Adsorption of Aliphatic Fatty Acids on Aquatic ... - ACS Publications

Two model surfaces were used (1) the mercury electrode .... near zero. At high positive or negative surface charge, the nonpolar organic molecules ...
1 downloads 0 Views 642KB Size
Adsorption of Aliphatic Fatty Acids on Aquatic Interfaces. Comparison between Two Model Surfaces: The Mercury Electrode and 6-AI2O, Colloids Hans-Jakob Ulrlch and Werner Stumm* Institute for Water Resources and Water Pollution Control, EAWAG, Swiss Federal Institute of Technology, CH-8600 Dubendorf, Switzerland

Boiena eosovl6 Center for Marine Research, Ruder Bogkovi6 Institute, 41001 Zagreb, Yugoslavia

Many organic substances in natural waters are amphipathic; i.e., they contain both a hydrophobic and a hydrophilic moiety. To assess the adsorptive behavior of such substances and to appreciate the factors that influence the distribution between particles and water, we need to understand how these substances interact with polar or nonpolar interfaces. To gain such an understanding, the adsorption of fatty acids of various chain lengths has been investigated on two model surfaces: (1) on a Hg electrode and (2) on 6-A1203particles. While hydrophobic expulsion dominates the adsorption on the nonpolar Hg surface, the adsorption of short-chain fatty acids (AI-OH H+ F? >Al-OH,+ (KaI-') (1) >Al-OH >A1-0- + H+ (KiJ

+

j~

The main objectives of this work were (1) to evaluate whether fatty acids adsorb on polar surfaces such as hydrous oxides or minerals by hydrophobic expulsion or by surface coordination (Figure 1)and (2) to describe the phase-selective ac polarography at a Hg electrode as a convenient method to study the extent and kinetics of adsorption of preponderantly hydrophobic organic species.

Experimental Section Phase-Selective ac Polarography. By use of phaseselective polarography, either the capacitive current (Le., the current needed to build up a certain potential on the electrode) or the Faradic current (i.e., the amount of current needed for electron transfer at the interface) can be measured selectively. The capacitive current reflects the extent of adsorption of the organic solute. Because of its fast response, the method can be used to evaluate the kinetics of adsorption. The principle is outlined here briefly: The capacitive behavior of the Hg/H20 interface is similar to that of two capacitors in series. The model of Bockris et al. (1) postulates an arrangement of two different layers adjacent to the electrode surface. The first layer, within the inner Helmholtz plane (IHP), consists of adsorbed water molecules and of contact-adsorbed anions of the supporting electrolyte. The second layer, between

0 1987 American Chemical Society

Environ. Scl. Technol., Vol. 22, No. 1, 1988

37

a)

bl

Flgure 1. Adsorption of fatty acids at two model surfaces. (a) At the nonpolar Hg electrode, the hydrophobic interaction between the hydrocarbon chain and the water phase leads to a displacement of the first-row H20 molecules and to adsorptlon at the Hg surface. (b) At the polar aluminum oxide surface, the fatty acid molecules become adsorbed because the functional carboxylic groups exchange for the surface OH groups (ligand exchange). At sufficiently hlgh chain length, adsorption due to the hydrophobic expulsion outweighs adsorption by coordlnatlve Interaction;the orientation of the fatty acid molecules may revert.

the inner and the outer Helmholtz plane (OHP), contains hydrated cations and water molecules with a lower degree of net orientation than the molecules that are contact adsorbed in the IHP layer. The capacitive behavior is mainly governed by the dielectric properties of the molecules contained in the adsorption layer. Thus, the interface can be interpreted in terms of a model of two capacitors in series, with the dielectric constants and t2, respectively, the inner Helmholtz plane and the adjacent plane, respectively (2, 3):

1

1

-(;

+ -1 = 1

c = c, c,

4

+

2)

