Effect of clays, metal oxides, and organic matter

on rhamnolipid biosurfactant sorption by soil

Francisco J. Ochoa-Loza

a,1

, Wouter H. Noordman

b,2

, Dick B. Jannsen

b

,

Mark L. Brusseau

a

, Raina M. Maier

a,*

a

Department of Soil, Water and Environmental Science, University of Arizona, Tucson, AZ, 85721, United States

b

Department of Biochemistry, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen,

Nijenborgh 4, 9747 AG Groningen, The Netherlands

Received 21 May 2005; received in revised form 22 July 2006; accepted 26 July 2006

Available online 11 September 2006

Abstract

Rhamnolipids produced by Pseudomonas aeruginosa have been proposed as soil washing agents for enhanced removal of metal and

organic contaminants from soil. A potential limitation for the application of rhamnolipids is sorption by soil matrix components. The
objective of this study is to empirically determine the contribution of representative soil constituents (clays, metal oxides, and organic
matter) to sorption of the rhamnolipid form most efficient at metal complexation (monorhamnolipid). Sorption studies show that
monorhamnolipid (R1) sorption is concentration dependent. At low R1 concentrations that are relevant for enhancing organic contam-
inant biodegradation, R1 sorption followed the order: hematite (Fe

2

O

3

) > kaolinite > MnO

2

illite Ca-montmorillonite > gibbsite

(Al(OH)

3

) > humic acid-coated silica. At high R1 concentrations, relevant for use in complexation/removal of metals or organics, R1

sorption followed the order: illite >> humic acid-coated silica > Ca-montmorillonite > hematite > MnO

2

> gibbsite

kaolinite. These

results allowed prediction of R1 sorption by a series of six soils. Finally, a comparison of R1 and R2 (dirhamnolipid) shows that the
R1 form sorbs more strongly alone than when in a mixture of both the R1 and R2 forms. The information presented can be used to
estimate, on an individual soil basis, the extent of rhamnolipid sorption. This is important for determining: (1) whether rhamnolipid
addition is a feasible remediation option and (2) the amount of rhamnolipid required to efficiently remove the contaminant.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: Biosurfactant; Sorption; Bioremediation; Remediation; Soil washing; Surfactant; Rhamnolipid

1. Introduction

Biosurfactants have gained attention as environmentally

compatible agents that can be used to enhance the remedi-
ation process (e.g.,

Bodour and Maier, 2002; Mulligan,

2005

). Rhamnolipids, produced by Pseudomonas aerugin-

osa, are one of the most widely studied biosurfactants.
Recent research has demonstrated that rhamnolipids or

rhamnolipid-producing isolates (

Dean et al., 2001

) have

several potential soil applications. They can increase degra-
dation of hydrophobic organic soil contaminants, enhance
removal of organics by soil flushing, facilitate removal of
soil-bound Cd

2+

, Zn

2+

, and Pb

2+

, and in addition, can

act as natural pesticides for zoosporic plant pathogens
(

Maier and Sobero´n-Cha´vez, 2000

). Pseudomonas sp. pro-

duce rhamnolipids either as the monorhamnolipid (R1)
form, or more frequently, as a mixture of the mono- and
dirhamnolipid (R2) forms. Due to their difference in struc-
ture, the R1 and R2 forms have differing efficacy in envi-
ronmental applications. For example, the R1 form is far
superior at metal complexation (

Ochoa-Loza et al., 2001

;

Ochoa-Loza and Maier, unpublished results).

0045-6535/$ - see front matter

Ó 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.chemosphere.2006.07.068

*

Corresponding author. Tel.: [phone]; fax: [phone].
E-mail address:

[e-mail]

(R.M. Maier).

1

Present address: Civil Engineering School, University of Sinaloa,

Culiacan, Sinaloa, Mexico 80040.

2

Present address: NIZO Food Research, Ede, The Netherlands.

www.elsevier.com/locate/chemosphere

Chemosphere 66 (2007) 1634–1642

It is generally considered desirable that surfactants used

for soil remediation have minimal sorptive interactions
once applied to the soil system, i.e., most of the surfactant
should remain in the aqueous phase. While it can be argued
that some sorption might be desirable for enhancing
desorption kinetics of hydrophobic contaminants, several
studies suggest that rhamnolipid sorption by mineral sur-
faces can be of sufficient magnitude to cause a major con-
straint in their application for contaminant removal (

Van

Dyke et al., 1993; Herman et al., 1995; Torrens et al.,
1998

). In fact, surfactant sorption in general is likely the

reason that high concentrations are often required for effec-
tive contaminant removal (

Van Dyke et al., 1993; Mata-

Sandoval et al., 2002

). Thus, sorption of biosurfactants

by the soil matrix is a serious limitation to successful bio-
remediation applications.

The objective of this research is to determine the extent to

which different soil constituents contribute to rhamnolipid,
in particular R1, sorption. Such information will have prac-
tical application for evaluating, on the basis of soil proper-
ties, the suitability and amounts of rhamnolipid required
for treatment. This paper describes a series of batch suspen-
sion experiments performed to determine sorption isotherms
for R1 in the presence of representative soil constituents
comprising the soil solid phase including a variety of clays,
metal oxides, and organic matter. In addition, sorption iso-
therms were measured for a series of soils to determine
whether clay, metal oxide, and organic matter content can
be used to predict R1 sorption. Finally a series of experi-
ments was performed with a R1/R2 mixture to determine
which rhamnolipid species is sorbed preferentially.

2. Materials and methods

2.1. Biosurfactant

Two types of rhamnolipids were used in this study.

