Rietveld Refinement of an Oil Field Cement Blend

B. Iverson1,R. Loghry1, and C. Edwards2

Halliburton, Duncan, Oklahoma, USA1 and Pampa, Texas, USA2

Keywords: Cement, X-ray Diffraction, Rietveld Refinement

Abstract

X-ray powder diffraction allows direct measurement of the phase content in cement;
however, quantification of component phases requires rigorous post-acquisition analysis. In this
work, an as-received cement was analyzed using a combination of Cu K X-ray diffraction,
Rietveld refinement, and X-ray fluorescence to quantify the initial cement composition.
Diffraction and fluorescence were shown to be complementary techniques for analyzing cement
The cement was then blended with varying amounts of crystalline silica and analyzed using the
Rietveld method. Results indicate that the blended cement compositions were in line with
calculated quantities of individual components of the cement blend, suggesting the Rietveld
method can be used to quantify all crystalline components in a cement blend simultaneously.

Introduction

Cement is arguably the most common and historically used manufactured material.
Modern hydraulic cements, materials that will harden and set when mixed with water, are
generally traced back to a British patent issued in 1824 covering the creation of artificial rock
with characteristics of typical Portland stone [1]. In 2009, approximately 2.8 billion metric tons
of cement were produced globally with the majority of the cement manufactured used in the
construction industry [2].
For oilfield applications, cement is one of the few cost-effective substances that can be
applied in a liquid form with the expectation that the material will eventually solidify. Cement is
often placed between casing, or steel pipe, and a subterranean formation to physically bond the
casing to the formation, to provide a barrier to prevent wellbore fluids from migrating throughout
the wellbore geometry, and to form a base for all subsequent well operations. The main
differentiator between typical oilfield cement and construction cement has less to do with
variations in the cement chemistry and more to do with the environment to which the cement is
to be exposed. In a wellbore environment, cement will often be exposed to elevated temperatures
and pressures, different rock formations of often widely varying composition, formation
stratification, and underground fluids that would typically not be encountered by construction
cements. Because of this, the demands on the chemistry of oilwell cements are typically much
more stringent than those placed on construction grade cement.
Cement clinker is manufactured through the clinkering of proportioned feedstock
components (usually quarried materials such as limestone, clay, sand and iron ore) in a high-
temperature kiln. Regardless of the class of cement used, there are four major components
typically found in cement clinkers:
3CaOSiO2 Tricalcium silicate, Alite or C3S
2CaOSiO2 Dicalcium silicate, Belite or C2S
 3CaOAl2O3 Tricalcium aluminate or C3A
4CaOAl2O3Fe2O3 Tetracalcium aluminoferrite or C4AF

These four components are often easy to identify when characterizing a cement due to
their high concentrations relative to other materials present and their distinct morphology and
crystallography. Gypsum, CaSO42H2O, is added to many cement clinkers as a grinding aid, as
well as to control the set of the cement. With the addition of gypsum, the clinker is ground to its
required particle size and the material is simply referred to as cement. For reference, an optical
micrograph of a polished section of a typical Portland cement grain is shown in Figure 1. In an
ordinary Portland cement, water is added to the cement, which starts a series of chemical
reactions between the water and these five components. The rates and products of the chemical
reactions determine the rheological and mechanical properties of the cement. The concentrations
of these major components, fineness of the cement powder, and the rheological and mechanical
property development ultimately determine the cement classification.