Eo

(2)

where E , is the vacuum permittivity (C2.N-1.m-2) and d, and d2 are the thicknesses of the inner adsorption layer and of the diffuse layer, respectively. The total differential capacitance C is thus determined primarily by the layer with the lowest capacitance (Cl). If organic molecules become adsorbed, the situation in the adsorption layer changes in such a way that the capacitance C1 of this layer will be lowered, (1) because the organic molecules have a typically lower dielectric constant than the water molecules which have been replaced and (2) due to the increase of the distance dl(IHP) between the "plates" of the hypothetical interface condenser. Furthermore, in presence of adsorbed organic molecules the analogue circuit model may be assumed to consist of the capacitors C1 and C2, which are connected in a parallel circuit with a capacitor C3 that is representing the amount of surface covered by organic molecules. The capacitance C3 is given by C3 = eOtorg/d.;, where tOrgis the dielectric constant ?f the adsorbed organic material. Because the distance d , is usually larger than dl and cow is lower than EH@, the capacitance C3 is expected to be significantly lower than that of C1. The total differential capacitance C is then (4)

c

(1- 6)(

4-

$)

-1

4- 6c3

(3)

where 6 is the degree of surface coverage with respect to organic adsorbates. Since C1/(C,- C,) 38

Environ. Sci. Technol., Vol. 22, No. 1, 1988

(5)

organrc

molecules

H20

Flgure 2. The ac polarogram of caprylic acid at pH 4. Capacitive current as a function of the electrode potential for different bulk Concentrations of the fatty acid. Organic molecules become adsorbed in the potential range between E = -0.2 V and E about -1.2 V [vs Ag/AgCI/KCI (3 M) reference electrode]. Concentrations (m~bdm-~): (0)0;(1) 5.6 x 10-5; (2) 8.3 x 10-5; (3) 1.1 x 10-4; (4) 1.67 x 10-4; (5) 2.78 x 10-4.

where C1 = capacitance of the interface (pF.cm-2) covered by water molecules, C = capacitance at any coverage by organic molecules, and C3 = capacitance of the interface totally covered by organic molecules. Since C3 is much lower than C1 (5))increasing adsorption of organic surfactants leads to a decrease of the differential capacitance of the interface. The extent of adsorption of organic substances is dependent on their concentrations and on the surface potential of the mercury electrode. The nonpolar molecules tend to adsorb more strongly when the surface charge is near zero. At high positive or negative surface charge, the nonpolar organic molecules become replaced at the electrode surface by the polar water molecules. This is reflected by the measurement of the capacitive current of the interface mercury/aqueous solution as a function of the electrode potential (Figure 2). The capacitive current i, is proportional to the differential interface capacitance C, and the surface coverage B may be calculated directly from the capacitive current (6) according to i,(U - iC(8) 6= (6) i m - iJ3) where i,(l) is the capacitive current (PA) corresponding to the interface capacitance C, (H20 only), i,(O) is the capacitive current (PA) corresponding to the interface capacitance C, and ic(3) is the capacitive current (PA) corresponding to the interface capacitance C3 (interface totally covered by organic molecules). The experiments were performed with a Metrohm E 506 and a PAR 170 polarograph, using a hanging mercury drop electrode (HMDE). The ac phase angle was selected to be -90". A potential of E = -0.6 V (vs Ag f AgCl reference electrode; 3 M KCl), Le., the potential where the difference between ic(l)and i,(3) is largest, was selected. This potential is near the potential of the electrocapillarymaximum (Eecm).The capacitive current, measured at the HMDE as a function of time, yields informations about the kinetics of the adsorption process (Figure 3) (7,8). Adsorption Experiments. pH was adjusted with 0.01 M acetate buffer (pH 4) or 0.01 M borate buffer (pH 9) in the case of caprylic, capric, and lauric acids (in the experiments at the Hg electrode) and with HC1 or NaOH (in the experiments with alumina). In the experiments with the Hg electrode, 0.5 M NaCl was used as a supporting electrolyte. The extent of adsorption on the Al,03

j p o l I

0

-556

E -.20 z - IO d o

-833 -111

-167

-278

0

150 300 450 600 750 900

10

tlme ( sec I

-m N

O

5 0

-7

-6

log c

-5

-3

-4

( mol / dm

)

