Professional Documents
Culture Documents
net/publication/300561563
CITATIONS READS
7 296
4 authors, including:
Some of the authors of this publication are also working on these related projects:
Pattern formation and mineral self-organization in highly alkaline natural environments (PROMETHEUS; ERC-2341) View
project
All content following this page was uploaded by Juan Manuel Garcia-Ruiz on 14 May 2016.
Abstract. Manganese oxide pattems known as "pyrolusite dendrites" are explained as the
result of the mineral record of flow structures in porous media. This interpretation is
supported by a) ftactal characterization, b) Mn profiles across the mineral pattem and the
matrix rocks, c) structures reminiscent offlow pattem observed at the scale afilie grain
size ofthe matrix rocks, d) the lack oflong-range arder ofthe manganese oxide particles
and e) the existence of clays and quartz grains of colloidal size intimately linked to the
manganese particles. This interpretation also explains other ftactal and non-ftactal
pattems accompanying the beautiful treelike ftactal forms associated with these
manganese oxide pattems.
Introduction
10.4 a
11,0l
9.8
9.2
8.6
Z 8.0
E 7.4 ()7
.6[)
6.8
62
5.6
5.0
0.0 0.3 0.6 0.9 12 1.5 1.8 2.1 2.4 2.7 3.0
In I
' . " . " " "
10.48
9.98 t b
o
8.98
o
8.48
948 ¡ ,I'.'r,\; '. ó
Z 7.98
E
7.48
6.98
6.48
5.98
5.48
0.0 03 0.6 0.9 1.2 1.5 18 2.1 2.4 2.7 3.0
In I
Fig. 1. Two digitized images of manganese dendritic pattems and their respective fractal
dimension obtained by fue box-counting method. The number ofboxes N(l) of sirle length
1 (pixels) is plotted versus 1. Sample a) is a dendritic pattem observed between two
laminae of [me-grained calcarenite from a flysh facies. Sample b) was found on a cracked
surface of a limestone. Deviation from fue straight line for low values of 1 in Fig lb are
interpreted as the effect of a inner cut-off at 1= 4.5 pixels.
mechanism. In ibis chapter we present data supporting the idea that manganese
and iron dendrites are the mineral record of flow instabilities. The driving force
for the flow may arise from any type of geological pressure gradient (for instance
in a hydrothermal field) but in the absence of such an external driving force for
injection, we propase a self-fed mechanism operating during early diagenesis and
we discuss the geological plausibility of ibis process.
TI Fonnation ofManganese Dendrites as the Mineral Record ofFlow Structures 309
Pattern Characterization
Fig. 2. High resolution electron microscopy view ofMn oxide particles showing the lack
of long-range order. The small particle showing lattice structure is a clay mineral with a
10 Á interplanar d-spacing.
slow exploration (static record during ten seconds for every 0.02°). When
attempting to obtain the degree of crystallinity using the line profile method
(Sanchez-Navas 1989) we found a crystallinity degree below the range of
resolution for that method. Therefore, in arder to obtain the coherent domain
size we used high resolution transmission electron microscopy. For this
purpose a suspension of the manganese mínerals was made and part of the
salid was dispersed on a holey carbon grid. A transmission electron
mícroscopeoperatedat 100kV, with an objectiveapertureof 90 !lIDwas used
in arder to add 000 transmítted beam and diffusion haloes. Low beam current
and mínimum exposure were used to avoid beam damage. The study showed
the absence of even short range three-dimensional periodical structure and the
clear colloidal nature of the Mn 'delimites (Figure 2). This means that the
kinetics of fue precipitation process has been fast and irreversible enough to
I
The Fonnation ofManganese Dendrites as the Mineral Record ofFlow Structures 311
Crystalline dendritic patterns are ordered fractals where the anisotropy of the
crystal structure governs the growth process. On the contrary, the so-called
random dendritic fractals are patterns where the fluid structure created in the
growth environment dominates crystal anisotropy. Therefore, it is a requirement
for the formation of the random dendrites that the kinetics of the phase transition
be rastel than the kinetics of the pattern formation and consequently, the
precipitation process must be a highly irreversible one occurring at conditions far
from equilibrium. It follows from this consideration that the precipitate formed
must be a low crystallinity product and that the whole growth pattern must lack
any geometrical relationships derived from the crystal structure. According to
observational facts 1) 3) and 4), the so-called manganese delimites belong to this
group of random fractals. Beyond systematic terminology, this distinction is
important in building a model of their morphogenetical mechanisms.
