Temperature dependence of trihalomethane adsorption on activated

Wadsworth Center for Laboratories and Research, New York State Department of Health, Empire State Plaza,. Albany, New York 12201. Adsorption isotherms...
1 downloads 0 Views 935KB Size
Environ. Scl. Technol. 1988, 22, 406-412

Temperature Dependence of Trihalomethane Adsorption on Activated Carbon: Implications for Systems with Seasonal Variations in Temperature and Concentration Katherine T. Aiben," Eugene Shpirt, and Joan H. Kaczmarczyk

Wadsworth Center for Laboratories and Research, New York State Department of Health, Emplre State Plaza, Albany, New York 12201 H Adsorption isotherms have been measured for chloro-

form, bromodichloromethane,dibromochloromethane,and bromoform on granular activated carbon (GAC, Calgon F400) at 4, 15, 30, and 45 "C. Coefficients of Langmuir and Freundlich isotherms are given and compared to isotherms reported in the literature for trihalomethanes. Isoteric heats of adsorption were found to be -7.9 f 2.4, -5.9 f 1.3, -5.2 f 2.2, and -4.1 f 2.2 kcal/mol for chloroform, bromodichloromethane, dibromochloromethane, and bromoform, respectively, with an average of -5.8 f 2.0 kcal/mol. Experimental precision and variations in experimental conditions chosen for the trihalomethanes prevented detection of an increase in heat of adsorption with increasing bromine substitution, although such a trend was otherwise expected theoretically. Long-term seasonal variations in temperature and chloroform concentration, which occur in surface water treatment, were shown to correspond to a relatively minimal change (26%) in GAC capacity; much larger long-term seasonal variations in GAC capacity would be expected for volatile organics with similar adsorption isotherms, if their seasonal variations in concentration are not determined by the kinetics of chlorination and its temperature dependence. Introduction It is generally believed that temperature does not have a major effect on the activated carbon adsorption of organics from water ( I , 2). Although Faust and Aly have argued the importance of seasonal variations in temperature (3),the temperature dependence of adsorption has been examined for only a few substances: benzene sulfonate (41, phenol (5, 6 ) ,p-nitrophenol (5), and biological waste effluents with mixed molecular weights (7). The temperature dependence of adsorption is of traditional interest because it provides access to thermodynamic functions such as the heat of adsorption. From a practical viewpoint, measurements of adsorption capacities as a function of temperature also provide a means to check the consistency of isotherm data for neutral compounds whose adsorption is unaffected by pH, an otherwise conveniently varied parameter: since adsorption is exothermic, experimentally determined adsorption capacities should increase as temperature decreases. For this paper, the temperature dependence of adsorption is examined for four trihalomethanes (THMs) on granular activated carbon (GAC, Calgon F400). Isotherm data as a function of temperature are interpreted thermodynamically to check for a correct temperature dependence and to investigate the effect of bromine substitution for chlorine atoms. An evaluation of the effect of temperature on CHC13 adsorption in water treatment is also made. Data obtained at Waterford, NY, are cited to establish the short time scale for CHC1, mass transport through a fixed-bed contactor, under nearly isothermal conditions, relative to the long period for a seasonal change in temperature to occur, accompanied by a change in mean CHC13 concentration. To quantitate the effect of tem406

Environ. Sci. Technol., Vol. 22, No. 4, 1988

perature on GAC capacities, the pathway for seasonal changes in temperature and CHC13 concentrations at Waterford, NY, is projected onto the CHCl,-GAC isotherms. The significance of temperature on GAC adsorption is also discussed for hypothetical seasonal variations representative of chemicals other than THMs and/or treatment systems other than Waterford, NY. Experimental Section THM Isotherm Measurements. Results in this paper are based on 12 X 40 mesh Calgon Filtrasorb 400, for application to various pilot column studies. All granular activated carbon (GAC) for isotherm experiments is powdered to 50 bm by grinding for 1-2 min in a rotary disc mill (Spex Shatterbox). The water used for isotherm experiments is laboratory tap water from the Albany Water Works, purified by passage through a 15 cm 0.d. X 72 cm long fixed bed of Calgon F400. The water is from a surface reservoir in the lower Hudson River drainage basin; treatment includes copper sulfate addition, aeration, powdered activated carbon addition, coagulation, settling, rapid sand filtration, chlorination, iron and manganese removal, and pH adjustment for corrosion control (8). Effluent from the GAC contactor in the laboratory is typically free of THMs, with less than 0.1 mg/L total organic carbon. To prepare the starting solution, THM-free water is spiked (concentration, C, = 300-1000 pg/L) by injecting a known ampunt of one of the THMs of interest, dissolved in methanol, through the septum of a filled stock bottle (1.9 L). The THMspiked water is used to fill smaller sample bottles (volume, V, = 0.155 L) containing preweighed amounts of GAC (dry weight, Do= 5-50 mg). Methanol was used for convenience to prepare THMspiked water. Guisti et al. have shown that hydrophilic, water-miscible compounds such as methanol have a relatively low affinity for GAC, on the order of 7 mg/g at 1000 mg/L (9). Dobbs and Cohen found that ethanol as a solvent, from 0.01 to 10 g/L in spiked water, has a very small effect on GAC capacities for adsorption of 2chlorophenol (IO). For isotherm experiments described in this paper, methanol is present at concentrations from 40 to 200 mg/L, 2-3 orders of magnitude higher than THM concentrations. To avoid having to account for the effect of methanol, we now prepare spiked water samples for GAC studies by direct addition of standards without a solvent such as methanol. Virgin GAC samples immersed in a water of known THM composition are placed on a wrist-action shaker (Burrell),allowing 7 days to reach equilibrium. The sample bottles and shaker are kept in a walk-in refrigerator (Hotpack) held at constant temperature (4,15, 30, or 45 "C). A t the stop time, sample bottles are left to stand in the refrigerator for approximately 24 h until the GAC settles. Water in individual sample bottles is then analyzed for the residual THM concentration (Cf, pg/L). The adsorbed THM concentration (Qex ,mg/g) is calculated from X Vo/do; a factor of is used Qexp= (C, - C,) X

