A mechanistic investigation of P-Xylene and water vapor sorption on soils and clay minerals

MISSING IMAGE

Material Information

Title:
A mechanistic investigation of P-Xylene and water vapor sorption on soils and clay minerals
Physical Description:
viii, 159 leaves : ill. ; 29 cm.
Language:
English
Creator:
Pennell, Kurt Davis, 1962-
Publication Date:

Subjects

Subjects / Keywords:
Soil Moisture -- Measurement   ( lcsh )
Clay minerals   ( lcsh )
Soil Science thesis Ph. D
Dissertations, Academic -- Soil Science -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1990.
Bibliography:
Includes bibliographical references (leaves 152-158).
Statement of Responsibility:
by Kurt Davis Pennell.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001750762
oclc - 26498485
notis - AJG3679
sobekcm - AA00004767_00001
System ID:
AA00004767:00001

Full Text









A MECHANISTIC INVESTIGATION OF P-XYLENE AND WATER VAPOR
SORPTION ON SOILS AND CLAY MINERALS














By

KURT DAVIS PENNELL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1990



























Copyright 1990

by

Kurt Davis Pennell












ACKNOWLEDGEMENTS


I have prospered from the knowledge and compassion of many, and I take

this opportunity to thank those persons. First, I would like to thank my advisory

committee chair, Dr. Arthur Hornsby, who not only convinced me to continue my

studies in soil science, but also provided financial and moral support at critical

junctures in my Ph.D. program. His scientific beliefs and convictions have been

an inspiration, and afforded me the freedom to develop a truly independent

research project. The remainder of my advisory committee, Dr. Dean Rhue, Dr.

Suresh Rao, Dr. Joseph Delfino, and Dr. Ramesh Reddy have all made significant

contributions to my graduate education. In particular, I thank Dean Rhue for

providing laboratory facilities and supplies, but more importantly for instilling in me

the value of solid research. I greatly appreciated the invaluable scientific guidance

and challenges given by Suresh Rao. In his spare time Suresh revived my ego

on the tennis court.

One of the finest attributes of the Soil Science Department is the respect

and professional opportunities offered to graduate students. In this regard, I thank

Dr. Randy Brown, Dr. Willie Harris, and Dr. Peter Nkedi-Kizza for their honesty and

encouragement. I would also like to thank Dr. Brian McNeal, whose support and







humor assured that my tenure as graduate student representative to the faculty

was an enjoyable and enlightening experience.

I thank Bill Reve for his initial tolerance and subsequent respect in the

laboratory, which has fostered a true friendship. Our numerous discussions on a

range of issues removed much of the tedium from my laboratory work. Linda Lee

and Ron Jessup also gave generously of their time and knowledge, for which I am

grateful.

I would like to acknowledge the financial support, in the form of a research

assistantship, provided by the State of Florida via the Soil Science Department,

and additional funding provided by the Florida Department of Environmental

Regulation.

I thank my mother and father for their pride in my accomplishments. Their

accepting nature has given me an appreciation for mutual respect which I could

not have otherwise obtained. Finally, I thank Page whose love has given me the

strength to excel and the tenderness to care.













TABLE OF CONTENTS


ACKNOWLEDGEMENTS ............................................................................ iii

A BSTRA C T ................................................................................................... vii

CHAPTERS

1 INTRODUCTION ................................................................... 1

Vapor-Phase Sorption ........................................................... 1
Adsorption on Oven-Dry Sorbents .................................... ..... 2
Competitive Adsorption ........................................................ 11
Sorption at High Relative Humidity................................... ..... 17
Research Sum mary ..................................................................... 19

2 VAPOR-PHASE SORPTION OF P-XYLENE AND WATER
ON SOIL MATERIALS AT HIGH RELATIVE HUMIDITIES.......... 23

Introduction ....................................................... ..................... 23
Materials and Methods ......................................................... 25
Results and Discussion ............................................. ........... 34
Sum m ary......................................................... ...................... 49

3 GAS CHROMATOGRAPHIC STUDIES OF P-XYLENE
SORPTION ON ANHYDROUS AND HYDRATED
QUARTZ SAND ..................................................................... 52

Introduction ....................................................... ..................... 52
Materials and Methods ......................................................... 55
Results and Discussion ............................................. ........... 58
Sum m ary........................................................................... ..... 68








4 COMPETITIVE ADSORPTION OF P-XYLENE AND WATER
VAPORS ON CA-, NA-, AND LI-SATURATED KAOLIN .............. 71

Introduction ....................................................... ..................... 71
Materials and Methods ......................................................... 75
Results and Discussion ............................................. ........... 77
Sum m ary....................................................................... ........ 95

5 THE EFFECT OF HEAT TREATMENTS ON THE TOTAL
CHARGE AND EXCHANGEABLE CATIONS OF CA-, NA-,
AND LI-SATURATED KAOLIN ................................................. 98

Introduction ....................................................... ..................... 98
Materials and Methods ................................................................ 101
Results and Discussion ............................................................... 105
Sum m ary ........................................................................................ 133

6 SUMMARY AND CONCLUSIONS .............................................. 135

APPENDICES

A WATER AND P-XYLENE SORPTION DATA.............................. 141

B SURFACE TENSION DATA........................................................ 150

REFERENC ES ............................................................................................. 152

BIOGRAPHICAL SKETCH ........................................................................... 159













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

A MECHANISTIC INVESTIGATION OF P-XYLENE AND WATER VAPOR
SORPTION ON SOILS AND CLAY MINERALS

By

Kurt Davis Pennell

December 1990

Chairman: Dr. Arthur G. Hornsby
Major Department: Soil Science


Vapor-phase sorption of p-xylene and water on soils and clay minerals was

studied using a flow-equilibration technique that allowed the amount of sorbed

water and p-xylene to be measured independently. Oven-dry sorbents exhibited

a sizable capacity to adsorb p-xylene vapors which was strongly correlated to

sorbent surface area. Increasing the relative humidity to 67 and 90% resulted in

dramatic reductions in p-xylene sorption and a shift from Type-II to Type-Ill

adsorption isotherms, regardless of the organic carbon content of the sorbent. The

observed increase in p-xylene sorption above relative vapor pressures of 0.6

suggests that hydrated sorbents can sorb significant quantities of organic vapors

near the contaminant source. Similar results were obtained for the sorption of p-







xylene vapors on quartz sand using a gas chromatographic technique, which

provided an efficient and versatile alternative to batch methods.

At relative humidities above which the sorbent surface was covered by at

least a monomolecular layer of water, p-xylene sorption by Webster soil was

primarily attributed to partitioning into organic carbon. However, predictions of the

surface excess based on the Gibbs equation indicated that essentially all p-xylene

sorption on sorbents of low organic carbon content was due to adsorption at the

gas-liquid interface. Therefore, it is recommended that vapor-phase sorption be

described by a multi-mechanistic model that incorporates adsorption onto exposed

mineral surfaces, adsorption at the gas-liquid interface, dissolution into adsorbed

water films, and partitioning into organic carbon.

The effect of cation saturation on the adsorption of water and p-xylene

vapors from single- and binary-vapor systems was also investigated. Water and

p-xylene adsorption on Ca- and Na-saturated kaolin was directly related to the

hydration energy of the exchangeable cation. However, Li-saturated kaolin

exhibited reduced adsorption which was attributed to a decrease in total charge

after heating. Therefore, Li-kaolin is not recommended as a reference surface for

the study of vapor-phase sorption. Predictions of the competitive adsorption of

water and p-xylene vapors using two-component Brunauer, Emmett, and Teller

models indicated that the preferential adsorption of water from p-xylene-water

systems was the result of cation hydration effects.












CHAPTER 1
INTRODUCTION


Vapor-Phase Sorption

The widespread detection of volatile organic chemicals (VOCs) in

groundwater and soil has prompted intensive study of vapor-phase sorption and

transport in porous media. Contaminants of this nature, such as petroleum-based

solvents, frequently enter the environment as a result of improper disposal

techniques, accidental spills, and loss from storage tanks. The subsequent

distribution and mobility of VOCs in the unsaturated zone can be directly

influenced by vapor-phase sorption. Previous studies have demonstrated that

anhydrous soils and clays have a sizable capacity to adsorb organic vapors, but

this capacity is greatly reduced in the presence of water (Call, 1957; Chiou and

Shoup, 1985; Rhue et al., 1989). However, a comprehensive understanding of the

processes responsible for VOC sorption in the unsaturated zone has yet to be

attained. This is particularly true for soils at high relative humidity (RH), at which

time a number of mechanisms may contribute to organic vapor retention, including

dissolution into adsorbed water films, partitioning into organic carbon (OC), and

adsorption at gas-liquid-solid interfaces.









Adsorption on Oven-Dry Sorbents

The adsorption of vapors on oven-dry sorbents has been studied in detail,

and provides a basis from which to investigate adsorption in more complex

systems. In his classic treatise on vapor-phase adsorption, Brunauer (1943)

defines adsorption as the accumulation of a chemical species from one bulk phase

at the surface of another bulk phase, without penetrating the structure of the

second phase. Based on the nature of adsorbate-adsorbent interactions,

adsorption can be classified as either physical or chemical (Table 1-1). The

adsorption of nonpolar organic vapors on mineral surfaces is generally considered

to result from nonspecific molecular interactions, such as van der Waals forces,

which are also responsible for vapor condensation and deviations from ideal

behavior. Chemical adsorption of aromatic hydrocarbons has been observed on

Cu-saturated montmorillonite using infrared and ultraviolet-visible spectroscopy

(e.g., Pinnavaia et al., 1974; Pinnavaia and Mortland, 1971). It was postulated that

the chemisorbed species was coordinated with exchangeable Cu(ll) ions via x
electrons. However, interactions of this nature are typically limited to clay minerals

saturated with certain transition metals under anhydrous conditions.

The equations of Freudlich, Langmuir, and Brunauer, Emmett, and Teller

(BET) are frequently used to interpret gas-solid adsorption data. Although the

Langmuir and BET models can be derived theoretically, the Freudlich adsorption

model is based solely on the following empirical equation:

S = kP1" (1-1)









Table 1-1. Comparison of physical and chemical adsorption of vapors on solids.


Property Physical Adsorption Chemical Adsorption


Enthalpy of
Adsorption

Reversibility
of Adsorption



Adsorbed
Layers

Adsorption
Rate

Adsorbate-
Adsorbent
Interaction


< 80 kJ/mole


Reduction in vapor
pressure results in
desorption.


Multilayer formation.


Instantaneous. May be
limited by diffusion.

Nonspecific. Van der
Waals forces.


2 200 kJ/mole


Stronger treatment
required for desorption.
Desorption may result
in chemical alterations.

Limited to monolayer.


Rapid or slow. May
require activation energy.

Specific. Chemical bond
formation.








4
where S is the amount adsorbed (mg/g), P is the equilibrium vapor pressure, and

k and n are constants. The constants k and n are indicative of the extent and

intensity of adsorption, respectively. At low vapor pressures the value of n

frequently approaches one, which results in a linear isotherm. However, the

Freudlich equation rarely fits gas-solid adsorption data over a range of vapor

pressures.

The Langmuir model was the first theoretical treatment of adsorption, and

has been applied with reasonable success to systems exhibiting chemical

adsorption. At equilibrium, Langmuir (1916) considers the rate of evaporation

(desorption) from occupied sites to be equal to the rate of condensation

(adsorption) on the bare surface, which yields the following equation:

S KP
= (1-2)
Sm 1+KP

where S is the amount adsorbed (mg/g), S, is the monolayer adsorption capacity

(mg/g), P is the equilibrium vapor pressure, and K is the rate of adsorption divided

by the rate of desorption. The Langmuir adsorption isotherm is linear at low vapor

pressures and reaches a limiting value at or above vapor pressures corresponding

to monolayer coverage (Figure 1-la).

Implicit in the kinetic derivation of the Langmuir equation are the following

assumptions: (1) the adsorbate behaves as an ideal gas in the bulk phase; (2)

adsorption is limited to monolayer coverage; (3) the energy of adsorption is

constant; (4) no adsorbate-adsorbate interactions occur; (5) adsorption is localized
















E
C)












E
W)


P (mm Hg)


Figure 1-1.


Gas-solid adsorption isotherms characteristic of the (a) Langmuir
model (Type I) and (b) BET model (Type II).


P
0








6
(i.e., site-specific). The assumption of constant adsorption energy and the

absence of adsorbate-adsorbate interactions are rarely valid (Ross and Olivier,

1964). However, the errors associated with these assumptions tend to nullify one

another, and thus the success of the Langmuir equation may be attributed, in part,

to this coincidence.

Brunauer, Emmett, and Teller (BET) extended the Langmuir model to

account for multilayer adsorption by assuming that the Langmuir equation is

applicable to each adsorbed layer (i.e., each layer has a fixed number of

adsorption sites). The first layer is assumed to have a heat of adsorption equal

to Q,, while all successive layers have heats of adsorption equivalent to the heat

of vaporization (Q0). Adsorption and desorption are considered to occur only from

exposed surfaces, and at equilibrium the amount of solute adsorbed on each layer

is at steady state. The BET equation is given as follows:

S C(P/Po)
= (1-3)
Sm (1-P/Po) [1-P/Po+C(P/Po)]

where S is the amount adsorbed (mg/g), Sm is the monolayer adsorption capacity

(mg/g), P is the equilibrium vapor pressure, Po is the saturated vapor pressure,

and Cis a parameter related to the heat of adsorption. The BET equation typically

yields Type-II adsorption isotherms which are characteristic of multilayer formation

resulting from physical adsorption (Figure 1-1b).








