On the quantitative thermogravimetric analysis of calcite content in

hydrated cementitious systems

VILLAGRÁN-ZACCARDI Yury1,2, a *, ÉGUEZ-ALAVA Hugo2,b,
DE BUYSSER Klaartje3,c, DE BELIE Nele2,d
1
Magnel Laboratory for Concrete Research, Ghent University, Zwijnaarde-Technologiepark 904,
9052, Ghent, Belgium
2
LEMIT, CONICET, 52 entre 121 y 122 s/n, 1900, La Plata, Argentina
3
Department of Inorganic and Physical Chemistry, Ghent University, Krijgslaan 281 S3, 9000,
Gent, Belgium
a
[email protected], [email protected], [email protected],
d
[email protected]
*corresponding author

Keywords: Quantitative analysis, thermogravimetry, calcite.

Abstract. Calcite is a compound that is present in hydrated cementitious materials when

carbonation of portlandite occurs or when limestone constituents are used. The quantification of its
content in cementitious systems is then frequently necessary.
Thermogravimetry (TG) measures the change in mass of a material (as a function of time) over a
temperature range using a predetermined heating rate. It can be applied to estimate calcite content in
the hydrated cement system, considering the temperature range at which it decomposes and releases
carbon dioxide. However, the quantification is not easy because the onset of this decomposition is a
function of many variables. The tangential method over the TG signal or the integration method
over the derivative TG curve (DTG) are usually used to discount the background over the
temperature range at which calcite decomposes.
However, consistent underestimation of compounds is frequently described. The reasons for this are
unclear and some hypothesis are discussed in this paper. Additionally, experimental quantitative TG
of cement paste and aggregate containing calcite and diluted with low contents of analytical calcite
are compared with the expected contents, as a calibration method regarding calcite content in the
samples is given to improve the reliability of the results.

Introduction
Thermogravimetry (TG) is a widely applied technique for characterization of hydration products
in cementitious systems. During TG the sample is heated while the weight loss is recorded. Calcite
content in hydrated cementitious materials is derived from the progression of carbonation of
portlandite, or limestone constituents. For the particular case of calcite, TG reflects its
decomposition at a particular temperature range. This reaction cause a weight change which is
linked with this particular compound. The quantification of calcium carbonate might however show
significant differences according to the method applied for this computation from the TG signal.
The thermodynamical basis of quantitative TG (QTG) is based on the decomposition of calcite
between 600 and 800°C [1] (Equation 1). However, some amorphous calcium carbonate may
partially decarbonate earlier, between 400 and 600°C, forming CaO and calcite [2]. The
crystallinity of CaCO3 also plays a significant role in the temperature at which it decomposes. Data
in the literature [3-5] shows that carbonation of cementitious materials leads to sequestration of CO2
that may form poorly crystallized carbonates. During TG, the decomposition of these compounds
occurs earlier and causes gradual mass loss that ends with a more intense peak around 800°C that
could be attributed to the decomposition of well-crystallised calcite. Then, the exact temperature for
the decarbonation peak(s) depends on the relative amount of calcium carbonate, proportion of hemi
and monocarbonates, and the fineness of carbonates. Also two consecutive overlapping peaks may
result from this process. Despite the advantage of mass of CO2 released per unit mass of calcium
carbonate not being depended on the various forms in which it can be present, it seems that poorly
crystallized compounds might be the main reason for underestimation of calcium carbonate content.

CaCO3(s) → CaO (s) + CO2 (g) (1)

The amount of calcite is computed from the corresponding weight loss considering that 100 g of
CaCO3 release 44 g of CO2. The main concern is how this weight loss is determined. For this, the
tangential method and the integration method are frequently applied.
The tangential method computes the mass loss from the distance between the tangents to the TG
curve after and before the change in the slope corresponding to the weight loss step.
Differentiation of the thermogravimetric data, namely derivative thermogravimetry or
differential thermogravimetry (DTG), allows a better resolution and identification of consecutive
weight losses. Here, the quantification of the weight loss can be obtained from integration of the
DTG over a determined temperature range or time period, discounting the background, this is the
area over the secant to the curve between the onset and end of the peak. For other thermal analysis
such as DTA, the interpretation and quantification of the results can be more difficult due to non
stable base line signal caused by the intrinsic properties of the products obtained during the thermal
decomposition of the sample. Then, the calcium carbonate decomposition can be better evaluated
by DTG than by DTA [6].
It should be mentioned that results of thermal analysis are also strongly influenced by setting up
of the measurement (vessel type, heating rate, sample amount, particle size, gas flow rate, purging
gas, pretreatments). This make the comparison between data in the literature quite difficult.
In this paper, a procedure for quantification of calcite in cementitious systems is presented. The
possible correction for underestimation by the tangential and integration methods is analyzed.
Cement pastes and gravel aggregate were tested with and without admixed pure calcite as internal
standard. From the results, the resultant quantified amounts are compared with the added amount
and correction factors for the systematic underestimation are derived.