-.i -1 1- -5

conceptratlop

5 -5

Figure 3. Differential capacitance C vs time, for dlfferent concentrations of caprylic acid at pH 4, at a constant potential E = -0.5 V at the Hg electrode (Inset, C vs t ) . From this set of data, the curves AC vs log c can be plotted. For each of the concentrations,the value AC is taken as the difference between the differential capacitance obtained for the supporting electrolyte solution (c = 0) and the differential capacltance obtained for the actual concentration, at three different accumulation periods: t,,, = 5, 30, and 900 s,respectlvely. 1

-4 5

log c

-4

-3 5

imoles.dT-31

-11- -5

5 -5

-4

log c

5

-4

-3 5

Imoles.dm-31

Figure 5. Data of Figure 3 plotted in terms of a double logarithmic plot of eq 8 show significant deviationsfrom the Langmuir isotherm. Panel a compares the experimental values with a theoretical Langmuir isotherm, using the same values for the adsorption constant B for both curves. Panel b shows that the adsorption process can be described by introducing the parameter a , which accounts for lateral interaction in the adsorption layer. Equation 8 postulates a linear relation between the ordinate [= log [&(l - e)] - 2aB/(ln lo)] and the abscissa (log c). If the correct value for a is inserted, a straight line results. For caprylic acld at pH 4, a value of a = 1.5 gives the best fit.

.B

.6 .4

species become adsorbed at much lower concentrations than the carboxylates. The latter cannot penetrate into the adsorption layer without being accompanied by positively charged counterions (Na+). The adsorption data do not fit a Langmuir equation:

e = Bc 1-8

log c

( mol / drn

)

Figure 4. Comparison of the adsorption of lauric and caprylic acld as a function of the total bulk concentratlon. At pH 4, this concentration Is made up predominantly of fatty acid molecules and at pH 9 predominantly by thelr conjugate bases. (0)Points at equilibrium (deldt = 0). (0)Points at which the equilibrium is not yet attained (dO/dt > 0). surface was determined with 14C-labeledfatty acids. After an equilibration period of 24 h, the alumina was separated from the solution with membrane filters (0.05 pm); then, the residual 14C activity of the filtrate was determined. To establish the pH dependence of the adsorption of fatty acids on alumina over a wide pH range, some adsorption experiments were carried out at pH 2 and pH 10, occurs at either pH; respectively. Slow dissolution of A1203 thus, adsorption equilibrium may not fully be attained; however, the specific adsorption of fatty acids is expected to retard the dissolution (9). Chemicals. The nonradioactive fatty acids were from Fluka (puriss.); the 14C-labeled acids were from New England Nuclear. 6-A1203 (Merck), characterized by a specific surface area of 120 m2.g-l, was used in concentrations of 1 gdm-3.

Results Mercury Surface. The adsorption of caprylic acid on the Hg surface, as measured by the decrease of the differential capacitance AC, is represented in Figure 3. The extent of adsorption at various accumulation times is read from differential capacitance vs time curves (inset of Figure 3). After 5 min adsorption equilibrium is essentially attained. In Figure 4 the adsorption data of lauric acid and its conjugate base are compared with those of caprylic acid and its conjugate base. As is to be expected the molecular

(7)

where B is the adsorption constant and c is the equilibrium bulk concentration of the adsorbate. The deviation from the Langmuir plot (as is shown for the data of Figure 4 in Figure 5a) is caused most likely by a two-dimensional association process of the fatty acids at the Hg surface subsequent to their accumulation at the Hg surface. A two-parameter equation is necessary to interpret the adsorption equilibria. The Frumkin (Frumkin-FowlerGuggenheim, FFG) equation has been specifically developed to take lateral interaction at the surface into account. In the FFG equation, the term t9/(l-e) is multiplied by the factor exp(-2 a e), which reflects the extent of lateral interactions: 8 exp(-2a8) = Bc 1-8 where a is the interaction coefficient. If a = 0, eq 8 reduces to the Langmuir equation; a > 0 indicates attraction, while a < 0 means repulsion. The values for the adsorption constant B and for the interaction coefficient a can be determined from the intercept and the slope of the resulting straight line in a plot of log [e/(l - ec] vs t9 (IO). Experimental data for the adsorption of different fatty acids fit well with the FFG isotherm. This is illustrated for the adsorption of caprylic acid (Figure 4) in Figure 5b. The values for the adsorption constant B and for the interaction coefficient a are given in Table I. A t pH 4, where the adsorbing species are uncharged acid molecules, attractive lateral interaction (a > 0) occurs. At high pH values, the species getting into the adsorption layer are the conjugate anions of the fatty acids. Under such conditions, repulsive lateral interaction is observed (a < 0). &-AluminumOxide. Figure 6 gives the adsorption isotherms of propionic acid (CJ, butyric acid (C4),caprylic acid (C&,and lauric acid (C12)at pH 2, pH 6, and pH 10. Environ. Sci. Technol., Vol. 22, No. 1, 1988 39