7
The values of Sm and C can be obtained from a least squares fit of

adsorption data to the linear form of the BET equation, given as follows:

P/Po 1 (C-1)P/Po
= + (1-4)
S(1-P/P,) SC SC

Estimated values of Sm and C for a range of adsorbate-adsorbent systems are

presented in Table 1-2. In theory, the approximate value of C is given by the

following equation:

(Q,- QC)/RT
C e (1-5)
where R is the gas constant and T is the temperature (K). Provided the heat of

vaporization (Q,) is known, the heat of adsorption (Q1) can be calculated from the

value of C (e.g., Chiou and Shoup, 1985; Jurinak and Volman, 1957). However,

Sing et al. (1985) concluded that the C parameter obtained from the BET model

does not provide a quantitative measure of heats of adsorption, but does indicate

the relative magnitude of adsorbate-adsorbent interactions. In general, the values

of C in Table 1-2 are indicative of the low interactions energies associated with

physical adsorption. The BET equation is considered to yield reliable estimates

of the monolayer adsorption capacity for systems exhibiting Type-II adsorption

isotherms. The surface area of the adsorbent can then be calculated from the

value of Sm and the cross-sectional area of the adsorbate, typically estimated from

the liquid density. This procedure has become standard practice for the

determination of surface area from N2 adsorption isotherms (Adamson, 1982).









Table 1-2.


BET parameters calculated from the adsorption of organic vapors on
anhydrous soil materials.


Adsorbate Adsorbent Sm C Reference"


Benzene
Benzene
Benzene
Toluene
Toluene
Toluene
Toluene
p-Xylene
p-Xylene
p-Xylene
p-Xylene
Ethylbenzene
Ethylbenzene
Dichloropropane
Dichloropropane
Dichloropropane
Dichloropropane
Chlorobenzene
m-Chlorobenzene
p-Dichlorobenzene
1,2,4-trichlorobenzene
Ethylene dibromide
Ethylene dibromide
Ethylene dibromide
Ethylene dibromide
Ethylene dibromide
Ethylene dibromide
Ethylene dibromide


Parsons silt loam
Weller silty loam
Woodburn silty clay
Bentonite
Kaolin
Silica Gel
Webster silty clay loam
Bentonite
Kaolin
Silica Gel
Lula aquifer material
Bentonite
Webster silty clay loam
Parsons silt loam
Weller silty loam
Bernow sandy loam
Summit silty loam
Woodburn silty clay
Woodburn silty clay
Woodburn silty clay
Woodburn silty clay
Yolo silty clay
Yolo loam
Salanis clay
Meloland clay loam
Hanford sandy loam
Aiken clay loam
Staten peaty muck


(1988); (4)


--mg/g--
7.3
11.3
5.6
7.3
4.2
65.3
2.3
7.9
4.6
64.7
1.8
7.8
2.6
8.2
12.7
1.2
8.1
7.5
7.4
5.5
9.5
35.9
18.7
23.9
26.3
3.7
26.9
11.7


1.8
20.0
13.6
10.0
27.0
12.0
12.0
10.0
18.0
17.0
45.0
17.0
14.0
7.8
32.7
18.1
8.1
22.8
24.4
78.5
27.5
73.4
96.2
59.9
41.3
15.0
17.1
12.6


a (1) Poe et al. (1988); (2) Chiou and Shoup (1985); (3) Rhue et al.
Jurinak and Volman (1957).








9
Rhue et al. (1988) reported that adsorption of p-xylene vapors on oven-dry

silica gel was considerably greater than that on kaolin, bentonite, and Lula aquifer

material (Figure 1-2a). However, when the adsorption data were normalized to the

monolayer adsorption capacity (Sm) of each adsorbent, the relative adsorption was

essentially the same for all adsorbents studied (Figure 1-2b). A similar trend was

observed for the adsorption of ethylene dibromide (EDB) on a number of soils

(Jurinak and Volman, 1957). The slight increase in relative adsorption noted for

Lula aquifer material corresponded to a larger value of C, suggesting that specific

adsorbate-adsorbent interactions may have occurred. However, the overall

similarity in relative adsorption is remarkable considering the variation in adsorbent

properties, and indicates that the adsorption of organic vapors on mineral surfaces

is primarily a function of surface area (Jurinak and Volman, 1957; Rhue et al.,

1988).

The significance of the monolayer adsorption capacity obtained from the

BET model has recently come under scrutiny. Rhue et al. (1988) suggested that

for systems exhibiting low values of C, the coverage of the entire surface with a

single monolayer may not occur at any relative vapor pressure. In fact, multilayer

adsorption of EDB on Ca-saturated montmorillonite and kaolinite was observed at

surface coverages of 0.65 and 0.35, respectively (Jurinak and Volman, 1957). The

corresponding values of C for the montmorillonite and kaolinite systems were 24

and 14, respectively. In theory, the onset of multilayer formation prior to

monolayer coverage would result in the formation of discrete adsorbate films or








100
Silica gel
80 Kaolin
o Bentonite
60 o Lula


CO 40-


20


0 0.1 0.2 0.3 0.4 0.5 0.6

2
o

(b) o
1.5 o m *
0

S1 o *
UO 0

0.5 .



0 0.1 0.2 0.3 0.4 0.5 0.6

p-Xylene (P/Po)

Figure 1-2. Adsorption of p-xylene vapors on oven-dry silica gel, kaolin,
bentonite, and Lula aquifer material, expressed as (a) amount
adsorbed and (b) relative amount adsorbed, reported by Rhue et al.
(1989).








11
"patches" on the adsorbent surface. This scenario appears to be reasonable for

the adsorption of nonpolar organic vapors on mineral surfaces based on the

similarity in their heats of adsorption and condensation (Rao at al., 1989).



Competitive Adsorption

Under natural conditions, the adsorption of organic vapors will usually occur

on surfaces containing at least some adsorbed water. Chiou and Shoup (1985)

reported that the magnitude of water, 1,2,4-trichlorobenzene, chlorobenzene, and

benzene adsorption on oven-dry Webster soil was positively correlated to

adsorbate polarity (Figure 1-3a). It should be noted that the relative adsorption

was similar for all adsorbates studied, which suggests that the monolayer

adsorption capacity determined from the BET model accounts for variations in both

adsorbate and adsorbent properties (Figure 1-3b). The fact that adsorbate polarity

enhanced vapor-phase adsorption indicates that water will compete strongly with

organic vapors for adsorption sites. The adsorption of water on mineral surfaces

has been attributed to cation-dipole interactions, hydrogen bonding, and weak

charge transfer, which fall within the extremes of physical and chemical adsorption

(Burchill et al., 1981). Jurinak and Volman (1961b) reported that the enthalpy of

water adsorption on Ca- and Ba-saturated kaolin was approximately -120 kJ/mole,

which corresponded to an enthalpy of cation hydration of approximately -250

kJ/mole. In contrast, Rao et al. (1989) reported that heats of adsorption for

trichloroethene (TCE), toluene, and cyclohexane vapors on oven-dry Oldsmar soil







35

30

25

20
E
15

10

5

0
0


Figure 1-3.


0.2


0.4


0.6


0.8


P/P
0
Adsorption of water, 1,2,4-trichlorobenzene, chlorobenzene, and
benzene on oven-dry Woodburn soil, expressed as (a) amount
adsorbed and (b) relative amount adsorbed, reported by Chiou and
Shoup (1985).


0.2 0.4 0.6 0.8 1


3

2.5

2


E

C)


1.5


0
0
-o


0

"00
S o



-i (b)

I I I _


1

0.5








13
were -23, -37, and -34 kJ/mole. Thus, as the relative humidity or soil moisture

content increases, water will displace organic vapors from adsorbent surfaces

resulting in the suppression of VOC sorption. This finding is supported by infrared

studies, which have demonstrated that p-xylene was immediately displaced from

the surface of montmorillonite when laboratory air was introduced into the sample

cell (C. T. Johnston, 1990, personal communication).

Prior to the attainment of monolayer coverage with water, the adsorption of

organic vapors may proceed, essentially unhindered, on exposed mineral surfaces.

Rhue et al. (1989) reported that p-xylene adsorption on silica gel and kaolin was

not suppressed until the RH was greater than that corresponding to the monolayer

adsorption capacity determined from the BET model (i.e., RH 18%) (Figure 1-4).
In contrast, Call (1957) observed a substantial decrease in ethylene dibromide

adsorption at 5% RH (Figure 1-5). Regardless of the exact point at which

competitive effects become evident, these data indicate that the presence of water

vapor not only reduces organic vapor sorption, but also results in a shift from

Type-II adsorption isotherms to ones that are essentially linear.

In order to predict the competitive adsorption of water and organic vapors

on mineral surfaces a two-component adsorption model must be employed.

Fortunately, the BET equation has been modified by Hill (1946a, 1946b), and more

recently by Rhue et al. (1989), to account for adsorption from two-component

systems. The only additional assumption required to derive the equation of Rhue

et al. (1989) is that each component adsorbs only onto exposed mineral surfaces






















0.2 0.4 0.6


0 0.2 0.4


p-Xylene (P/Po)


Figure 1-4.


The effect of RH on the adsorption of p-xylene vapors on
(a) kaolin and (b) silica gel reported by Rhue et al. (1989).


0)
E
Cl)


2

0


140

120

100


0.8


0)


80

60

40


0.6









15

LO
10



c;

r~
60 0
t:
.0



xO
0, a,
C,








oo




0c

a
0










III I
S0,
a,



















o



0 0 0 0 -


(6/6w) C



C
0I
.u








16
or adsorbed layers of itself. This implies that the two components form immiscible

fluids on the surface, which is reasonable for adsorbates of low mutual solubility,

such as water and a nonpolar organic vapor. The total adsorbed mass of

component "a", Ma (mg/g), can be calculated from the following equation:

Ma = a C, So [Xa/(1-X,) (1-6)

where a. is the mass of component "a" occupying a unit area of surface (mg/m2),

Ca is the BET parameter related to the heat of adsorption determined from single-

component adsorption isotherms, Xa is the relative vapor pressures of component

"a", and So is the area of bare surface per unit mass of adsorbent (m2/g).

Unfortunately, the value of So is unknown in a binary system, and thus the amount

of each component adsorbed could not be expressed on a per-mass-of-adsorbent

basis (Rhue et al.,1989). However, the two-component model was evaluated by

comparing the measured and predicted mass fraction of component "a" (F,) on the

surface using the following equation:

Ma
Fa =. (1-7)
M + Mb

The multi-component BET equation of Hill (1946a, 1946b) is based primarily

on the BET postulate which states that molecules of the second and higher

adsorbed layers possess evaporation-condensation properties of the bulk liquid.

If this assumption is valid, it follows that adsorption and desorption from the

second and succeeding layers of a binary system should approximate the

properties of the liquid mixture. In addition, Hill assumed that area fraction of








17
component "a" (y.) is related to the mole fraction (Na) by the following equation:

Na
Y, = (1-8)
Na + (1-N,) (vdv,)23

where va and Vb are the mole fractions of component "a" and "b", respectively. If

component "a" is assumed to adsorb only onto bare mineral surfaces and an

adsorbed layer of itself, Hill's equation can be reduced to:

S, XJ[C.(1-Xb) + XbCJ
(1-9)
Sm (1-Xa-Xb) [1+X,(C,-1) + Xb(Cb-1)]
where S, is the amount of component "a" adsorbed (mg/g), S,, is the monolayer

adsorption capacity of component "a" (mg/g), X, and Xb are the relative vapor

pressures of component "a" and "b", respectively, and C, and Cb are the BET

parameters related to the heat of adsorption for component "a" and "b",

respectively. The values of Sam, C,, and Cb are determined from single-component

adsorption isotherms. If one of the components is absent (i.e., X, or Xb = 0), Hill's

equation reduces to the single-component BET equation. Although the derivation

referred to here is limited to a two-component system, in theory, the Hill equation

can accommodate an infinite number of components.

Due to the lack of suitable adsorption data, the two-component models of Rhue

et al. (1989) and Hill (1946a, 1946b) have rarely been tested.









Sorption at High Relative Humidity

At RHs approaching saturation (i.e. RH > 90%), the sorption of organic
vapors has been attributed to partitioning into organic carbon (Chiou and Shoup,

1985). This hypothesis is based on (1) the linear nature of benzene and

chlorobenzene vapor adsorption isotherms on Woodburn soil at 90% relative

humidity, and (2) the similarity in the magnitude of sorbate uptake from the

aqueous and vapor phase. However, linear isotherms have been obtained for a

range of sorbents, including those with trace levels of organic carbon (OC), such

as kaolin, silica gel, and Ca-saturated montmorillonite (Call, 1957; Rhue et al.,

1989). In addition, vapor-phase sorption coefficients, normalized for OC (Kc),

calculated for sorbents of low OC content have repeatedly been found to be orders

of magnitude greater than literature values (Rhue et al., 1989; Peterson et al.,

1988). Although these data suggest that other mechanisms may contribute to the

vapor-phase sorption at high RHs, the sorption of organic vapors on hydrated soils

is widely described by dissolution into adsorbed water films and subsequent

partitioning into OC (Baehr, 1987; Glotfelty and Schomburg, 1989; Jury et al.,

1990).

