SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 765

0038-075X/04/16911-765–775 November 2004

Soil Science Vol. 169, No. 11
Copyright © 2004 by Lippincott Williams & Wilkins, Inc. Printed in U.S.A.

A TRANSMISSION ELECTRON MICROSCOPY STUDY

OF THE FRACTAL PROPERTIES AND AGGREGATION
PROCESSES OF HUMIC ACIDS
F. R. Rizzi1, S. Stoll2, N. Senesi1, and J. Buffle2

Transmission electron microscopy (TEM) image analysis has been ap-

plied to the evaluation of the fractal dimensions and aggregation mech-
anisms of soil and peat humic acids (HAs) in aqueous solutions at realis-
tically small concentrations and at various pH values and equilibration
times. TEM image analysis based on fractal concepts and the determina-
tion of fractal dimensions suggests a cluster-cluster aggregation mecha-
nism for HAs in any condition examined. At pH
3, HA clusters exhibit
relatively compact structures characterized by time-dependent values of
the fractal dimension, which are consistent with an underlying diffusion-
limited aggregation (DLA) mechanism controlled by the diffusion of ag-
gregates, which is predominant in any condition. At pH
6, the HA ag-
gregates are characterized for short times by open, elongated structures
that become larger and more compact with increases in the equilibration
time and feature fractal dimension values consistent with a reaction-
limited aggregation (RLA) mechanism. The aggregate chemical reactiv-
ity, which is controlled by the solution pH affecting the electrostatic
charge of HAs, plays an important role in this mechanism. (Soil Science
2004;169:765–775)

Key words: Soil, peat, humic acids, transmission electron microscopy,

fractal properties, aggregation processes.

is a complex physicochemi- is a central issue of colloidal science. Because of

A GGREGATION
cal process in which initially dispersed basic
units (atoms, particles, cells, etc.) stick together as
a result of the action of attractive forces of vari-
ous types and strengths to form aggregates whose
size increases with time ( Jullien and Botet,
1987). Aggregation processes can be studied by a
kinetic and/or a geometric approach. The ki-
netic study provides the quantitative description
of the time evolution of the mean size and size
distribution of colloidal aggregates, whereas the
geometric approach provides quantitative infor-
mation on aggregate structures. Understanding
the morphology and how aggregates are formed
1Dipartimento di Biologia e Chimica Agroforestale ed Ambientale, University of
Bari, Bari, Italy. Dr. Senesi is corresponding author. E-mail: senesi@agr.uniba.it
2Département de Chimie Minérale, Analytique et Appliquée, University of
Geneva, Geneve, Switzerland
Received April 16, 2004; July 30, 2004.
DOI: 10.1097/01.ss.0000148737.88830.c1
the complexity of aggregation mechanisms,
computational chemistry has been used as a pow-
erful tool for the investigation and understanding
of aggregation and flocculation processes in nat-
ural systems (Stoll and Buffle, 1995). Computer
simulation, combined with experimental verifi-
cation of data, often provides useful insights that
are critical for the understanding of experimen-
tal results.
Witten and Sanders (1981) introduced an ag-
gregation model for the interpretation of growth
processes governed by Laplace equations in which
aggregates are grown in an infinitely diluted sys-
tem by the addition of single particles to a grow-
ing germ. Although a large number of applica-
tions (electrical breakdown, electrodeposition,
biological growth patterns, etc.) have been found
for this model, attempts have been made to de-
velop more realistic aggregation models, in partic-
ular for colloidal systems. Thus, the cluster-cluster

