Prediction of adsorption isotherms of organic compounds from water on activated carbons

Prediction of Adsorption Isotherms of Organic Compounds from Water on Activated Carbons V. Multiple Linear Regression Analysis Method An empirical adsorption equation is presented for predicting the adsorption isotherms from aqueous solution for a wide variety of hydrophobic porous adsorbent-adsorbate systems. The equation, which is based on the multiple linear regression analysis, comprises four parameters: surface area (S), mean pore diameter (D), molecular refraction (Mr), and number of hydrophilic functional group (Nf). Mr of adsorbate offers a basis for expressing quantitatively the adsorbent-adsorbate interactions associated with the dispersion forces. Nj in an adsorbate molecule represents the strength of the adsorbate-water interactions. S and D of adsorbent represent the adsorptive capacity and the adsorptive strength, respectively. The ability to predict the adsorption isotherm for a given adsorbent from a knowledge of the physical properties of the adsorbate is an important objective in adsorbent-adsorbate interaction research. In a previous paper, (1) a method was presented for predicting the adsorbability of a wide variety of organic compounds (alcohols, esters, ketones, aldehydes, ethers, amines, fatty acids, glycols, amino acids, saccharides, aromatics, etc.) from aqueous solution onto an activated carbon. The method requires only the knowledge of the adsorbate's molecular refraction and the number of hydrophilic functional group in the molecule. Furthermore, a method was presented for predicting the adsorbability of methylene blue on hydrophobic porous adsorbents with various porosity (specific surface area: 260-1420 m2/g, pore volume: 0.161-1.46 ml/g, mean pore diameter: 19.6-96.5 A) (2). The method requires 0nly the adsorbent's surface area and the mean pore diameter. In this paper the prediction method of the adsorption isotherm for various hydrophobic porous adsorbent-adsorbate systems has been investigated. EXPERIMENTAL The adsorbates studied were 12 organic compounds shown in Table I. All of the adsorbates came from commercial sources (stated minimum assay 99%) and were used without further purification. The adsorbents studied were 20 activated carbons with greatly different pore size distribution. Table II shows the properties of the adsorbents. The specific surface area (S) was determined from the BET nitrogen adsorption method. The pore volume (V) was calculated from the limiting vapor adsorption P/Po -~ 1. The mean pore diameter (D) was calculated from D = 4V/S

[11

the pore system being assumed to be made up of uniform cylindrical nonintersecting capillaries (3).

Equilibration took place in 50-rnl double-stoppered flasks, which were shaken for a minimum of 14 hr at 25 °C in a thermostated bath; check experiments at longer shaking times established that the shaking time sufficed for eq~ibrati0n. In order to eliminate loss through evaporation, pressure filtration was chosen for removing the carbon from the solution. The concentration of solute was analyzed by determining the total organic carbon in a Shimadzu Model TOC-10A analyzer. RESULTS AND DISCUSSION The adsorption equilibria of the 12 organic compounds from aqueous solution onto the 20 activated carbons at 25°C were measured. The adsorption data were approximated by the Freundlich equation with the following units: log X = log K + (I/N) log C [2] where X is the amount of solute adsorbed (mg/g of adsorbent), Cis the equilibrium concentration of solute (rag/ liter), and K and I / N are constants. For hydrophobic adsorption a good linear relationship is obtained between the logarithm of K and 1IN (4). The equation of the line for various activated carbon-organic compound systems is 1/N = -0.186 log K + 0.572 [3] From Eqs. [2] and [3] the adsorption isotherm can be expressed by the following equation with one parameter: log K = (log X - 0.572 log C)/(1 - 0.186 log C)

[4]

We now propose the following equation for predicting the log K: log K = A~M, + A2Ny + A3S + A,I) + Ao

[5]

where Mr, Ny, S, and D are molecular refraction, the number ofhydrophilic functional group, specific surface area, and mean pore diameter, respectively, and Ao-A4

588 0021-9797/84 $3.00 Copyright © 1984 by Academic Press, Inc. All rights of reproduction in any form reserved.

Journal of Colloid and Interface Science, VoL 99, No. 2, June 1984

NOTES

and --COOH, etc.) in the molecular represents the strength of the adsorbate-water interactions. The surface area is associated with the adsorptive capacity of the adsorbent. The mean pore diameter gives a measure o f the adsorptive strength of the adsorbent. An increase of m e a n pore diameter causes a decrease in the adsorptive strength since the dispersion forces is inversely proportional to the sixth power of the distance of adsorbent and adsorbate. The adsorption data of 385 points for the 20 adsorbents and the 12 adsorbates have been given to a NEC PC-8801 computer to determine the constants Ao-A4 in Eq. [6] by means of multiple regression analysis.