Monorhamnolipid was produced by and purified from Pseu-
domonas aeruginosa ATCC 9027 as described earlier (

Zhang

and Miller, 1992

). This monorhamnolipid contains 80%

C20-R1 and 15% C18-R1 (CXX = number of carbon atoms
in the lipid moiety; RX = number of rhamnose units). A
mixture of R1 and R2 was produced by Pseudomonas aeru-
ginosa UG2 as described by

Van Dyke et al. (1993)

and puri-

fied by column chromatography over Sephadex LH20 with
methanol as the eluent (

Noordman et al., 2000b

). The R1/

R2 mixture contained 8% C18-R1, 25% C20-R1, 7% C18-
R2, 55% C20-R2, and 5% C22-R2. The average molecular
weights of these biosurfactants are 504 for the ATCC 9027
R1 and 588 for the UG2 R1/R2 mixture. For both biosurf-
actants the critical micelle concentration (CMC) is 0.1 mM
under the conditions used in this study.

2.2. Matrices

A well crystallized kaolinite (Georgia) and Ca-mont-

morillonite (Texas) were obtained from the Clay Mineral

Society reference collection, Department of Geology, Uni-
versity of Missouri and an illite (No. 35. Fithian, Illinois)
was obtained from Ward’s Natural Science (Rochester,
N.Y.). Hematite (Fe

2

O

3

, 99.6% purity), MnO

2

(99.8% pur-

ity), and Al(OH)

3

(100% purity) were purchased from J.T.

Baker (Phillipsburg, NJ). Iron oxide-coated silica (FeOx-
Si) with an Fe

2

O

3

content of 18.4 g kg

1

was obtained from

Sigma Labs (Tucson, AZ). The iron minerals used (hema-
tite and iron oxide-coated silica) may be very different;
no further analyses were performed to identify the iron
oxide minerals coating the silica. Humic acid-coated silica
(HA-Si) was prepared with Acros humic acid (Acros, NJ)
as described previously (

Noordman et al., 2000a

). The

organic carbon content was 22 g kg

1

, as determined using

elemental analysis. The silica (230–400 mesh) was obtained
from Merck (Darmstadt, Germany) and is characterized by
a particle size of 40–60 lm and a pore size of 60 A

˚ . Proper-

ties of these matrices are shown in

Table 1

.

A set of six soils was selected in order to provide a wide

range of soil particle distribution, predominant clay min-
eral, total organic content (TOC), and iron oxide content.
Prior to the experiments, all six soils were air-dried and
sieved through a no. 10 standard sieve (2 mm openings).
Properties of these soils are shown in

Table 2

.

2.2.1. Preparation of rhamnolipid solutions

To prepare a rhamnolipid solution, a known mass of

purified R1 or R1/R2 mixture was dissolved in a 5 mM
KNO

3

background electrolyte solution and the solutions

were adjusted to pH 6.8 by addition of 0.1 M KOH
(

Torrens et al., 1998

). The rhamnolipid solutions were

filter-sterilized using a 0.22-lm cellulose acetate filter
(Falcon Easy-Flow filters, Benton Dickinson Labware,
Lincoln Park, NJ) and stored in autoclaved 100-ml glass
bottles.

2.2.2. Rhamnolipid sorption isotherms

Batch experiments were conducted to measure adsorp-

tion of the rhamnolipid by all solid matrices. R1 sorption
by clays (0.5 g), by Al(OH)

3

(0.2 g), and by hematite and

MnO

2

(0.1 g) was measured using solid:solution ratios of

1:10, 1:25, and 1:50, respectively. The R1 sorption experi-
ments for soils (2.5 g) were carried out using a solid:solu-
tion ratio of 1:2, except for the Molokai soil in which a
solid:solution ratio of 1:13 was used. Samples of known
mass of the solid matrices were placed into 40-ml polypro-
pylene centrifuge bottles and conditioned by washing with
5 ml of 5 mM KNO

3

(pH 6.8), prior to addition of the

rhamnolipid, for 2 d at room temperature on a shaker
(100 rpm). Next, the bottles were centrifuged (15 000 rpm,
20 min) and the supernatant discarded. The remaining pel-
lets were autoclaved to inhibit any biodegradation of R1,
and then pellets from triplicate bottles were suspended in
5 ml of 0, 0.62, 1.25, 2.5, 5.0, 6.5, 8.0, or 10 mM R1 pre-
pared as described earlier. After incubating the samples
for 3 d at room temperature at 100 rpm, the bottles were
centrifuged

and

the

rhamnolipid

concentration

was

F.J. Ochoa-Loza et al. / Chemosphere 66 (2007) 1634–1642

1635

Table 2
Chemical properties of the soils used

Soil

Sand

a

(g kg

1

)

Silt
(g kg

1

)

Clay
(g kg

1

)

Clay min.

b

pH

c

TOC

d

(g kg

1

)

Fe

e

(g kg

1

)

Al

e

(g kg

1

)

Exchangeable bases

f

(g kg

1

)

Sum of bases
(cmol

c

kg

1

)

Exchangeable
acidity

g

(cmol

c

kg

1

)

Sum of acidity
(cmol

c

kg

1

)

CEC

h

(cmol

c

kg

1

)

Surface area

i

(m

2

g

1

)

Na

+

K

+

Ca

2+

Mg

2+

Al

3+

H

+

Pierre

96

409

495

MT4,IL3,KK2,QZ2

7.5

14.6

21.4

1.9

0.1

1.5

65

14

81

–

–

–

81

161

Barnes

395

359

246

MT3,IL3,KK2

5.6

32.6

12.9

1.9

tr

j

2.5

28

10

40

–

–

10

50

73.4

Bonify

912

55

33

KK2,GEl,GI1

5.4

3.2

2.9

1.9

0.1

0.1

1.4

0.2

1.8

–

1.8

1.8

3.6

10.5

Gaston

355

254

391

KK5,HE2,GE2,GI2

5.7

11.2

64.4

9.5

tr

j

0.8

9.6

3.6

1.4

–

7.2

7.2

21

62.9

Molokai

125

306

569

KK4,GI3,1L2

6.2

7.9

230

3.2

0.1

1.3

4.8

1.2

7.4

4.2

0.3

4.5

12

167

Andisol

380

523

97

AL2,IM2

6.2

36.3

38.2

11.5

0.1

3.1

48

2.9

51

5.1

0.3

5.4

56

262

a

The pipet method was used for Pierre, Barnes, Bonify, and Gaston, the hydrometer method was used for Molokai and Andisol (