765
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 766
766 RIZZI, STOLL, SENESI, AND BUFFLE
aggregation (CCA) model was introduced by
Kolb et al. (1983) and Meakin (1983). At the ini-
tial stage, according to this model, elementary par-
ticles are distributed randomly in a periodic box.
These particles are then chosen randomly and
moved in all space directions in order to mimic
Brownian motion.When they come into a neigh-
boring position, particles stick together to form
clusters that are able to diffuse and stick together
to form larger structures. An important result of
the CCA model is that the aggregate morphology
can be described using fractal geometry concepts.
Fractal geometry is able to describe an in-
credibly broad range of structures and processes
occurring in nature (Mandelbrot, 1982). The ap-
plication of fractal geometry has recently become
popular and is now used successfully in the study
of the influence of aggregation mechanisms on
the structures of resulting clusters. Not only sim-
ulations or theoretical studies, but also numerous
numerical and experimental studies, which have
used techniques such as light scattering, small an-
gle neutron scattering, and image analysis, have
shown that the structure of the clusters formed
by colloid aggregation can be characterized well
by fractal dimensions or scaling laws. The values
of the fractal dimension, which describes the spa-
tial mass distribution within an aggregate, have
been found to depend on whether the aggrega-
tion is controlled by the diffusion rate of the clus-
ters (diffusion limited aggregation, DLA) or by
their chemical reactivity (reaction limited aggre-
gation, RLA) (Family et al., 1985; Meakin et al.,
1985). An important parameter related to the
chemical reactivity of the different species to
form permanent bonds is the sticking probability
Po. The value of Po is equal to one for a DLA
process and is between zero and one for a RLA
process (Stoll and Buffle, 1995). When the prob-
ability of attachment approaches unity, very ten-
uous structures are formed, which are representa-
tive of a DLA process. This morphology can be
explained by the rapid attachment of colloidal
particles either to other free colloidal particles,
with which they collide, or to the aggregate sur-
face; thus, they do not enter into the pre-existing
aggregates. When the sticking probability ap-
proaches zero (RLA process), colloidal particles
need to collide many times before sticking, which
allows them to enter into pre-existing aggregates
and thus increase their density and compactness.
Humic substances (HS) are typical soil col-
loids characterized by a physical and chemical
heterogeneous nature that derives from (a) the
absence of a discrete structural organization at
SOIL SCIENCE
both the molecular and supramolecular level; (b)
the wide variety of sizes and shapes assumed in
the solid and colloidal states; (c) the occurrence of
complex aggregation and dispersion phenomena
in aqueous media; and (d) the various degrees of
irregularity and roughness of the exposed sur-
faces (Hayes et al., 1989a and b; Swift, 1989).
These properties play an important role in deter-
mining the physical, chemical, and biological re-
activity of HS towards mineral surfaces, metal
ions, organic chemicals, plant roots, and microor-
ganisms in soil (Senesi and Boddy, 2002).
Fractal geometry represents a useful means for
achieving a quantitative description of the physical
and chemical behavior and properties of natural
HS, including soil HS (reviewed by Senesi, 1994,
1996, 1999). The major contribution that fractal
geometry can make to the understanding of the
two main fractions of HS, humic acids (HAs) and
fulvic acids (FAs), is in a general, simple, and suc-
cinct representation of their complex structures
through the parameterization by a single number,
that is, the fractal dimension D (Österberg and
Mortensen, 1992, 1994; Österberg et al., 1994;
Rice and Lin, 1993, 1994; Ren et al., 1996; Senesi
et al., 1996, 1997; Tombácz et al., 1997).
Österberg and Mortensen (1992) first re-
ported that small-angle neutron scattering from
HA solutions obeys a power law, revealing that
these materials in solution may have a fractal
structure. Atomic force microscopy (AFM) data
obtained in a succeeding study (Ikai and Öster-
berg, 1996) were consistent with previous data in
suggesting fractal formation and a CCA process
for HAs analyzed in similar conditions. Further,
these authors emphasized that caution should be
exercised before any final conclusion based on a
study performed in the dry state might be drawn
about the possible occurrence of cluster-cluster
interactions in HA solutions.
Transmission electron microscopy (TEM) has
been used in the past to study soil HAs and FAs
(e.g., Stevenson and Schnitzer, 1982; Chen and
Schnitzer, 1989; Schnitzer, 1991), but high reso-
lution observations of colloidal sized HAs and
FAs have been limited not only by the poor elec-
tron density of HS, which makes visualization by
TEM difficult, but also by the inadequacy of the
techniques used for sample preparation. The
preparation of samples for microscopic observa-
tion (Buffle and Leppard, 1995a and b) needs to
take several aspects into consideration. These in-