0013-936X/88/0922-0406$01.50/0

0 1988 American Chemical Soclety

Table I. Summary of Experimental Conditions and Coefficients of Langmuir and Freundlich Isotherms for THM Adsorption on Calgon F400n

compd CHC1,

CHBrClz

CHBrzCl

CHBr,

temp, T,OC

initial concn, C,, pg/L

final concn, Cf, pg/L

4 15 30 45 4 15 30 45 4 15 30 45 4 15 30 45

526 526 526 526 326,648 326 326 326 453,676 453 453 453 467,1040 467,675,1040 467 675

20-256 24-184 47-376 62-359 10-265 17-85 12-130 15-127 10-170 10-90 10-107 10-127 10-122 10-178 11-97 11-184

Langmuir coeff fractional surface no. of data Qo, b, coverage points, N mg/g L/pg

0.17-0.72 0.15-0.58 0.21-0.68 0.08-0.34 0.18-0.85 0.15-0.47 0.10-0.55 0.06-0.36 0.26-0.86 0.21-0.70 0.08-0.49 0.07-0.50 0.41-0.90 0.21-0.83 0.24-0.74 0.20-0.80

12 12 12 12 18 6 12 12 16 8 9 12 10 12 5 10

8.36 8.77 6.62 16.1 10.2 14.8 9.87 15.8 17.1 11.6 21.2 18.0 16.0 17.9 13.7 12.4

0.0102 0.00750 0.00553 0.00142 0.0216 0.0103 0.00944 0.00448 0.0210 0.0265 0.0090 0.0078 0.0703 0.0268 0.0291 0.0224

Freudlich coeff regression Kf, coeff, r [(mg/g)(L/pg)]'/" l/n

0.94 0.97 0.93 0.99 0.92 0.98 0.97 0.95 0.86 0.93 0.99 0.97 0.76 0.95 0.99 0.92

0.240 0.0806 0.119 0.053 0.525 0.303 0.164 0.152 0.790 0.641 0.344 0.231 3.35 1.34 0.892 0.503

0.611 0.855 0.639 0.790 0.578 0.710 0.747 0.729 0.605 0.592 0.733 0.775 0.358 0.491 0.561 0.654

"Conversion factors from Q1 (mg/g) and C1 (pg/L) to Qz (mmol/g) and C2 (pmol/L) are K z = K1(MW)'In-', l/n2 = l/nl,b2 = bl QOi/MW, where MW = molecular weight.

to convert from micrograms to milligrams. Values for Cf are corrected for incomplete recovery of THMs during a single-step liquid-liquid extraction: for this paper, an average value of 80% recovery was assumed, to correct raw data for THM concentrations in water samples (11). THM Sample Preparation, Chromatographic Analysis, and Data Reduction. Water samples (2'7 mL) are solvent-extracted using reagent-grade hexane (2 mL) and following procedures described elsewhere (11). The hexane layer with extracted THMs is recovered after phase separation from water and analyzed by packed-column gas chromatography with electron capture detection. THM concentrations in individual water samples are determined from linear calibration curves based on THM peak areas relative to the area of an internal standard: the range of quantitation is typically 8-400 pg/L. Sample concentrations are also corrected by substracting the concentration of THMs in solvent blanks: average values were 5.3, 3.4, 0.0, and 0.0 pg/L for.CHCl,, CHBrC12, CHBr2Cl, and CHBr3, respectively. Details of methods for data processing, using a program written in BASIC, have been given previously (11, 12). Monitoring of Real Time Mass-Transfer Zones for CHC13 on a GAC Pilot Column at Waterford, NY. Data for CHC1, adsorption-desorption are based on a GAC pilot column operated for 60 weeks at the Waterford Water Works, from November 1982 through January 1984. Details are given elsewhere for pilot column design, operation, monitoring, and data interpretation (13). In brief, the pilot column (15 cm 0.d. X 152 cm long) was filled with 8750 g of Calgon F400 to a bed depth of 107 cm. The GAC was initially backwashed but not during operation as an adsorber to remove organics. Influent to the pilot column was pumped from the treatment plant clearwell, prior to terminal chlorination. The average flow rate was 3.10 f 0.46 L/min 6.81 gpm), corresponding to a linear velocity of 17.0 f 2.5 cm/min (hydraulic loading 4.3 gpm/sq ft) and an empty-bed contact time of 6.3 & 8.9 min, with the uncertainty given as f one standard deviation about the average. Trihalomethane concentrations were determined by analysis of GAC samples collected weekly at six sequential bed depths (3.5, 24, 44, 64, 85, and 105 cm). Methanol extracts of the GAC samples were analyzed for CHC13by a GC-ECD procedure similar to that used for hexane extracts of water samples (11): the range of quantitation for

