the fuel and lubricant had a strong effect on particle-bonded PAH emission; in comparison to this, the lead content of the fuel was of minor importance. The lubricant itself had only slight influence on particle-bonded PAH emission. This varied with the air:fuel ratio in a manner similar to.particulate matter and to unburned hydrocarbons, showing high emissions a t very rich as well as very lean mixtures. On increasing the engine load, the particle-bonded PAH emission increased. H

Lung cancer has been one of the highest causes of death of humans in the industrialized countries (1, 2 ) . The existence of an urban factor in the risk of lung cancer seems a t least partly to be caused by atmospheric pollution (2-4). In general, 0013-936X/80/0914-71$01 .OO/O

@ 1980 American Chemical Society

the atmospheric levels of polycyclic aromatic hydrocarbons (PAHs) are orders of magnitude higher in urban areas than in rural areas ( 5 , 6 ) . Some of the PAHs are considered to be strong carcinogenics (5, 7). The main part of the carcinogenic PAHs in the air has been found to be associated with particles ( 4 9 ) .Since particles in the low micrometer range are deposited in the pulmonary region of the lungs while larger particles are impacted in the upper respiratory tract, removed by mucociliary action, and swallowed (IO),the health hazards associated with inhalation of the two size ranges are different. The emission of PAHs from automotive sources has been estimated to be 2% of the total amount emitted to the atmosphere in the U S . ( 5 ) .Changes in the traffic intensity and in the extent to which coal and fuel oil are used for domestic heating and power generation might influence this significantly; in areas close to congested roads and streets, automotive emissions can be the dominant source (5,11-13). The worldwide trend to reduce lead addition to gasoline in order to reduce lead emission to the atmosphere is likely to have the effect of increasing the content of aromatic hydrocarbons in order to maintain the knock resistance of the fuel (14);a number of investigations have shown that this leads Volume 14, Number 1, January 1980 71