clude: the preservation of HA conformation; the
absence of induced coagulation as a result of sam-
ple preparation; and the absence or limitation of
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 767
VOL. 169 ~ NO. 11 FRACTAL PROPERTIES AND AGGREGATION OF HUMIC ACIDS
microbial activity. Significant advances made in
the preparation (Perret et al., 1991) and observa-
tion (Perret et al., 1995;Wilkinson et al., 1995) by
TEM of environmental samples have brought
about marked reduction in the potential for arti-
fact production and have increased resolution by
using hydrophilic resins and small concentrations
of ionic stains. In this work, the same preparation
technique has been used to enable the non-per-
turbing observation of HAs. An extended de-
scription of the limitations resulting from the op-
erating principles of the TEM technique, sample
preparation technique, precautions to minimize
potential artifacts, and any possible pitfalls, are de-
scribed in Perret et al. (1991) and Wilkinson et al.
(1999) and are beyond the scope of this paper.
The objective of this work is to combine
TEM analysis, fractal geometry principles, and
cluster-cluster aggregation modeling concepts to
find new insights into the formation kinetics and
structural characteristics of HA aggregates as a
function of HA source, solution pH, and equili-
bration time.
MATERIALS AND METHODS
Four HA samples were used in this work,
three of which were obtained from the HA Stan-
dard and Reference Collection established by the
International Humic Substances Society (IHSS).
These samples originated from a mollic horizon
of the Elliot silt loam soil, Joliet, Illinois (IHSS
code, 1S 102H); the Summit Hill grassland soil,
New Zealand (IHSS code, 1R 106H); and the
Pahokee peat, Okachobee, Florida (IHSS code,
1S 103H). The fourth HA sample was isolated
from a typical alluvial soil.
The HA samples were extracted and purified
according to the procedure designed specifically
by the IHSS for the preparation of the HA and
FA standard and reference collection (Swift,
1996). The HA samples were obtained in ho-
mogenized, freeze-dried H
form. The ash con-
tent was in all cases below 1.0%. Solutions of each
HA were prepared in deionized distilled water at
a concentration of 1 mg L1, and the pH was ad-
justed immediately to a value of 6 or 3 by adding
small amounts of HCl or NaOH. The HA solu-
tions were then equilibrated at 4 C for 4, 10, and
28 h. After the equilibration time, the pH re-
mained constant in all cases.
High-resolution TEM observations were
made with a TEM-ZEISS EM 109 apparatus. The
TEM copper grids were first coated with a collo-
dion film (10–50 nm thickness), and a uniform
767
layer of amorphous carbon (5–10 nm thickness for
a good transparency) was then deposited by evap-
oration (Hayat, 1985). The carbon layer allows
good thermal and electrical conduction of elec-
trons under the beam of the microscope, whereas
the collodion film serves as an optimal planar sup-
port for the carbon film (Perret et al., 1991).
Four TEM copper (200 mesh, collodion, and
carbon-coated) grids were placed on each ce-
ramic plaque and covered with 5 L of a hy-
drophilic resin (nanoplast). Each plaque was then
inserted in a 4-mL ultracentrifuge tube. Succes-
sively, each HA solution (2.5 mL) was prepared as
described above and added to each tube immedi-
ately before centrifugation (24 h, 124,000 g) onto
the grids. This procedure allowed the embedding
of HA particles into the thin layer of resin. Fi-
nally, the grids were air dried for 24 h before
TEM observation. The resin-embedded material
was shown not to be modified during this step
(Perret et al., 1991), because of the peculiar hy-
drophilic properties of the nanoplast resin. Thus,
four grids for each of the eight HA solutions
(four HA samples at two pH values) were ana-
lyzed by TEM (quadruplicate experiment).
Although particles smaller than about 10 nm
are not collected by ultracentrifugation, this does
not represent an important limitation for the de-
termination of fractal dimensions provided that
these cover at least two orders of magnitude. This
occurs for HA aggregates whose sizes range from
20 to 2000 nm, as can be seen in TEM im-
ages shown in Figs. 1 2, and 3 and discussed later.
According to the procedure described by For-
rest and Witten (1979), TEM images were digi-
tized to obtain the numbers and coordinates of
HA aggregates. At least eight TEM images taken
from the four grids were analyzed for each HA so-
lution to permit statistical analysis of the results.An
appropriate processing software (MOCHATM)
was used to count the total number of pixels (M)
within each aggregate in each TEM image and to
measure the length (l) corresponding to the major
distance between pixels belonging to the same ag-
gregate (Weitz and Oliveria, 1984; Tence et al.,
1986). The following geometric scaling relation-
ship (power law)
M
lD (1)
between the aggregate mass M and aggregate
length l was used to calculate the mass fractal di-
mension D of the aggregates, based on their TEM
pictures (Forrest and Witten, 1979; Weitz and
Oliveria, 1984; Tence et al., 1986; Zhang and Buf-
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 768