regression coeff, r

0.99 0.95 0.94 0.95 0.96 0.97 0.98 0.96 0.94 0.97 0.99 0.99 0.51 0.96 0.98 0.96 X

MW, Qo2=

THMs on GAC samples was typically 0.027-0.54 mg/g. Monitoring of Seasonal Variations in Chloroform Concentrations and Temperature at the Waterford Water Wprks, Waterford, NY. Data for seasonal variations in chloroform concentrations and temperature from 1983 to 1985 are based on samples collected from the treatment plant clearwell, prior to terminal chlorination. These samples are representative of Hudson River water that has been through preaeration, coagulation (alum, activated silica), flocculation (with powdered activated carbon added), prechlorination, sedimentation, and rapid filtration through anthracite and sand (8). Water samples (27 mL) are dechlorinated with 50 mg of Na2S03at the time of collection to remove a 1 mg/L chlorine residual. Samples are stored at 5 "C until analysis, about 1week later. Trihalomethane concentrations are determihed as described above, with linear calibration curves from 8 to 160 pg/L. Fourier analysis of these data for seasonal and higher frequency fluctuations is described in detail elsewhere (14, 15). Data for CHCl, concentrations in 1985 are based on 1460 samples collected 4 times per day for 1 year, at intervals (8 am, 12 noon, 4 pm, and 8 pm) approximating the residence time of water in the treatment plant. Data for temperature in 1985 represent 365 daily samples of finished water. Results for chloroform and temperature in 1984 and 1983 are for 52 samples collected from the clearwell once per week.

Results and Discussion Trends in Adsorption of THMs on Calgon F400: Effect of Bromine Substitution for Chlorine Atoms and Effect of Temperature. Coefficients of Langmuir and Freundlich isotherms obtained from experimentaldata are given in Table I. In Figure 1, Freundlich isotherms at 4, 15, 30, and 45 "C are plotted for each of the four THMs of interest. Coefficients of isotherms reported in the literature for THM adsorption to Calgon F400 are summarized in Table 11; these data are also plotted in Figure 1 to show the extent of agreement. Qualitatively, our data indicate that adsorption capacities increase with substitution of Br for C1 atoms, as observed for isotherms reported in ref 16 (CHC13,CHBr,Cl, and CHBr, at 10 "C) and 17 (CHC13,CHBrCl,, and CHBr, at 20 "C). Quantitatively, the equilibria described by our data are closest to isotherms from ref 16: in both cases, concentrations of Environ. Sci. Technol., Vol. 22,

No. 4, 1988 407

40m 2'oxz-j

50.0 7 c

1

CHCI,

L

L

-

i

L

50.0

0

'

o

~

0

4

05 1

02

10.0

5.0

0

.lo

05

1.0

10

Flgure 2. Langmuir Isotherms for THMs with coefficients given in Table

0.5

I.

0.2

5 10

50 100

500 5

c

50 100

IO

500

(lJg/g)

Flgure 1. Freundlich Isotherms for THMs determined experimentally at 4, 15, 30, and 45 O C for this paper compared to THM isotherms reported in the literature at 10 ( 18), 20 ( I I ) , and 23 O C ( 75).

Table 11. Summary of Experimental Conditions and Coefficients of Published Freundlich Isotherms for THM Adsorption on Calgon F400

compd

05

I/C (ug/L]'

Freundlich coeff temp, final concn, Kf, T,OC Cf, P d L [(mg/d(L/rdl'/" l / n

CHCla CHBrCl, CHBrzCl CkBr,

10 20 23 20 10 10 20

8-1180 4-118 10-757 92-1830 40-2000 5-380

0.285 0.0418 0.114 0.142 1.266 1.80 0.970

0.532 0.726 0.704 0.746 0.517 0.563 0.624

ref 16 17 15 17 16 16 17

THMs are determined by liquid-liquid extraction of water samples with an ofganic solvent. Data for THMs in ref 17 are based an a headspace technique. Taken together, these results illustrate the limitations in accuracy of experimental isotherms. Experimental uncertainty is evidefit in our own data, in terms of inconsistent trends in slopes and crossing of Freundlich isotherms: sources of experimental uncertainty are discussed in greater detail in a later section. Qualitatively, capacities are also observed to increase with decreasing temperature, particularly if our experi-