Figure 1: Optical micrograph of a typical Portland cement grain

Because the composition of the non-hydrated cement is tied to the overall performance of
the cement in terms of both its rheological slurry properties and its set properties, various
methods have been developed to quantify the specific overall cement chemistry [4]. The
historical method by which to quantify the different concentrations of these components is a
method originally labeled the Bogue calculation [5], later modified to the Taylor-Bogue
calculation [6]. The former method assumes that the four main components comprising the
cement are present in pure phases, while the latter assumes a typical impurity profile for each
component. An oxide analysis is conducted using X-ray florescence (XRF) (sometimes
supplemented by inductively coupled plasma (ICP) atomic absorption spectroscopy), and the
results of the oxide analysis are used to calculate the specific concentrations of the individual
cement components. The Bogue calculation makes several assumptions, which speed up the
calculation but decrease the overall accuracy of the results. The ferrite phase is assumed to be the
only phase containing iron, which therefore fixes the ferrite concentration. The aluminate phase
can be calculated by subtracting the alumina content in the ferrite phase from the total alumina
content in the cement. All silica is initially assumed to be in the form of C2S, with the
expectation that excess lime will be present in the material. Finally, the lime surplus is allocated
to the formation of C3S. The assumptions made in quantifying the content of the cement
introduce significant systematic errors and reduce the overall accuracy of the method to more of
an approximation. Despite its flaws, the Bogue calculation is the most commonly used method in
industry for quantifying cement compositions.
The Taylor-Bogue method refines the Bogue calculations fundamental assumptions by
introducing impurity profiles into each major phases composition, as well as providing a means
of accounting for native sulfate and alkali content. Because of its accounting for native sulfate
content, the Taylor-modified Bogue calculation is only valid for clinkers without supplemental
knowledge of the state of the interground gypsum.
Alternative methods for identifying and quantifying the cement components include
optical microscopy, as seen in Figure 1, and X-ray diffraction (XRD) methods, including
Rietveld refinement. Because the main components of the cement are all crystalline materials,
XRD can be useful for identifying the presence or absence of specific phases in a cement.
Application of the Rietveld method to cement analysis is a relatively new approach that allows
for the quantification of the individual components [7]. Furthermore, many of the additives that
are commonly used to enhance the properties of the cement are crystalline, which allows for both
their identification and quantification using the same process. For instance, crystalline silica, or
quartz, is one of the most commonly used cement additives and is a requirement for stabilizing
the cement chemistry at temperatures greater than approximately 110C [8]. This process also
allows for tracking the development of oilfield cement from the initial starting components
through the long-term hydration products. The major downside is the expertise required to
conduct and interpret the Rietveld analysis.
In this study, typical oilfield cement was analyzed using both the Bogue calculation and
XRD with accompanying Rietveld refinement. Comparisons are made between the two methods.
Five cement blends containing varying amounts of crystalline silica were analyzed using XRD
with accompanying Rietveld analysis. This quantified both the cement composition as well as the
concentration of non-cementitious components. Ramifications of this type of testing are
discussed.

Experimental

A class G cement (Dyckerhoff AG, Lengerich, Germany) was used for all testing in this
study. Oxide analysis was conducted using XRF data collected using a PANalytical MiniPal unit
with 12-position sample changer. The Bogue calculation was conducted on the results from the
oxide analysis to convert the information into the cement phase concentrations. Crystallographic
analysis was conducted using XRD data collected via a PANalytical X'PERT diffraction system
with Cu K X-rays over a 2 range from 4 to 80. The scan rate for the system was conducted at
0.133/s with a curved X-ray detector. Rietveld analysis was conducted using AMD Jade version
9.0. Crystal structures for the individual phases were obtained from literature for C3S, C2S, C3A,
C4AF, gypsum and quartz [914]. During the Rietveld analysis, a fourth-order background
function was applied while refining the Caglioti shape parameters, X-ray wavelength,
experimental geometry, lattice parameters, atomic positions, site occupancy, and phase fractions.
Refinement was continued until the weighted residual R was minimized [15]. Final values for R
and E (weighted residual minus the background) are given. All results from Rietveld analysis
are given in volume percent.
After analyzing the base cement, quartz was blended with the cement to create five
separate cement blends with quartz (Halliburton SSA-1 cement additive, average particle size
of 17 microns) concentrations of 40, 50, 60, 70, and 80 % quartz by weight of cement (bwoc).
XRD and accompanying Rietveld refinement were conducted on these samples as well to
quantify the total concentration of components in the blended system.

