101
Journal of Paleolimnology 25: 101–110, 2001.
© 2001
K
l
uwer Academic Publishers. Printed in the Netherlands.
Loss on ignition as a method for estimating organic and carbonate content in
sediments: reproducibility and comparability of results
Oliver Heiri
1
, André F. Lotter
1, 2
& Gerry Lemcke
2, 3
1
Geobotanical Institute, University of Bern, Altenbergrain 21, CH-3013 Bern, Switzerland
(E-mail: heiri@sgi.unibe.ch)
2
Swiss Federal Institute of Environmental Science and Technology (EAWAG), CH-8600 Dübendorf, Switzerland
3
Schweizerische Rückversicherung, Mythenquai 50/60, CH-8022 Zürich, Switzerland
Received 27 September 1999; accepted 18 November 1999
Key words: loss on ignition, lake sediment, carbonate, organic matter, quality control
Abstract
Five test runs were performed to assess possible bias when performing the loss on ignition (LOI) method to estimate
organic matter and carbonate content of lake sediments. An accurate and stable weight loss was achieved after 2 h
of burning pure CaCO
3
at 950 °C, whereas LOI of pure graphite at 530 °C showed a direct relation to sample size
and exposure time, with only 40–70% of the possible weight loss reached after 2 h of exposure and smaller samples
losing weight faster than larger ones. Experiments with a standardised lake sediment revealed a strong initial weight
loss at 550 °C, but samples continued to lose weight at a slow rate at exposure of up to 64 h, which was likely the
effect of loss of volatile salts, structural water of clay minerals or metal oxides, or of inorganic carbon after the
initial burning of organic matter. A further test-run revealed that at 550 °C samples in the centre of the furnace lost
more weight than marginal samples. At 950 °C this pattern was still apparent but the differences became negligible.
Again, LOI was dependent on sample size.
An analytical LOI quality control experiment including ten different laboratories was carried out using each
laboratory’s own LOI procedure as well as a standardised LOI procedure to analyse three different sediments. The
range of LOI values between laboratories measured at 550 °C was generally larger when each laboratory used its
own method than when using the standard method. This was similar for 950 °C, although the range of values tended
to be smaller. The within-laboratory range of LOI measurements for a given sediment was generally small.
Comparisons of the results of the individual and the standardised method suggest that there is a laboratory-specific
pattern in the results, probably due to differences in laboratory equipment and/or handling that could not be eliminated
by standardising the LOI procedure.
Factors such as sample size, exposure time, position of samples in the furnace and the laboratory measuring
affected LOI results, with LOI at 550 °C being more susceptible to these factors than LOI at 950 °C. We, therefore,
recommend analysts to be consistent in the LOI method used in relation to the ignition temperatures, exposure
times, and the sample size and to include information on these three parameters when referring to the method.
Introduction
Sequential loss on ignition (LOI) is a common and
widely used method to estimate the organic and
carbonate content of sediments (e.g., Dean, 1974;
Bengtsson & Enell, 1986). In a first reaction, organic
matter is oxidised at 500–550 °C to carbon dioxide and
ash. In a second reaction, carbon dioxide is evolved
from carbonate at 900–1000 °C, leaving oxide. The
weight loss during the reactions is easily measured by
weighing the samples before and after heating and is
closely correlated to the organic matter and carbonate
Eawag_08980
102
content of the sediment (Dean, 1974; Bengtsson &
Enell, 1986). Dean (1974) evaluated the method and
concluded that LOI provides a fast and inexpensive
means of determining carbonate and organic contents
of clay-poor calcareous sediments and rocks with
precision and accuracy comparable to other, more
sophisticated geochemical methods. Bengtsson & Enell
(1986), in their instructions to the technique, mention
that the method gives a rough indication of the organic
matter and carbonate content of sediments. Depending
on the ignition temperature, various losses of volatile
salts, structural water and inorganic carbon may occur
(Dean, 1974; Bengtsson & Enell, 1986; Sutherland,
1998) and it is important to check the ignition tem-
perature carefully for organic matter determination.