mental data are plotted as Langmuir isotherms in Figure 2. As expected, adsorption of the THMs is exothermic. These conclusions do not change with the use of mass or molar concentration units; conversion factors are given in the footnote to Table I. Experimental Heats of Adsorption. Experimental values of the isoteric heat of adsorption are given in Table 111, calculated from the temperature dependence of the b coefficient of the Langmuir isotherms, b a exp(-AHlRT). The data from which values of AH were obtained are plotted in Figure 3. The experimental edthalpies of adsorption are noted to have a relatively large uncertainty, derived from the y deviation of the fit In b = uo ul(l/T). Somewhat more well-behaved values of AH are obtained from a regression of In (bQo)on 1/T: values thus obtained are AH = -5.7 f 2.2, -4.9 f 0.6, -5.1 f 0.8, and -5.3 f 1.2 kcal/mol for CHCl,, CHBrCl,, CHBr2C1,and CHBr,, respectively, or -5.2 f 1.2 kcal/mol overall. Corresponding values of the correlation coefficients are 0.99, 0.99,0.99, and 0.94. This interpretation is valid if the Langmuir coefficient Qo is independent of temperature: Shoemaker and Garland note that, in general, the number of sites Qo to form a complete monolayer over a sorptive surface does not vary with temperature (18). For results presented in this paper, differences in the two methods of calculation are not statistically significant. In either case, the main observation is that experimental precision for values of b (or bQo) is not sufficient to observe an increase in AH with increasing Br number, although values of related thermodynamic functions do show this behavior: heats of condensation and net energies of ad-

+

Table 111. Thermodynamic Interpretation of Temperature Dependence of Langmuir Isotherms CHCl,

Bo mg/g PmoI/g In b 4 "C 15 "C 30 OC 45 "C linear regression, In b vs 1/T (K) a0

a,

x IO3

r Ydev

AH, kcal/mol

CHBrCl,

CHBr,Cl

10.0 i 4.2 84 i 35

12.7 i 3.1 78 f 19

17.0 3= 4.4 82 f 21

15.0 i 2.4 59 i 10

-4.58 -4.89 -5.20 -6.56

-3.84 -4.58 -4.67 -5.40

-3.86 -3.63 -4.71 -4.85

-2.66 -3.62 -3.54 -3.80

av 13.7 i 3.0 76 i 11

-10.4 13.1 -14.7 -18.7 2.61 2.06 2.99 3.98 0.83 0.88 0.96 0.93 0.29 0.16 0.25 0.24 -7.9 i 2.4 (30%) -5.9 f 1.3 (22%) -5.2 i 2.2 (42%) -4.1 f 2.2 (54%) -5.8 i 2.0 (34%)

OMolecular weights (MW) used are CHCl,, 119.4, CHBrCl,, 163.8; CHBr2C1,208.3; CHBr,, 252.8. 408

CHBr,

Environ. Scl. Technol., Vol. 22, No. 4, 1988

~

Table IV. Thermodynamic Properties of THMs CHCla mcond, kcal/mol

AHocond, kcal/mol

ET,kcal/mol

-7.08 -7.04 -10.3 6.2

CHBrClz -7.62 -10.4 7.5

CHBrzCl -8.23 -10.4 8.3

CHBr, -8.86 -11.3 9.3

av

ref 19

* *

-7.9 0.8 -10.6 0.5 7.8 & 1.3

a

20* 23

"Calculated from Trouton's rule: iW,,,,d = AHvsp= Tb (K) AS, where Tbp is the boiling point in K for CHCI, (62 "C), CHBrCi, (90 "C), CHBrzCl (119 "C), and CHBr, (149 "C) and S * 21 cal/mol - K. BStandard enthalpy of vaporization. Table V. Standard E r r o r of Estimate for Langmuir a n d Freundlich Isotherms

-7

temp, "C

I

L

-3

1

CHC13

4 15 30 45

CHBrClZ

4 15 30 45

CHBrzCl

4 15 30 45

Inb ............. CHBrZCI

CHBr,

4 15

30 45 -4

-5

I

I 0,0032

0.0034

Langmuir relative error error

Freundlich relative error error

0.025 0.071 0.046 0.068 0.052" 0.171 0.025 0.103 0.097 0.127n 0.060 0.048 0.024 0.050 0.050" 0.041 0.025 0,010 0.047 0.036"

0.032 0.079 0.071 0.066 0.062" 0.056 0.025 0.103 0.097 0.082n 0.094 0.056 0.043 0.042 0.067" 0.206 0.060 0.114 0.131 0.138"

0.090 0.218 0.024 0.147 0.120" 0.203 0.108 0.137 0.164 0.192" 0.231 0.139 0.104 0.092 0.164n 0.427 0.130 0.070 0.278 0.276"

0.066 0.416 0.126 0.201 0.202" 0.203 0.106 0.137 0.164 0.168" 0.162 0.134 0.102 0.175 O.15ln 0.189 0.085 0.076 0.114 0.128"

Average.