Results and Discussion

The initial oxide analysis and results from the Bogue calculation are shown in Table 1.
Trace amounts of elements that are not readily identifiable as a component of the four main
phases are often present because of the typically impure nature of the feedstock materials used in
cement-clinker manufacturing. These materials can be incorporated at low levels into the
structures of the more common phases. Calcium sulfate dihydrate, or gypsum, is artificially
added to the cement to control the initial set of the cement, as well as to serve as a grinding aid.
Because the oxide analysis does not provide any information about the form of the minor
constituents, the cement concentration was recalculated based only on the presence of the four
identified major components plus the addition of gypsum. This concentration can be seen in
Table 2.
Table 1 Cement component concentration from oxide
analysis and subsequent Bogue calculation
Oxide % Cement %
Na2O 0.15 C3S 55.07
MgO 0.64 C2S 20.27
Al2O3 3.85 C4AF 16.62
SiO2 21.36 C3A 1.06
SO3 2.39 CaSO4 4.1
K 2O 0.74 Na2O 0.15
CaO 63.44 MgO 0.65
TiO2 0.19 K 2O 0.74
MnO 0.71 TiO2 0.19
Fe2O3 5.41 MnO 0.71
ZnO 0.02 ZnO 0.02
SrO 0.16 SrO 0.16
LOI 0.95 Free Lime 0.25

TOTAL 100 TOTAL 100

 The diffraction pattern for the cement with accompanying Rietveld refinement can be
seen in Figure 2. Two separate Rietveld analyses were conducted on this cement, one including
and one excluding the C3A phase. The Class G cement used was intended to have a near-zero
concentration of C3A. The initial Rietveld analysis of the cement is given in Table 2. Because of
the difference between the Bogue calculation C3A concentration and the value obtained by
Rietveld, a second Rietveld refinement was conducted where the C3A was removed from the
analysis. Results for all three tests can be seen in Table 2.

Figure 2: XRD pattern of a Class G cement with accompanying phase analysis using Rietveld refinement
 Table 2. Comparisons among cement
analyses conducted using Rietveld and the
Bogue calculation. All concentration values
are given in volume percent
Rietvelda Rietveld Bogue
C3S 51.8 53.8 56.7
C2S 19.8 23.3 20.9
C4AF 18.4 19.2 17.1
C3A 6.3 -- 1.1
Gypsum 3.7 3.7 4.2
R 4.1 4.01 --
E 3.2 3.21 --
a
Initial Rietveld results including C3A content

The initial Rietveld analysis appeared to overstate the C3A concentration. This was most
likely caused by a peak overlap in the diffraction pattern. The Bogue calculation typically
overestimates the concentration of C3A in the cement [6], which, calculated at 1.1%, indicated
the C3A should be much lower than the 6.3% initially calculated using the Rietveld method.
Other than the C3A content, the values obtained for the other four components were in line with
values for a typical Class G cement. While the C3A content was overestimated by both the
Bogue calculation and the Rietveld method, the structure could be removed from the Rietveld
refinement, which resulted in a reduced residual value. There is currently no method for
removing the major phases from the Bogue calculation in situations where the analysis
overestimates the concentration.
The C3A concentration of the second Rietveld analysis was also verified during the
Rietveld analysis of the samples blended with quartz. The diffraction patterns for the five
samples containing quartz can be seen in Figure 3. The base cement in each case was identical to
that described in Tables 1 and 2. The intense peak observed at approximately 26.7 2 is the
characteristic intensity peak for quartz. This peak obscures some of the minor diffracted peaks
for the cement because of its relative intensity. A Rietveld analysis was conducted on all five
diffraction patterns to determine how closely the calculated values from Rietveld match the
theoretical values of the cement blend.
The XRD pattern with accompanying Rietveld analysis for the sample containing 40%
quartz bwoc can be seen in Figure 4. Initial results from the Rietveld analysis are given in Table
3. The C3A values calculated for each cement blend were near the 6.3% value initially calculated
in the cement blend. Because more material was added to the cement, these values should have
dropped unless the initial value was also overstated. This indicated that the original Rietveld
analysis of the cement blend was inaccurate to include the C3A content and the second analysis
was more representative of the cement composition. Because of this inconsistency, the five
blends were subjected to a second Rietveld analysis, results of which can be seen in Table 4.
 Figure 3: Diffraction patterns for the five cement blends containing crystalline silica in the form of quartz