Experimental
First, pure calcite was analysed by TG, as a procedure for calibration of the device for
quantitative analysis. This pure calcite used as reference material was an analytical reagent grade
chemical commodity. Next, OPC and OPC+limestone pastes, and siliceous gravel (containing about
15% of carbonate), with and without known amounts of extra calcite added were tested.
The tested pastes were prepared with pure OPC (CEM I 42.5) and OPC+10% limestone powder
and cured for 28 days. After the curing period, they were manually ground with mortar and pestle
and tested for TG. The difference between the measurements of pure samples, and samples admixed
with 5 or 10% of standard calcite, is compared with the added amount of calcite. This comparison
was additionally made with samples consisting in a ground gravel aggregate containing carbonates
with and without internal standards.
Thermogravimetric analysis were carried out using a Netzsch STA 449F3 apparatus. Samples of
approximately 50mg were tested in alumina open crucibles and exposed under an inert atmosphere
of nitrogen to increasing temperatures, ranging from 20 °C to 1100 °C at a rate of 10 °C/min.

Results
After the measurement, a correction for buoyancy effect is always necessary. This correction
signal is obtained by measuring an empty crucible under exactly the same conditions that are used
for the analysis of samples. Then, the apparent mass increase due to decreasing gas pressure in the
chamber during heating is discounted. However, some little discrepancies are always present, and it
can be seen that the correction using one or another curve will always result in dissimilar results.
Making an immediate blank determination before measuring the actual sample is the most
appropriate procedure and would normally reduce the effect of this variation in the blank
In this respect, TG results for calcite are presented in Figures 1 (a) uncorrected and (b) corrected
for buoyancy. In Figure 1 (b), the mere weight losses in the ranges 20-105°C, 105-600°C and 105-
1000°C are showed. Also, the weight loss by the tangential (Marsh) method is computed for the
step corresponding to calcite decomposition. Here, the weight loss of 2.53% observed in the range
of 105-600°C is not connected with decomposition of calcite, and cannot be linked with the
composition of the material either. Instead, it seems that this reduction is an artifact caused by the
correction for buoyancy, which is not completely the same for the blank and the measurement of the
samples. This artifact would be relatively higher for smaller amounts of sample being tested.
Identifying this issue and choosing a more convenient correction curve is possible in this case, but
not for paste samples where this effect overlaps with weight loss from dehydration of C-S-H during
the whole temperature interval from 20 to 600°C and no individual distinction is possible. Then,
despite a weight loss virtually induced by the correction by the blank, the real extent of this
influence cannot be determined. In this sense, a double measurement of the sample with and
without an admixed amount of internal standard could give some more clarification on the issue.

(a) Uncorrected signals.

(b) Corrected signal for buoyancy.

Figure 1. TG measurements for calcite. Blank correction for buoyancy (blue), and calcite (red).
 Figures 2, 3 and 4 present results with/without added calcite from OPC and OPC+limestone
pastes, and gravel aggregate, respectively. For pastes, peaks for DTG around 120, 170, 450 and
750°C are noted in association with decomposition of ettringite, AFm, portlandite and calcite,
respectively. All the remaining weight loss is associated with dehydration of C-S-H. Naturally, the
inclusion of calcite increases the intensity of the corresponding peak and weight loss step in the TG
signal. For the limestone blended paste admixed with calcite, a double peak for calcite is noted,
presumably due to differences in crystallinity and grain sizes for calcite from limestone and the
calcite added. This difference is not noted in the signals for OPC paste with added calcite. It can
also be noted that the addition of calcite causes a change in the position of the peak in the DTG.
This shift towards higher temperatures is in accordance with the added amounts.

LA1A285C
LA1A2810C
LA1A

Figure 2. TG for limestone blended paste.

LA1A: pure paste; LA1A2810C: paste+10% calcite; LA1A285C: paste+5% calcite.

P12810C
P1285C
P1

Figure 3. TG for OPC paste.