According to observational fact 1), some of the patterns displayed by
manganese delimites have a visual appearance and a fractal dimension which are
typical of those patterns derived from the Laplace equation. Today we know
312 García-Ruiz, 1M. et al.
several chemical and physical Laplacian growth processes (Vicsek 1989) but on1y
two of them are relevant to the formation of the Mn dendrites. The first of these
processes leading to two-dimensional objects with fractal dimension around 1.7
are irreversible growth processes governed by the diffusional control of the
nutrient phase, the so-called diffusion-limited aggregation (DLA) model (Witten
& Sander 1983). The second one is the viscous fingers (VF) process (Van
Damme 1989, Feder 1988, Vicsek 1989) which is known to produce DLA-type
parterns when the less viscous fluid invades a more viscous and non-Newtonian
one. Despite the fact that one of the most important findings of the new physics
of partero formation is that both DLA and VF display the same fractal dimension
and that they can be described by the same formal equations (Daccord et al. 1986,
Paterson 1984), it is important to note that from a phenomenological viewpoint,
DLA and VF are c1early difIerent. As suggested by Van Damme, to reveal the
origin of the pyrolusite dendrites reduces to the choice of one of these growth
phenomena.
unstable interface between the two fluids bifurcates iteratively and therefore a
branched treelike structure reminiscent of manganese dendrites is obtained. The
main parameter controlling the interfacial pattern is the critical wavelength
b2
A= -*-, r
(jJ2 vV
where b and 'UJare the thickness and the width of the cell respectively, /l is the
difference in fluid viscosity, y the surface tension and V the velocity of the
injected fluido Today it is well known that viscous fingering have a fractal
behaviour when either the pushed fluid is a non-Newtonian one or it occurs luto
porous media. The type of fractal obtained depends on the applied flow cates and
the surface tension between the fluids (Lenormand & Daccord 1988). In porous
media, for very low values of V (or according to the previous equation, for larger
cell thickness) the fluid provokes an invasive percolation pattern with a
calculated fractal dimension D = 1.82 (Stauffer 1985). This pattern changes for
larger V values and after a crossover it becomes a fractal structure with a
D = 1.70, visually rerniniscent of DLA patterns (Lenormand & Daccord 1988).
This situation accounts very well for the geometrical properties of manganese
dendrites and their related structures. First, as described above, the fractal
dimension of typical Mn dendrites with a DLA-like morphology matches the one
expected for viscous fingers at high injection cates. Because of their beauty, Mn
dendrites are by far the best known patterns, but as stated in observational fact
nol, there are many other accompanying deposits which show other kinds of
patterns. Among them, invasive percolation and other patterns with fractal
dimensions and visual geometry ilifficult to explain by aggregation under
diffusional control are the most developed. These accompanying non-dendritic
patterns also can be explained by the viscous fingering mechanism. For a given
value of the injection flow, the parameter b may adopt small local changes but
large enough to produce the well-known change of regime provoked by variations
in V. Thus, the VF mechanism explains not only the beautiful treelike fractal
growth, but also the fractal and non-fractal patterns accompanying them, which
constitute the majority of manganese oxides. Unlike a DLA-type mechanism, the
viscous fingering approach is consistent with the manganese profiles across the
rack matrix containing the dendrites (observational fact nOS).Morec:vu, the SEM
views (Figure 3) of Mn dendrites suggest the existence of flo\.v patterns
surrounding the quartz and calcite grains of the rack and high resolution electron
microscopy shows that the Mn oxides closely adhere to clay particles a!¡d quartz
grains with a colloidal size which also suggests that a high viscosity colloidal
fluid was pushed by the Mn2+-rich injected fluido
The above discussion indicates a genetic mechanism for the formation of
manganese dendrites based on the mineralization of fluid structures. These fluid
structures occur during the invasion of sedimentary discontinuities or in general,
low attachment surfaces, by a fluid rising through cracks. The cracks (or
sedimentary laminations) are the conduits for Mn2+ and Fe2+, and the surfaces
316 García-Ruiz, J.M. et al.
oí low cohesion associated with them allow the development oí narrow gaps
which in many cases are the ¡oei íor pattern íormation. It is thereíore reasonable
to think that the existence oí rough impermeable pairs oí quasi-parallel surfaces
with a gap between them led to the íormation oí natural Hele Shaw cells.