Klute, 1986

).

b

Clay mineralogy: AL = allophane; GE = goethite; GI = gibbsite; HE = hematite; IL = illite; IM = imogolite; KK = kaolinite; MT = montmorillonite; QZ = quartz. The number indicates the

amount of each mineral present; (5) very large, (4) large, (3) moderate, (2) small, (1) very small.

c

Soil:water ratio 1:1 (

Sparks et al., 1996

).

d

Determined by Walkley–Black method (

Sparks et al., 1996

).

e

As Fe

2

O

3

and Al

2

O

3

, respectively. Iron and aluminum extracted by the dithionate-citrate method (

Sparks et al., 1996

).

f

Exchangeable bases determined by NH

4

/Na-acetate, pH 7.0 (

Sparks et al., 1996

).

g

Exchangeable acidity determined by extraction with 1 M KC1 (

Sparks et al., 1996

).

h

CEC = effective cation exchange capacity (

Sparks et al., 1996

).

i

Determined by EGME (ethylene glycol monoethyl ether) method (

Kennedy et al., 2002

).

j

tr = trace.

Table 1
Chemical and physical properties of soil constituents used

Soil
constituent

pH

a

Exchange bases

b

Sum of bases
(cmol

c

kg

1

)

Exchangeable acidity

c

Sum of acidity
(cmol

c

kg

1

)

CEC

d

(cmol

c

kg

1

)

PZC

f

Surface area
(m

2

g

1

)

g

Na

+

(cmol

c

kg

1

)

K

+

(cmol

c

kg

1

)

Ca

2+

(cmol

c

kg

1

)

Mg

2+

(cmol

c

kg

1

)

Al

3+

(cmol

c

kg

1

)

H

+

(cmol

c

kg

1

)

Kaolinite

4.52

0.05

0.6

0.3

0.1

1.1

1.5

0.6

2.1

3.2

e

4.6

11.0

Illite

7.57

0.06

1.9

79

1.3

83

–

–

–

83

e

2.4

157

Ca-Mont.

7.44

1.7

3.0

150

19

170

–

–

–

170

2.5

708

Hematite

6.75

–

–

–

–

–

–

0.04

0.04

1.4

b

8.3

8.2

MnO

2

4.22

–

–

–

–

–

–

0.06

0.06

0.7

b

7.2

2.4

Al(OH)

3

6.90

–

–

–

–

–

–

0.04

0.04

1.1

b

6.5

1.6

FeOx-Si

3.90

–

–

–

–

–

–

5.0

5.0

16

b

–

328

HA-Si

5.50

–

–

–

–

n.d.

h

n.d.

h

n.d.

h

n.d.

h

n.d.

h

–

219

a

Soil:water ratio 1:4 (

Sparks et al., 1996

).

b

Determined by NH

4

/Na-acetate, pH 6.8 (

Sparks et al., 1996

).

c

Determined by extraction with 1 M KC1 (

Sparks et al., 1996

).

d

CEC = effective cation exchange capacity (

Sparks et al., 1996

).

e

Sum of exchangeable bases and exchangeable acidity.

f

PZC = point of zero charge. From

Schwarzenbach et al. (1993), Stumm and Morgan (1981) and Gebhardt and Fuerstneau (1984)

.

g

Determined by EGME (ethylene glycol monoethyl ether) method (

Kennedy et al., 2002

).

h

Not determined.

1636

F.J.

Ochoa-L

oza

et
al.
/

Chemosph

ere

66
(2007)

1634–1642

determined as described below. The amount of rhamnoli-
pid adsorbed was obtained from the difference between
rhamnolipid recovered from the supernatant and rhamnoli-
pid originally added. Biosurfactant loss due to sorption to
container surfaces was negligible.

The isotherms for R1 and the C20-R1 and C20-R2 com-

ponents from the R1/R2 mixture in untreated silica (0.2 or
0.5 g for R1 and R1/R2 mixture, respectively), FeOx-Si
(0.2 g), and HA-Si (0.2 g) were performed in a solid:solu-
tion ratio of 1:10. These isotherms were measured using

the procedure described above with some modifications.
The solid matrix was transferred in triplicate into 8-ml
pyrex tubes stoppered with aluminum coated septa, and
suspended in 2 ml of 5 mM KNO

3

(pH 6.8) containing 0,

0.2, 0.4, 0.6, 1.2, 1.8, and 2.4 mM R1 or R1/R2 mixture.
An additional R1 concentration of 4 mM was used for
the untreated silica isotherm. After incubation for 16 h in
an end-over-end shaker, the samples were centrifuged
(3000 rpm, 12 min) and the supernatant removed for direct
analysis by HPLC. Preliminary experiments showed that

Table 3
Monorhamnolipid (R1) isotherms and parameters for mineral constituents and soils used in this study

a

Mineral or soil

Ratio

b

Freundlich

Langmuir

Regression analysis

K

f

SE

c

n

SE

K

1

SE

b

SE

r

2

F-test

a

Kaolinite

1:10

1790

597

0.33

0.04

0.97

150

0.01

Illite

1:10

7.2

4.5

1.67

0.12

0.99

396

0.01

Ca-Mont.

