Document [original]

Provenancing of cement by the means of isotope

techniques

vorgelegt von

M. Sc.

Anera Kazlagic

an der Fakultät VI – Planen Bauen Umwelt

der Technischen Universität Berlin

zur Erlangung des akademischen Grades

Doktorin der Naturwissenschaften

- Dr. rer. nat. -

genehmigte Dissertation

Promotionsausschuss:

Vorsitzender: Prof. Dr. Frank Rackwitz

Gutachter: Prof. Dr. habil. Dietmar Stephan

Gutachterin: PD Dr. Franziska Emmerling

Tag der wissenschaftlichen Aussprache: 10. November 2023

Berlin 2023

Acknowledgements

The pathway to success would not be the same without the strength that we carry within

ourselves.

However, there are some people in life who help whilst paving – in my case very thick, resistant,

concrete path.

My sincere gratitude goes in the first place to my PhD mentor, prof. dr. rer. nat. habil. Stephan

Dietmar (TU Berlin) and my BAM supervisor, Dr. Vogl. It was because of them that I learnt so

much, and I got to know how important it is to have a continuous support in mentoring. Your

support as mentor and supervisor is invaluable.

I would like to express my deepest appreciation to Dr. Possolo Antonio (NIST Fellow & Chief

Statistician), for providing help concerning statistical data mining and overall support.

This PhD thesis is a result of research conducted at the Federal Institute for Materials Research

and Testing (BAM) in Berlin and was funded under project number MIT1-2019–8. I am grateful

for being surrounded by people who always offered help and that is primarily Dr. Meermann,

the head of the division 1.1 – thank you. It would not be the same without my colleagues, as

well as BAM Staff - Maren, Dorit, Antje, Heike and of course, remarkable Mittagsessen dates

Silvana, Marcus O., and Martin (IsoAnalysis). The time spent with you at BAM was very special,

and I will always remember it. In addition, I want to say thanks, to my best friend Anne, for

making these last 3,5 years in Berlin memorable. Furthermore, I would like to thank my friends

from Sarajevo, as well as those who I had a chance (Marija, Emina) to meet and get acquainted

to during my stay in Berlin.

The fuel for making the clinker, later cement, later concrete – to serve later to build my

pathway to success was my Odej. I cannot imagine how would the pathway look without you.

Thank you for reminding me of my strengths and helping me accomplish my goals. I am very

grateful for having you in my life!

My mother Mirjana, both of my sisters Armana and Anesa with their families, are the people

to whom I owe my deepest gratitude. Being apart from you and my dogs all this time was just

a matter of kilometres, but in fact we were very close.

Finally, I want to thank my father.

Tata, hvala ti. Znam da si oduvijek vjerovao u mene.

Tebi posvećujem ovaj rad.

“Curiosity keeps leading us down new paths. “

Walt Disney

Abstract

Concrete is the most important human-made material because it serves as the basis of our

built environment. Since the properties of concretes are dominated by their key compound,

cement, devising a way to determine the cement’s origin, known as provenancing, is of great

importance for answering different provenance-related questions. These questions range

from liability issues when damage occurs to concrete-made structures, to forensic

investigations where cement particles are found at crime scenes. This thesis showcases the

use of isotope techniques to answer these provenance-related questions. Conventional

87Sr/86Sr and 143Nd/144Nd isotope and elemental ratios consisting of Ca, Sr, K, Mn, Mg, and Ti

are used as fingerprints for ordinary Portland cement (OPC) provenancing. The first part of

this thesis describes research previously conducted in fingerprinting cementitious materials,

providing an overview of provenance studies of cement and the main approaches commonly

used. In several studies, the origin of clinker for certain locations was determined via different

approaches. However, clinker is an intermediate product, which is available only at the

production site and therefore, the practical relevance is rather limited. Furthermore, the use

of Sr and Nd isotope systems, together with elemental fingerprints are presented as state of

the art in the field. Therein, the principal approach for the overall study is sketched. For the

second part of the study, a sample preparation technique for Sr isotopes in Portland cement

was developed. The aim was to find the most appropriate sample preparation procedure for