The most intuitive source (and probably the most general Que) oí Mn2+ and
Fe2+ is the rige oí enriched solutions through the cracks. However, even in the
absence oí such an "external" source, we propase a genetic model using a self-íed
system that could account íor those geological scenarios where the existence oí an
external source oí manganese and iron solutions is difficult to support. We start
from a sedimentary material undergoing fracturing and compaction during early
diagenesis. It íollows that the host rock body to be injected must be at least
partially cemented and must preserve some bedding discontinuities. Also, the
injected fluid must displace another oí higher viscosity. Thus, the pattern
íormation should be generated during diagenesis, when a colloidal suspension
fills the interbedding laminations and cavities oí the sedimentary materials.
Under these conditions, the injection oí a rising solution rich in Mn2+ into a
natural Hele Shaw cell with rough surfaces containing a colloidal suspension wiIl
produce either the dendritic pattern with 0=1. 70, the invasive percolation pattern
wiili 0=1.82, or an intermediate pattern somewhere in the range between these
two values (Lenormand & Daccord 1988). The width and thickness oí the gap,
the roughness oí the surfaces, the pressure oí the injection fluid and the viscosity
oí the pushed fluid are the parameters controIling the great variety oí dendritic
and non-dendritic patterns observed in field studies and extending the range oí
possible D values. From the Eh-pH diagram íor Mn and Fe oxides, sediments
containing Mn4+ and Fe3+ compounds are reduced after being buried below the
anoxic-oxic interphase, and thus enriched in Mn2+ and Fe2+. Compaction and
cracking events provoke the upward injection oí these solutions which travel
through the cracks oí the sedimentary body and invade the upper a1ka1ineand
oxygenated zone, pushing the colloidal suspension that fills the sedimentary
laminations and consequentIy forming the dendritic patterns. Under such
conditions Mn2+ will precipitate as y-MnOOH (via Mn(OH>2 Mn304 plus 6-
MnOOH) and Fe2+ as y-FeOOH (Giovanoli 1980), both in the íorm oí low
crystallinity colloidal precipitates. The Mn3+ oxides are intermediate phases that
later convert to manganese oí higher oxidation states. This mechanism oí
íormation also explains the Mn-concentration profile obtained by electron
microprobe chemical analysis (Fig. 4), which has a negative gradient towards the
host rock ayer a narrow band oí a íew micrometers and attains a zero-value
outside the precipitate (a similar behaviour was íound íor Fe3+). Finally,
preservation oí the pattern can be produced by three mechanisms working
simultaneously: a) the cementation oí the host rocks which obstructs the
intraporous diffusion oí reducing agents, b) the plugging oí the system by the
íormation oí colloidal iron oxides, and c) the exhaustion oí the source oí reducing
agents. We have simulated this mechanism in the laboratory. A reservoir was
filled with a colloidal suspension oí yFeOOH and yMnOOH oxide partic1es.
The Formation ofManganese Dendrites as fue Mineral Record ofFlow Stl1lctmes 317
Three 5 mm thick glass disks were piled and the surfaces between them coated
with tooth paste, a material made 99% of colloidal silica. The pile was immersed
in the manganese solution and then a set of fractures was provoked by impacting
a hammer on the glass píleo After a couple of minutes, dendritic patterns were
observed in the silica thin layer (Fig. 5). A difIerent chemical system waS also
tried: the colloidal silica was replaced by a suspension of day with BaO solution
and the pile was now immersed in a true solution ofFe2+ and Mn2+. Details on
the similarities and difIerences of these laboratory model and real manganese
dendrites will be published elsewhere.
Acknowledgements. This work was carried out under financial support from
CICYT Project PB92-1137 and the Junta de Andalucía. We acknowledge Tamás
Vicsek, Eugene Stanley and Michael Russell for their useful comments on this
work.
r 318 García-Ruiz,lM. et al.
References