768 RIZZI, STOLL, SENESI, AND BUFFLE SOIL SCIENCE

Fig. 1. TEM images of centrifuged Alluvial soil humic acid at pH 3 (a) and pH 6 (b) after 4 h equilibration, and cen-
trifuged Summit Hill soil humic acid at pH 3 (c) and pH 6 (d) after 10 h equilibration.

fle, 1996). The slope of the corresponding log-log
plot of M vs. l provides the value of the fractal di-
mension D. This method allows a good statistical
estimate of D for the whole of aggregates that are
present in the TEM pictures, i.e., for a large num-
ber of structures.
An alternative method, which consists of
choosing a given position in the central part of
each aggregate and then considering larger and
larger circular areas centered on this position in
order to determine the variation of the aggregate
mass as a function of the size of the circles, was
not attempted here because of the finite size and
relatively small number of large aggregates. In-
deed, the latter method provides a good estimate
of D only when the center of the aggregate can
be analyzed far away from the edges, and this is
possible only for large aggregates and when a
cutoff function is used to interpret the resulting
curves.
However, both methods encounter difficul-
ties when they are applied to aggregates having
fractal dimensions larger than two because the
fractal dimension of their TEM projection is
bounded by two ( Jullien et al., 1994). Thus,
methods based on TEM image analysis are valid
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 769

VOL. 169 ~ NO. 11 FRACTAL PROPERTIES AND AGGREGATION OF HUMIC ACIDS 769

Fig. 2. TEM images of centrifuged peat humic acid at pH 3 (a) and pH 6 (b) after 10 h equilibration, and centrifuged
Elliot soil humic acid at pH 3 (c) and pH 6 (d) after 10 h equilibration.

only for aggregates with a fractal dimension
lower than the critical value of 2, in agreement
with current mathematics for which the dimen-
sion of a projected object is always limited to 2.
It should be noted here that microscopy tech-
niques yield a two-dimensional (2D) projection
of the object under observation.Various methods
were proposed regarding how to extract the frac-
tal dimension of aggregates by electron mi-
croscopy and the sources of errors analyzed
(Tence et al., 1986; Jullien et al., 1994; Vormoor,
2001). These methods were compared with com-
puter simulations of the cluster-cluster aggrega-
tion model, where 2D projections of 3D aggre-
gates were considered. It was demonstrated that
the “scaling method”, which consists of integrat-
ing the number of pixels of 2D projections into
the entire aggregate and in relating the obtained
value to the measure of the size of the aggregate,
provides fractal dimensions that are consistent
with their counterparts in 3D. Thus, the projec-
tion of a 3D aggregate to 2 dimensions does not
affect its fractal dimension value if calculated by
the scaling method used here for values of D
smaller than 2. For D values larger than 2, this
method cannot provide more than an effective
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 770
770 RIZZI, STOLL, SENESI, AND BUFFLE
Fig. 3. TEM images of centrifuged Alluvial soil (a) and Elliot soil (b) humic acids at pH 6 after 28 h equilibration.
3D fractal dimension equal to 2. This method
was also used successfully to investigate the frac-
tal and projected structure properties of soot ag-
gregates (Koylu et al., 1995).
RESULTS AND DISCUSSION
Some representative TEM images of HA solu-
tions as a function of pH and equilibration times
are presented in Figs. 1 to 3. Aggregates are appar-
ent after an equilibration time of 4 h for the allu-
vial soil HA and after 10 h for the other three HAs.
The aggregation process is more pronounced
at pH 3 than at pH 6, with a large number of
small, open aggregates present at pH 6 and a small
number of large, compact aggregates at pH 3.Ac-
cording to these observations, and as a first ap-
proximation, the sticking probability Po at pH 3
should be larger than at pH 6 and is possibly close
to unity. This hypothesis is consistent with the
fact that at pH 3, protonation of most acidic
groups of HA occurs with reduction of repulsive
negative charges on HA macromolecules and,
subsequently, induces an aggregation process be-
tween HA particles by weak, short-range attrac-
tive forces (Verwey and Overbeck, 1984). The
aggregation process would thus occur quickly
and be of the diffusion-limited (DLA) type, that
is, controlled only by diffusion processes and not
by chemical reactivity (Kolb et al., 1983; Stoll and
Buffle, 1995).
SOIL SCIENCE
The TEM images obtained at pH 6 after 4
to 10 h equilibration (Figs. 1b, 1d, 2b, 2d) sug-
gest a slow aggregation rate between HA parti-
cles. This effect can be ascribed to a value of Po
lower than that hypothesized at pH 3, possibly
due to a greater number of deprotonated acidic
groups on HA particles with a predominance of
repulsive electrostatic forces over attractive
forces. If the thermal motion is not sufficient to
overcome the electrostatic repulsive barrier, HA
particles are expected to bounce on each other
several times before sticking together, thus slow-
ing down the aggregation rate. In these condi-
tions the aggregation process would be of the
reaction-limited (RLA) type, that is, mainly
controlled by the aggregate chemical reactivity.
A comparison of average aggregate sizes ob-
tained from TEM images at pH 6 for the alluvial
soil HA after 4 h (Fig. 1b) and for the Elliot soil
HA after 10 h equilibration (Fig. 2d) with those
obtained from the corresponding TEM images
after 28 h equilibration (Fig. 3a, 3b) demonstrates
qualitatively that aggregation proceeds with in-
crease in the equilibration time. These observa-
tions are consistent with the presence of a slow
aggregation process of the RLA type at pH 6
( Jullien and Botet, 1987).
A similar conclusion was reached in previous
studies (Österberg and Mortensen, 1992, 1994;
Österberg et al., 1994) of soil HA solutions at pH
 5 analyzed by small angle neutron scattering
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 771