0.0036

T ' ( K)

Figure 3. Temperature dependence of the b coefficient of Langmuir Isotherms for THMs: heavy line Indicates linear fit of data for In b vs 1/T; light lines indicate fydeviation of linear fit to experimental data.

sorption reported in the literature for THMs are tabulated in Table IV for comparison. Uncertainty and Bias in Experimental Isotherms. To identify limitations of our isotherm data and directions for improvement, it remains of interest to look at the overall uncertainty of individual isotherms, as measured by the standard error of estimate u = [byP- y;h)2/1v]1/2 calculated from residuals for N data points or by the relative standard error of estimate based on residuals normalized to yth. Table V summarizes values for the standard error of estimate for Langmuir and Freundlich isotherms. The least precise isotherm data are for CHC1, at 15 "C ahd for CHBrC12,CHBr2C1,and CHBr, at 4 "C. This fact suggests the potential bias in our experimental data base, which can lead to problems when looking for smooth trends in Langmuir or Freundlich coefficients as a function of temperature and/or bromine substitution for chlorine. As already noted, inconsistent trends in slopes and crossing behavior are evident in Freundlich isotherms, particularly for CHC1, at 15 "C and for CHBrC1, data at 4 OC, as shown in Figure 1. The sensitivity of Freundlich isotherms is greatest to data at high concentrations: these data are obtained for the lowest amounts of GAC weighed out (down to 5 mg, weighed to the nearest 1/100th mg). These considerations suggest that increasing the amount of GAC by tenfold, to amounts that can be weighed and transferred with greater analytical precision, could result

in a higher quality data base for calculating Freundlich isotherms. In contrast, Langmuir isotherms are most sensitive to data points with high values of 1/C, which are those nearest analytical detection limits. To avoid problems in the Langmuir isotherms from using values of 1/C > 0.1, a minimum equilibrium concentration of the aqueous phase was taken as 10 pg/L for all four THMs (cf. Table I), which was above the minimum quantitation limit (8 pg/L). Future work to determine Langmuir isotherms would benefit from use of higher C, and corresponding Cf values than in Table I, so that low Cf concentrations characterized by poor precision do not dominate the results of regression analysis. To minimize the impact of uncertainties in spiking to obtain a given C,, it would also be preferable if each isotherm at a particular temperature in Table I is the average of three separate determinations, rather than the result of a single experiment. It is of interest that our Freundlich isotherm for CHC1, at 23 "C, on the basis of six separate experiments over a wide range of concentrations (12), agrees well with data obtained for this paper as shown in Figure 1 (present 15 "C data excepted). Sources of bias and inconsistencies in the experimental data base must also be carefully eliminated as the data base is refined. For example, from Table I, it can be seen that the fractional surface coverage 0 ranges somewhat higher for CHBr, than for the other THMs; likewise in going to higher temperatures there is a shift to a lower range of fractional surface coverages, from 0.26-0.83 at 4 "C to 0.10-0.50 at 45 "C. Ideally all results in this paper should be equally distributed about 0 = 0.5, for example, from 0.25 to 0.85. Only after doing these experiments did Environ. Scl. Technol., Voi. 22,

No. 4, 1988 409

0.81

I

0.8

I

85cm

105cm

I

I TlME(wk)

0

m

INFLUENT CHC13(Ug/LI

40

20

m

M M

++YE%k-

60

m

M

m

M

II

15 20 20 15

Figure 4. Experimental data for CHCi, adsorption and desorption at six different bed depths of pilot column at Waterford, NY. Influent CHCI, concentrations and temperature are given on horizontal scales, subdivided to indicate values corresponding to a particular time: Mand m refer to maximum and minlmum values, respectlvely. Data for CHCI, adsorbed on GAC samples are taken from ref 13. Concentrations of adsorbed CHCI, are glven in mg/g, assuming 100% recovery ( 7 7): a factor of 80% was used previously ( 73). Reprinted from 7984 AWWA Annual Conference Proceedings, by permission. Copyright 1984 American Water Works Association.

it become clear that C, should be steadily increased in going from CHC1, to CHBr, or in going from 45 to 4 "C. Since this was not consistently done, there is a decrease in the number of data points N used to calculate isotherms of THMs more heavily substituted with Br. Time Scales for THM Mass Transport through a GAC Pilot Column at Waterford, NY, Relative to Seasonal Variations of Temperature and THM Concentrations. Given the fact that adsorptive capacities for THMs on GAC decrease with increasing temperature, it remains of interest to determine the impact on GAC when used in a treatment plant. Figure 4 gives concentrations of chloroform on GAC at sequential bed depths in the pilot column operated at Waterford, NY, in 1983. Concentrations of CHC1, on GAC are plotted as a function of time the pilot column was in operation. Three stages in the pilot column's history are of interest: the advance of the mass-transfer zone for adsorption of CHCl,; the advance of the mass-transfer zone for desorption of CHC1,; the long plateau in concentrations of CHC1, on the GAC at deep bed depths, particularly 85 and 105 cm. For each of these developmental stages, there are well-defined operating conditions for the pilot column, in terms of temperature and influent CHCl, concentrations. Coefficients for rates of adsorption and desorption are given elsewhere (13). To define the operating conditions, values for temperature and influent chloroform concentration, corresponding in time, are also shown on the horizontal axis in Figure 4: these results were obtained by Fourier and statistical analysis, and details of the calculations are reported elsewhere with the complete data base (14). A seasonal change in temperature is evident, with a minimum of 2 "C a t 11weeks (February 1,1983) and a maximum of 25 "C at 37 weeks (August 2,1983). Similarly, a seasonal change in influent chloroform concentration is indicated, with a 410