The theoretical concentrations (in volume percent) of quartz added to the cement blend
and the concentration calculated from Rietveld are highlighted in Table 4. The theoretical and
calculated values both increased as the amount of quartz added to the cement increased. In all
five cases, the differences between the calculated value and the theoretical value was less than or
equal to 3%. This indicated that the Rietveld method can be utilized to identify and quantify the
concentrations of non-cement, phase-specific materials. When analyzed in this manner, the
cement phase-specific materials did not have the same ratios between phases, as was observed
when the cement was analyzed in the absence of quartz. They were similar if the values were
normalized by removing the quartz concentration. The presence of quartz in the diffraction
pattern appeared to skew the overall intensities in the diffraction pattern, which would result in a
difference in the calculated values for the cement phases when quantified in this manner. This
indicated that a cement analysis conducted on a sample containing externally added crystalline
material could have errors introduced into the quantification because of the presence of materials
with higher intensity peaks or peaks that overlap with those of the specific cement phases.
Figure 4: XRD pattern and accompanying Rietveld analysis of the cement sample blended with 40% quartz refined
without C3A.

Table 3 Rietveld refinement results for the five cement/quartz

blends containing C3A in the refinement
Quartz 40% 50% 60% 70% 80%
C3S 37.3 35.6 33.9 32.4 30.3
C2S 10 8.1 8.6 6.5 6
C4AF 11.5 10.1 8.4 8.7 8
C3A 6.1 6.7 7.2 6.6 6.5
Gypsum 1.8 1.6 1.1 1.1 1.2
Quartz 33.2 37.9 40.9 44.8 48.1
R 4.09 4.12 4.44 4.26 4.28
E 3.13 3.13 3.09 3.09 3.05
 Table 4: Final Rietveld refinement results for the five
cement/quartz blends
Quartz 40% 50% 60% 70% 80%
C3S 38.7 37.5 35.9 33.5 31.7
C2S 13.4 10.7 11.1 9.2 8.3
C4AF 12.5 11.4 9.8 10.1 9.4
Gypsum 1.9 1.7 1 1.2 1.2
Quartz 33.6 38.7 42.1 46 49.4
R 4.1 4.16 4.5 4.33 4.43
E 3.13 3.13 3.09 3.09 3.05
Theory 32.6 37.7 42 45.8 49.2

This information highlights some of the differences between the two methods for
quantifying crystalline phases within a cement as opposed to a cement blend. The Bogue
calculation provides a more detailed description of the elemental makeup of the cement at the
expense of assuming all materials are present in pure phases. The Bogue calculation is useful for
cements that have not been modified with externally added materials. The Bogue calculation
provides only an estimation of the concentration of materials because of the assumptions that are
made while conducting the calculation. The addition of any crystalline materials to the cement
blend also renders the Bogue calculation useless. While the Rietveld method neither assumes any
phases are present nor assumes any purity of the specific phases, a detailed examination of the
diffraction pattern is required to help ensure that phases that cannot positively be identified are
not included in the results. This can make a Rietveld analysis prone to overstating the
concentrations of materials that should be present in small concentrations or not at all. The
Rietveld method provides information on the structures of the cement phases present and does
not assume that only pure phases are present. While not included in this study, elemental
substitutions could be incorporated into the cement crystal structures during the refinement. The
Rietveld method also allows for the quantification of crystalline materials externally blended into
the cement system. Analysis in this manner resulted in a good correlation between the theoretical
and calculated concentrations of the externally blended quartz, but resulted in a slightly different
ratio of the cement phases components from when the sample was analyzed in the absence of
quartz. Because of this difference, a Rietveld analysis of the cement in both the presence and
absence of the blended material is suggested to better quantify the total concentration of
crystalline solids.