VOL. 169 ~ NO. 11 FRACTAL PROPERTIES AND AGGREGATION OF HUMIC ACIDS 771

Fig. 4. Log-log plot of M vs l for Alluvial soil humic acid at pH 3 (a) and pH 6 (b) after 4 h equilibration and at pH 6
after 28 h equilibration (c), and for Summit Hill soil humic acid at pH 3 (d) and pH 6 (e) after 10 h equilibration.
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 772

772 RIZZI, STOLL, SENESI, AND BUFFLE SOIL SCIENCE

Fig. 5. Log-log plot of M vs l for peat humic acid at pH 3 (a) and pH 6 (b) after 10 h equilibration, and for Elliot soil
humic acid at pH 3 after 10 h equilibration (c) and pH 6 after 28 h equilibration (d).

(SANS) after 48 h at 4 C and 11 C, and after 60
h at 22 C.
Quantitative TEM image analysis of the four
HAs studied in any condition indicates the exis-
tence of a power law between the aggregate mass
M and its characteristic length l. These results im-
ply that the HAs studied can be described as frac-
tal objects in any condition used in this work and
that their fractal dimension D can be obtained di-
rectly from the slope of the experimental straight
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 773
VOL. 169 ~ NO. 11 FRACTAL PROPERTIES AND AGGREGATION OF HUMIC ACIDS
lines of log M vs. log l (Figs. 4 and 5 refer to one
out of four replicates for each sample). Table 1 lists
the mean values  standard errors of D obtained
at pH 3 and pH 6 after various equilibration times
for the four replicates of each HA solution.
At pH 3, aggregates are characterized by frac-
tal dimensions that are very similar to the D value
of a DLA-type process (D  1.78) (Family et al.,
1985; Meakin et al., 1985). At pH 6, after 4 h
equilibration for the alluvial soil HA (Fig. 1b) and
10 h equilibration for the Summit Hill soil HA
(Fig. 1d) and peat HA (Fig. 2b), the aggregates
appearing in TEM images are characterized by
open, rather elongated structures that are consis-
tent with the small fractal dimension values ob-
tained by image processing analysis (Table 1). In
regard to Elliot soil HA at pH 6 and after 10 h
equilibration (Fig. 2d), the fractal dimension was
not calculated because images were obtained in
nonideal conditions, and, consequently, the im-
age analysis software was not able to distinguish
between dispersed and aggregated HA.
The experimental D values of the various
HAs at pH 6 after 4 h or 10 h equilibration are
smaller than the theoretical D value of RLA- and
even DLA-type aggregates (D  2.09) (Family et
al., 1985; Meakin et al., 1985). However, the allu-
vial soil HA at pH 6 after 28 h equilibration is
characterized by a fractal dimension much larger
than that measured after 4 h (Table 1) and is, thus,
closer to the theoretical value of an RLA-type
process. This result is consistent with the presence
of aggregates larger and more compact (Fig. 3a)
than those present after 4 h (Fig. 1b).A similar ef-
TABLE 1
Mean values  standard errors of the fractal dimensions
of the humic acids (HAs) studied at various
pH values and equilibration times
Sample
pH 3 pH 6
Equilibration time
Alluvial soil HA
4h 1.77  0.05 1.52  0.08
28h † 1.95  0.09
Summit Hill soil HA
10h 1.79  0.03 1.64  0.03
Peat HA
10h 1.79  0.05 1.55  0.07
Elliot soil HA
10h 1.72  0.01 ‡ characterized by a D value close to that of RLA.
28h † 1.80  0.07
†Not determined.
‡Image analysis not possible because the system is very
dispersed.
773
fect can be observed for the Elliot soil HA (Fig.
3b vs. Fig. 2d).
The differences observed between experi-
mental D values for alluvial soil HA, and espe-
cially Elliot soil HA, at pH 6 after 28 h equilibra-
tion, and the theoretical D value for an RLA-type
process may reflect a heterogeneous aggregation
mechanism, e.g., flocculation involving the pres-