Environ. Sci. Technol., Vol. 22, No. 4, 1988

minimum of 10 pg/L at 13 weeks (February 15,1983) and a maximum of 20.4 pg/L at 39 weeks (August 16, 1983). A positive correlation coefficient r = 0.49 was found for paired values of temperature and influent chloroform concentration in 1983; the correlation coefficient for these parameters in 1984 and 1985 was 0.61 and 0.68. Other values for r reported in the literature are 0.50490 for three plants on the Ottawa River in Canada (21),0.40 for water from the plant at Oneida, NY (22), and 0.74 for water from the Grasse River in Canton, NY (22). Thus, seasonal variations in chloroform concentration and temperature are expected for a surface water supply such as that at Waterford, NY. High-frequency, low-amplitude fluctuations on a daily time scale are also noted for the THMs: daily variations in CHC1, concentration at Waterford, NY, were found to have a 17% relative standard deviation from analysis of the 1985 data base (14). With respect to GAC adsorption, the interesting aspect of seasonal variations is that they occur in concert, as indicated by their positive correlation coefficient. While an increase in temperature alone would imply a loss of GAC capacity, this shift is counterbalanced by the simultaneous increase in CHC1, concentrations with increasing temperature: the net effect is to stabilize GAC capacity during seasonal changes in temperature and concentration. These concepts are discussed quantitatively in the following section, with temperature-dependent CHC1, isotherms. From Figure 4, it is evident that mass transport of CHC1, through the GAC pilot column is rapid, relative to the slow seasonal changes in water quality described above. The adsorption mass-transfer zone advanced through the pilot column in 13 weeks. During this time, the temperature decreased from 8 to 2 "C. Saturation of individual bed depths occurred in an even narrower, almost isothermal, temperature range. The significance of temperature for this period of pilot column operation is primarily to define the relatively static mean chloroform concentration and corresponding theoretical GAC capacity given by the isotherm relevant to that temperature. Thus in our original interpretation of these data, the capacity of the pilot column was concluded to be 85% of the theoretical capacity, predicted by using a single room temperature isotherm (13). This estimate is reduced to 64% if analytical recoveries for extraction of THMs from water (80%) and GAC samples (100%) are taken into account (11). This estimate is further reduced to 27% if operating temperatures for the pilot column are taken into account, and the theoretical capacity is predicted by using temperature-dependent Langmuir isotherms for CHC1, adsorption, as given in Table I. Desorption of CHC13from the pilot column occurred at sequential bed depths on a somewhat longer time scale than adsorption. However there is no consistent relationship between desorption occurring and the change in temperature at each bed depth. Net temperature changes during desorption were -4, -2, -8, +18/-18, -8, and -8 "C at 3.5, 24,44,64,85, and 105 cm, respectively. Desorption occurred at sequential bed depths on the pilot column regardless of whether temperature was decreasing and/ or increasing. Also the driving force for desorption on the pilot column is considered unrelated to the changes in temperature. As noted previously, changes in temperature coupled with seasonal variations in THM concentrations would also tend to stabilize GAC capacities and not lead to desorption. However, as seen in Figure 4, the long period that GAC remained at maximum capacity at the 85- and 105-cm bed

Table VI. Effect of Temperature and Seasonal Variations in Concentration on Adsorption Capacities of Calgon F400 winter temp, T1, "C

aq concn,

case

c1, MIL

AA' BB' CC'

12.2 28.3 20.2 12.2

DD'

1 1 1 13

I

sorbent concn, Qi, mg/g

aq concn, c 2 t MdL

1.63 3.14 2.45 1.03

28.3 12.2 20.2 28.3

I

0

0.5 I/C

1 .o

(UQ/LP

Figure 5. Pathways for seasonal varlations in GAC capacities resulting from seasonal Variations in temperature and CHCI, concentration as defined In Table V I .