The adsorption of insoluble and sparingly-soluble organic vapors has also

been investigated by gas chromatography (GC). In general, Type-Ill isotherms

have been obtained for the adsorption of nonpolar vapors on hydrated sorbents,

with heats of adsorption that are smaller than heats of vaporization (Dorris and

Gray, 1981; Karger et al., 1971 a, 1971b). These data suggest that the gas-liquid








19
interface of adsorbed water films acts as a low energy surface toward nonpolar

vapors. Once the surface is covered with several monomolecular layers of water

it appears that the sorbent has no effect on vapor-phase adsorption. Karger et al.

(1971a, 1971b) concluded that adsorbed water films between 1.5 and 200 nm in

thickness have properties similar to bulk water. Measurements of the change in

surface tension of bulk water with vapor pressure also indicate that nonpolar

organic vapors are adsorbed at the gas-liquid interface (Blank and Ottewill, 1964;

Cutting and Jones, 1955; Hauxwell and Ottewill, 1968). Due to the large surface

area of adsorbed water films, adsorption at the gas-liquid interface of hydrated

soils may contribute significantly to VOC sorption. In fact, Call (1957) attributed

the sorption of EDB on moist soils to adsorption at the gas-liquid interface and

dissolution into adsorbed water films. It is apparent that a number of mechanisms

may contribute to vapor-phase sorption at high RHs, including partitioning into OC,

adsorption at the gas-liquid interface, and dissolution into adsorbed water films.

In addition, the extent to which sorbate and sorbent properties determine the

importance of each mechanism remains largely unknown.



Research Summary

The purpose of this work was to investigate, from a mechanistic perspective,

the sorption of organic vapors on soils and clay minerals. The study of soil

processes, particularly sorption, is limited by our inability to measure soil

phenomena on a microscopic scale. Thus, a macroscopic approach was







20
employed here, in which experiments were specifically designed to ascertain the

mechanism responsible for sorption under various sorbate-sorbent regimes.

A single volatile organic chemical, p-xylene, was utilized in these studies.

Although this may be viewed as a limitation, it allowed for intensive study of the

factors affecting vapor-phase sorption and for the development of experimental

techniques to study VOC sorption. Relevant physicochemical properties of p-

xylene are presented in Table 1-3. The sorbents used in these studies were

selected to provide a range of physical and chemical properties. In addition, the

OC content and cation saturation of certain sorbents were modified in order to

study the effect of such treatments on vapor-phase adsorption.

In Chapter 2, a flow-equilibration apparatus was utilized to measure the

sorption of p-xylene vapors on soils and mineral surfaces at 0, 67, and 90% RH.

Experiments were designed to determine the amount of p-xylene sorption

attributable to adsorption at the gas-liquid interface, dissolution into adsorbed water

films, and partitioning into OC. Adsorption at the gas-liquid interface of hydrated

sorbents was estimated from measurements of the surface tension of bulk water

exposed to p-xylene vapors using the Gibbs adsorption equation.

Gas chromatography techniques were used to study the sorption of p-xylene

vapors on anhydrous and hydrated quartz sand in Chapter 3. The effect of salt

treatments on vapor-phase sorption, and a comparison of batch and GC data are

also presented. In addition, the advantages and limitations of GC methods for the

study of vapor-phase sorption on soil materials are discussed.









Selected physicochemical properties of p-xylene.


Property


Valuea


Structure
Molecular Weight
Boiling Point
Density
Solubility
Vapor Pressure
Henry's Law Constant


C6H4(CH),2
106.17 g/mole
138.40C
0.86 g/mL (200C)
198 mg/L (250C)
8.14 mm Hg (240C)
7.44 X 10- m3- atm/mole


a Data obtained from Verschueren (1983) and Weast (1987).


Table 1-3.







22
In Chapter 4, the adsorption of p-xylene vapors on Ca-, Na-, and Li-

saturated kaolin was measured using the flow-equilibration method. The effect of

cation saturation on the competitive adsorption of water and p-xylene vapors was

also studied. These data were used to evaluate the two-component BET

adsorption models of Hill (1946a, 1946b) and Rhue et al. (1989).

Li-saturated kaolin was utilized as a reference surface for the study of water

and organic vapor sorption in Chapter 4. However, there has been considerable

debate in the literature as to whether or not Li-kaolin actually represents a surface

free of cation hydration effects. Thus, the mechanism responsible for reduced

water adsorption on Li-kaolin was investigated by gravimetric, spectroscopic, and

ion extraction techniques in Chapter 5.













CHAPTER 2
VAPOR-PHASE SORPTION OF P-XYLENE AND WATER ON
SOIL MATERIALS AT HIGH RELATIVE HUMIDITIES

Introduction

Vapor-phase sorption influences the mobility and distribution of volatile

organic chemicals (VOCs) in the unsaturated zone as well as the atmospheric

transport and deposition of VOCs. Previous studies have demonstrated that oven-

dry soils and clay minerals have a sizable capacity to adsorb organic vapors,

which can be described by the Brunauer-Emmett-Teller (BET) model (Call, 1957;

Chiou and Shoup, 1985; Jurinak and Volman, 1957; Poe et al., 1988; Rao et al.,

1989; Rhue et al., 1988). However, under natural conditions the sorption of

organic vapors usually occurs on surfaces containing at least some adsorbed

water. As the soil moisture content or relative humidity (RH) increases VOCs are

displaced from the surface, resulting in the suppression of VOC sorption (Call,

1957; Chiou and Shoup, 1985; Hollist and Foy, 1971; Rao et al., 1989; Rhue et

al., 1988, 1989; Spencer and Cliath, 1970, 1972, 1974, Wade, 1954). It is

generally agreed that water effectively competes with organic vapors for mineral

surfaces due to the polar nature of water and mineral adsorption sites (Valsaraj

and Thibodeaux, 1988).

At RHs above which the sorbent surface is occupied by at least a

monolayer of water, the specific mechanisms responsible for the sorption of VOCs








24
remain unclear. It has been postulated that the sorption of organic vapors at high

RH can be described by partitioning into organic carbon (OC) (Chiou and Shoup,

1985). This hypothesis was based on the linear nature of benzene and

chlorobenzene vapor sorption isotherms obtained for Woodburn soil (1.9% organic

matter) at 90% RH, and the similarity between the magnitude of solute uptake from

the aqueous and vapor phase (Chiou and Shoup, 1985). However, at 90% RH,

Call (1957) obtained linear isotherms for the sorption of ethylene dibromide (EDB)

vapors on Ca-saturated montmorillonite which contained no detectable OC. Rhue

et al. (1989) also reported linear isotherms for p-xylene vapor sorption on sorbents

with low OC contents; e.g., kaolin (0.65 g OC/kg) and silica gel (0.45 g OC/kg), at

67% relative humidity. In addition, sorption coefficients normalized for OC (K)

calculated from p-xylene sorption data and solubility considerations were 2 to 5

orders of magnitude greater than p-xylene Ko values reported in the literature. A

similar trend was noted by Peterson et al. (1988), who reported that linear sorption

coefficients measured for trichloroethylene (TCE) vapor over a range of

unsaturated conditions were 1 to 4 orders of magnitude greater than values

obtained under saturated conditions. Chiou and Shoup (1985) also found that the

vapor-phase sorption of benzene, m-dichlorobenzene, and 1,2,4-trichlorobenzene

was consistently greater than sorption from aqueous solution, even though the

sorbent surface should have been covered with several monolayers of water at

90% RH. These data suggest that in addition to partitioning into soil organic

carbon, other mechanisms contribute to VOC sorption at high RH.








25
Dissolution into adsorbed water films and adsorption at the gas-liquid

interface were considered by Call (1957) to be the dominant mechanisms

responsible for EDB sorption on moist soils. For sparingly-soluble VOCs, such as

p-xylene, the magnitude of sorption is far greater than dissolution into sorbed

water, based on solubility limits in bulk water (Rhue et al., 1989). However,

adsorption at the interface between the vapor phase and bulk water has been

reported for a number of sparingly-soluble VOCs (Blank and Ottewill, 1964; Cutting

and Jones, 1955; Drozd et al., 1982; Hauxwell and Ottewill, 1968). Due to the

large surface area to volume ratio of adsorbed water films, adsorption at the gas-

liquid interface of hydrated soils could contribute significantly to vapor-phase

sorption.

The experiments reported here were designed to determine the nature of

p-xylene vapor sorption isotherms at high RH and to determine the amount of p-

xylene sorption attributable to adsorption at the gas-liquid interface, dissolution into

adsorbed water films, and partitioning into OC. Adsorption at the gas-liquid

interface was estimated from surface tension measurements of distilled water

exposed to p-xylene vapors using the Gibbs adsorption equation.



Materials and Methods

Sorbents

The sorbents used in this study were selected to provide a range of surface

area and OC content. Colloidal kaolin (K-6, Lot #731063) was obtained from








26
Fisher Scientific Products. The kaolin had a cation exchange capacity (CEC) of

4.23 cmoljkg at pH 5.5 and was predominantly Na-saturated with only trace

amounts of Ca, K, and Mg (Rhue et al., 1989). Silica gel (Syloid 244), average

particle size of 3 pm and a pore volume of 1.4 cm3/g, was obtained from the

Davison Chemical Division, W.R. Grace & Co., Baltimore, MD. The Webster soil,

a silty clay loam (Typic Haplaquoll), was collected from the surface horizon (0-30

cm) of a site in Iowa and ground to pass a 106 pm mesh screen. The Webster
soil had a CEC of 51.11 cmolJkg and was predominantly Ca-saturated. A portion

of the Webster soil (Webster HP) was treated with hydrogen peroxide buffered to

pH 5.0 with acetic acid to remove OC.

Organic carbon content was measured by the Wakley-Black heat-of-dilution

method (Donohoe, 1983). The sorbent surface areas were determined from N2

adsorption data (Advanced Materials Research Center, University of Florida).

Cation exchange capacity was determined by washing the sorbents with 1.0 M

CaCI, 5 times, removing excess salts with 95% ethanol until a negative chloride

test was achieved using AgNO3, and extracting exchangeable Ca with 0.5 M

Mg(NO3)2. The amount of Ca present in the Mg(NO3)2 extract was then measured

by atomic absorption spectroscopy. These and other sorbent properties are

presented in Table 2-1.

Vapor-Phase Adsorption Studies

Vapor-phase sorption of p-xylene was determined on Webster soil and

Webster HP at 0 and 90% RH, and on kaolin and silica gel at 0 and 67% RH. The























I-
0
a-

a)
CD
a)






4.
C

La
0





0



4-











a)
Q-
ci
o
-2



'0









a
cc






a)
a
0





0






-c
a-




ca
I-


0D

E c


0)

CO





a



CO



*C
ODO











x

co
.0
CO


















Wa-
o 2
0















ml)
oc
0o













Wa)











o
00
4-



a )
0 c



(0 -
2(
00




u,


1




E
6)




o
0











E
0




























0)
0
I
O
O
,
:
.


cu




o
CD
















o
0


L0


LO







































4-
an-


co




CM





C4












4-
U)

Bm








28
flow-equilibration apparatus used to measure p-xylene and water vapor sorption

was similar to that described in previous studies (Rhue et al., 1988, 1989).

However, the p-xylene flow stream consisted of two gas washing bottles in series;

the first contained water and the second contained p-xylene and water (Figure 2-

1). This allowed the RH to be maintained at either 67 or 90% while the p-xylene

relative vapor pressure (P/Po) could be varied from 0 to 0.9, where P is the

equilibrium vapor pressure and Po is the saturated vapor pressure. The P, of p-

xylene at 240C is 8.14 mm Hg, which is equivalent to a vapor concentration 46.64

mg/L (Weast, 1987). For sorption experiments conducted at 0% RH, valve V3 was

closed and valves V, and V2 were adjusted to obtain relative vapor pressures of

p-xylene ranging from 0 to 0.9.

The concentration of water and p-xylene vapor in the flow stream was

determined by passing the flow stream through three polyethylene tubes in series.

The three tubes contained magnesium perchlorate, activated charcoal, and a

mixture of activated charcoal and magnesium perchlorate, respectively. The

difference between the initial and final trap weights yielded the total mass of vapor

trapped. The p-xylene vapor concentration was then determined by bubbling the

flow stream through two 40-mL glass centrifuge tubes containing 20 mL of

methanol. P-xylene trapped in these solutions was measured with a Perkin Elmer

Model 320 UV-VIS spectrophotometer. The concentration of water vapor in the

flow stream was determined by subtracting the concentration of p-xylene from the

total vapor concentration.


























I












Ir


o
LI
-j

WI



, -- -
0 I


RI



0 I


!0 I


,)

0
cm
z


C



0)


12
C
0
0




c
0


0
0



3


C
CL
.
cc
0








0


I-
a





-0W
0



oc

'6
04-' [








30
Approximately 1 g of sorbent, which had been oven-dried to 1300C, was

placed in glass centrifuge tubes and capped with teflon-backed septa. Four tubes

containing sorbent, and four blank tubes were placed on the flow stream in series.

The gas flow stream passed through the centrifuge tubes via hypodermic needles

at a rate of approximately 1.0 mL/sec. In general, an exposure time of 24 hours

was sufficient to attain sorption equilibrium.