1:10

41.2

14.0

1.20

0.06

0.99

1010

0.01

Hematite

1:50

2100

995

0.50

0.06

0.97

163

0.01

MnO

2

1:50

434

261

0.62

0.08

0.96

129

0.01

Al(OH)

3

1:25

167

195

0.73

0.15

0.92

56.5

0.01

FeOx-Si

d

1:10

547

266

0.27

0.08

0.91

19.9

0.05

HA-Si

d

1:10

0.0063

0.04

2.69

1.19

0.83

14.3

0.05

Si

1:10

0.00104

0.0008

12006

5709

0.87

33.2

0.01

Pierre

1:2

5.84

7.32

1.61

0.28

0.96

105

0.01

Barnes

1:2

330

49.1

0.37

0.02

0.99

562

0.01

Bonifay

e

1:2

1.53

2.74

1.01

0.23

0.94

c

494

d

0.01

Gaston

1:2

0.0031

0.0005

14438

1159

0.99

741

0.01

Molokai

1:13

0.0308

0.0066

80706

7195

0.97

189

0.01

Andisol

1:2

152

23.2

0.51

0.02

0.99

1940

0.01

a

For each case, the data were analyzed using the Freundlich, Langmuir, and Linear isotherm equations. The coefficients are reported only for the

isotherm that provided the most robust fit.

b

Solid:solution.

c

SE = standard error.

d

Only C20-R1 concentration determined.

e

For the Bonifay soil the linear isotherm had an r

2

-value equivalent to the Freundlich isotherm.

Table 4
A comparison of the best isotherms and parameters for selected soil constituents and the C20-R1 and C20-R2 species of an R1/R2 mixture with the R1
species alone

Soil constituent

Ratio

Freundlich

Langmuir

Linear

Regression analysis

K

f

SE

n

SE

K

1

SE

b

SE

Coefficient

SE

r

2

F-test

a

Kaolinite

C20-R1

a

1:40

0.047

0.03

2669

562.7

0.76

12.9

0.05

C20-R2

b

1:40

0.120

0.08

1272

130.3

0.55

4.9

0.1

R1

c

1:40

0.033

0.02

7699

1537

0.79

11.6

0.05

FeOx-Si

C20-R1

1:10

0.054

0.02

290.8

44.9

0.89

30.9

0.01

C20-R2

1:10

0.088

0.03

308.5

21.1

0.91

38.6

0.01

R1

1:10

547

266

0.27

0.08

0.91

19.9

0.05

HA-Si

C20-R1

1:10

3.92

1.34

1.08

0.08

5.45

0.17

0.99

d

431.8

e

0.01

C20-R2

1:10

5.78

4.89

0.91

0.16

3.61

0.24

0.94

d

67.2

e

0.01

R1

1:10

0.0063

0.04

2.69

1.19

0.83

14.3

0.05

a

C20-R1 = R1 species from R1/R2 mixture.

b

C20-R2 = R2 species from R1/R2 mixture.

c

R1 = All of the R1 species in the R1.

d

For both models.

e

Only for Freundlich model.

F.J. Ochoa-Loza et al. / Chemosphere 66 (2007) 1634–1642

1637

rhamnolipid sorption reached equilibrium within 1 h on
the silica matrices.

2.2.2.1. Analytical procedures. The rhamnolipid concentra-
tions in stock solutions were measured using the 6-deoxyhex-
ose assay with L-rhamnose as a standard (

Chandrasekaran

and BeMiller, 1980

). For most experiments, aqueous rhamn-

olipid concentrations were determined by surface tension
analysis using a surface tensiomat (Model 21, Fisher Scien-
tific) that employs the Du Nouy ring method. All solutions
were diluted to below the CMC prior to surface tension mea-
surement. A calibration curve was prepared under identical
experimental conditions using rhamnolipid standards and
relating rhamnolipid concentration (mg l

1

) to surface ten-

sion (mN m

1

).

For experiments with the R1/R2 mixtures, in particular

for FeOx-Si, HA-Si and kaolinite, concentrations of spe-
cific rhamnolipids (C20-R1 and C20-R2) were determined
by HPLC (

Noordman et al., 2000b

). The detection limit

was 5 mg l

1

for each rhamnolipid. The relative concentra-

tions of the rhamnolipid species (C20-R1 and C20-R2)
were determined from their respective areas in the calibra-
tion sample at 1000 mg l

1

. Separate calibration curves

were determined for each species.

2.2.2.2. Sorption isotherm analysis. A nonlinear least
squares regression analysis program was used to fit sorp-
tion isotherm data to two of the more commonly used
adsorption models, the Freundlich equation and the Lang-
muir equation. If the isotherm data approximated a
straight line for either of these equations, then a regression
using a linear equation was also performed. Using the cri-
teria of r-squared, standard error, and F-value of the model
regression analysis, the equation that fit each particular iso-
therm best was chosen (

Tables 3 and 4

). The fitting param-

eters were used to define the relative importance among the
different solid matrix characteristics with regard to rham-
nolipid sorption.

3. Results and discussion

The results and discussion section is organized into three

parts. In the first part, sorption results of R1 by clays,
metal oxides, and humic acid is presented. In the second
part, the relative contribution of each soil constituent is
discussed and used to interpret sorption of R1 by the differ-
ent soils tested. Finally, the sorption of different rhamnoli-
pid species (R1 and R2) by a representative clay (kaolinite),
metal oxide (FeOx-Si), and organic matter (HA-Si) is
discussed.

3.1. Part 1 – sorption of monorhamnolipid to clays, metal
oxides, and humic acid

Three clay minerals were tested as sorbents; illite, mont-

morillonite, and kaolinite. Note that R1 was added in eight

concentrations: 0, 0.62, 1.25, 2.5, 5.0, 6.5, 8.0, or 10 mM,
all substantially above the CMC of 0.1 mM. Thus, these
solutions contain a mixture of surfactant monomers (equal
to the CMC concentration) and aggregates that will exhibit
complex behavior when added to a sorbent. Note that after
sorption equilibrium, the solution-phase rhamnolipid con-
centration may be less than CMC in which case, only
monomers would be present.