ence of filamentous structures (often invisible by
TEM because of small electron density) between
large clusters (Wilkinson et al., 1995). From a
structural point of view, the role of these fibrils
would become predominant at pH 6. Since HAs
usually consist of semiflexible structures bearing
significant amounts of ionizable groups, their
conformation may change dramatically with
charge density, i.e., as a function of pH and/or
ionic strength, passing from dense coil conforma-
tions to extended conformations when pH in-
creases (Swift, 1989). Furthermore the observed
increase of the fractal dimension with time at pH
6 may be the result of the occurrence of simulta-
neous aggregation and fragmentation processes
followed by internal restructuring, as shown by
numerical models (Shih et al., 1987).
CONCLUSIONS
In this paper, application of quantitative TEM
image analysis to environmental HA solutions at
realistically low concentration (1 mg L1) pro-
vides new information about the kinetics of ag-
gregation processes occurring for HA systems in
various conditions.
TEM image analysis shows the existence of a
power law between the mass and the characteris-
tic length of HA aggregates measured at various
pH values and equilibration times, which implies
an underlying fractal structure for HAs.
The fractal dimensions of HAs calculated in
various conditions allows a quantitative descrip-
tion of the HA aggregate structures. At pH 3 in
particular, HA clusters exhibit large and compact
structures that are, in any case, consistent with an
underlying cluster-cluster DLA-type process. At
pH 6, the HA aggregates at short times (about 10
h) are characterized by open, elongated struc-
tures, whose fractal dimension is smaller than that
typical of an RLA-type process. However, with
an increase in the equilibration time, larger and
more compact structures are formed that are
Large aggregates are likely not formed by addi-
tions of single HA particles to existing aggregates
but rather by collisions of HA aggregates. Aggre-
gate attachment and collision efficiency are ap-
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 774
774 RIZZI, STOLL, SENESI, AND BUFFLE
parently controlled by solution pH, which affects
the electric charge on the colloidal material.
The difficulty of studying very complex ma-
terials such as HAs and the novelty of the present
approach help us understand that the conclusions
presented in this paper are simplified as a result of
the intrinsic limitations of the applied technique
and the sample size, which are not suited for great
precision estimates of fractal dimensions. A more
precise description of HA aggregation processes
would require a complete electrochemical and
hydrodynamic characterization of HAs as well as
an approach based on CCA models that also con-
sider aggregate fragmentation and restructuring
processes.
AKNOWLEDGMENTS
The authors thank Dr. K.J. Wilkinson for
useful suggestions. The senior author also ex-
presses her appreciation to the European Envi-
ronmental Research Organization (EERO) and
the Swiss National Foundation for partial sup-
port of her work at the University of Geneva.
REFERENCES
Buffle, J., and G. G. Leppard. 1995a. Characterization of
aquatic colloids and macromolecules. 1. Structure
and behavior of colloidal material. Environ. Sci.
Technol. 29:2169–2175.
Buffle, J., and G. G. Leppard. 1995b. Characterization of
aquatic colloids and macromolecules. 2. Key role of
physical structures on analytical results. Environ.
Sci. Technol. 29:2176–2184.
Chen, Y., and M. Schnitzer. 1989. Sizes and shapes of
humic substances by electron microscopy. In Hu-
mic Substances II: In Search of Structure. M.H.B.
Hayes, P. MacCarthy, R.L. Malcom, and R.S. Swift