Adsorbed concentrations of water and p-xylene were measured by

extracting the sorbents with 20 mL of methanol containing CaCI2. The methanol

solution was made by adding 10 mL of CaCI2-saturated methanol per 1 L of

methanol. Water in the methanol extract was measured by Karl Fisher (KF)

titration. The KF reagent was diluted with KF diluent to a strength of approximately

0.5 mL titer per mg of water. The visual endpoint was established by adding titer

to 20 mL of CaCI2-methanol solution until the desired endpoint was obtained. In

a second centrifuge tube, 5 mL of methanol solution was pretitrated to the visual

endpoint and 10 gL of deionized water was injected and titrated to the endpoint to
give the exact strength of the titer. A known volume of the methanol extract was

titrated in a similar manner to determine the amount of adsorbed water.

The concentration of p-xylene in the methanol extract of kaolin and silica gel

was measured using a Perkin Elmer Model 320 UV-VIS spectrophotometer. The

Webster soil and Webster HP released methanol-soluble compounds that

interfered with direct UV spectrophotometric analysis. Therefore, p-xylene extracts

from these sorbents were analyzed by high-performance liquid chromatography








31
(HPLC) techniques. The HPLC system consisted of a Gilson 302 pump, Waters

450 detector, Hewlett Packard 3392A integrator, and a 10-cm Waters RCM C-18

column with a 3-cm Brownlee guard column.

Surface Tension Measurements

The surface tension of deionized water exposed to p-xylene vapors was

measured using the "drop weight" method. The technique is based on Tate's law,

which considers the weight of a drop falling from a small diameter tube to be

proportional to the radius (r) of the tip (cm), and the surface tension (y) of the liquid

(g/s2),

weight per drop = 2 n r y. (2-1)
Tate's law assumes that a spherical drop will form at the tip, but in reality the drop

tends to elongate before it detaches from the tip. Harkins and Brown (1919)

recognized the importance of this discrepancy and developed a correction factor

(CF) based on the ratio of the tip radius to the length of the drop,

CF = f(rN'") (2-2)

where V is the volume of the drop (Table 2-2). In general, the values of rN/V3 were

between 0.45 and 0.50. The following equation was then used to calculate surface

tension:

mg
Y = 2, (2-3)
2 ?rCF

where m is the mass of the drop (g), and g is the acceleration due to gravity

(cm/s2).








32
Table 2-2. Drop-weight surface tension correction factors (CF) adapted from
Harkins and Brown (1919).


rN"3 CF rN/13 CF


0.30 0.7256 0.75 0.6032

0.35 0.7011 0.80 0.6000

0.40 0.6828 0.85 0.5992

0.45 0.6669 0.90 0.5998

0.50 0.6515 0.95 0.6034

0.55 0.6362 1.00 0.6098

0.60 0.6250 1.05 0.6179

0.65 0.6171 1.10 0.6280

0.70 0.6093 1.15 0.6407








33
A 10-mL graduated burette was used forthe surface tension measurements.

After carefully grinding the tip flush, an ocular microscope was used to measure

the diameter of the tip (0.37 cm). Drops of water falling from the tip were collected

in a 40-mL centrifuge tube, capped with a teflon-backed septa. A 0.4-cm diameter

hole was cut in the septa, and the height of the centrifuge tube was adjusted to

achieve a tight seal between the septa and the burette. A total of 18-20 drops, at

a rate of 4-6 drops per minute, were collected in the centrifuge tube. The surface

tension of water was then calculated from the volume and weight of liquid

collected. Solutions of NaCI were used to calibrate the diameter of the tip (0.3724

cm).

p-Xylene vapor, at relative vapor pressures from 0.1 to 0.8, was bubbled

through a 50-mL centrifuge tube capped with teflon-backed septa. The tube

contained approximately 40 mL of deionized water. No detectable change in the

p-xylene concentration was measured after one, two or four days of bubbling.

Therefore, the system was allowed to equilibrate for at least 24 hours. The

surface tension of deionized water exposed to p-xylene vapors was determined in

the same manner as described previously except that the p-xylene vapor was

allowed to flow through the collection tube for at least 2 minutes prior to initiating

the flow of drops.









Results and Discussion

Adsorption at 0% RH

Figures 2-2 and 2-3 show equilibrium isotherms for p-xylene vapor sorption

on silica gel, kaolin, Webster soil, and Webster HP at 240C. Sorption data are

expressed as milligrams of p-xylene sorbed per gram of sorbent (S) versus the p-

xylene relative vapor pressure (P/Po). At 0% RH, the p-xylene isotherms of all four

sorbents conformed to Type-II BET adsorption isotherms. This type of isotherm

is typical of unrestricted monolayer-multilayer adsorption of gases on nonporous

or macroporous (pore width > 0.05 pm) sorbents (Sing et al., 1985). These
sorption data were fit by a least squares procedure to the linear form of the BET

equation,

P/Po 1 (C-1)P/Po
= + (2-4)
S(1-P/Po) SC S,C

where Sm is the nominal monolayer adsorption capacity (mg/g), and C is a

parameter related to the heat of adsorption. Estimated values of S,, C, and the

P/Po associated with monolayer coverage are presented in Table 2-3.

The BET equation is considered to give reliable estimates of Sm for surfaces

exhibiting Type-11 adsorption isotherms. However, the value of C does not provide

a quantitative measure of the heat of adsorption, but does give an indication of the

relative magnitude of sorbent-sorbate interactions (Sing et al., 1985). The values

of C obtained for these systems are indicative of low sorbate-sorbent interactions

associated with physical adsorption. For systems exhibiting low values of C,










180
160
140
0i 120
C 100
80
0C 60
40
20
0


60

50


0 0.2 0.4 0.6 0.8


p-Xylene (P/Po)


Figure 2-2.


Sorption of p-xylene on (a) silica gel and
RH.


(b) kaolin at 0 and 67%


0 0.2 0.4 0.6


0)

E


40

30

20

10

0


1.0























00
cO




oc_

o
d











CM
ac






0


LI o
r-11-


(6/6wu)




























0 0
co a



o o
o 0o





oco
dd
66 V


o

c%
r_
L.




0












0



o













-2
oo
Cu





















0


to T-


U)
0 0
o 0











V- T-











L) 0
4 0
' CM


1. CM
.*t ,*-


C c

". 0





7 CC
c
O 0)


r-
ui





C
0







C/,
0


0n CM
N- T


- c0 CO
T- T- N


co
CD



0
0




0

d
















cr
CM



CO




a-
at
c

0x


0)
c-
0
cc
0






E


0
4-
0

a








38
discrete regions of multilayer sorption (i.e., patches) may form prior to the

attainment of complete monolayer coverage (Jurinak and Volman, 1957; Sing et

al., 1985).

The amount of sorbent surface available for the adsorption of a given

molecule (SA) can be estimated from the Sm value and the area occupied by each

adsorbed molecule. The cross-sectional areas (a) of water and p-xylene were

calculated to be 0.105 and 0.380 nm2, respectively, using the following equation

(Karnaukhov, 1985; McClellan and Harnsberger, 1967),

am = 1.09(MW/tA)m (2-5)

where MW is the molecular weight, I is the liquid density, and A is Avogadro's
number. This equation assumes that the sorbate molecules are oriented in a

hexagonal close packing at a density similar to that of the bulk liquid. The surface

areas occupied by either p-xylene or water molecules at monolayer coverage (S,)

are shown in Table 2-3.

The values of S, for oven-dry kaolin and Webster HP were similar to those

determined from N2 adsorption isotherms. These data are consistent with the

findings of Jurinak and Volman (1957) and Rhue et al. (1989), who concluded that

the N2 surface area provides a reasonable estimate of the area available for the

sorption of organic vapors on predominantly mineral surfaces. It is interesting to

note that the surface area of Webster soil, as determined from N2 and p-xylene

adsorption isotherms, increased following the hydrogen peroxide treatment. These

data suggest that both p-xylene and N2 vapors have a greater affinity for mineral








39
surfaces than for organic matter at 0% RH. The difference between the N2 (2.62

m2/g) and p-xylene/BET surface area (12 m2/g) of Webster soil may have been

due to greater sorption of p-xylene vapor by soil organic matter. This hypothesis

is supported by the work of Chiou and Shoup (1985), who found that the

adsorption of vapors on oven-dry Woodburn soil increased with the polarity of the

sorbate. The discrepancy between the N2 and BET surface area of silica gel has

been discussed in a previous paper (Rhue et al., 1989)

The relative sorption (S/Sm) of p-xylene on oven-dry kaolin, silica gel,

Webster soil, and Webster HP is shown in Figure 2-4. The use of S/S, allows for

the comparison of the adsorptive capacities of various sorbents on a unit-surface-

area basis. At low values of P/Po, the relative sorption of p-xylene on Webster HP

and Webster soil was slightly greater than that on kaolin and silica gel. The

increased sorption corresponds to an increase in the value of C for Webster soil

(C = 38) and Webster HP (C = 73), which suggests that specific sorbate-sorbent

interactions may have occurred. However, the overall similarity in relative sorption

for the sorbents studied here suggests that the adsorption of p-xylene vapors at

0% RH was primarily a function of surface area.

Sorption at High RH

Vapor-phase sorption of p-xylene on silica gel and kaolin decreased

dramatically when the RH was raised to 67% (Figure 2-2). The observed

suppression of vapor-phase sorption at high RH is in agreement with the findings

of others (e.g., Call, 1957; Chiou and Shoup, 1985; Rhue et al., 1989), and lends



































0

CL


c
U)

X
r_


0 L


S/S


0)

0

c,
0
C




I-

0

E



0



E


(0
0


oo


&a.
0(















0- .
ES














4-
CI
4-







co

0
Q.

Cu
0.-








41
further support to the contention that water effectively competes with organic

vapors for mineral surfaces. Sorption of p-xylene vapors on Webster soil and

Webster HP was also reduced at 90% RH (Figure 2-3). In addition, the sorptive

capacity of Webster soil at 90% RH was greater than that of Webster HP, while

the reverse was true at 0% RH. These data suggest that OC contributed to vapor

sorption on Webster soil at high RH, while adsorption on mineral surfaces was the

dominant mechanism at 0% RH.

At both 67 and 90% RH, p-xylene sorption isotherms were linear until the

P/Po reached approximately 0.5, above which, the amount of p-xylene sorbed

increased sharply, resulting in Type-Ill adsorption isotherms. The linear nature of

the sorption isotherms below 0.5 P/Po was consistent with isotherms obtained by

Call (1957) for EDB, and by Chiou and Shoup (1985) for benzene and

chlorobenzenes. Increased sorption at high relative vapor pressures has not been

previously reported for batch adsorption studies. However, researchers studying

sorption of hydrocarbons on hydrated silica and soil materials by gas

chromatography techniques have consistently obtained retention data which yield

Type-III adsorption isotherms (Dorris and Gray, 1981; Karger et al., 1971a;

Okamura and Sawyer, 1973). Therefore, hydrated soils appear to have a sizable

capacity for organic vapor sorption when the relative vapor pressure approaches

one.

The sorption of organic vapors at high RH is commonly described by

dissolution of organic vapors into sorbed water films, using Henry's Law constants








42
(KH), and subsequent solute partitioning into organic carbon, using sorption

coefficients normalized for OC content (KoJ. The justification for such an approach

comes primarily from a single article by Chiou and Shoup (1985). In this work, the

authors conclude that solute partitioning into organic carbon is the dominant

mechanism responsible for organic vapor sorption at high RH based primarily on

two pieces of evidence; (1) the linearity of the sorption isotherms and, (2) the

similarity in sorptive capacity from the aqueous and vapor phase. However, the

data presented here demonstrate that below 0.5 P/Po linear isotherms can be

obtained for sorbents with OC contents ranging from trace levels to 41 g OC/kg.

Thus, the mere existence of a linear isotherm was not sufficient evidence to

conclude that partitioning into organic carbon had occurred. In addition, the

dramatic increase in sorption above 0.5 P/Po was not consistent with partitioning

theory.

In order to compare sorption from the aqueous and vapor phase, Ko values

were estimated from measured p-xylene sorption data for P/Po below 0.5 (linear

portion of the isotherm) and the aqueous concentration of p-xylene determined

using a KH value of 7.44 X 103 m3- atm/mole (Table 2-4). Predicted p-xylene Ko

values for silica gel, kaolin, and Webster HP were 4, 2, and 1 orders-of-magnitude

greater, respectively, than Ko values determined from column studies (105-176

mUg) (Brusseau et al.; Gamerdinger et al., submitted for publication in

Environmental Science and Technology) and estimated from a log octanol-water

partition coefficient of 3.15 (573 mUg) (Karickhoff, 1984). However, the predicted








43

Table 2-4. Sorption coefficients normalized for organic carbon content (K)
calculated from p-xylene sorption data for P/P of 0.06 to 0.50.


Sorbent


Silica Gel


Kaolin


Webster HP


Webster Soil


RH


-(%)--

67


67


90


90


OC Content


---g/kg---

0.45


0.65


2.27


41.36


KOc


-----ml/g-----

2.35 X 106


9.10 X 104


1.53 X 103


2.17 X 102


0.936


0.989


0.967


0.907


---


----








44
p-xylene K. for Webster soil was within the range of literature values. Thus, as

the sorbent OC content increased, the predicted Ko value approached those

reported in the literature. These data suggest that the utility of Ko values for

predicting vapor sorption was limited to sorbents with relatively high OC contents,

at relative vapor pressures less than 0.5.