Sorption of R1 to the clay minerals was nonlinear and

the clay-R1 isotherms fitted well with the Freundlich equa-
tion (

Table 3

). The R1 isotherms for kaolinite, illite, and

Ca-montmorillonite exhibit different shapes as well as mag-
nitudes of associated sorption (

Fig. 1

A). The kaolinite

isotherm is concave (n = 0.33), whereas those for illite
and Ca-montmorillonite are convex (n = 1.67 and 1.22,
respectively). At low R1 concentrations kaolinite had the
strongest affinity for R1, however, as the R1 concentration
was increased, sorption increased greatly for illite and Ca-
montmorillonite in contrast to kaolinite. The behavior for
illite and Ca-montmorillonite is typical of two-step adsorp-
tion behavior that has been observed for synthetic anionic
surfactants. In this case, following sorption of a layer of
individual monomers, a significant increase in sorption
occurs as surfactant hemimicelles form on the surface
(e.g.,

Zhu and Gu, 1991

).

Four different metal oxides were used in this study; hema-

tite (Fe

2

O

3

), MnO

2

, gibbsite (Al(OH)

3

), and iron-oxide

coated silica (FeOx-Si). The R1-metal oxide isotherms were
nonlinear and followed the order: hematite > MnO

2

> gibb-

site > FeOx-Si for the entire range of rhamnolipid concen-
trations tested (

Table 3

,

Fig. 1

A). They all fitted best with

the Freundlich model and have similar shapes (n-values < 1).
The fact that the iron oxide matrices occupy the first (hema-
tite) and the last (FeOx-Si) position of this sequence is due, in
part, to the difference in their Fe

2

O

3

content (995 and

18.4 g kg

1

, respectively). It should be noted that a lower

R1 concentration range was used for the FeOx-Si isotherm
(0–2.4 mM) and only the sorption of the C20-R1 fraction
was quantified.

Finally, R1 sorption by an untreated and humic acid-

coated silica (HA-Si) was evaluated. The R1-HA-Si iso-
therm was nonlinear and convex indicating a high affinity
of R1 for humic acid at high rhamnolipid concentrations
(

Table 3

,

Fig. 1

A). These data fitted best with the Freund-

lich equation (n-value = 2.69). For untreated silica, sorp-
tion was similar to HA-Si at low R1 concentrations, but
at high R1 concentrations was much lower. The R1-
untreated silica isotherm had an S-shape which fitted the
Langmuir equation best (

Table 3

,

Fig. 1

A). The sorption

behavior for these two types of silica is similar to a previ-
ously reported rhamnolipid sorption study that was carried
out under saturated flow conditions (

Noordman et al.,

2000a

). In this study, the magnitude of sorption for R1

at 20 mg l

1

(0.04 mM) was similar for silica and HA-silica,

but at 500 mg l

1

(1 mM), sorption was much higher for

HA-silica. The untreated silica isotherm can also be com-
pared to a R1 isotherm using Accusand 40/50 mesh (0.3–

1638

F.J. Ochoa-Loza et al. / Chemosphere 66 (2007) 1634–1642

0.42 mm) that was reported by

Bai et al. (1997)

. The

Accusand study reported 10-fold less R1 sorbed than for
the silica tested in this study. This difference can be
explained by the specific surface areas of these matrices
that differ by over two orders of magnitude (

Noordman

et al., 2000c

).

A comparison of the rhamnolipid sorption isotherms for

all the soil parameters tested indicates that both aluminos-
ilicates and iron oxide minerals are critical to the sorption
of this anionic biosurfactant and that since sorption is non-
linear, there are large differences in sorption at low and high
concentrations of rhamnolipid. At low R1 concentrations,
the relative order of importance is: hematite > kaolinite >

MnO

2

illite Ca-montmorillonite > gibbsite > FeOx-

Si > HA-Si (

Fig. 1

B). However, at rhamnolipid concentra-

tions greater than 400 mg kg

1

(0.8 mM), the order

changes to: illite >> HA-Si > Ca-montmorillonite > hema-
tite > MnO

2

> gibbsite

kaolinite (note that for the pur-

pose of this comparison we have extrapolated the illite,
HA-Si, and Ca-montmorillonite isotherms beyond the
experimental data points collected). The fact that the clay
fraction has a significant contribution to the sorption of
an anionic biosurfactant is at variance with what has been
previously reported in some early studies (

Law and Kunze,

1966; Law et al., 1966

), but is similar to results reported

more recently by

Mata-Sandoval et al., 2002

). This sorption

may be considered to be analogous to sorption of humic
components to soil clay minerals. It has been demonstrated
that sorption of humic and fulvic acid to clays is enhanced
by the presence of di- and trivalent cations occupying the
exchange positions (

Greenland, 1971; Theng and Scharpen-

seel, 1975; Theng, 1976; Davis, 1982; Murphy et al., 1992

).

Of the iron oxides, hematite shows a significant contribu-
tion to sorption at all surfactant concentrations, which is
not surprising since electrostatic interactions are likely
involved. These are due to the high zero point of charge
which results in a net positive charge at the neutral pH that
these experiments were performed. Of the metal oxides,
gibbsite [Al(OH)

3

] has the lowest contribution to R1 sorp-

tion likely because it has the lowest point of zero charge
and the lowest surface area (

Table 1

). The results with

HA-Si suggest that the humic acid content in soil, at least
up to a TOC content of 20 g kg

1

, is not critical for deter-

mining rhamnolipid sorption at low rhamnolipid concen-
trations. A similar conclusion was reached in a study of
rhamnolipid sorption to four sandy soils (

Noordman

et al., 2000b

). In general, we conclude that hematite, illite,

kaolinite, and Ca-montmorillonite are the most important
soil constituents exerting the greatest effect on R1 sorption.