~~

Table I. Characterization of Fatty Acid Adsorption on the Hg Surface in Terms of the Adsorption Constant B and the Interaction Coefficient a of the Frumkin-Fowler-Guggenheim Equation (eq 8)

Table 11. Comparison of the Free Energies of Adsorption of Free Fatty Acids at the Mercury Surface and on the Surface of 6-A120, Colloids AG,ds, kJ-mol-’ (298 K)”

PH 4 1% B, mol~dm-~ a

acid propionic (C3) butyric (C,) valeric (C,) caprylic (C,) capric (Clo) lauric (C12)

0.35

0.6 1.2 2.1 1.5 1.5 1.7

0.8 1.6 3.1

4.1 5.2

PROPIONIC ACID

PH 9 1% B, m~l-dm-~ a

2.5

( C 3 )

4.1

-1.0 -1.5

5.1

-1.2

BUTYRIC ACID

( C 4 )

-2.3 -5.1 -8.6 -17.7 -23.2 -30

A1203C -17.3 -17.1

-18.3

-27

Calculated from the adsorption constant B (Hg electrode) by

-2

-4

Hg surfaceb

AG = -RT(ln B ) and AG = -RT(ln sK) (alumina), respectively.

I

PH 6

-2

acid C,H,COOH C8H7C00H C4HgCOOH CTH&OOH CgHigCOOH CllH&OOH

bBased on the data obtained at pH 4 (where “free” fatty acids prevail). “he equilibrium constant nKcorresponds to that defined by eq 10. Because of the pH- and surface-dependentcharge, the constants given are conditional for the surface charge encountered at 1 = 0.02 and pH 6.

-4

-6

-6

-6

-35

I.35

-8

-8

-10

-10

-8 log c

-

-6

-2

-4

( mol / dm 3

CAPRYLIC ACID n .

-10

0

1

log c

( C 8

1

.

.

-6

-4

IOQ c

4

( C 12 )

n

-10

( mol / dm

-4

-4

-64

-6

,

,

- i o -a

-6

;-a{

,

IOQ

)

-8

,

-10

-10

0

-2

-21

-p

-10 ,

0

( mol / dm 3 )

LAURIC ACID

-8

-8

-2

n

-6

,

-4

I -

-2

-10

-6

.

C

(

mol / dm

- ,

I

3

5

7

9

1

0

1

-

1

PH

Flgure 7. Dependence of the distribution coefficient Qedsupon the pH value, for different allphatic fatty acMs at 6-A1203(1 g.dmg): proplonlc (C,), butyric (C ), caprylic (Ca),and lauric (C12). Qads= adsorption coefflcient (dm’sg-‘).

In Figure 7, the distribution ratio is plotted for these acids as a function of pH. The distribution ratio Qadsis given by

where x,ds is the amount of acid adsorbed at the 6-A120, particles (mol-g-l), caq is the total concentration in the aqueous phase (m~l-dm-~), and [HA] and [A-] are the bulk concentrations of the fatty acids and their conjugate anions, respectively (mol.dm-9. The results can be quantified 40

Environ. Scl. Technol., Vol. 22, No. 1, 1966

1

2

3

4

a,: 5

log P

I! 0

6

0

2

4 4

, 8,

6

n u m e r a’

0 10

;I

52

14

C - atons

0 / W - par:ition coeff

Flgure 8. (a) Adsorption on the mercury electrode vs h;drophobic properties (as measured by the octanoVwater partition coefficient) of adsorbing species at pH 4. B,, is the adsorption constant (dm3.rnol-’). (b) The free energy of adsorption as a function of the number of carbon atoms (n,) in the fatty acid, on the mercury electrode and on the alumlna.