depths is significant. Apparently the cause of desorption was moving progressively more slowly through the GAC bed and did not reach these bed depths until 40 and 50 weeks, respectively. At 85 cm, GAC remained at maximum capacity for 30 weeks, during an increase in temperature from 2 to 20 "C and an increase in CHC13 concentration from 11to 20 pg/L. Similarly, at 105 cm, GAC remained at maximum capacity for 37 weeks, during an increase in temperature from 2 to 20 OC followed by a decrease to 10 "C and an increase in CHC1, concentration from 11to 20 pg/L, followed by a decrease to 15 pg/L. During this time, the GAC capacity was fairly stable, except for fluctuations about the mean; this type of behavior is expected for GAC adsorption with slow seasonal variations in temperature accompanied by a seasonal change in concentration of the adsorbate. Seasonal Variations in Concentration and Temperature: Effect on GAC Capacity Predicted from Temperature-Dependent Isotherms. The long-term effect of relatively slow changes in temperature on GAC adsorption can be quantitatively assessed by temperature-dependent isotherms. Four cases are evaluated as defined in Table VI and plotted in Figure 5, for hypothetical seasonal variations superimposed on CHCl, isotherms. Concentrations and temperatures in the first case AA' are taken from the data base acquired at Waterford, NY, in 1985. Concentrations of CHC1, in 1985 and their correlation with temperature are somewhat greater than those found for 1983, during pilot column operation, or for 1984 (14). By using the CHC13 isotherms, concentrations of CHC13 on GAC at saturation in winter vs summer are calculated to differ by 26% relative to the mean value, the smallest variation of cases considered in Table VI. The actual pathway in going from winter to summer would be somewhat wider than shown for AA' in Figure 5 to allow for short-term daily variations in THM concentrations, which have a relative standard deviation averaging 17% (14). The main conclusion for seasonal variations AA' in GAC capacities is that the effect of increasing CHC13 concentrations in summer is counteracted by a less favorable adsorption isotherm at 25 OC. For GAC contactors such as the pilot column run for 60 weeks at Waterford

summer temp, Tz, O C 25 25 25 13

sorbent concn, Q2, mg/g

A(Q2 - Q1)/

1.25 0.58 0.92 2.00

-26 -138 -91 64

QW,%

in 1983, long-term seasonal variations in temperature and influent THM concentrations would represent a relatively small perturbation: long-term seasonal variations in CHC13 concentrations could not account for the slow, nearly complete, desorption of CHC13that occurred at sequential bed depths. Other cases are also of interest to consider, insofar as they pertain to the adsorption of THMs in treatment systems with different seasonal variations or to the behavior of compounds other than THMs at Waterford. There are compounds (for example, those causing taste and odor) with low mass inputs that approach breakthrough and saturation on a relatively long time scale. For example, in the hypothetical case BB', seasonal variations in aqueous concentrations are taken opposite in a sense to seasonal variations in temperature. This could be the case for compounds whose concentrations in water are determined by the temperature dependence of partitioning between the aqueous and vapor phases, rather than by the temperature dependence of the kinetics of chlorination. One would expect such compounds to exist, among synthetic volatile organics for example, having adsorption isotherms (10) and heats of adsorption (23) similar to the THMs. For such compounds, the transition from winter to summer would result in a 138% relative loss in GAC capacity, much greater than predicted in case AA' for CHC1,. For type B compounds, aeration (20) would be of interest to counterbalance summertime losses in efficiency of GAC treatment (the loss in driving force, for aeration with concentrations decreasing in summer, would still have to be evaluated). Case CC' is interesting since it allows only for seasonal variations in temperature, with no variation in concentration. Compounds of this type are expected to exist, with properties similar to case BB'. The effect of temperature for CC' is sufficient to account for a 91% relative loss in GAC capacity in going from winter to summer. In this case, aeration and GAC adsorption would be expected to have significant complementary roles. Finally case DD' applies to variations in influent concentrations under isothermal conditions, which can occur for groundwater systems. An 80% increase in aqueousphase concentration (relative to the average) corresponds to a 64% increase in GAC capacity and conversely. This is the only case in which changes in aqueous-phase concentrations and GAC capacities are in the same direction.

Conclusions In summary, although it is generally believed that temperature does not have a major effect on GAC adsorption in water treatment (1,2),this must be qualified for specific treatment systems by an understanding of seasonal variations in temperature and concentrations of individual adsorbates, as well as their rates of mass transport through GAC. In this paper, it has been shown that the characteristic decrease in GAC capacities with increasing temperature is significant, even for the THMs, which are weakly adsorbed and have a relatively small heat of adsorption, Environ. Sci. Technol., Vol. 22,