Conclusions

The Rietveld method was compared to the classical Bogue calculation used for
quantifying the concentrations of C3S, C2S, C3A, C4AF, and gypsum in an as-received cement.
Results indicated that both the Rietveld and Bogue calculation can be prone to overstating the
C3A concentration. This is an issue inherent in the Bogue methodology caused by a difficulty in
positively identifying the presence of C3A in the XRD pattern; whereas, the Rietveld method is
readily able to correct this calculation by removing the structure from the refinement. The
additions of crystalline material blended to the cement can be quantified using the Rietveld
method. Using this method, both the cement composition and the cement-blend composition can
be quantified simultaneously. The cement composition should be quantified in a sample that has
not had other materials added to it because of the possibility of XRD peaks of the added material
skewing the overall intensities of the cement components, thus altering relative concentrations
when compared to a pure sample.

Acknowledgements

The authors acknowledge the late Bill Caveny for the polished cement sample used in
Figure 1 and Halliburton for permission to present this research.

References

[1] J. Aspdin, An Improvement in the Modes of Producing Artificial Stone, British Patent No.
5022

[2] U.S. Geological Survey, Mineral Commodity Summaries, January 2010

[3] P. C. Hewlett, Leas Chemistry of Cement and Concrete, Elsevier, 1998, p 783-811

[4] G. Polkowski, Correlation of Oilwell Cement Performance with Cement Characteristics,

Society of Petroleum Engineers paper no. 16651

[5] R. H. Bogue, Calculation of the Compounds in Portland Cement, Industrial and Engineering
Chemistry Analytical Edition, 1929, Vol 1 (No. 4), p 192-197

[6] H. F. W. Taylor, Modification of the Bogue Calculation, Advances in Cement Research,

1989, Vol 2 (No. 6), p 73-77, 154-156

[7] K. L. Scrivener, T. Fullmann, E. Gallucci, G. Walenta, and E. Bermejo, Quantitative Study of

Portland Cement hydration by X-ray diffraction/Rietveld analysis and Independent Methods,
Cement and Concrete Research, Vol 34, 2004, p 1541-1547

[8] D. Stiles, Effects of Long-Term Exposure to Ultrahigh Temperature on the Mechanical

Parameters of Cement, Society of Petroleum Engineers paper no. 98896

[9] . G. de la Torre, S. Bruque, J. Campo and M. A. G. Aranda, The Superstructure of C3S

from synchrotron and neutron powder Diffraction and its role in Quantitative Phase Analyses,
Cement and Concrete Research, Vol 32, 2002, p 1347-1356

[10] T. Tsurumi, Y. Hiran, H. Kato, T. Kamiya, and M. Daimon, Crystal Structure and Hydration
of Belite, Locality: Synthetic, Ceramic Transaction, Vol 40, 1994, p 19-25

[11] A. A. Colville and S. Geller, The Crystal Structure of Brownmillerite, Ca2FeAlO5, Acta
Cryst., Vol B27, 1971, p 2311-2315

[12] B. F. Pedersen and D. Semmingsen, Neutron Diffraction refinement of the structure of

Gypsum, CaSO42H2O, Acta Cryst., Vol B38, 1982, p 1074-1077
[13] A. F. Wright and M. S. Lehmann, The Structure of Quartz at 25 and 590C Determined by
Neutron Diffraction, Journal of Solid State Chemistry, Vol 36, 1981, p371-380

[14] Y. Takeuchi, F. Nishi and I. Maki, Zeit. Krist., Vol 152, 1980, p 259-307

[15] L. B. McCusker, R. B. von Dreel, D. E. Cox, D. Louer and P. Scardi, Rietveld Refinement
Guidelines, Journal of Applied Crystallography, Vol 32(1), 1999, p 36-50