(eds.). Wiley, New York, pp. 621–638.
Family, F., P. Meakin, and T. Vicsek. 1985. Cluster size
distribution in chemically controlled cluter-cluster
aggregation. J. Chem. Phys. 83:4144–4148.
Forest, S., and T.Witten. 1979. Long-range correlations
in smoke-particle aggregation. J. Phys. A. 12:L109–
117.
Hayat, M. A. 1985. Basic techniques for Transmission
Electron Microscopy. Academic Press, San Diego.
Hayes, M. H. B., P. MacCarthy, R. L. Malcolm, and
R. S. Swift. 1989a. The search of structure: Setting
the scene. In Humic Substances II: In Search of
Structure. M.H.B. Hayes, P. MacCarthy, R.L. Mal-
com and R.S. Swift (eds.). Wiley, New York, pp.
3–30.
Hayes, M. H. B., P. MacCarthy, R. L. Malcolm, and R.
S. Swift. 1989b. The search of structure: Setting the
scene. In Humic Substances II: In Search of Struc-
ture. M.H.B. Hayes, P. MacCarthy, R.L. Malcom
SOIL SCIENCE
and R.S. Swift (eds.). Wiley, New York, pp. 689–
730.
Ikai, A., and R. Österberg. 1996. Atomic force mi-
croscopy of humic acids. Scanning Microsc. 10:
947–951.
Jullien, R., and R. Botet. 1987. Aggregation and Frac-
tal Aggregates. World Scientific, Singapore.
Jullien, R., R. Thouy, and F. Ehrburger-Dolle. 1994.
Numerical investigation of 2-dimensional projec-
tions of random fractal aggregates. Phys. Rev. E. 50:
3878–3882.
Kolb, M., R. Botet, and R. Jullien. 1983. Scaling of ki-
netically growing clusters. Phys. Rev. Lett. 51:
1123–1126.
Koylu, U. O., G. M. Faeth, T. L. Farias, and M. G. Car-
valho. 1995. Fractal and projected structure proper-
ties of soot aggregates. Combust. Flame 100:621–
633.
Mandelbrot, B. B. 1982. The Fractal Geometry of Na-
ture. Freeman, San Francisco.
Meakin, P. 1983. Formation of fractal clusters and net-
works by irreversible diffusion-limited aggregation.
Phys. Rev. Lett. 51:1119–1122.
Meakin, P., T. Vicsek, and F. Family. 1985. Dynamic
cluster-size distribution in cluster-cluster aggrega-
tion: Effects of cluster diffusivity. Phys. Rev.
B–Condensed Matter 31:564–570.
Österberg, R., and K. Mortensen. 1992. Fractal dimen-
sion of humic acids, a small angle neutron scatter-
ing study. Eur. Biophys. J. 21:163–167.
Österberg, R., and K. Mortensen. 1994. Fractal geom-
etry of humic acids. Temperature-dependent
restructuring studied by small-angle neutron scat-
tering. In Humic Substances in the Global Envi-
ronment and Implications on Human Health. N.
Senesi and T.M. Miano (eds.). Elsevier,Amsterdam,
pp. 127–132.
Österberg, R., L. Szajdak, and K. Mortensen. 1994.
Temperature-dependent restructuring of fractal
humic acids:A proton-dependent process. Environ.
Int. 20:77–80.
Perret, D., G. G. Leppard, M. Muller, N. Belzile, R. De
Vitre, and J. Buffle. 1991. Electron microscopy of
aquatic colloids: Non perturbing preparation of
specimens in the field. Water Res. 25:1333–1343.
Perret, D., C-P. Lienemann, and D. Mavrocordatos.
1995. EELS-ESI identification of heterogeneous
suspensions of aquatic microparticles. Microsc. Mi-
croanal. Microstruct. 6:41–51.
Ren, S. Z., E. Tombácz, and J. A. Rice. 1996. Dynamic
light scattering from power-law polydisperse frac-
tals: Application of dynamic scaling to humic acid.
Phys. Rev. E. 53:2980–2983.
Rice, J., and J. S. Lin. 1993. Fractal nature of humic ma-
terials. Environ. Sci. Technol. 27:413–414.
Rice, J., and J. S. Lin. 1994. Fractal dimension of humic
materials. In Humic Substances in the Global Envi-
ronment and Implications on Human Health. N.
Senesi and T.M. Miano (eds.). Elsevier,Amsterdam,
pp. 115–120.
SS_11-04_154998_Rizzi 11/11/04 5:06 PM Page 775
VOL. 169 ~ NO. 11 FRACTAL PROPERTIES AND AGGREGATION OF HUMIC ACIDS
Schnitzer, M. 1991. Soil organic matter—The next 75
years. Soil Sci. 151:41–58.