cement provenancing and selection was realised by comparing the 87Sr/86Sr isotope ratios of

differently treated OPCs with those of the corresponding clinkers. Based on these findings, the

third part of the study focused on the measurements of Sr and Nd isotope ratios, together

with elemental ratios, to establish a reliable technique for OPC provenancing. The outcomes

of this final stage are then used to establish a procedure for fingerprinting cements. This

becomes possible with the use of Sr and Nd isotope ratios and geochemical profiles. To

perform isotope ratio measurements and obtain reliable data, it was necessary to establish a

quality control procedure. Thus, an interlaboratory comparison (ILC) was organised to

characterise 87Sr/86Sr isotope ratios in geological and industrial reference materials by

applying the conventional method for 87Sr/86Sr isotope ratios. As reference material, four

cements (VDZ 100a, VDZ 200a, VDZ 300a, IAG OPC-1), one limestone (IAG/CGL ML-3) and one

slate (IAG OU-6) were selected, thus covering a wide range of Sr isotope signatures.

Zusammenfassung

Beton dient als Grundlage unserer gebauten Umwelt und ist deshalb einer der wichtigsten von

Menschenhand geschaffenen Materialien der Welt. Da die Eigenschaften von Beton von

seinem Hauptbestandteil Zement dominiert werden, ist die Entwicklung einer Methode zur

Bestimmung der Herkunft des Zements, von großer Bedeutung für die Beantwortung

verschiedener herkunftsbezogener Fragen. Diese Fragen reichen von Haftungsfragen bei

Schäden an Betonkonstruktionen bis hin zu forensischen Untersuchungen, wenn

Zementpartikel an Tatorten gefunden werden. In dieser Arbeit wird der Einsatz von

Isotopentechniken zur Beantwortung dieser herkunftsbezogenen Fragen nachgewiesen.

Konventionelle 87Sr/86Sr und 143Nd/144Nd-Isotopensysteme und Elementverhältnisse, die aus

Ca, Sr, K, Mn, Mg und Ti bestehen, werden als Fingerabdrücke für die Herkunft von

gewöhnlichem Portlandzement (OPC) verwendet. Der erste Teil dieser Arbeit beschreibt die

bisherige Forschung zum „Fingerprinting“ von zementhaltigen Materialien und gibt einen

Überblick über die Herkunftsbestimmung des Zements und die wichtigsten, üblicherweise

verwendeten Ansätze. In mehreren Studien wurde die Herkunft des Klinkers für bestimmte

Standorte mit unterschiedlichen Ansätzen bestimmt. Da es sich bei Klinker jedoch um ein

Zwischenprodukt handelt, das nur am Produktionsstandort verfügbar ist, ist die praktische

Relevanz eher begrenzt. Darüber hinaus wird die Verwendung von Sr- und Nd-

Isotopensystemen zusammen mit elementaren Fingerabdrücken als Stand der Technik auf

diesem Gebiet dargestellt. Darin wird der grundsätzliche Ansatz für die Gesamtstudie skizziert.

Für den zweiten Teil der Studie wurde eine Probenvorbereitungstechnik für Sr-Isotope in

Portlandzement entwickelt. Ziel war es, das am besten geeignete

Probenvorbereitungsverfahren für die Zementherkunft zu finden. Die Auswahl erfolgte durch

den Vergleich der 87Sr/86Sr-Isotopenverhältnisse von unterschiedlich behandelten OPCs mit

denen der entsprechenden Klinker. Auf der Grundlage dieser Erkenntnisse konzentrierte sich

der dritte Teil der Studie auf die Messung der Sr- und Nd-Isotopenverhältnisse zusammen mit

den Elementverhältnissen, um eine zuverlässige Technik für die Zementherkunft zu

entwickeln. Die Ergebnisse dieses Teils werden verwendet, um ein Verfahren für den

Fingerabdruck von Zementen zu entwickeln. Das Verfahren des Fingerabdrucks ist durch die

Verwendung von Sr- und Nd-Isotopenverhältnissen und geochemischen Profilen möglich. Um