No. 4, 1988 411

found experimentally to average -5.8 f 2.0 kcal/mol. However pilot column data indicate that the mass-transfer zone for THM adsorption advances rapidly through a fixed-bed contactor, relative to slow seasonal changes in temperature and THM concentrations. Operating conditions for THM adsorption are nearly isothermal. Therefore, the primary significance of temperature-dependent isotherms for the THMs is to accurately predict GAC capacities for relevant values of the influent concentration and temperature of operation. However the THMs also penetrate rapidly to deep bed depths and are retained on a contactor for a long period of time, during substantial seasonal changes in temperature and mean influent concentration. By using the adsorption isotherms, GAC capacities for THMs are still predicted to remain stable through seasonal changes in operating conditions, and this behavior is confirmed by data for deep bed depths of a pilot column operated at Waterford, NY. THMs are in a sense unique: the loss in GAC capacity expected from an increase in temperature alone is counteracted by the THMs more favorable kinetics of formation and increasing aqueous concentrations, which also occur with an increase in temperature. Conceptually, it is possible to identify characteristics of compounds whose GAC adsorption would be most sensitive to the effects of slow seasonal variations in temperature: strong adsorption with large negative low influent concentration; slowly advancing mass-transfer zone; maximum concentrations in winter and minimum concentrations in summer or no seasonal concentration change. Laboratory experiments with model compounds are of interest to further test the effect of temperature on GAC adsorption and to determine whether a compound’s adsorption can be reversed by dynamic changes in temperature. To quantitatively evaluate the effect of temperature on GAC adsorption in water treatment, more measurements are needed for seasonal variations of specific compounds (synthetic organics; disinfection byproducts; substances causing taste and odor) and the temperature dependence of their GAC isotherms. Registry No. C, 7440-44-0; CHC13, 67-66-3; CHBrC12, 75-27-4; CHBr2Cl, 124-48-1;CHBr3, 75-25-2.

Literature Cited (1) Weber, W. J. Physicochemical Processes for Water Quality Control; Wiley-Interscience: New York, 1972; pp 208-209, 236. (2) Fed. Regist. 1985, 50, 46880-46933. (3) Faust, S. D.; Aly, 0. M. Chemistry of Water Treatment; Butterworth: Boston, MA, 1983; pp 209-210.

412

Environ. Sci. Technol.. Vol. 22, No. 4, 1988

(4) Morris, J. C.; Weber, W. J. Adsorption of Biochemically Resistant Materials from Solution 1; Environmental Health Series AWTR-9; U.S. Department of Health Education and Welfare: Washington, DC, 1964. (5) Snoeyink, V. L.; Weber, W. J.; Mark, H. B. Environ. Sei. Technol. 1969, 3, 918-926. (6) Zogorski, J. S. Ph.D. Thesis, Rutgers University, New Brunswick, NJ, 1975; as cited in ref 3. (7) Arbuckle, W. B. In Activated Carbon Adsorption; McGuire, Michael J., Suffet, Irwin H., Eds.; Ann Arbor Science: Ann Arbor, MI, 1980; Vol. 2, pp 237-252. (8) New York State Bureau of Public Water Supply Inventory: Community Water Systems with Sources; New York State Department of H e a l t h Albany, NY, 1974. (9) Giusti, D. M.; Conway, R. A,; Lawson, C. T. J.-Water Pollut. Control Fed. 1974, 46, 947-965. (10) Dobbs, R. A.; Cohen, J. M. Carbon Adsorption Isotherms for Toxic Organics; US. Environmental Protection Agency: Cincinnati, OH, April 1980; EPA-60018-80-023. (11) Alben, K. T.; Kaczmarczyk, J. H. Anal. Chem. 1986,58, 1817-1822. (12) Alben, K. T.; Kaczmarczyk, J. H. J. Chromatogr. 1986,351, 497-500. (13) Alben, K. T.; Shpirt, E.; Kaczmarczyk, J. H. 1984 Annual Conference Proceedings of the American Water Works Association; American Water Works Association: Denver, CO, 1984; pp 1555-1571. (14) Alben, K. T.; Shpirt, E.; Kaczmarczyk, J. H.; Berger, H.; Farrar, D. Proceedings of the 1986 Water Quality Technology Conference; American Water Works Association: Denver, CO, 1986; p p 881-894. (15) Alben, K. T.; Shpirt, E.; Kaczmarczyk, J.; Berger, H.; Farrar, D. J.-Am. Water Works ASSOC.,in review. (16) Crittenden, J. C.; Luft, P.; Hand, D. W.; Oravitz, J. L.; Loper, S. W.; Arl, M. Environ. Sei. Technol. 1985, 19, 1037-1043. (17) Weber, W. J.; Pirbazari, M. in Viruses and Trace Contaminants in Water a n d Wastewater; Borchardt, J. A,, Cleland, J. K., Redmand, W. J., Oliver, G., Eds.; Ann Arbor Science: Ann Arbpr, MI 1977; pp 125-141. (18) Shoemaker, D. P.; Garland, C. W. Experiments in Physical Chemistry; McGraw-Hill: New York, 1967; p p 258-262. (19) Lunge’s Handbook of Chemistry; Dean, J., Ed.; McGrawHill: New York, 1967; Chapter 9, pp 96, 140. (20) Nicholson, B. C.; Maguire, B. P.; Bursill, D. B. Environ. Sci. Technol. 1984, 18, 518-521. (21) Otson, R.; Williams, D.; Bothwell, P.; Quon, T. Environ. Sci. Technol. 1986,15, 1075-1080. (22) Edzwald, J.; Becker, W.; Wattier, K. J.-Am. Water Works ASSOC.1985, 77(4), 122-132. (23) McGuire, M. J. Ph.D. Thesis, Drexel University, Philadelphia, PA, 1977, pp 386-393. Received for review J a n u a r y 15, 1987. Revised manuscript received August 24, 1987. Accepted November 3, 1987.