P1: pure paste; P12810C: paste+10% calcite; P1285C: paste+5% calcite.
 AGG
AGG5C
AGG10C

Figure 4. TG for aggregate with and without internal standards.

AGG: pure aggregate; AGG5C: aggregate+5% calcite; AGG10C (1): aggregate+10% calcite.

Discussion
From the weight losses obtained from the tangential and integration methods, the corresponding
computed amounts of calcite are presented in Table 1. For samples consisting in calcite and
aggregate with/without added calcite, only one decomposition reaction takes place during heating.
Then, both methods should theoretically result in the same amount to that obtained from the mere
weight loss (i.e. the start and the end parts of the TG signal should be horizontal in the graph), but
this is not the case. Moreover, it is notorious that only when the quantification is based on the mere
weight loss in the TG of calcite (44.02%), 100% of calcite is obtained. However, when the
tangential and integration methods are applied the amount of calcite is underestimated by factor of
0.912 and 0.914, respectively.
The previous values were used as correcting factors applied to determinations on pastes (Table
1). Still, the differences between the measurements with and without added calcite are lower than
the actual added amounts. Consequently, there seems to be a source of error additional to that of
buoyancy correction, that could not be discerned at this stage.
In all cases, better results are obtained for the integration method in comparison with the
tangential method. Although from a mathematical point of view the tangential and integration
methods are the same procedure, the method based on the integration over the DTG curve is more
precise for low contents, as the absolute local minimums of this curves in the ranges near the onset
and end temperatures are easier to be determined than the change in the slope of the TG required for
the tangential method. Moreover, the integration method offers a better traceable quantification than
the tangent method and is less affected by the noise in the signal.
 Table 1. Contents of calcite determined from weight losses from the tangential and integration
methods and corrected by factors obtained from the TG of analytical calcite [g/100 g sample].
Sample Tangential method Integration method
Without Corrected Without Corrected
correction correction
Standard Calcite 91.23 - 91.39 -
AGG 13.84 15.17 14.66 16.04
AGG5C 18.75 20.55 18.84 20.62
AGG10C 22.57 24.74 22.75 24.89
P1 1.18 1.30 1.05 1.14
P1285C 5.75 6.30 5.68 6.22
P12810C 9.93 10.89 9.95 10.89
LA1A 6.73 7.37 6.14 6.71
LA1A285C 11.00 12.06 10.75 11.76
LA1A2810C 14.23 15.59 14.93 16.34

Conclusions
Increased reliability of the quantification of calcite in hydrated cement pastes when the
measurements are compared to those of samples with added calcite is achieved. Still, the amounts
obtained for added calcite were systematically underestimated by TG. Among the potential causes
of this underestimation, the imperfect correction for buoyancy seems to be responsible for a high
proportion of the bias, however, other sources of error still remain unexplained.
Higher accuracy was obtained for quantifications based on the integration over DTG in
comparison with the tangential method. The integration method is more precise than the tangential
method because the onset is more easily identified by the operator as a difference in the vertical
magnitude in the DTG curve rather than as a difference in the angle of the slope in the TG curve.
Therefore, traceable and more precise results can be expected from DTG, especially considering
that calcite in cementitious materials can show very variable crystallinity and grain size, with an
extended range of temperatures for its decomposition.

References
[1] B. Lothenbach, P. Durdziński, K. De Weerdt, Thermogravimetric Analysis, in: K. Scrivener,
R. Snellings, B. Lothenbach (Eds.), A Practical Guide to Microstructual Analysis of Cementitious
Materials, CRC Press, Boca Raton, 2015, pp.177- 211.
[2] S. Goto, K. Suenaga, T. Kado and M. Fukuhara, Calcium silicate carbonation products, J. Am.
Ceram. Soc. 78 (1995) 2867–2872.
[3] W.F. Cole and B. Kroone, Carbon dioxide in hydrated portland cement, J. Am. Concr. Inst. 31
(1960) 1275-1295.
[4] B. El-Jazairi and J.M. Illston, A simultaneous semi-isothermal method of thermogravimetry
and derivative thermogravimetry, and its application to cement pastes, Cem. Concr. Res. 7 (1977)
247-257.
[5] J.M. Bukowsky, R.L. Berger, Reactivity and strength development of CO2 activated non-
hydraulic calcium silicates, Cem. Concr. Res. 9 (1979) 57-68.
[6] J. Dweck, P.M. Buchler, A.C. Vieira Coelho, F.K. Cartledge, Hydration of a Portland cement
blended with calcium carbonate, Thermochim. Acta 346 (2000) 105-113.