Adsorption at the Gas-Liquid Interface

Adsorption at the gas-liquid interface was considered as a possible

mechanism to account for the sorption p-xylene vapors on sorbents with low OC

contents and at high relative vapor pressures. The adsorption of insoluble and

sparingly soluble hydrocarbons on water surfaces has been estimated by

measuring the change in surface tension with the partial pressure of the organic

vapor (Baumer and Findenegg, 1982; Blank and Ottewill, 1964; Cutting and Jones,

1955; Hauxwell and Ottewill, 1968; Jho et al., 1978). This approach is based on

the Gibbs adsorption equation,

dy
F = (2-6)
dL
where r is the surface excess (mol/cm2), 7 is the surface tension, and g is the

chemical potential. If the vapor is assumed to obey the ideal gas law, equation (2-

6) can be written as,

P dy
r = (2-7)
RT dP

where P is the partial pressure of the organic vapor, R is the gas constant, and T

is the temperature. The surface tension of distilled water exposed to p-xylene







45
vapor, as determined by the drop-weight method, is shown in Figure 2-5a. These

data were fit by a least squares regression procedure to yield a slope which could

be then used to calculate the surface excess as a function of the partial pressure

using equation (2-7) (Figure 2-5b). The Type-III adsorption isotherm generated

from this procedure were typical of those obtained for other aromatic hydrocarbons

and indicates that water acts as a low energy surface toward the nonpolar vapor

(Cutting and Jones, 1955; Vidal-Madjar et al., 1976). The amount of p-

xylene adsorbed at the gas-liquid interface was estimated from the surface excess

and the N2 surface area of each sorbent (Table 2-5). In addition, the mass of p-

xylene partitioned into organic carbon was predicted using a K. of 200 ml/g and

a KH of 7.44 X 103 m3- atm/mole, and while mass of p-xylene dissolved in

adsorbed water was predicted using the KH value and the measured amount of

water adsorbed. The sum of the estimated values for these adsorption

components represents the total predicted adsorption, which is compared to the

measured adsorption data in Table 2-5. Adsorption on mineral surfaces was not

included in the predicted total because water is generally assumed to displace

organic vapors from mineral surfaces at high RH. However, the appropriateness

of this assumption at 67% RH is discussed.

The predicted data indicated that the gas-liquid interface was the dominant

mechanism responsible for vapor sorption on silica gel and kaolin. Neither

partitioning into organic carbon nor dissolution into sorbed water contributed

significantly to p-xylene sorption, due to the low OC content of these sorbents and














(M

L,


0 1 2 3 4 5 6 7


E


E
0


X
L,


0 1 2 3 4

P (mm Hg)


Figure 2-5.


5 6 7


Surface tension (a) of deionized water exposed to p-xylene vapors
at 240C and surface excess (b) of p-xylene at the gas-liquid
interface calculated using the Gibbs adsorption equation.









Table 2-5.


Predicted and measured p-xylene sorption on Silica Gel, Kaolin,
Webster HP, and Webster Soil at 90% RH.


Sorption by Components Total Sorption
Sorbent P/P, Gas-Liquid Organic Sorbed Predicted Measured
Interface Carbon Water

--------- -------------------mg/g---------------------...----


Silica
Gel





Kaolin







Webster
HP





Webster
Soil


0.081
0.145
0.274
0.282
0.439
0.610

0.069
0.164
0.258
0.286
0.415
0.600
0.896

0.099
0.201
0.206
0.497
0.601
0.777

0.099
0.197
0.206
0.384
0.497
0.610
0.777


1.802
3.921
10.052
10.514
21.522
37.706

0.084
0.267
0.523
0.614
1.120
2.094
18.262

0.324
0.870
0.903
3.677
5.228
14.014

0.026
0.067
0.072
0.190
0.292
0.415
1.113


0.001
0.002
0.004
0.004
0.006
0.004

0.001
0.003
0.005
0.006
0.008
0.012
0.018

0.007
0.014
0.014
0.034
0.043
0.054

0.126
0.251
0.262
0.488
0.632
0.776
0.988


0.001
0.001
0.002
0.003
0.004
0.008

0.000
0.000
0.000
0.000
0.001
0.001
0.001

0.001
0.003
0.003
0.007
0.008
0.009

0.002
0.003
0.003
0.005
0.008
0.008
0.010


1.804
3.924
10.058
10.521
21.532
37.718

0.085
0.270
0.528
0.620
1.129
2.107
18.281

0.332
0.887
0.920
3.71
5.279
14.077

0.154
0.321
0.337
0.683
0.932
1.199
2.111


11.61
20.74
41.95
36.73
80.19
173.97

0.45
1.41
2.28
2.77
3.76
7.71
52.19

0.05
0.08
0.10
0.28
0.91
3.88

0.21
0.26
0.25
0.61
0.63
1.31
4.36








48
the low solubility of p-xylene, respectively. However, the total predicted sorption

of p-xylene on silica gel and kaolin was approximately one-quarter of the measured

amount. This discrepancy may have been due to adsorption of p-xylene vapors

on exposed mineral surfaces. The average amount of water adsorbed on silica

gel and kaolin at 67% RH was 57.9 and 9.3 mg/g, respectively, which is equivalent

to 1.7 and 2.2 monolayers of water, based on BET estimates of S, (Table 2-3).

Due to the low values of C (20 and 18) for these sorbents, it was possible that

patches of water formed on the mineral surface (Jurinak and Volman, 1957; Sing

et al., 1985), thereby allowing for adsorption on exposed mineral surfaces. Even

if monolayer coverage of water was attained, the mineral surface could still have

exerted a surface effect on sorbed water films (Fowkes, 1968), resulting in greater

sorption than would be predicted for bulk water surfaces. These hypotheses are

supported by the work of Dorris and Gray (1981), who reported that the sorption

coefficient for n-heptane on water-coated silica increased significantly at water

contents less than those achieved at 88% RH. Therefore, it is quite possible that

additional adsorption of p-xylene vapors occurred on exposed mineral surfaces

and/or surface-affected water films of silica gel and kaolin.

Predicted p-xylene sorption on Webster HP was considerably greater than

that measured with the flow-equilibration apparatus. Dissolution into sorbed water

films and partitioning into organic carbon contributed very little to the overall

estimate of p-xylene sorption on Webster HP. Thus, the difference between

predicted and measured sorption was primarily the result of the estimate of








49
adsorption at the gas-liquid interface. The surface area would have to be reduced

from 33.8 m2/g to approximately 4 m2/g in order to bring the estimated adsorption

at the gas-liquid interface into agreement with the measured sorption data. In

contrast, the predicted sorption of p-xylene on Webster soil was similar to the

measured values. Partitioning of p-xylene into OC on Webster soil (41 g OC/kg)

accounted for the majority of sorption at relative vapor pressures less that 0.6.

Sorption at the gas-liquid interface of Webster soil was much less than that of

Webster HP because the N2 surface area was only 2.62 m2/g.



Summary

The transport of VOCs in the unsaturated zone is directly influenced by

partitioning of the organic chemical between the liquid, vapor, and solid phase. Of

particular concern is the vapor phase, in which VOC transport may occur at a rate

greater than that in the liquid phases. Thus, it is imperative that the mechanisms

responsible for vapor-phase sorption are correctly identified and incorporated into

VOC models.

For all sorbents studied here, vapor-phase sorption of p-xylene increased

dramatically at relative vapor pressures greater than 0.5, resulting in Type-Ill

adsorption isotherms. Although it is generally accepted that oven-dry sorbents

have sizable sorption capacity for organic vapors, these data suggest that hydrated

soil materials can also sorb significant quantities of organic vapors in regions of

high vapor concentration (e.g., near the contaminant source). In addition, the data








50
presented here indicate that the use of K. values to predict vapor-phase sorption

at high RH appears to be only valid for sorbents with high OC content, at relative

vapor pressures less than 0.5. These findings are particularly important in light of

the fact that most VOC and multi-phase transport models either fail to consider

vapor-phase sorption (Pinder and Abriola, 1986; Sleep and Sykes, 1989) or

describe vapor-phase sorption by dissolution into soil water and partitioning into

OC (Baehr, 1987; Jury et al., 1990).

In order to describe vapor-phase sorption for different sorbate-sorbent

systems over a range of relative vapor pressures, one must consider all

mechanisms which contribute to VOC sorption. The data presented here indicate

that a multi-mechanistic approach should include adsorption on mineral surfaces,

partitioning into OC, dissolution into sorbed water films, and adsorption at the gas-

liquid interface. Surface tension data suggest that adsorption at the gas-liquid

interface could contribute significantly to the sorption of sparingly-soluble VOCs in

the unsaturated zone. Of particular interest is the sizable surface excess

estimated for aromatic hydrocarbons of environmental concern such as toluene,

benzene, and o-xylene (Blank and Ottewill, 1964; Cutting and Jones, 1955;

Hauxwell and Ottewill, 1968; Vidal-Madjar et al., 1976). The task at hand now is

to further define the limits of applicability of each mechanism based on sorbate,

sorbent, and environmental considerations, and to incorporate this knowledge into

current modeling efforts. An initial attempt at this approach has been attempted

by Shoemaker et al. (1990), who described vapor-phase sorption of TCE using an








51
"effective sorption" term that included sorption coefficients for the solid-liquid and

the solid-gas interface.












CHAPTER 3
GAS CHROMATOGRAPHIC STUDIES OF P-XYLENE SORPTION ON
ANHYDROUS AND HYDRATED QUARTZ SAND


Introduction

Gas-liquid chromatography (GLC) is recognized as an efficient and versatile

method for investigating the sorption of volatile organic compounds (VOCs) on

hydrated sorbents. The interpretation of GLC data requires a consideration of

several retention mechanisms including (1) adsorption at the gas-liquid interface,

(2) adsorption at the solid-liquid interface, and (3) partitioning into the liquid phase.

Conder et al. (1969) proposed the following equation to describe the net retention

volume (VN) of a solute at infinite dilution:

VN = KLVL + KA, + KsAs (3-1)

where KL (mUmL), K, (mUm2), and Ks (mLm2) are the distribution coefficients for

the liquid phase, gas-liquid interface, and solid-liquid interface, respectively, VL

(mUg) is the volume of the liquid phase, and A, (m2/g) and A. (m2/g) are the areas

of the gas-liquid interface and solid-liquid interface, respectively. Previous studies

have demonstrated that aliphatic hydrocarbon retention on water-coated supports

is due solely to adsorption at the gas-liquid interface; whereas, aromatic

hydrocarbons and other weakly polar solutes are simultaneously adsorbed at the

gas-liquid interface and partitioned into the liquid phase (Karger et al., 1971a,








53
1971 b; Martin, 1961). At water contents above 3 to 4% by weight, the effect of the

solid support on solute retention appears to vanish, and thus equation (3-1) can

be applied without the KsAs term (Dorris and Gray, 1981; Okamura and Sawyer,

1973). Comparisons of partition coefficients and heats of solution determined by

GLC and static methods indicate that adsorbed water films between 1.5 and 200

nm in thickness have properties similar to those of bulk water (Karger et al.,

1971a, 1971b; Chatterjee et al., 1973). Thus, gas chromatography (GC) appears

to be an ideal method to study vapor-sorption at the gas-liquid interface, and can

be utilized to measure sorption on anhydrous soil materials. A detailed review of

the theoretical and experimental application of GC methods to the study of VOC

sorption has been presented by Rhue and Rao (1990).

The adsorption of insoluble and sparingly-soluble hydrocarbons on water

surfaces has also been estimated from measurements of the change in surface

tension of bulk water with the partial pressure of the vapor, using the Gibbs

adsorption equation. The surface excess of organic vapors calculated in this

manner typically yields Type-Ill adsorption isotherms. In addition, heats of

adsorption for n-hexane and toluene on bulk water were found to be greater than

corresponding heats of vaporization (Hauxwell and Ottewill, 1968). Adamson

(1967) attributed this behavior to the rearrangement of surface water molecules to

accommodate the hydrocarbon, which resulted in larger heats of adsorption.

Although measurements of vapor adsorption on water-coated supports by GLC

yield Type-Ill adsorption isotherms (Dorris and Gray, 1981; Karger et al., 1971b;








54
King et al., 1972), heats of adsorption estimated from GLC data are generally

smaller than heats of vaporization. These findings suggest that the gas-liquid

interface of water acts as a low energy surface toward nonpolar vapors (Chatterjee

et al., 1972; Dorris and Gray, 1981; Hartkopf and Karger, 1973; Karger et al.,

1971a, 1971b). The discrepancy between heats of adsorption obtained from GLC

and surface tension data has been attributed to uncertainty in surface tension

measurements at low surface coverages (Dorris and Gray, 1981; Karger et al.,

1971a, 1971b); whereas, flame-ionization detectors (FID) used in GLC studies

provide accurate measurements of hydrocarbon adsorption.

In addition to high sensitivity, gas chromatography allows for the rapid

collection of sorption data over a range of temperature and moisture content

regimes. Despite the apparent advantages of GC techniques over conventional

batch adsorption methods, relatively few studies have been conducted using soil

material as the solid support phase (Bohn et al., 1980; Okamura and Sawyer,

1973; Rao et al., 1988). Thus, the purpose of this work was to test the utility of

gas chromatography for the study of p-xylene sorption on anhydrous and hydrated

quartz sand. The effect of CaCI2 treatments on p-xylene adsorption was also

investigated by GC, and was compared to adsorption data obtained by flow-

equilibration and surface tension methods.