3.2. Part 2 – monorhamnolipid sorption by different soils

In light of the results obtained with the soil constituents

tested, it seems that R1 sorption in a natural system will
depend largely on the sorbent properties including: chemi-
cal (mineralogical and organic) composition, the relative
proportion of the constituents comprising the solid phase,
as well as the concentration of rhamnolipid in the aqueous
phase. Thus, it is reasonable to hypothesize that the magni-
tude of R1 sorption by soils may be predicted based on the
knowledge of their mineral and chemical composition. To
test this hypothesis, the sorption of R1 by six different soils
was measured (

Fig. 2

). Based upon the data in

Table 2

, in

particular the clay (amount and type) and iron content, we
predicted the following order of sorption: Molokai >
Gaston > Pierre > Barnes > Andisol > Bonify.

Each of the soil-R1 isotherms fitted either the Freund-

lich or Langmuir equation well (

Table 3

). Rhamnolipid

sorption generally followed the order: Molokai > Pierre >
Gaston > Barnes

Andisol > Bonify which was similar

R1 in Solution (mg l

-1

)

0

10

20

30

40

50

60

R1 Sorbed (m

g kg

-1

)

0

2000

4000

6000

8000

10000

0

100

200

300

400

500

0

20000

40000

60000

80000

100000

Illite
Montmorillinite
Kaolinite
Hematite
MnO

2

Gibbsite
Fe-Ox-silica
HA-silica
Silica

HA-Si*

Montmorillonite*

Hematite

MnO

2

Gibbsite

Kaolinite

FeOx-Si

Hematite

MnO

2

HA-Si

Gibbsite

FeOx-Si

Illite

Kaolinite

Montmorillonite

R1 Sorbed (m

g kg

-1

)

A

B

Illite*

Fig. 1. A comparison of R1 sorption by all soil constituents tested. The
symbols indicate experimentally determined data points. The lines
represent isotherms that are based on the equation that best fitted each
experimentally determined soil constituent isotherm. (A) The experimental
data and fitted isotherms for the entire data set. The starred isotherms
(Illite

*

, HA-Si

*

, and Montmorillonite

*

) have been extrapolated beyond the

experimental data points collected. Note that not all experimental points
are represented because the abscissa is cut off at 500 mg l

1

. (B) The fitted

isotherm lines at low solution phase R1 concentrations (<60 mg l

1

). Note

that there are no experimental data points at these low concentrations.

F.J. Ochoa-Loza et al. / Chemosphere 66 (2007) 1634–1642

1639

to our prediction. The lowest sorption of R1 was observed
for the Bonify soil, the soil with the lowest clay (33 g kg

1

),

iron (2.9 g kg

1

) and TOC (3.2 g kg

1

) content (

Table 2

). In

contrast, the greatest R1 sorption corresponded to the Mol-
okai and Pierre soils. The Molokai soil is characterized by a
high iron content (230 g kg

1

) and high clay content

(569 g kg

1

) dominated by gibbsite and kaolinite. The

Pierre soil is distinguished from the other soils by its rela-
tively high clay content (495 g kg

1

) and in particular, its

large component of layer silicate minerals predominated
by montmorillonite and illite (

Table 2

). In fact, the Pierre

soil isotherm was very similar to that obtained for illite sug-
gesting that this aluminosilicate mineral dominated the
sorption process in this soil (

Table 3

). The Gaston soil

sorbed R1 much less strongly than either Molokai or Pierre
despite its relatively high clay (391 g kg

1

, dominated by

kaolinite) and iron (64.4 g kg

1

) content. The Barnes and

Andisol soils exhibited moderate R1 sorption in compari-
son to the Molokai and Pierre soils due to their lower com-
bined iron and clay contents. These two soils have the
highest TOC content of the soils studied (

Table 2

). This sug-

gests that soil organic matter is much less important than
clays and iron oxides in the sorption of R1 as noted above.

These results are in good agreement with previous work

from our lab demonstrating the ability of rhamnolipids to
remove soil-bound metals under saturated flow conditions
(

Torrens et al., 1998

). Removal of Cd

2+

from soils ranged

from 44% to 102% using a 10 mM solution of an R1/R2
mixture produced by Pseudomonas aeruginosa IGB83.
The metal removal was most rapid and complete in the soil
with the lowest clay, iron oxide, and TOC content. The
poorest removal was in the soil with highest clay content.

3.3. Part 3 – sorption of specific rhamnolipid constituents

The R1 biosurfactant produced by Pseudomonas aeru-

ginosa ATCC 9027 has been characterized as a mixture
of four R1 forms that differ only in the length of the two
fatty acid tails (

Zhang and Miller, 1994

). The predominant

species in this mixture (80%) contains two C10 fatty acids
and is referred to as the C20-R1 species. However, other
strains of Pseudomonas aeruginosa produce rhamnolipid
mixtures containing both R1 and R2 form. Recent research
in both our lab and other labs with R1/R2 combinations
suggests that it may be advantageous to use such mixtures
(

Torrens et al., 1998; Mata-Sandoval et al., 2002

). We pos-

tulated that a mixture of R1 and R2 might sorb less

R1 in Solution (mg l

-1

)

0

100

200

R1 Sorbed (mg k

g

-1

)

0

20000

40000

60000

Pierre

Bonify

Gaston

Molokai

Andisol, Barnes

R1 (mg l

-1

)

0

10

20

30

40

50

R1 (m

g K

g

-1

)

0

1000

2000

3000

Molokai

Bonify

Pierre

Gaston

Barnes

Andisol

Fig. 2. A comparison of R1 sorption by the six soils tested. Symbols
represent experiment data points and lines represent isotherms based on
the equation that best fitted each experimentally determined soil isotherm.
Note that in the inset there are no experimental data points.