Isotopenverhältnismessungen durchzuführen und zuverlässige Daten zu erhalten, war es

notwendig, ein Qualitätskontrollverfahren einzuführen. Daher wurde ein Ringversuch

iii

organisiert, um die 87Sr/86Sr-Isotopenverhältnisse in geologischen und industriellen

Referenzmaterialien unter Anwendung der herkömmlichen Methode für 87Sr/86Sr-

Isotopenverhältnisse zu charakterisieren. Als Referenzmaterial wurden vier Zemente (VDZ

100a, VDZ 200a, VDZ 300a, IAG OPC-1), ein Kalkstein (IAG/CGL ML-3) und ein Schiefer (IAG

OU-6) ausgewählt, wodurch ein breites Spektrum an Sr-Isotopensignaturen abgedeckt wurde.

Table of Contents

Acknowledgements ........................................................................................................... i

Abstract ............................................................................................................................. ii

Zusammenfassung ............................................................................................................ iii

1. Introduction .................................................................................................................. 1

1.1. General............................................................................................................................. 1

1.2. Previous cement provenance studies .............................................................................. 3

1.3. Isotopes in cement provenancing – The correlation ....................................................... 6

1.3.1. The background ........................................................................................................ 6

1.3.2. Sr seawater curve ................................................................................................... 11

1.4. Chemical separation and purification of analytes ........................................................ 13

1.5. MC-TIMS Technique ...................................................................................................... 14

1.6. MC-ICP-MS Technique .................................................................................................. 16

1.7. Other techniques for cement provenancing ................................................................. 17

1.8. Research objectives ....................................................................................................... 20

1.9. Research outline ............................................................................................................ 21

References ........................................................................................................................... 24

2. Publications ................................................................................................................ 34

2.1. Provenancing of cement using elemental analyses and isotope techniques –

the state-of-the-art and future perspectives ........................................................................... 34

2.2. Development of a sample preparation procedure for Sr isotope analysis of

Portland cements ................................................................................................................. 48

2.3. Fingerprinting Portland Cements by means of 87Sr/86Sr and 143 Nd/ 144Nd Isotope Ratios

and Geochemical Profiles ......................................................................................................... 74

3. Interlaboratory comparison on conventional 87Sr/86Sr isotope ratios as a

quality control indicator ................................................................................................ 106

3.1. lntroduction ................................................................................................................. 107

3.2. Study design and 87Sr/86Sr isotope ratio ...................................................................... 109

3.2.1. Study design .......................................................................................................... 109

3.2.2. 87Sr/86Sr isotope ratio ........................................................................................... 110

3.3. Reference Materials..................................................................................................... 111

3.4. Analytical procedures – General Overview ................................................................. 112

3.5. Study design by Lab15 ................................................................................................. 113

3.5.1. Chemicals .............................................................................................................. 113

3.5.2. Sample preparation .............................................................................................. 114

3.5.3. Measurements ...................................................................................................... 115

3.6. Results and discussion – Lab15 .................................................................................... 116

3.6.1. 87Sr/86Sr isotope ratios and associated uncertainties .............................................. 116

3.6.2. MC-ICP-MS and MC-TIMS comparison ..................................................................... 122

3.7. Conclusion .................................................................................................................... 123

References ......................................................................................................................... 124

4. Main results and discussion ....................................................................................... 129

4.1. Literature review and method proposal...................................................................... 129

4.2. Preliminary study ........................................................................................................ 130

4.3. The sample preparation method for Sr isotope ratios in OPC .................................... 131

4.4. The sample preparation method for Nd isotope ratios in OPC ................................... 133

4.5. Fingerprinting Portland Cements ................................................................................ 134

4.6. Recycled concrete– An example for fingerprinting future building materials ............ 137

4.7. Quality Control via performing an Interlaboratory Comparison Study ....................... 138

References ......................................................................................................................... 140

5. Conclusions and future recommendations ................................................................. 142

References ......................................................................................................................... 145

Bibliographic information ............................................................................................. 146

List of abbreviations ...................................................................................................... 147

Chapter 1

Introduction

This chapter presents the current research in cement provenance studies, previous attempts

in cement provenance determination, as well as the necessity of developing a new approach

for geographical origin determination of Portland cements. The relation between isotopes and

provenancing, including the importance of sample preparation procedure and isotope

measurements are also elucidated. The main technique used in this research is explained, and

other techniques which can be used for provenancing purposes are given. Finally, the ultimate

objectives of this research and a research outline are also detailed.