TABLE I Properties of Adsorbates Studied Adsorbate

M#

Nf

Methyl acetate Butyl acetate Butyric acid Hexanoic acid 1-Pentanol 2-Pentanone 2-Butanone 1,2-Butanediol Triethanolamine Tetraethylene glycol D-(-)-Mannitol Raffinose

17.922 31.866 22.280 31.576 26.820 25.434 20.786 23.696 38.310 47.640 39.088 105~236

1 1 1 1 1 1 1 2 4 5 6 16

589

(log X -

0.572 log C)/(1 - 0.186 log C) = AtMr + A2Nf + A3S + A4D + Ao

[6]

The data could be expressed by the following equation:

These physical constants are taken from Vogel, A. I., Cresswell, W. C., Jeffery, G. H., and Leicester, J., J. Chem. Soc. 514 (1952). are constants. The molecular refraction offers a basis for expressing quantitatively the adsorbent-adsorbate interactions associated with the dispersion c o m p o n e n t of the Van der Waals forces. The n u m b e r o f hydrophilic functional group ( - - O H , - - O - - , )C~-O, - - C H O , - - C O O - - ,

( l o g X - 0.572 log C)/(1 - 0.186 log C) = (0.127 ± 0.004)Mr + (-0.651 ± 0.002)Nf + (0.000283 _+ 0.000097)S + (-0.0285 + 0.0041)D + ( - 0 . 9 7 5 + 0.134)

[7]

r = 0.955, s = 0.207, F = 982 In this equation, the errors are in the 95% confidence intervals, r is the multiple correlation coefficient, s is the

TABLE II Properties o f Activated Carbons Studied

Carbon

[A] IS] [C] [DI [E] [S] [G] [H] [I] [Jl [K] [L] [M] [NI [0] [P] [Ol JR] [S] [T]

Raw material

Coal Coconut Coconut Wood Wood Coal Coal Coconut Coconut Coal Coal Coconut Coal Coconut Coconut Coconut Coal Coconut Coal Coconut

shell shell

shell shell

shell shell shell shell shell shell

Use

Specific surface area S (m'/g)

Pore volume V (ml/g)

Mean pore diameterD (A)

Decolorization Solvent recovery Waste water treatment Decolorization Solvent recovery Waste water treatment Waste water treatment Decolorization Water treatment Waste water treatment Waste water treatment Decolorization Waste water treatment Deeolorization Catalyst Waste water treatment Decolorization Water treatment Waste water treatment Solvent recovery

1440 978 704 1420 1164 1015 1215 941 953 824 831 834 707 994 1374 658 1010 1182 923 788

1.15 0.477 0.347 1.44 0.543 0.576 0.520 0.514 0.511 0.379 0.490 0.390 0.440 0.440 0.680 0.431 0.579 0.527 0.597 0.477

31.9 19.3 19.7 40.6 18.7 22.7 17.1 21.8 21.4 18.4 23.6 18.7 24.9 17.7 19.8 26.2 22.9 17.8 25.9 24.2

Journal of Colloid and Interface Science, Vol. 99, No. 2, June 1984

590

NOTES o

Z

2

o~ 3

o

~O

o

oo

, -c9~ 0

o

~-o

/ -1 -1

o

from the molecular refraction and the number of hydrophific functional group. Furthermore, the surface area and mean pore diameter of a given activated carbon can be predicted from the adsorption data of several compounds on the activated carbon.

o Ooo

~Co o o

°

I

I

I

0

i

2

REFERENCES 1. Abe, I., Hayashi, K., Kitagawa, M., and Hirashima, T., Bull. Chem. Soc. Jpn. 56, 1002 (1983). 2. Abe, I., Hayashi, K., Hirashima, T., Kitagawa, M., and Kuroki, N., J. Colloid Interface Sci. 93, 572 (1983). 3. Gregg, S. J., and Sing, K. S. W., "Adsorption, Surface Area and Porosity," p. 208. Academic Press, New York, 1967. 4. Abe, L, Hayashi, K., Hirashima, T., and Kitagawa, M., J. Amer. Chem. Soc. 104, 6452 (1982). IKUO ABE

log K cal

KATSUMI HAYASHI TSUNEAK/ HIRASHIMA

FIG. 1. Relationship between log K observed and log K calculated using Eq. [7].

standard deviation, and F-value represents the overall goodness of fit. The coefficients of Mr and S were positive values and those for Ny and D were negative. Figure 1 shows a plot of log Kobserved (Eq. [4]) and log K calculated using Eq. [7]. It can be seen from Fig. 1 that Eq. [7] gives a good prediction of the isotherms for many adsorbentadsorbate systems. If the surface area and the pore volume of a given activated carbon are known, the adsorption isotherms of many organic compounds can be predicted

Journal of Colloid and Interface Science,

Vol. 99, No. 2, J u n e 1984

Osaka Municipal Technical Research Institute Morinomiya, Joto-ku Osaka 536, Japan MUTSUO

K1TAGAWA

Society for Activated Carbon Research Morinomiya, Joto-ku Osaka 536, Japan Received August 31, 1983;acceptedNovember 16, 1983