0

100

200

300

0

500

1000

1500

2000

2500

3000

0

100

200

300

0

2000

4000

6000

8000

C20-R1 (ATCC 9027)
C20-R1 (UG2 mix)
C20-R2 (UG2 mix)

Kaolinite

FeOx-Si

Rhamnolipid in Solution (mg l

-1

)

0

100

200

300

S

o

rbed Rhamnoli

pid (mg kg

-1

)

0

2000

4000

6000

8000

10000

HA-Si

Fig. 3. A comparison of the sorption of R1 alone (from ATCC 9027) with
R1 when it is in a R1/R2 mixture (from UG2) by kaolinite, iron oxide-
coated silica (FeOx-Si), and humic acid-coated silica (HA-Si). In each
case, only the C20-R1 and the C20-R2 forms were measured. For ATCC
9027, the C20-R1 form comprises 80% of the rhamnolipid. For UG2
mixture, the C20-R1 and C20-R2 forms comprise 25% and 55%,
respectively. The isotherms are based on the equation that best fitted
each experimentally determined isotherm.

1640

F.J. Ochoa-Loza et al. / Chemosphere 66 (2007) 1634–1642

strongly than R1 alone due to the more hydrophilic nature
of R2. In order to gain more insight into the behavior of
these rhamnolipid species, and the possible impact of spe-
cies mixtures on sorption, a series of sorption experiments
were performed using kaolinite, FeOx-Si, and HA-Si and
an R1/R2 mixture produced by Pseudomonas aeruginosa
UG2. The R1/R2 mixture consisted mainly of the C20-
R1 component (25%) and the C20-R2 component (55%)
(

Noordman et al., 2000b

).

An analysis of the sorption behavior of the R1 and R2

species by kaolinite, FeOx-Si and Ha-Si shows that the
R1 form alone sorbed more strongly to all materials than
the R1 or R2 components when added as a mixture (

Table

4

,

Fig. 3

). Specifically, for kaolinite the C20-R1 form

sorbed 3-fold more strongly alone than when in the pres-
ence of C20-R2. For FeOx-Si and HA-Si, the C20-R1 form
sorbed 10-fold and 30-fold more strongly alone than when
in the presence of C20-R2. Further, the sorption of the
R1reached a plateau at much lower aqueous concentra-
tions when the C20-R2 was present. These results show
that the sorption of the R1 and R2 components are not
independent of each other. These data clearly suggest that
despite the fact that the R1 form is more effective than the
R2 form for some applications such as metal removal,
application of rhamnolipids to soil may be most effective
as a mixed R1/R2 system. Use of a R1/R2 mixture should
decrease sorption of the R1 form by up to several orders of
magnitude depending on the predominant soil constituent
present.

4. Conclusions

The efficiency and success of biosurfactants in facilitat-

ing removal of organic and inorganic contaminants from
soil systems will depend largely on the amount of rhamn-
olipid present in the aqueous phase. These results show
that sorption of R1 by soil can be semi-quantitatively pre-
dicted by considering the iron oxide (Fe

2

O

3

) content as well

as the clay content and predominant clay type. Soils that
have a low content of aluminosilicate minerals and iron
oxides are likely to exhibit relatively low sorption of R1.
This is in good agreement with previous research reported
about the adsorption of organic anions by iron and alumi-
num hydrous oxides (

Evans and Russell, 1959; Greenland,

1971; Parfitt et al., 1977a,b,c

), and of an anionic synthetic

surfactant by soils (

Fink et al., 1970

).

The findings of this study suggest that there may be seri-

ous limitations in the application of R1 as an in situ tech-
nology for remediation of certain soils (e.g., those with
high iron and/or illite content). However, the results pre-
sented herein, concerning the impact of soil mineralogical
and chemical composition on R1 sorption, will allow
improved consideration of whether rhamnolipid may serve
as a good reagent option and what rhamnolipid dose will
be required to maximize the aqueous phase concentration
of rhamnolipid. It is also suggested that addition of R1
as an R1/R2 mixture will increase the aqueous phase con-

centration of the R1 component thus making it more avail-
able for remediation.

Acknowledgements

This work was supported in by grant CHE-0133237

from the National Science Foundation, grant E504940
from the Superfund Basic Research Program, National
Institute of Environmental Health Sciences, NIH, and by
grant SIR 14-1548 from the Dutch organization for scien-
tific research (NWO).

References

Bai, G., Brusseau, M.L., Miller, R.M., 1997. Biosurfactant-enhanced

removal of residual hydrocarbon from soil. J. Contam. Hydrol. 25,
157–170.

Bodour, A.A., Maier, R.M., 2002. Biosurfactants: types, screening

methods, and applications. In: Bitton, G. (Ed.), Encyclopedia of
Environmental Microbiology. John Wiley and Sons, pp. 750–770.

Chandrasekaran, E.V., BeMiller, J.N., 1980. Constituents analysis of

glycoaminoglycans. In: Whistler, R.L., BeMiller, J.N. (Eds.), Methods
in Carbohydrate Chemistry, vol. 8. Academic Press, Inc., New York,
pp. 89–96.

Davis, J.A., 1982. Adsorption of natural dissolved organic matter at the

oxide/water interface. Geochim. Comochim. Acta 46, 2381–2393.

Dean, S.M., Jin, Y., Cha, D.K., Wilson, S.V., Radosevich, M., 2001.

Phenanthrene degradation in soils co-inoculationed with phenan-
threne-degrading and biosurfactant-producing bacteria. J. Environ.
Qual. 30, 1126–1133.

Evans, L.T., Russell, E.W., 1959. The adsorption of humic acid and fulvic

acids by clays. J. Soil Sci. 10, 119–132.

Fink, D.H., Thomas, G.W., Meyer, W.J., 1970. Adsorption of anionic

detergents by soils. J. Water Pollut. Control Fed. 42, 265–271.

Gebhardt, J.E., Fuerstneau, D.W., 1984. The effect of preadsorbed

polymers on adsorption of sodium dodecylsulfonate on hematite. In:
Rosen, M.J. (Ed.), Structure/Performance Relationships in Surfac-
tants, Am. Chem. Soc. Symposium Series, 253. American Chemical
Society, Washington, DC, pp. 291–310.