I

-4 -2

Flgure 6. Adsorption on 6-Ai203. The isotherms for some aliphatlc capryllc (ca),and laurlc (CI2), fatty acids, propionic (C3), butyric (C4), at pH 2, pH 6, and pH 10, respectively.

1

, ,

0 0

-4

i

-8

in terms of a surface complex formation reaction (Figure lb): [>Al-A J 8K = [>AI-OH][HA] The concentration of nonprotonated >AI-OH groups can be calculated from mass balance considerations, the quantity of fatty acids adsorbed, and the microscopic acidity constants of reactions (eq 1). Values of the free energy of adsorption for some fatty acids are given in the last column of Table 11. As shown in Figure 6, the adsorption isotherms on alumina, different from those on the Hg electrode, plot nearly linearly in a double log plot (quantity adsorbed vs residual equilibrium concentration); thus they can be represented by a Freundlich-type equation; since the slope is nearly 1,they can also be fitted by a Langmuir equation that can be derived from eq 10 (11). The deviation observed at higher total concentrations is caused by surface saturation.

Discussion The adsorption of fatty acids on the nonpolar Hg electrode is dominated by their hydrophobic properties. The extent of adsorption increases with increasing chain length. An excellent correlation is observed between the adsorption constant B and the hydrophobic properties (as defined by the octanol/water coefficient) of the fatty acids (Figure 8). The following relationship of the free energies of adsorption AGA and the number of C atoms n, of the fatty acids can be established: -RT In B = AGads = 7 - 31n, (kJ-mol-l) (11) The interaction coefficient a influences the shape of the adsorption isotherms; it is a measure for the two-dimen-

Environ. Sci. Technol. 1888, 22, 41-46

sional (lateral) interaction (attractive or repulsive) of the adsorbate at the surface. As Table I illustrates, this coefficient a tends to be positive (0.6-2) for the free acids and negative (-1 to -1.5) for their conjugate bases. The repulsive interaction observed for carboxylates reflects the repulsion of the equally charged hydrophilic groups. The excess of counterions in the outer adsorption layer leads to a loss of entropy. The attraction observed for the free acids may be caused by hydrogen bonding between the carboxylic groups and by van der Waals attraction between the hydrocarbon chains. As seen in Figure 4, the adsorption of lauric acid ((212) is slow because of slow transport (diffusion) at concentrations smaller than lo4 M. In case of sodium caprylate (C,) the attainment of equilibrium is delayed most probably by structural rearrangements at the Hg surface. In case of anions, such association reactions are slower (repulsive lateral interaction) than with free acids. Apparently the longer chain fatty acid anion (C1J becomes rearranged in the surface layer somewhat faster than the shorter anion ((3,). In the case of 6-A120, the adsorption of fatty acids results from coordinative interaction (Figure lb) of the carboxylic groups with the A1 ions, the Lewis centers, in the surface layer. Mass law considerations (eq 10) on the pH dependence of the adsorbens and the adsorbate predict a maximum extent of surface binding around a pH value close to 5-6, Le., close to the pK value of the fatty acids (12). Figure 7 illustrates a pH dependence in accordance with that predicted. The values of the log 8K values obtained experimentally are comparable with those observed for other carboxylic acids; e.g., Kummert and Stumm report a value for benzoic acid of log *K = 3.7 (11). While -AGeh for the Hg surface increases systematically with chain length, this is not the case for -AGed, on the

alumina surface. Plausibly, the adsorption energy due to coordinative interaction, among short-chain fatty acids is not dependent on chain length; with molecules such as caprylic acid (C,) and larger ones a free energy contribution due to the hydrophobic effect of the longer hydrocarbon moiety becomes preponderant. As has been shown before for carboxylic acids (11),more than one layer may become adsorbed, presumably because of hemimicelle formation.

