ÐÓ°ÉÔ­´´

Ground water is a precious resource. The time it has spent filtering
through soil and rock usually ensures that this water is good to drink,
free of contamination from parasites and harmful bacteria. But during its
time underground the water can pick up a variety of elements from soil and
rocks. Some are innocuous, or even good for you, as the purveyors of mineral
water are anxious to point out. Others are harmful. Aluminium, for example,
is thought to lead to degenerative disorders of the nervous system, such
as Alzheimer’s disease.

Chemical analysis of ground water in the tropics usually reveals aluminium
at levels of less than 0.1 parts per million, the maximum recommended by
the World Health Organization. But this picture is deceptive. My colleague,
Lorenzo Giusti, of the University of Oxford and I, together with Des Bowden
of Newman and Westhill College, of the University of Birmingham, found levels
of aluminium over 100 times as high when we looked at the chemistry of water
from wells in rural Malawi. It seems that the method normally used to treat
the water before analysis for aluminium levels hides potentially dangerous
contamination of the water people drink.

We think that the source of the aluminium is a clay mineral, kaolinite,
which is broken down by the tropical weathering of Africa’s bedrock. The
cores of the continents are made of granitic rocks, which consist mainly
of the minerals quartz, feldspar and mica. Granite weathers slowly in the
cool climate of high latitudes, but in a humid tropical climate it suffers
a much more aggressive attack. Feldspar and mica soon turn to clay minerals
that only weakly hold the quartz grains together. The result is a layer
of soft, crumbly material called saprolite, which forms a layer tens or
even hundreds of metres thick. Towards the top it is dominated by kaolinite,
a clay mineral made of aluminium, silicon, oxygen and hydrogen. Kaolinite
is very stable and, together with iron minerals, survives in the saprolite,
while other elements such as sodium, magnesium, calcium and potassium, leach
away as water soaks through.

This is not the limit to tropical weathering. Close to the surface of
the saprolite the leaching is so intense that even kaolinite begins to break
down. In a few special situations, silica (SiO₂) is leached out
of the kaolinite leaving the aluminium behind, a process known as incongruent
dissolution. The residue of this process is bauxite, the ore that provides
most of the world’s aluminium.

But incongruent dissolution is not the usual fate of this mineral. It
is far more common for all constituents of the kaolinite to disappear; both
the silica and aluminium vanish through congruent dissolution. When this
happens, the saprolite collapses and the land subsides into the plains so
characteristic of tropical Africa. The surface layer or ‘mantle’ of laterite
that remains can be tens of metres thick and is enriched in minerals that
are even more resistant to weathering and leaching than kaolinite – mainly
iron minerals and quartz. The iron gives both tropical soils and laterites
their striking red colour. For each metre of laterite that remains, several
metres of kaolinite have disappeared.

So where does the kaolinite go? Researchers once thought that the saprolite
collapsed because the kaolinite had been physically eroded and transported
away as particles, either at the surface in runoff or by water flushing
through at shallow depths underground. But if this were the case, we would
expect to find deposits of kaolinite that settled out where this water could
no longer carry it, perhaps in river valleys. Deposits formed in this way
would be vast, far more substantial than the mantles of laterite left behind,
but they are not there.

The alternative hypothesis is that the kaolinite dissolves in the ground
water, leaching both silicon and aluminium from the surface layers of saprolite.
The problem with this idea is that it contradicts experimental evidence;
if you mix silicon and aluminium together in a solution, they form insoluble
aluminosilicates that immediately precipitate out. Another hitch is that
water analyses in tropical areas are remarkably consistent in finding a
few parts per million of silicon but almost no aluminium.

The kaolinite and its aluminium must go somewhere. But where? The scale
of the problem is illustrated by a typical laterite in Malawi, in which
a mantle formed when the saprolite collapsed as a result of the removal
of kaolinite. We measured the concentration of elements such as zirconium,
that resist leaching. In fact, zirconium proved to be one of the most resistant;
there is nearly five times as much of it in the mantle as there is in the
saprolite below. Manganese and niobium are also resistant; we found that
their concentrations in the mantle were respectively four and five times
that in the saprolite. By contrast, aluminium is slightly less concentrated
in the mantle than in the saprolite, indicating that it has been leached
away. We calculated that about 80 per cent of the aluminium that was once
in the saprolite is missing from the mantle.

The key to the mystery comes from Malawi’s seasonally-waterlogged bottomlands,
known as dambos. The dambos are fed by ground water, which evaporates each
year in the dry season, leaving a residue of aluminosilicate minerals. Clearly,
aluminium is being supplied in the ground water, and weathering of the saprolite
on higher land is an obvious source. This supports our hypothesis of congruent
dissolution of kaolinite. And because the aluminosilicates are forming today,
water now moving through the ground must contain aluminium.

This evidence for aluminium in ground water raises two further questions:
first, why does aluminium appear to dissolve in ground water? Aluminium
is most soluble (as a hydroxide) when water is either very acid or very
alkaline. Malawi water is close to neutral, and aluminium would not be expected
to dissolve in these conditions. Secondly, why have earlier analyses detected
so very little aluminium in ground water, while the minerals forming in
the dambos show that it is there and that it does move?

Both questions could be answered if the aluminium in ground water is
not carried as a true solution. It could move if the aluminium is ‘packaged’
in some way, perhaps with each ion surrounded by organic matter. This way,
aluminium could be leached out of the saprolite and carried in ground water
along with silicon without precipitating. One source of such organic matter
is microorganisms, which are known from work in Uganda to be able to break
down kaolinite in laterite. When we examined the laterites of Malawi, we
found that they too contained microorganisms that could break down kaolinite.
This encouraged us to pursue the idea that it is organic binding that allows
aluminium to be carried in ground water.