Greenland, D.J., 1971. Interactions between humic and fulvic acids and

clays. Soil Sci. 111, 34–41.

Herman, D.C., Artiola, J.F., Miller, R.M., 1995. Removal of cadmium,

lead, and zinc from soil by a rhamnolipid biosurfactant. Environ. Sci.
Technol. 29, 2280–2285.

Kennedy, M.J., Pevear, D.R., Hill, R.J., 2002. Mineral surface control of

organic carbon in black shale. Sci. 295, 657–660.

Klute, A. (Ed.), 1986. Methods of Soil Analysis: Part 1 Physical and

Mineralogical Methods, second ed. ASA-SSSA, Madison, WI.

Law Jr., J.P., Kunze, G.W., 1966. Reactions of surfactants with

montmorillonite: adsorption mechanisms. Soil Sci. Soc. Am. Proc.
30, 321–327.

Law Jr., J.P., Bloodworth, M.E., Runkles, J.R., 1966. Reactions of

surfactants with montmorillonitic soils. Soil Sci. Soc. Am. Proc. 30,
327–332.

Maier, R.M., Sobero´n-Cha´vez, G., 2000. Pseudomonas aeruginosa rham-

nolipids: biosynthesis and potential environmental applications. Appl.
Microbiol. Biotechnol. 54, 625–633.

Mata-Sandoval, J.C., Karns, J., Torrents, A., 2002. Influence of rhamn-

olipids and Triton X-100 on the desorption of pesticides from soils.
Environ. Sci. Technol. 36, 4669–4675.

Mulligan, C.N., 2005. Environmental applications for biosurfactants.

Environ. Poll. 133, 183–198.

Murphy, E.M., Zachara, J.M., Smith, S.C., Phillips, J.L., 1992. The

sorption of humic acids to mineral surfaces and their role in
contaminant binding. Sci. Total Environ. (117/118), 413–423.

F.J. Ochoa-Loza et al. / Chemosphere 66 (2007) 1634–1642

1641

Noordman, W.H., Bruining, J.-W., Wietzes, P., Janssen, D.B., 2000a.

Facilitated transport of a PAH mixture by a rhamnolipid biosurfactant
in porous silica matrices. J. Contam. Hydrol. 44, 119–140.

Noordman, W.H., Brusseau, M.L., Janssen, D.B., 2000b. Adsorption of a

multicomponent rhamnolipid surfactant to soil. Environ. Sci. Technol.
34, 832–838.

Noordman, W.H., DeBoer, G.J., Volkering, F., Janssen, D.B., 2000c.

Assessment of the use of partitioning and interfacial tracers to
determine the content and mass removal rates of non-aqueous phase
liquids. Environ. Sci. Technol. 34, 4301–4306.

Ochoa-Loza, F.J., Artiola, J.F., Maier, R.M., 2001. Stability constants for

the complexation of various metals with a rhamnolipid biosurfactant.
J. Environ. Qual. 30, 479–485.

Parfitt, R.L., Farmer, V.C., Russell, J.D., 1977a. Adsorption on hydrous

oxides I. Oxalate and benzoate on goethite. J. Soil Sci. 28, 29–39.

Parfitt, R.L., Fraser, A.R., Russell, J.D., Farmer, V.C., 1977b. Adsorption

on hydrous oxides II. Oxalate, benzoate and phosphate on gibbsite. J.
Soil Sci. 28, 40–47.

Parfitt, R.L., Fraser, A.R., Farmer, V.C., 1977c. Adsorption on hydrous

oxides III. Fulvic acid and humic acid on goethite, gibbsite, and
imogolite. J. Soil Sci. 28, 289–296.

Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 1993. Environ-

mental Organic Chemistry. John Wiley and Sons, New York.

Sparks, D.L., Page, A.L., Helmke, P.A., Loeppert, R.H., Soltanpour,

P.N., Tabatabai, M.A., Johnston, C.T., Sumner, M.E. (Eds.), 1996.

Methods of Soil Analysis: Part 3 Chemical Methods. SSSA, Madison,
WI.

Stumm, W., Morgan, J.J., 1981. Aquatic Chemistry. John Wiley and Sons,

New York.

Theng, B.K.G., 1976. Interactions between montmorillonite and fulvic

acid. Geoderma 15, 245–251.

Theng, B.K.G., Scharpenseel, H.W., 1975. The adsorption of

14

C-labeled

humic acid by montmorillonite. In: Proc. Int. Clay Conf., Mexico City,
pp. 643–653.

Torrens, J.L., Herman, D.C., Miller, R.M., 1998. Biosurfactant (rhamn-

olipid) sorption and the impact on rhamnolipid-facilitated removal
of cadmium from various soils. Environ. Sci. Technol. 32, 776–
781.

Van Dyke, M.I., Couture, P., Brauer, M., Lee, H., Trevors, J.T., 1993.

Pseudomonas aeruginosa UG2 rhamnolipid biosurfactants: structural
characterization and their use in removing hydrophobic compounds
from soil. Can. J. Microbiol. 39, 1071–1078.

Zhang, Y., Miller, R.M., 1992. Enhanced octadecane dispersion and

biodegradation by a Pseudomonas rhamnolipid surfactant (biosurfac-
tant). Appl. Environ. Microbiol. 58, 3276–3282.

Zhang, Y., Miller, R.M., 1994. Effect of a Pseudomonas rhamnolipid

biosurfactant on cell hydrophobicity and biodegradation of octade-
cane. Appl. Environ. Microbiol. 60, 2101–2106.

Zhu, B.Y., Gu, T.R., 1991. Surfactant adsorption at solid liquid interfaces.

Adv. Colloid Interface Sci. 37, 1–32.

1642

F.J. Ochoa-Loza et al. / Chemosphere 66 (2007) 1634–1642