Materials and Methods

Column Preparation

The solid support material was collected from the Bh horizon of an Oldsmar

soil (Alfic Arenic Haplaquod) located in Collier County, Florida. Mechanical

analysis indicated that the sand-size fraction (diameter > 50 pm) accounted 90%
of the soil sample. The Oldsmar soil sample had a cation exchange capacity of

5.2 cmoljkg and was predominantly Ca-saturated. The N2 surface area was 10.05

m2/g, while the organic carbon content was 10.9 g OC/kg (Rhue et al., 1988). The

soil sample used in the GC experiments was sieved to pass a 250 pm screen and
will be referred to as Oldsmar sand. A portion of this sample was washed 3 times

with 0.1 M CaCIl to test the effect of salt on vapor-phase sorption.

The Oldsmar sand was packed in 6.35 mm o.d. glass columns,

approximately 16 and 90 cm in length. The 90-cm columns were used to measure

sorption on hydrated sorbents; the additional length was necessary to obtain

adequate solute separation. These columns were crafted to form a U-shape which

matched the inlet and outlet ports of the gas chromatograph. As sand was added

to the column, the column was gently vibrated to achieve uniform packing. Glass

wool was placed at the ends of the columns to maintain support integrity.

Gas Chromatoaraphy Experiments

A Tracor 222 gas chromatograph equipped with a FID was used for the

sorption studies. Certified grade p-xylene (99.8% purity) and high purity grade

methane (99.97% purity) were obtained from Fisher Scientific Products. The FID

was calibrated for p-xylene at three N2 flow rates, with air and hydrogen flow rates








56
maintained at 0.65 mL/sec and 5.0 mL/sec, respectively. During calibration the p-

xylene vapor was introduced onto a straight glass column by passing the N2 flow

stream over a thermostated sample of liquid p-xylene. When a constant mV

reading was obtained, the flow stream was bubbled through a 40-mL centrifuge

tube containing 20 mL of methanol. The concentration of p-xylene in the methanol

was measured by UV-VIS spectroscopy or HPLC techniques.

The adsorption of p-xylene on untreated and salt-treated Oldsmar sand was

measured under anhydrous and hydrated conditions at room temperature (= 24C).

For anhydrous experiments, the column was allowed to equilibrate with "dry" N2

for several hours prior to the initiation of p-xylene injections. This was analogous

to the flow-equilibration method in which "dry" N2 was used as the carrier gas. For

hydrated systems, Okamura and Sawyer (1971, 1973) recommend that the desired

water content be reached by desorption of a saturated column in order to attain

uniform water coverages. However, this procedure resulted in the introduction of

excessive quantities of water into the FID and associated column fittings. Dorris

and Gray (1981) obtained reproducible solute retention data at water contents of

0.6, 1.4, and 3.8% by equilibrating the column with carrier gas at 26, 62, and 88%

RH. For the experiments described here, relative humidities of 90 and 98% were

achieved by bubbling the N2 flow stream through a gas-washing bottle containing

deionized water in a manner similar to that described in Chapter 2. The relative

humidity of the N2 flow stream was measured at the column inlet and outlet by

trapping the water vapor in two magnesium perchlorate traps arranged in series.







57
In addition, the back pressure at the column inlet was measured using a pressure

transducer, for which the mV output had been calibrated against pressure heads.

Adsorption isotherms for p-xylene were obtained by the eluted-pulse method

of Dorris and Gray (1981). The retention time of methane (to) and p-xylene (t,)

vapors, which were simultaneously injected on the column, was recorded with a

Hewlett Packard 3390A integrator. The retention time of air injections, which gave

a negative response, were identical to those of methane. These data indicate that

methane was not retained by the stationary phase. Thus, the net retention volume

(VN) of p-xylene was calculated from the difference between t, and to, using the N2

flow rate. Injections volumes ranging from 1 LL to 2 mL of p-xylene vapor were
used to obtain a range of net retention volumes.

If the adsorption of p-xylene is assumed to occur on mineral surfaces for

anhydrous sorbents, and solely by adsorption at the gas-liquid interface of water-

coated supports, the net retention volume (V,) required to elute a solute is given

by,

VN = RT(dF/dP)A = RT(dS/dP)w (3-2)

where R is the gas constant, T is the temperature, r is the surface concentration
(mol/cm2), P is the partial pressure of the vapor, A is the surface area, S is the

amount adsorbed (mg/g), and w is the weight of the column packing (Dorris and

Gray, 1981). Adsorption isotherms can then be obtained by integrating equation

(3-2) as follows:

S= 1/RTw VN(P) dP. (3-3)

The peak-maxima method of Dorris and Gray (1981) was employed to obtain a








58
chromatographic envelope which could be integrated by equation (3-3). The

calibration curve described previously was used to convert mV readings, obtained

from peak height measurements, to partial pressures of p-xylene. However, the

use of equation (3-2) implies that the sorption effect and pressure gradient along

the column were negligible.



Results and Discussion

Adsorption on Anhydrous Oldsmar Sand

Chromatographic peak maxima obtained from injections of p-xylene vapors

on untreated and salt-treated Oldsmar sand are presented in Figure 3-1. Net

retention volumes of p-xylene on untreated Oldsmar sand were larger than those

of salt-treated soil, indicating greater solute retention. Adsorption in the Henry's

Law region, characterized by highly symmetric peaks and net retention volumes

independent of sample size, was not attained for untreated Oldsmar sand, even

at p-xylene partial pressures of 2.5 X 103 mm Hg. In contrast, the salt-treated

Oldsmar sand exhibited Henry's region adsorption at a net retention volume of

approximately 5.2 mL. As greater quantities of p-xylene were injected, the net

retention volume decreased until the peaks became asymmetric, at which point the

net retention volume began to increase. Although not shown here, the position of

the leading edge of the asymmetric peaks was similar, indicating that adsorption

equilibrium was attained (Dorris and Gray, 1981). The connection of peak maxima







7

6

5

14-
E
E3
02-

1

0-
0

4


3

I
E2-
E

1-


0-
0


Figure 3-1.


1 2 3 4 5
VN(mL)


Chromatographic peak maxima at room temperature (-240C) for p-
xylene vapor on (a) untreated Oldsmar soil and (b) salt-treated
Oldsmar soil.








60
formed a chromatographic envelope which was integrated by equation (3-3). The

resulting p-xylene adsorption isotherms are presented in Figure 3-2.

The adsorption of p-xylene on untreated Oldsmar sand yielded a Type-II

isotherm, indicative of multilayer formation. A similar isotherm was obtained for

the salt-treated sand, although the adsorption capacity was reduced. These data

have yet to be confirmed by batch techniques; however, p-xylene adsorption on

250-425 gim Oldsmar sand measured by the GC and flow-equilibration methods
were in close agreement (R. D. Rhue, 1990, personal communication).

The effect of the salt treatments on p-xylene adsorption by Webster soil was

also studied using the flow equilibration apparatus described in Chapter 2 (Figure

3-3). Webster soil and Webster soil treated with hydrogen peroxide (Webster HP)

was extracted with methanol containing CaCI2 to measure water and p-xylene

sorption. Following methanol extraction, the Webster soil and Webster HP

contained 12.01 (0.65) and 15.18 (0.44) mg CaCI2/g, respectively. Subsequent

adsorption experiments indicated that in the presence of salt p-xylene vapor

adsorption was reduced by 6.95 and 5.98 mg/g on Webster HP and Webster soil,

respectively. These data were in agreement with the observed decrease in the

adsorptive capacity of Oldsmar sand following a salt treatment, measured by the

eluted-pulse method. From a mechanistic perspective, the salt may have coated

the adsorbent surface or reduced the surface charge, such that the magnitude of

adsorbate-adsorbent interactions was decreased.









61





CO




-o
0








\.

0
c as 0











I-
c
\ \

C



0



C) 0
.-











0 \ C) 0


(8/CD C
C.s

0.
L -I ----^ i-^ -- ^| o
co m sj i -^ i o 0







0i








20


15

0)
E. 10
C)
5


0


0.2


0.4
p-Xylene


0.6


0.8


(P/Po)


Figure 3-3.


Adsorption of p-xylene vapors on Webster soil and Webster HP at
(a) 0% RH and (b) 90% RH.


0.1 0.2 0.3 0.4


5

4


0)
02
C/)

1

0


0.5


o Webster Soil
*Webster Soil/CaCI 2
0 Webster HP
* Webster HP/CaCI2




(b)
--- 0/
~ "









Adsorption on Hydrated Oldsmar Sand

The net retention volume of p-xylene injected onto 90-cm columns packed

with Oldsmar sand decreased from approximately 297 mL to less than 8 mL when

the RH was increased to 90%. This reduction was indicative of the effect of water

on the retention of p-xylene vapors. Net retention volumes of p-xylene on salt-

treated Oldsmar sand at 90% RH, and on untreated Oldsmar sand at 90 and 98%

RH are presented in Figure 3-4. Henry's region adsorption occurred at smaller

retention volumes, and for greater p-xylene partial pressures, than under

anhydrous conditions. In addition, the value of VN increased continuously, except

in the Henry's Law region, which is characteristic of adsorption yielding Type-III

isotherms (Dorris and Gray, 1981). At N2 flow rates of 0.42 and 0.33 mL/s, the p-

xylene net retention volumes were similar, demonstrating the internal consistency

of the eluted-pulse method. Adsorption isotherms obtained from chromatographic

envelopes of these data are presented in Figure 3-5.

The presence of water not only precipitated a shift from Type-Il to Type-Ill

adsorption isotherms, but also resulted in a substantial reduction in p-xylene

adsorption. These findings are consistent with data reported in Chapter 2 for p-

xylene sorption on silica gel and kaolin at 67% RH, and on Webster soil and

Webster HP at 90% RH. At 98% RH an additional reduction in p-xylene

adsorption on Oldsmar sand was observed. These data suggest that sorbent

surface was not completely covered with water or that the surface continued to

exert an effect on adsorption, even though the RH of the carrier gas at the column





























0
Ol


0
U8
0

0
* 0


If


cP O
0



a 90% RH
0.33 ml/s
o 90% RH
0.42 m/s
* 98% RH
0.31 ml/s


VN (ml)


Figure 3-4.


Chromatographic peak maxima at room temperature (- 240C) for
p-xylene vapor on (a) salt-treated Oldsmar soil at 90% RH and (b)
untreated Oldsmar soil at 90 and 98% RH.


5


4


S3
E
E2

10
1


(a)




90% RH
0.32 ml/s


/_____


41-


I
E
E
(0_


(b)








65



(0




"-n \
'o



6 0
I-
0 6



O 1 c

oo ) V


2 d
00 C0" 0) 0
E E E E ca


\ a)
000 0V 1 .


o)) ) C) V







0 | o o or
.O \
0 0 C 0. 0. x

M It I Lm \ d






c 0 0 co 0 ca \
S 0) 0)) C0) a









C)
Co





)


0)
T-








66
inlet was 90%. Subsequent measurements revealed that inlet relative humidities

of 90 and 98% corresponded to relative humidities at the column outlet of 60 and

70%, respectively. Since the back pressure measured at the column inlet ranged

from only 1.04 to 1.07 atm, it was unlikely that the RH drop was due solely to

pressure gradients. Apparently, the 8-hour equilibration period was not sufficient

to completely hydrate the support. Future experiments will be conducted after

equilibrating the columns for several days at 90% RH.

The presence of salt also reduced p-xylene adsorption on Oldsmar sand,

which was consistent with batch data obtained for the sorption of p-xylene vapors

on Webster soil and Webster HP at 90% RH (Figure 3-3b). Based on water

adsorption data, the concentration of CaCI2 in adsorbed water films of the Webster

soils was approximately 1.0 M. To further investigate the adsorption at the gas-

liquid interface, the surface tension of 1.0 M CaCI2 exposed to p-xylene vapors

was measured by the drop-weight method, as described in Chapter 2. These data

were expressed as surface pressure (n), which is equivalent to the difference
between the surface tension of the pure solution (y) and the film-covered surface

(y) (Figure 3-6). The surface excess (F) can be calculated in the same manner
as described in Chapter 2 by simply replacing y in equation (2-7) with n (Blank and
Ottewill, 1964). However, the data presented in Figure 3-6 indicated that the

surface excess calculated for 1.0 M CaCI2 and deionized water would be identical.

This finding was consistent with the data of Blank and Ottewill (1964), who

reported that the surface excess of benzene, toluene, and o-xylene on 0.1 M NaCI

was similar to that obtained using distilled water. Thus, the surface tension





















4 .


*

4


4.


. I


| .- I


LO) '- CO C-

( s/6) J11


I I


PD


co1



o
E
2



CL
co
0



C
a,

D I
x



0
CV 2
E0



O
Cu
L.
0



N
0


0
'ID
CL


0-
o M

3/ R








68
measurements suggested that the presence of salt would have no effect on the

adsorption of p-xylene at the gas-liquid interface.



Summary

The measurement of vapor-phase adsorption on soil materials by batch

techniques is typically an arduous task, requiring several time-consuming

experiments in order to obtain an adsorption isotherm. In contrast, gas

chromatography can be efficiently employed under a range of temperature and

moisture regimes. Vapor-phase adsorption of p-xylene on Oldsmar sand was

measured by the eluted-pulse method of Dorris and Gray (1981). Type-II

isotherms obtained for p-xylene adsorption on anhydrous Oldsmar sand were in

agreement with preliminary batch data. At high RH, adsorption isotherms shifted

from Type-II to Type-Ill, and the magnitude of p-xylene adsorption was reduced.

A similar effect has been observed for the sorption of several organic vapors on

hydrated soils (Call,1957; Chiou and Shoup, 1985; Rhue et al., 1989). However,

some difficulty was encountered at high RH, and it is recommended that the

columns be equilibrated at 90% RH for several days. Additional experiments will

be conducted at high RH, and batch experiments are planned to further verify GC

data.