Nonetheless, a short survey of recently published
palaeolimnological studies shows that there are still
substantial differences in the methods used. Exposure
times vary from 1 to 4 h (e.g. Spaulding et al., 1997;
Henderson & Last, 1998) and ignition temperatures from
500 to 550 °C (e.g. Virkanen et al., 1997; Korsman et
al., 1999). Cautioned by the discovery that a defective
furnace produced a strong and immediately noticeable
change of LOI results, we decided to run several tests
to estimate possible bias of the method with a new
digital-display furnace. In particular, we wanted to
answer the following questions:
i) How long must samples be exposed in the furnace
for a complete reaction?
ii) Does the position of crucibles in the furnace
influence the LOI results?
iii) Does sample size affect the results?
iv) Are LOI results measured in different laboratories
comparable?
This last point is all the more important as palaeo-
limnological projects reach regional or even continental
scales (e.g. Wathne et al., 1995; Korhola et al., 1999)
and LOI as a quick and inexpensive estimate of
sedimentary organic matter is increasingly measured
at different institutes. Interpretation of results is then
only possible if an estimate of a laboratory specific
error is available and real trends in the data can be
viewed in relation to measurement errors.
The basic LOI method
Determination of weight percent organic matter and
carbonate content in sediments by means of LOI is
based on sequential heating of the samples in a muffle
furnace (see Dean, 1974; Bengtsson & Enell, 1986 for
more details on the method, but note that in the equations
for calculation of LOI in the latter publication the
weight loss is related to the wet weight of the sediment
instead of the dry weight). After oven-drying of the
sediment to constant weight (usually 12–24 h at ca.
105 °C) organic matter is combusted in a first step to
ash and carbon dioxide at a temperature between 500
and 550 °C. The LOI is then calculated using the
following equation:
LOI
550
= ((DW
105
–DW
550
)/DW
105
)*100 (1)
where LOI
550
represents LOI at 550 °C (as a per-
centage), DW
105
represents the dry weight of the sample
before combustion and DW
550
the dry weight of the
sample after heating to 550 °C (both in g). The weight
loss should then be proportional to the amount of
organic carbon contained in the sample and Dean
(1974) showed a strong correlation between LOI at
550 °C and organic carbon content determined chro-
matographically in lake sediments.
In a second step, carbon dioxide is evolved from
carbonate, leaving oxide and LOI is calculated as:
LOI
950
= ((DW
550
–DW
950
)/DW
105
)*100 (2)
where LOI
950
is the LOI at 950 °C (as a percentage),
DW
550
is the dry weight of the sample after com-
bustion of organic matter at 550 °C, DW
950
represents
the dry weight of the sample after heating to 950 °C,
and DW
105
is again the initial dry weight of the sample
before the organic carbon combustion (all in g).
Assuming a weight of 44 g mol
–1
for carbon dioxide
and 60 g mol
–1
for carbonate (CO
3
2–
), the weight loss
by LOI at 950 °C multiplied by 1.36 should then
theoretically equal the weight of the carbonate in the
original sample (Bengtsson & Enell, 1986). Again,
LOI shows a good correlation with other methods of
determining carbonate content of lake sediments
(Dean, 1974).
Experimental setting
All LOI analyses were carried out in a Nabertherm
®
Controller C6 muffle furnace with digital temperature
display and thermostatic temperature control (Naber-
therm
®
, Lilienthal/Bremen, Germany). For this study
we used LOI temperatures close to those proposed by
103
Dean (1974) and Bengtsson & Enell (1986), i.e. 530
and 550 °C for organic matter and 950 °C for car-
bonate. Care was taken that no humidity remained in
samples before weighing. Therefore, empty crucibles
and wet sediment were dried at 105 °C overnight and
all samples were cooled to room temperature in a des-
iccator before any measurements were made. To avoid
overheating, crucibles were put into the furnace only
after a constant temperature was reached. For the
longer time series crucibles were heated, cooled to
room temperature, measured, and the same sediment
was then again exposed to the respective temperature.
For the test runs, three artificial sediments were
produced: Cores from several lakes on the Swiss
Plateau were mixed to produce the first test sediment
standard with a low organic matter and intermediate
carbonate content. A short Kajak core of Nydalasjön,
northern Sweden, provided the source for a sediment
standard with a high organic matter content. A third
sediment standard with a high carbonate content was
produced from a mix of mid-Holocene sediment from
Rotsee, central Switzerland. The three sediments are
referred to as ‘mixed sediment’, ‘high organic matter
sediment’ and ‘high carbonate sediment’ respectively.