Senesi, N. 1994. The fractal approach to the study of
humic substances. In Humic Substances in the
Global Environment and Implications on Human
Health. N. Senesi and T.M. Miano (eds.). Elsevier,
Amsterdam, pp. 3–41.
Senesi, N. 1996. Fractals in general soil science and in
soil biology and biochemistry. In Soil Biochem-
istry, Vol. 9. G. Stotzky and J.M. Bollag (eds.).
Dekker, New York, pp. 415–472.
Senesi, N. 1999. Aggregation patterns and macromole-
cular morphology of humic substances: A fractal
approach. Soil Sci. 164:841–856.
Senesi, N., and L. Boddy. 2002. A fractal approach for
interactions between soil particles and microorgan-
isms. In Interactions of Soil Particles and Microor-
ganisms and Their Impact on the Terrestrial Envi-
ronment. P.M. Huang, J.M. Bollag and N. Senesi
(eds.). IUPAC Series on Analytical and Physical
Chemistry of Environmental Systems,Vol. 8.Wiley,
New York, pp. 41–83.
Senesi, N., F. R. Rizzi, P. Dellino, and P. Acquafredda.
1996. The fractal dimension of humic acids as a
function of pH and time. Soil Sci. Soc. Am, J. 60:
1773–1780.
Senesi, N., F. R. Rizzi, P. Dellino, and P. Acquafredda.
1997. Fractal humic acids in aqueous suspensions at
various concentration, ionic strengths, and pH val-
ues. Colloids Surf. A—Physicochem. Eng. Aspects
127:57–68.
Shih, W. Y., I. A. Aksay, and R. Kikuchi. 1987. Re-
versible-growth model: Cluster-cluster aggregation
with finite binding energies. Phys. Rev. A. 36:
5015–5019.
Stevenson, I. L., and M. Schnitzer. 1982. Transmission
electron microscopy of extracted fulvic and humic
acids. Soil Sci. 133:179–185.
Stoll, S., and J. Buffle. 1995. Computer simulations of
colloids and macromolecules aggregate formation.
Chimia 49:300–307.
Swift, R. S. 1989. Molecular weight, size, shape, and
charge characteristics of humic substances: Some
775
basic considerations. In Humic Substances II: In
Search of Structure. M.H.B. Hayes, P. MacCarthy,
R.L. Malcom, and R.S. Swift (eds.). Wiley, New
York, pp. 449–465.
Swift, R. S. 1996. Organic matter characterization. In
Methods of Soil Analysis, Part 3, Chemical Meth-
ods. D.L. Sparks (ed.). ASA, Madison, WI, pp.
1011–1069.
Tence, M., J. P. Chevalier, and R. Jullien. 1986. On the
measurement of the fractal dimension of aggre-
gated particles by electron microscopy: Experimen-
tal method, corrections and comparison with nu-
merical models. J Phys. 47:1989–1998.
Tombácz, E., J. A. Rice, and S. Z. Ren. 1997. Fractal
structure of polydisperse humic acid particles in
solution studied by scattering methods. ACH–
Models Chem. 134:877–888.
Verwey, E. J. W., and J. Th. G. Overbeck. 1984. Theory
of the Stability of Lyophobic Colloids. Elsevier,
Amsterdam.
Vormoor, O. 2001. Grayscale projection of finite size
fractal clusters. Eur. Phys. J. E. 5:551–556.
Weitz, D. A., and M. Oliveria. 1984. Fractal structures
formed by kinetic aggregation of aqueous solu-
tions. Phys. Rev. Lett. 52:1433–1437.
Wilkinson, K. J., S. Stoll, and J. Buffle. 1995. Character-
ization of non-colloid aggregates in surface waters:
Coupling transmission electron microscopy stain-
ing techniques and mathematical modelling. Frese-
nius J. Anal. Chem. 351:54–61.
Wilkinson, K. J., E. Balnois, G. G. Leppard, and J. Buf-
fle. 1999. Characteristic features of the major com-
ponents of freshwater colloidal organic matter re-
vealed by transmission electron and atomic force
microscopy. Colloids Surf. A: Physicochem. Eng.
Aspects 155:287–310.
Witten, T. A., and L. M. Sanders. 1981. Diffusion-lim-
ited aggregation: A kinetic critical phenomenon.
Phys. Rev. Lett. 47:1400–1403.
Zhang, J., and J. Buffle. 1996. Multi-method determi-
nation of the fractal dimension of hematite aggre-
gates. Colloids Surf. A–Physicochem. Eng.Aspects.
107:175–187.