Testing the treatment

The next stage was to look at the methods that had been used to prepare
water for analysis to find out why the aluminium was not showing up there.
Usually, water quality measurements start by filtering out the larger particles,
those more than 0.45 micrometres across. If they are left in the water even
for a few days, certain elements, including aluminium, tend to become attatched
to them. When the particles settle out, they take these elements with them.
The water left behind contains far less of these elements than it originally
did, so they are already under-represented in the analysis. Then the water
is further stabilised with acid, which helps to hold elements in solution
for the hours or days before it is analysed. If water is neutral or alkaline,
some elements tend to precipitate out.

We thought that this standard sequence of treatment, intended to stabilise
the composition of the water for analysis, might have precisely the opposite
effect, removing the aluminium before the water is analysed. Filtration
would remove large particles of aluminium bound to organic matter. Smaller
particles would be broken down by the acid, which attacks the bonds that
hold the aluminium in its organic complexes. Without its organic packaging,
the aluminium is free to react with silicon already in solution to form
insoluble compounds that precipitate. The water left at the end of this
process would be without aluminium to be detected by the normal analytical
procedure.

To test this, we collected water from a well and treated it differently
before analysis. We did not filter samples immediately they were taken;
instead we mixed each sample ultrasonically just before analysis to resuspend
any material bound by organic matter that could have settled out. This treatment
of samples resulted in the highest levels of aluminium, well above the WHO
limits. We analysed the samples again after leaving them overnight for large
particles to settle out and found less aluminium, but values were still
three times the recommended maximum. We then tried a version of the standard
technique, in which as well as filtering and acidifying the samples, we
mixed them ultrasonically before the analysis. We used ultrasonic mixing
because we suspected that the acid would cause some precipitation, after
which the particles would separate out to give spuriously low values in
the solution. This sequence of treatment further lowered the aluminium levels,
and we think this was because the filtering removed not only particles bearing
aluminium that were big enough to settle out overnight, but also some that
were still in suspension.

As a last test we filtered and acidified the samples as before, and
then filtered again to see if the acidification had led to any precipitation
of aluminium combined with silicon in particles big enough to be caught
in the filter the second time. It had, because after the second filtration
the remaining aluminium levels in the water were low enough to be acceptable.
These results confirm our hypothesis that the normal procedure of filtering
and adding acid removes aluminium from the water samples.

In all, we analysed 35 samples of water. Most of the wells they were
taken from were 1 or 2 metres deep, usually hand-dug. For comparison, we
included samples from a few government-funded wells that were between 3
and 6 metres deep, and some from deeper boreholes. Only three samples contained
less aluminium than the 0.1 parts per million (ppm) recommended by the WHO;
15 of them, nearly half, had more than 10 times the recommended concentration,
and of these five had aluminium at 100 times the recommended level or more.
The highest values, including one of 24 ppm, came from the shallow wells,
but all of the government wells we sampled also exceeded the limit, one
reaching 4.3 ppm. Two of the five boreholes sampled had the lowest levels,
but one was high, at 4.5 ppm. Filtering reduced the levels, as did acidification.

The levels of aluminium in samples at three stages, before filtering,
after filtering and after both filtering and acidification showed a clear
trend; a progression from 90 per cent of samples having more than the recommended
0.1 ppm aluminium, through 30 per cent down to 3 per cent. In short, the
standard procedure of water treatment before analysis – filtration followed
by acidification – could hardly be better designed as a system for masking
the aluminium that is present in the water.

As further support for our model, there is a link between the electrical
conductivity of the water samples and the amount of aluminium we measured.
Shallow water, from the upper part of a weathering profile, has a low conductivity
because most of the soluble elements – whose ions conduct electricity –
have already been leached out. Water from the deeper parts of the profile,
where such ions are abundant, has higher conductivity. The association of
high aluminium content with low conductivity fits our deduction that the
aluminium comes from the upper part of the profile, by congruent dissolution
of kaolinite.

In practical terms, this means that the shallow wells, usually dug by
local people, are the most heavily contaminated with aluminium. Almost every
household in Malawi’s rural areas draws water from one of these wells. The
people who use this water do not filter or acidify it before drinking. So
the real aluminium levels to which users are exposed remain uncertain.

Although we focussed on aluminium in this pilot study, there are clear
indications that other metals may be bound by organic matter in the same
way, and leach from the upper parts of the weathering profile. Conventional
procedures used before analysis of shallow ground water in the tropics may
result in serious understatements of the true levels of titanium, chromium
and even zirconium, as we have found they do for aluminium. Researchers
have assumed that these elements do not move in groundwater because they
do not expect them to dissolve. In Malawi, we calculated that more than
46 per cent of the titanium has been leached out of the saprolite that collapsed
to form the iron-rich mantle – and there is titanium in the minerals forming
in the dambos. More than 75 per cent of the chromium once in the saprolite
has also been lost.

The levels of aluminium we found in Malawi were so pervasive and so
much higher than previous results that there could be a significant aluminium
problem in such ground water, previously masked by the procedures used to
stabilise water for analysis. This could also apply throughout the tropical
laterite belt – from 30 degrees north to 30 degrees south – where most people
depend on untreated ground water.

We need to know more about the extent of the problem and, at the same
time, develop water treatment methods that can reach individual households
in remote rural areas where poverty prevails. Though this is a daunting
task, there are some hopeful prospects. For example, in ‘Africa at a watershed’
(New ÐÓ°ÉÔ­´´, 23 March), Fred Pearce described how the seeds of the horseradish
tree are sometimes added to water. This is an effective way to make organic
particles flocculate and settle out. But the first problem is to find the
true levels of aluminium and other elements in the water that people are
already drinking.

Marty McFarlane works in the School of Geography at the University of
Oxford.