The sediments were dried, finely ground and hom-
ogenised by hand stirring and shaking in a closed
container. Due to different amounts of material available
sample size varied between 0.4 and 3.9 g dry weight
(see figure legends for average dry weight and number
of samples per test run).
Results and discussion
Exposure time and sample size
In the first two test runs, we exposed samples of pure
graphite (Merck #104206, Merck KgaA, Darmstadt,
Germany) and of pure calcium carbonate (CaCO
3
,
Merck #102064) to 530 and 950 °C to assess if the
reactions are complete within the 2 h we previously
used for LOI analysis. Theoretically, graphite, as a
source of totally combustible carbon, should give a
weight loss of 100% at 530 °C. Assuming a weight of
99.96 g/mol for CaCO
3
and 55.96 g/mol for CaO,
calcium carbonate should produce a weight loss of
44.02% at 950 °C (molar weights according to
Mortimer, 1983). Only 40–70% of the graphite was
combusted in most samples after 2 h of exposure at
530 °C (Figure 1), whereas 5 h of exposure were
needed for a median weight loss of 98.3%. The range
of 26.7–99.0% LOI was unexpectedly high after 2 h of
exposure but decreased to 63.8–99.8% after 5 h of
exposure. This high variability is only partly due to the
fact that larger samples needed more time to combust,
as can be seen if the samples are separated into weight
classes and plotted vs. time (Figure 2). A median of
99.8% of the total weight loss (range: 99.7–99.9%) was
reached after exposing the samples for 2 h to 950 °C,
confirming that the graphite we used is entirely com-
bustible and that in many samples the reaction at
530 °C was not finished even after 5 h of exposure.
However, graphite is an inadequate substitute for
sedimentary organic matter, as it has different chemical
properties and is presumably more refractory than
autochthonous algal remains that commonly constitute
most of the lacustrine sedimentary organic carbon.
Still, our results indicate that 1 or 2 h of exposure as
generally recommended for combustion of organic
matter (Dean, 1974; Bengtsson & Enell, 1986) may not
be sufficient for larger sediment samples with a high
organic content and that LOI at 550 °C may depend on
weight (Figure 2).
Figure 1. Box plots of LOI values of 56 samples of graphite heated
for 2, 3, 4 and 5 h to 530 °C (average sample dry weight ± S.D.:
1.12 ± 0.22 g). The central horizontal line in the box marks the
median of the samples, the box edges (hinges) the first and third
quartile. The interquartile range within the box includes the central
50% of the values. The whiskers show the range of observed values
that are not within the first and third quartile but not further away
than 1.5 times the interquartile range from the hinges. Values
between 1.5 and 3 times the interquartile range from the nearest
hinge are marked by asterisks, values farther than 3 times the
interquartile range from the next hinge are marked by open circles.
104
Using CaCO
3
, no significant weight loss was det-
ected at 530 °C, whereas at 950 °C a median of 42.7%
of the total weight loss was reached after 2 h of
exposure. This represents the equivalent of 97% of the
theoretically possible weight loss, indicating that the
evolution of carbon dioxide from the carbonate was
largely finished within the exposure time.
In a third test we combusted samples of the ‘mixed
sediment’ at 550 °C, exposing them for 0.5–64 h to
assess how long the organic matter needs to be
completely ashed. In order to test the reproducibility
of the results, this series was repeated in three separate
runs of short (0.5, 1, 1.5, 2, 2.5 h), intermediate (4.5,
6.5, 8, 10 h) and long (16, 24, 33, 40, 48, 56, 64 h) time
intervals. Afterwards, each set of samples was heated
to 950 °C for 2.5 h to calculate the total carbon content.
The results indicate that LOI at 550 °C was slightly
lower in the samples used for the intermediate time
intervals (approximately 0.15%; Figure 3). Still, the
method is precise enough to give a clear overall
impression of the weight loss through time. Continuing
weight loss was registered up to the last measurement
at 64 h of exposure (Figure 3). As expected, LOI was
strongest in the first 2–2.5 h whereupon weight loss per
time diminished to approach a more or less constant
value of approximately 0.02% per h.