Salt treatments resulted in decreased p-xylene adsorption on anhydrous and

hydrated Oldsmar sand. These data were consistent with batch studies of p-

xylene adsorption on Webster soil and Webster HP exposed to CaCIl, and suggest








69
that vapor-phase sorption will be significantly reduced in salt-effected soils. Under

anhydrous conditions, the reduced adsorption capacity of salt-treated Oldsmar

sand may have resulted from the formation of salt coatings on the sorbent surface.

In Chapter 2 it was postulated that sorption on hydrated sorbents occurred by (1)

partitioning into OC, (2) dissolution into adsorbed water films, (3) adsorption at the

gas-liquid interface. Given the low solubility of p-xylene in water and the relatively

low OC content of Oldsmar sand it is unlikely that any alteration of these

components by salt would significantly effect adsorption. Thus, it was postulated

that the salt-treatment reduced the adsorptive capacity of the gas-liquid interface.

However, the surface concentrations of p-xylene on water and 1 M CaCIl

calculated from measurements of the change in surface tension of deionized water

and 1 M CaCl2 exposed to p-xylene vapors were identical. This finding was

somewhat puzzling, and prompted a reformulation of the Gibbs adsorption

equation to account for three components (i.e., p-xylene, CaCI2, and water). If the

surface excess of water is assumed to be constant, the Gibbs equation can be

expressed as:

1 dy dlna2
r, ---- ---- (3-4)
RT dlna, dlna,

where r, and 1, are the surface concentrations of p-xylene and CaCI2 (mol/cm2),

respectively, a, and a2 are the activities of p-xylene and CaCIl, respectively, R is

the gas constant, and T is the temperature. This equation indicates that if CaCI2

moved to the gas-liquid interface, it could compensate for any reduction in the








70
surface tension arising from the accumulation of p-xylene at the gas-liquid

interface. Unfortunately, the surface concentration of CaCI2 (F) was unknown, and

thus it was impossible to calculate the surface concentration of p-xylene. This line

of reasoning may explain why no difference was observed in the change in surface

tension of water and CaClI exposed to p-xylene vapors.

Additional studies indicated that the utility of GC techniques under

anhydrous conditions may be limited to supports of low sorptive capacity. p-

Xylene vapors injected on coated sands and kaolin were strongly adsorbed and

eluted solute peaks were not discernable. Planned experiments will focus on the

dilution of such sorbents, by either adding low-sorptive-capacity materials to the

column or by coating sands with clay films. It should be recognized that batch

studies are difficult to conduct for materials of low sorptive capacity and, thus, GC

techniques are complimentary in this regard.













CHAPTER 4
COMPETITIVE ADSORPTION OF P-XYLENE AND WATER VAPORS
ON CA-, NA-, AND LI-SATURATED KAOLIN


Introduction

Relative humidity (RH) or soil moisture content is one of the most important

factors influencing the sorption of volatile organic chemicals (VOCs) in the

unsaturated zone. Previous studies have demonstrated that water effectively

competes with nonpolar organic vapors for mineral surfaces, resulting in the

suppression of VOC sorption on soils and clay minerals. (Call, 1957; Chiou and

Shoup, 1985; Rhue et al., 1989). This phenomenon has been attributed to the

relatively strong interactions between water and mineral surfaces (Chiou and

Shoup, 1985; Valsaraj and Thibodeaux, 1988), which may result from cation-dipole

interactions, hydrogen bonding, and weak charge transfer interactions (Burchill et

al., 1981). Of particular interest is cation hydration, the energy of which has been

directly related to the degree of water adsorption on kaolinite (Keenan et al., 1951;

Jurinak, 1963). However, the effect of exchangeable cations on the competitive

adsorption of water and organic vapors is largely unknown. In addition, models

capable of predicting vapor-phase adsorption from binary systems have rarely

been tested due, primarily, to the lack of suitable data.







72
Cation saturation has been shown to indirectly effect the competitive

adsorption of water and ethylene dibromide (EDB) vapors by altering the surface

area of montmorillonite. Following exposure to P20s, Jurinak (1957) observed that

water retention by Mg-, Ca-, and Na-saturated montmorillonite was directly related

to the hydration energy of the exchangeable cation. The retained water expanded

the interlayer space, and thus Mg-montmorillonite exhibited the greatest surface

area and EDB adsorption, followed by Ca- and Na-montmorillonite. A similar

phenomenon was noted by Call (1957), who reported that EDB sorption on Ca-

saturated montmorillonite was greater at 5, 10 and 20% relative humidity (RH) than

at 0% RH (Figure 4-1). This effect was attributed to expansion of the clay lattice,

which increased from 3A at 0% RH to 9A at 5-10% RH. Apparently, EDB

molecules were only able to enter the interlayer space after the clay lattice had

expanded to 9A. However, as the number of water molecules continued to

increase, competition between EDB and water became greater, resulting in the

suppression of EDB adsorption.

Although these data clearly demonstrate that cation saturation influenced

EDB adsorption on montmorillonite, the effect of specific interactions between

water and exchangeable cations was obscured by changes in surface area. In

addition, Jurinak (1957) reported that montmorillonite forms porous aggregates or

floccules during dehydration. This process, which has been observed by electron

micrography (Grim, 1953), restricted the adsorption of EDB on Mg-, Ca-, and Na-

montmorillonite (Jurinak, 1957). In contrast, EDB adsorption on kaolinite occurred









73







Ln
C0

0
0


NO C)I
O



oc






o .










Oo o
(L6 0
0 o
o 0



o


0 a






0
E





If)0
0
CL

CL










C\l







ilx







74
on free surfaces or in pores whose size was far greater than that of the EDB

molecule. Thus, kaolin appears to provide a surface absent of surface area and

porosity effects which complicated the interpretation of EDB adsorption on

montmorillonite.

Recently, Rhue et al. (1989) used a methanol extraction procedure to

simultaneously measure water and p-xylene adsorption on predominantly Na-

saturated kaolin. At low RHs and relatively high p-xylene vapors pressures, Rhue

et al. (1989) observed enhanced water adsorption based on comparisons between

measured and predicted data. The predicted values were calculated using a

modified Brunauer, Emmett, and Teller (BET) equation which accounted for two

adsorbate species. The purpose of this study was to determine if the preferential

adsorption of water could be attributed cation hydration effects. Initially, single-

adsorbate isotherms were obtained for p-xylene and water vapor adsorption on

Ca-, Na-, and Li-kaolin. The linear form of BET equation was utilized to obtain the

monolayer adsorption capacity and value of C. These data were then used to

predict competitive adsorption of p-xylene and water vapors based on the two-

component BET equations of Hill (1946a, b) and Rhue et al. (1989).



Materials and Methods

Kaolinite Samples

Colloidal kaolin, obtained from Fisher Scientific Products (K-6, Lot# 731063),

used in this study was identical to that described in Chapter 2. The kaolin had a







75
cation exchange capacity (CEC) of 4.2 cmoljkg at pH 5.5, and was predominantly

Na-saturated with trace amounts of Ca, Mg, and K (Rhue et al., 1989). Prior to

adsorption experiments, the kaolin was washed with 1 M NaAOc buffered to pH

4.11 with acetic acid to remove carbonates. Approximately 2.5 g of kaolin was

placed in individual polyethylene centrifuge tubes, to which 20 mL of 1 M NaAOc

was added. The samples were mixed until the kaolin was completely dispersed,

and separated by centrifugation at 2500 rpm for 5 minutes. When the pH of the

supernatant was less than 5.0, generally after one wash, the kaolin was washed

five times with 1 M NaCI to removed entrained NaAOc. The supernatant was

analyzed for Ca and Mg using a Perkin Elmer model 603 atomic absorption

spectrophotometer. If either Ca or Mg was detected in the supernatant, the

NaAOc treatment was repeated, otherwise the kaolin was saturated with the

desired cation.

Ca-, Na-, and Li-saturated Fisher kaolin was prepared by washing the kaolin

with the appropriate 1 M chloride salt until no other cations could be detected in

the supernatant by atomic absorption spectroscopy. Following cation saturation,

the kaolin was repeatedly washed with 95% ethanol until a negative chloride test

was achieved using AgNO3. The kaolin was air-dried at room temperature and

ground with an agate motar and postal. Prior to adsorption experiments, the kaolin

was oven-dried at approximately 1200C for at least two weeks. The N2 surface

area of Ca, Na, and Li-kaolin was 15.8, 15.6, and 15.5 m2/g, respectively, prior to








76
heating, and 15.5, 15.4, and 15.2 m2/g, respectively, after heating (Advanced

Materials Research Center, University of Florida).

Vapor-Phase Adsorption Experiments

Single and mixed-vapor adsorption experiments were conducted at 240C

using the flow-equilibration apparatus described in Chapter 2. The concentration

of water in the flow stream was determined by Karl Fisher (KF) titration.

Approximately 20 mL of CaCI-saturated methanol was added to two 40 mL glass

centrifuge tubes, and pretitrated to the visual KF endpoint. The tubes were placed

on the flow stream in series, via hypodermic needles, for a measured time period.

The solutions were then removed from the flow stream and immediately retitrated.

A small amount of titer was consistently needed to retitrate the second trap in the

series, which was attributed to loss of volatile compounds from the KF reagent

(Rhue et al., 1988). Therefore, the volume of titer used for the second trap was

subtracted from that of the first trap, and the corrected value was used to calculate

the amount of water trapped in the methanol. p-Xylene vapors were trapped by

bubbling the flow stream through two glass centrifuge tubes containing methanol,

as described in Chapter 2.

Single-adsorbate isotherms were determined for p-xylene and water vapors

at relative vapor pressures ranging from 0.1 to 0.5. The competitive adsorption of

p-xylene and water vapors was measured at 10 and 20% RH. Adsorbed

concentrations of p-xylene and water vapors from single- and binary- sorbate

systems were measured following the procedures described previously.









Preliminary Adsorption Experiment

A preliminary study was conducted to test the effect of cation saturation on

the competitive adsorption of water and p-xylene vapors on kaolin. A portion of

the original Fisher kaolin was saturated with Li by washing the kaolin with 1 M LiCI.

Excess salt was removed by repeatedly washing the kaolin with 95% ethanol until

a negative chloride test was achieved with AgNO3. The adsorption of water and

p-xylene vapors on Na- and Li-saturated kaolin was measured for single- and

binary-sorbate systems as described previously.



Results and Discussion

Preliminary Adsorption Experiment

Water adsorption on Na-kaolin was in agreement with the data obtained by

Rhue et al. (1989), and was significantly greater than that on L-kaolin (Figure 4-

2a). In contrast, p-xylene adsorption on Na- and L-kaolin was similar (Figure 4-

2b), suggesting that exchangeable cations had only a minor effect on p-xylene

adsorption. The adsorption data were fit by a least squares procedure to the linear

form of the BET equation,

P/Po 1 (C-1)P/Po
+ (4-1)
S(1-P/Po) SmC SC

where Sm is the monolayer adsorption capacity (mg/g), and C is a parameter

related to the heat of adsorption. Estimated values of S, and C are presented in

Table 4-1. BET parameters for Na-kaolin were calculated using a combined data











0 (a)


S.... -r Na-Kaolin, Rhue
et al. (1989)
o Na-Kaolin
Li-Kaolin

0.1 0.2 0.3 0.4 0.5 0.6 0.7


Water


(P/Po)


0.1 0.2 0.3


0.4


p-Xylene (P/Po)


Figure 4-2.


Vapor-phase adsorption of (a) water and (b) p-xylene on Na- and
Li-saturated Fisher kaolin at 240C.


E
C)


7

6

5
04
0)
E
3

2

1

0








79
Table 4-1. BET parameters for Na- and Li-kaolin calculated from single-
sorbate isotherm data (Rhue et al., 1989).


P/PO
Sorbent Sorbate Sm C r2 Range Used


-mg/g-

Na-kaolin p-xylene 4.5 14 0.969 0.065-0.424

water 4.2 20 0.988 0.056-0.481


Li-kaolin p-xylene 4.4 9 0.997 0.074-0.365

water 2.6 52 0.997 0.112-0.289







80
set, since the values obtained for water and p-xylene adsorption were similar to

those reported by Rhue et al. (1989).

The monolayer adsorption capacity of water on U-kaolin was considerably

less than that on Na-kaolin. It has been proposed that exchangeable Na on kaolin

hydrates readily; whereas, exchangeable Li does not hydrate, despite the fact that

Li has a slightly greater hydration energy than Na (Keenan et al., 1951).

Apparently, Li ions form inner-sphere complexes or are strongly adsorbed on

kaolin, such that hydration does not occur even at high RHs (Keenan et al., 1951;

Martin, 1959). If the amount of water adsorbed by Li-kaolin is considered to be

independent of cation hydration effects, then the difference between the monolayer

adsorption capacities of Na- and U-kaolin provides an estimate of the amount of

water hydrating Na. Based on this assumption, approximately 1.8 molecules of

water were associated with each exchangeable Na ion, which was consistent with

the range of values (1.5 to 1.8 molecules of water) reported by Keenan et al.

(1951) for Na-, K-, and Rb-saturated kaolinite.