Literature Cited (1) Bockris, J. O'M.; Devanathan, M. A. V.; Muller, K. Proc. R. SOC.London, A 1963,274, 55-79. (2) Bockris, J. G. M.; Reddy, A. K. N. Modern Electrochemistry; Plenum: New York, 1970; pp 758-759. (3) Bard, A. J.; Faulkner, L. R. Electrochemical Methods; Wiley: New York, 1981; pp 506-516. (4) Hansen, R. E.; Minturn, R. E.; Hickson, D. A. J. Phys. Chem. 1966, 60, 1185; 1957,61, 953. (5) Payne, R. Adv. Electrochem. Electrochem. Eng. 1970, 7, 1-76. CosoviE, B.; Kozarac, Z. Mar. Chem. 1983,14, (6) KrznariE, Do; 17-29. (7) Batina, N.; RdiE, J.; CosoviE, B. J. Electroanal. Chem. 1985, 190,21-32. (8) RdiE, J.; Ulrich, H. J.; CosoviE, B., submitted for publication

in J. Colloid Interface Sci.

(9) Furrer, G.; Stumm, W. Geochim. Cosmochim. Acta 1986, 50, 1847-1860. (10) Trasatti, S. J. Electroanal. Chem. 1974,53, 335-363. (11) Kummert, R.; Stumm, W. J. Colloid Interface Sci. 1980, 75(2), 373-385. (12) Stumm, W.; Furrer, G.; Kunz, B. Croat. Chem. Acta 1983, 56(4), 593-611.

Received for review February 19, 1986. Revised manuscript received November 3, 1986. Accepted September 7, 1987.

Effect of Storm Type on Rainwater Composition in Southeastern North Carolina Joan D. Wllley,' Ramona I. Bennett, Jeanne M. Wllllams, Robert K. Denne, Cynthia R. Kornegay, Mark S. Perlotto, and Beth M. Moore Department of Chemistry, University of North Carolina at Wilmington, Wilmington, North Carolina 28403-3297

a function of storm origin or type. During 1983-1987, the most acidic rain and highest sulfate and nitrate concentrations occurred in rain from local summer thunderstorms, followed by rain from continental frontal storms, with the least acidic rain coming from coastal storms. Seasonal variation was observed for rainwater pH (although not for sulfate or nitrate concentrations) from continental storms, with the most acidic rain in the summer. Thunderstorm nitrate concentrations were high enough to affect seasonal averages for nitrate concentration because thunderstorms are a warm-season type of rain. Coastal storm rainwater did not show seasonal changes; this type of rainwater is similar in pH, sulfate, and nitrate concentrations to rainwater in remote areas of the world. Sulfate from sea spray was a small percentage of the total sulfate except in coastal storm rainwater. Large annual differences in rainwater composition were observed.

information was then available about rainwater composition in this area, which ieceives precipitation from several types of storm systems. The study area includes the land-sea interface, which is a source of seawater, a basic and well-buffered solution, and salt marshes, which produce gases that eventually oxidize in air to become acids. There are, therefore, natural processes that could affect rainwater pH in opposite directions. This study area also includes the city of Wilmington, an industrialized urban area with a population of approximately 109000, the only coastal city in North Carolina. Southeastern North Carolina, including Wilmington, has enjoyed extensive industrial and population growth over the last decade, as has much of the southeastern US. This region as a whole may be experiencing changes in the composition of rainwater (1-41, with resulting changes in surface water composition (4-7), so the patterns of variation in rainwater composition are of interest.

Introduction In 1983, a project was initiated to study rainwater composition and variability in coastal North Carolina. Little

Methods Sampling. Rainwater was collected on an event basis throughout this study; this allows investigation of the

rn Rainwater composition in Wilmington, NC, varies as

0013-936X/88/0922-0041~01.50/0 0 1987 American Chemical Society

Environ. Sci. Technol., Vol. 22, No. 1, 1988

41