This small but continuing weight loss at 550 °C was
an unexpected result and may be explained by the high
inorganic content of the mixed sediment. Different
authors (e.g. Ball, 1964; Dean, 1974; Bengtsson &
Enell, 1986) warn that clay may lose structural water
during LOI. According to Ball (1964) this may happen
at temperatures as low as 500 °C, thus causing a weight
loss of up to 20% in clay minerals. Sutherland (1998)
mentions that structural water may be lost by metal
oxides at temperatures as low as 280–400 °C and that
inorganic carbon may be lost at temperatures between
425 and 520 °C in minerals such as siderite, magnesite
and rhodochrosite (Weliky et al., 1983; Sutherland,
1998). Finally, Bengtsson & Enell (1986) mention the
possible influence of loss of volatile salts at 550 °C on
LOI results. In our opinion one or several of these
processes taking place at a lower reaction speed than
the combustion of organic matter is the most likely
cause of the continuing weight loss observed in our
experiment.
Two hours of exposure at 950 °C were sufficient for
the total loss of carbon dioxide to be completed and
further exposure led to no detectable subsequent
reduction in weight. The total weight loss (at 550 and
950 °C) proved to be reproducible with an average (±
S.D.) of 20.32 ± 0.03% for short time intervals, 20.31
± 0.03% for the intermediate, and 20.38 ± 0.04 for the
long time intervals.
Position in the furnace
In a fourth test run, again using the ‘mixed sediment’,
we assessed the influence of the position of the
crucibles in the furnace on LOI (see Figure 4a for the
experimental setup). LOI at 550 °C was consistently
highest in the centre of the furnace (exemplified for 2
and 18 h of exposure in Figures 4b & 4c). To compare
how these differences change with increasing exposure
time, the median LOI value and range of the twelve
outermost samples and the six innermost samples are
plotted in Figure 5a (see Figure 4a for the experimental
setup). Again samples lost weight up to the maximum
exposure time of 18 h. Differences due to the position
Figure 2. Median and range of LOI at 530 °C of three weight classes
of samples in the graphite experiment. Circles: 44 samples between
0.7 and 1.2 g dry weight; triangles: 7 samples between 1.2 and 1.5
g dry weight; diamonds: 5 samples between 1.5 and 1.8 g dry weight.
Figure 3. Median and range of LOI of 5 samples of the ‘mixed
sediment’ exposed to 550 °C in irregular intervals up to 64 h
(average sample dry weight ± S.D.: 3.93 ± 0.03 g).
105
of the crucibles became more distinct with longer
exposure to 550 °C, starting with 0.42% difference of
median weight loss between the innermost and outer-
most samples after 2 h (maximum difference between
crucibles 1.24%) to 2.36% difference after 18 h (max-
imum difference 3.70%). At 950 °C weight loss was
also highest in the middle of the tray but the difference
of 0.08% between median values of the innermost and
outermost crucibles (maximum difference 0.48%) was
negligible at an average total weight loss of 20.10%.
As in the experiment using graphite, smaller samples
again lost weight faster than larger ones (Figure 5b).
Figure 4. Spatial distribution of LOI at 550 °C within the furnace.
(a) Experimental setup and position of the crucibles on the tray:
The measurements were carried out on 72 samples of the ‘mixed
sediment’ (average sample dry weight ± S.D.: 1.30 ± 0.24 g). Black
and shaded circles indicate the position of the six innermost and
twelve outermost samples used to calculate LOI for Figure 5; (b)
LOI after 2 h of exposure; (c) LOI after 18 h of exposure. For (b)
and (c) LOI is interpolated using the program SURFER
®
and
triangulation/linear interpolation.
Figure 5. (a) Median and range of LOI of the six innermost (circles)
and the twelve outermost samples (triangles) on a tray of 72 samples
of the ‘mixed sediment’ exposed for 2, 3, 4, 5, 6, 7, 8, 9, 10 and
18 h to 550 °C (see Figure 4a for the experimental setup); (b) Median
and range of LOI at 550 °C of three weight classes of samples in
the same experiment. Circles: 0.8–1.0 g dry weight (7 samples);
triangles: 1.0–1.5 g dry weight (48 samples); squares: 1.5–1.8 g
dry weight (17 samples).