The binary-vapor experiment was conducted at p-xylene and water relative

vapor pressures of 0.395 and 0.095, respectively. The amount of water and p-

xylene adsorbed on Na-kaolin in this experiment was 2.46 (0.00) and 6.56 (0.05)

mg/g, respectively. In contrast, Li-kaolin adsorbed 2.06 (0.01) and 6.04 (0.05)
mg/g of water and p-xylene, respectively. Rhue et al. (1989) modified the BET

equation to account for the competitive adsorption of two sorbate species. The








81
total adsorbed mass of sorbate "a" on the surface (M,, mg/g) can be calculated

using the following equation:

M,a = a C, S, [Xa/(1-Xa) (4-2)

where a is the mass of compound "a" occupying a unit area of surface (mg/m2),

C, is the BET parameter related to heat of adsorption calculated from single-vapor

adsorption isotherm for sorbate "a", X, is the relative vapor pressures of sorbate

"a", and S, is the area of exposed surface area per unit mass of adsorbent (m2/g).

The a values for water and p-xylene were calculated to be 0.31 and 0.50 mg/m2,

respectively, based on the following equation:

a = 1.091(MW/a, A) (4-3)

where MW is the molecular weight, a, is the cross-sectional area of sorbate

determined in Chapter 2, and A is Avogadro's number (Rhue et al., 1989). The

mass of sorbate "b" (M) can be calculated in a similar manner. Unfortunately, the

mass of either adsorbate on the surface cannot be calculated because S, is

unknown in the binary vapor system (Rhue et al., 1989). However, the fraction of

adsorbate "a" (F,) on the surface can be calculated using the following equation:

Ma
Fa = (4-4)
Ma + Mb

The calculated versus the measured fraction of water adsorbed on Na-kaolin (Na)

and Li-kaolin (Li) are presented in Figure 4-3. These data indicate that Na-kaolin

adsorbed considerably more water than was predicted by the two-component BET

equation. The fraction of water adsorbed on Na-kaolin was previously reported by









C
0




mo
>oI

\ 0 o
cI





O
A-
aa









O o
\ .z


\ c
\0 0z





\-- 00

0Z 3



*\ -a


\ o*Co
"-- c




io z 0


as*2 00
*LLL


\ cir



\ 0










uo p l
-a,



UOipoeJd pollnoluO
L.~








83
Rhue et al. (1989), and is denoted by (n) in Figure 4-3. In contrast, the measured

and calculated fractions of water adsorbed on Li-kaolin were almost identical.

These data suggest that the preferential adsorption of water relative to p-xylene

at low water fractions (below 0.4) was due to Na hydration.

One possible explanation for the difference between the measured and

calculated fraction of water adsorbed on Na-kaolin was the similarity in the values

of C estimated from water and p-xylene single-sorbate isotherms. Since the

values of C were essentially the same, the two-component BET equation of Rhue

et al. (1989) did not account for preferential adsorption of water over p-xylene on

Na-kaolin. In contrast, the values of C for water and p-xylene adsorption on Li-

kaolin were 52 and 9, respectively. The large value of C for water adsorption on

Li-kaolin was not anticipated, because Li-kaolin is generally considered to

represent a mineral surface free of cation hydration effects. If a C value of 20 was

used to estimate the mass of water adsorbed on Li-kaolin, then the calculated

fraction of water adsorbed on Li-kaolin would be 0.12 rather than 0.28. Thus, Li-

kaolin would also have exhibited a small degree of preferential water adsorption

if the measured value of C had been in the range expected for water adsorption

on Fisher kaolin.

Vapor-Phase Adsorption on Ca-, Na-, and Li-kaolin

Isotherms for the adsorption of water vapor on Ca-, Na-, and U-kaolin are

presented in Figure 4-4a. Ca-kaolin adsorbed the greatest amount of water at all

RHs studied, followed by Na- and Li-kaolin. The monolayer adsorption capacities








6



.4
0)
E
CO 2



0
0


0.1 0.2 0.3 0.4 0.5
p-Xylene (P/Po)


Figure 4-4.


Vapor-phase adsorption of (a) water and (b)
and Li-saturated Fisher kaolin at 240C.


p-xylene on Ca-, Na-,


0.1 0.2
Water (P/Po)


0)

E
0)


0.3


0.6








85
of water on Ca-, Na-, and Li-kaolin derived from the single-sorbate adsorption data

were 5.0, 3.9, and 2.8 mg/g, respectively (Table 4-2). These data indicate that

cation saturation had a considerable effect on water adsorption, and were similar

to values of Sm obtained for Na- and Li-kaolin in the preliminary study. The value

of C for water adsorption on Li-kaolin (12) was less than that measured previously,

but was now within the range expected for water adsorption on Fisher kaolin.

Thus, the value of C was subject to uncertainty, even when the value of S, derived

from the same adsorption data appeared to give a reasonable estimate of the

monolayer adsorption capacity. The reasons for uncertainty in the value of C are

unclear, but the evaluation of numerous adsorption isotherms indicates that small

differences in adsorption data can result in substantial changes in the value of C,

particularly if the isotherm consists of only a few data points.

In general, Ca-kaolin adsorbed the largest amount of p-xylene, followed by

Na- and Li-kaolin (Figure 4-4b). However, p-xylene adsorption on Na-kaolin was

slightly greater than that on Ca-kaolin at relative vapors pressures below 0.05 and

above 0.5. The p-xylene monolayer adsorption capacities of Ca-, Na-, and U-

kaolin were 6.5, 5.5, and 5.0 mg/g, respectively (Table 4-2). These data indicate

that cation saturation influenced the adsorption of p-xylene vapors, but the

similarity between the values of S, indicates that the effect of cation saturation was

minimal. In order to compare the monolayer adsorption capacities for water and

p-xylene, the value of Sm was expressed as the amount of surface occupied by

sorbate molecules per gram of kaolin (m2/g), as described in Chapter 2 (Table 4-








86
Table 4-2. BET parameters for Ca-, Na- and Li-kaolin calculated from single-
sorbate isotherm data.


Sorbent


Ca-kaolin




Na-kaolin


Sorbate


p-xylene

water


p-xylene

water


Li-kaolin p-xylene

water


C P


Sm

-mg/g-

6.5

5.0


5.5

3.9


5.0

2.8


0.999

0.993


0.998

0.978


0.997

0.994


SA

-m2/g-

14.0

17.6


11.9

13.7


10.8

9.8


P/Po
Range Used


0.045-0.301

0.126-0.245


0.044-0.301

0.126-0.241


0.047-0.303

0.115-0.241







87
2). The BET surface areas determined in this manner indicate that when the

molecular weight and surface packing are taken into consideration, the p-xylene

monolayer adsorption capacities were less than those measured for water. In

addition, the effect of the saturating cation and CEC the BET surface areas were

also smaller than that observed for water.

In order to utilize the method of Keenan et al. (1951) for estimating the

amount of water and p-xylene associated with exchangeable cations, the CEC of

kaolin used in the adsorption experiments was measured. The CEC of oven-dried

Ca-, Na-, and Li-kaolin at pH 4.11 was 3.9, 3.5, and 1.9 cmol/kg, respectively. To

confirm these values, the CEC of the kaolin used in the preliminary adsorption

study was also measured. At pH 5.5, the CEC of oven-dried Na- and Li-kaolin

was 4.23 and 2.19 cmol/kg, respectively. Apparently, the heat-treatment resulted

in a sizable reduction in the CEC of Li-kaolin. These findings suggest that heat-

treated Li-kaolin represents a mineral surface of reduced charge, rather than a

surface free of cation hydration effects as proposed by Keenan et al. (1951) and

Martin (1959). Therefore, the use of Li-kaolin as a reference mineral surface may

not be valid, and could result in incorrect estimates of the amount of water

associated with each exchangeable cation.

Competitive Adsorption of p-Xvlene and Water on Ca-, Na-, and Li-kaolin

Data for the adsorption of p-xylene vapors on Ca-, Na-, and Li-kaolin at 0,

10, and 20% RH are presented in Figure 4-5. An increase in RH from 0% to 10

and 20% had a relatively minor effect on p-xylene adsorption regardless of the












E


E
cl


0
12


E
C/)


p-Xylene (P/Po)


Figure 4-5.


Vapor-phase adsorption of p-xylene on (a) Ca-, (b) Na-, and (c) Li-
saturated Fisher kaolin at 0, 10, and 20% RH.







89
cation saturation. Rhue et al. (1989) reported that sizable reductions in p-xylene

adsorption were not observed until the RH was greater than that required to attain

monolayer water coverage, as estimated from the single-sorbate BET equation.

The RHs corresponding to monolayer coverage of water on Ca-, Na-, and Li-kaolin

were 20, 22, and 22%. Thus, the slightly greater displacement of p-xylene from

Ca-kaolin was consistent with the fact that a lower RH was required to achieve

monolayer coverage on Ca-kaolin than on Na- and Li-kaolin.

The measured fraction of water adsorbed on Ca-, Na-, and Li-kaolin, versus

the fraction calculated from the two-component BET equation of Rhue et al.

(1989), is presented in Figure 4-6a. Below 0.4, enhanced water adsorption was

exhibited by all exchangeable cations studied, but was less for Li-kaolin than Ca-

and Na-kaolin. The degree of preferential water adsorption was consistent with

the CEC values previously reported for oven-dried Ca-, Na-, and Li-kaolin. These

data further support the contention that Li-kaolin represents a surface of reduced

charge.

The fraction of water adsorbed on kaolin was also predicted using the two-

component BET equation proposed by Hill (1946a, 1946b). If the sorbate is

assumed to adsorb only onto the mineral surface or sorbed layers of itself, Hill's

equation can be simplified to:

S, Xa [Ca (1-Xb) + XCJ
= (4-5)
Sam (1 X,- Xb)[1 + X(C,-1)+ Xb(Cb-1)]
where Sa is the amount of sorbate "a" adsorbed (mg/g), Sam is the monolayer









7/

01.


0.8

o

O




0o
50.6

ILL
a2
D 0.4



0




S0.2

0
t0.8
C

!0


(D0.4



(z 0.2


L


0.2


0.2


0.4


0.4


(a)









, (


0.6


0.6


Measured Fraction

Figure 4-6. Measured fraction of water adsorbed on Ca-, Na-, and Li-
saturated Fisher kaolin versus the fraction calculated using the
two-component BET equation of (a) Rhue et al. (1989), and (b)
Hill (1946a, b).


- I/
*
* U


Elm


-


-



^-C
(b) E -




U, ,




/


0.8


0.8


* Ca-kaolin
* Na-kaolin
* Li-kaolin
'







91
adsorption capacity of sorbate "a" (mg/g), Xa and X, are the relative vapor

pressures of sorbate "a" and "b", respectively, and Ca and Cb are the BET

parameters related to the heat of adsorption of sorbate "a" and "b", respectively.

The predicted adsorption of water and p-xylene on Ca-, Na-, and Li-kaolin is

presented in Table 4-3. The adsorption of water and p-xylene on kaolin was well

predicted, regardless of the cation saturation. However, the predicted values for

Li-kaolin were more accurate than those estimated for Ca- and Na-kaolin.

The measured fraction of water adsorbed on Ca-, Na-, and Li-kaolin versus

the fraction predicted using the two-component BET equation of Hill (1946a,

1946b) are presented in Figure 4-6b. The measured and calculated fractions of

water adsorbed on Li-kaolin were similar for all mixed-vapor systems studied. At

fraction less than 0.3, Ca- and Na-kaolin exhibited enhanced water adsorption.

However, the magnitude of the preferential water adsorption observed here was

far less than that obtained using the two-component BET equation of Rhue et al.

(1989). This was primarily due to the fact that Hill's two-component BET equation

incorporates the value of S,, determined from single-sorbate adsorption isotherms.

Thus, the adsorption of water and p-xylene predicted by Hill's equation is based,

in part, on the monolayer adsorption capacity of each sorbent. In contrast, the

equation of Rhue et al. (1989) does not include a monolayer adsorption capacity

term. Therefore, the value of C must be related to the cation hydration energy in

order for the fraction of water calculated using the two-component BET equation

to reflect differences in the monolayer adsorption capacity due to cation saturation.









Table 4-3.


Comparison of measured p-xylene and water adsorption on Ca-,
Na-, and Li-kaolin from mixed-vapor systems with values predicted
using the multi-sorbate BET equation of Hill (1946a, 1946b).


P/Po Measured Predicted
Sorbent Water PX Water PX Water PX

------------------------------------- mg/g ----------.--------------------


Ca-kaolin






Na-kaolin


0.117
0.101
0.112
0.179
0.212
0.184

0.117
0.101
0.112
0.179
0.212
0.218


Li-kaolin 0.117
0.101
0.112
0.179
0.212
0.218


0.093
0.201
0.395
0.096
0.217
0.403

0.093
0.201
0.395
0.096
0.217
0.410

0.093
0.201
0.395
0.096
0.217
0.410


2.64
2.64
3.24
4.13
4.18
3.42

2.31
2.38
2.27
3.56
3.09
3.48

2.02
1.56
1.24
2.93
2.72
2.34


3.58
5.72
8.00
3.22
5.14
7.89

3.40
5.42
7.94
3.32
5.33
7.89

2.57
4.09
6.85
2.39
3.99
6.97


3.06
2.34
2.33
4.07
4.26
3.91

1.93
1.42
1.47
2.71
2.82
3.28

1.56
1.21
1.23
2.13
2.28
2.61


2.51
4.88
9.17
2.28
4.79
9.80

2.75
4.84
8.45
2.54
4.84
9.67

1.91
3.71
7.04
1.77
3.72
8.08