1.1. General

Concrete has played a major role in human life for several decades and continues to be

predominantly used in the present-day. The need and consumption of concrete only continues

to increase globally. Take for example that approximately 47.000 m3 of concrete was needed

for the construction of the Empire State building in New York [1], whereas in construction of

the Burj Khalifa, the tallest building in the world to date, around 330.000 m3 of concrete was

used. [2] These numbers symbolise the significance of concrete for the modern building

industry. Besides being a symbol of power, economic potential, and wealth, concrete is

necessary for providing protection against environmental influences, and the means for many

forms of traffic. Thus, it is not surprising that the production of its key compound, cement, is

constantly increasing. Cement is an important product in our daily lives, and it is an

irreplaceable part of numerous products. In Germany alone, almost 27 Mt of cement are used

every year [3] and the demand for cement is globally predicted to rise.

Knowing information about the geographical origins of a cement and having the possibility to

identify a particular cement sample would help in solving many issues relating to various fields.

Due to the impacts of weather and the environment we live in, concrete ages and therefore,

countless structural damages to houses, buildings and bridges occur worldwide. However,

these damages can also occur for other reasons, such as erroneous applications due to faulty

constructions or incorrect mixture ratios of the raw materials. Intending to prevent high costs

and those events which lead to the endangerment of human lives, cement provenancing, in

turn, can help in identifying the cause of failure in concrete structures. In most structural

failure cases, analysis and determination of cement origin is one of the key issues for

understanding the quality and the strength of the structure in question. The extensive body

of analytical data produced could thus be used when concrete failures are investigated, if

unequivocal identification of the cement present in a concrete is possible.[4]

Cement provenancing is also of great importance for other purposes, such as the recycling of

rubble or in the restoration of historic buildings [5-7] and monuments. [8] Testing laboratories,

e.g., those being part of governmental institutes and agencies, are also interested in cement

provenance studies, since they are often tasked in determining whether two cementitious

samples are identical or are of the same origin. [9] Besides researchers, questions of cement

provenance are also of interest to forensic investigators. Major crime laboratories located

worldwide often conduct crime-scene investigations involving the analysis of human-made

materials derived from geological raw materials, such as cement. [10] In forensic science,

traces or larger quantities of cement found at a crime scene can hold evidentiary significance.

The goal of investigating forensic samples is to determine a link between a sample at a crime

scene and a possible source of known origin.[11] Identifying and finding the origin of cement

can be beneficial for locating or reconstructing the crime scene, or it can serve as physical

evidence for use in court, by linking a suspect(s) to the victim(s) and/or the crime scene.

Furthermore, in forensic cases, when carrying out cement comparisons it is important to first

define the word “comparable” because no two physical objects can ever, in a theoretical

sense, be the same. [12] Similarly, a cement sample cannot be said, in an absolute sense, to

derive from one single place. Nevertheless, it is possible to establish with a high degree of

probability that a sample was or was not derived from a given origin. [12]

Therefore, in order to identify the origin of a cement sample, a reliable procedure with

estimated uncertainty was developed. In other words, cement provenancing requires

appropriate methods for sample preparation, chemical extraction/separation and

purification, analysis, and finally, interpretation of the results.

1.2. Previous cement provenance studies

To gain an understanding of the cement-provenance status quo and to identify the previously

applied methodologies in the field, it is important to first discuss already existing research

studies relevant to this topic. The first attempt to find the origin of a cement dates back to

1993., when Goguel and St John published an article about characteristic differences among

New Zealand cements in minor and trace element chemistry.[4] During their investigation,

they found that the composition of cement produced from individual plants in New Zealand

had varied little over time. Furthermore, they revealed that differences in major components

among the cements are insufficient to identify them in concrete. Thus, the authors focused

on identifying minor and trace elements, especially rare earth elements (REE) in cements,

since these elements are tightly fixed to the solid phases of aggregates and cements under

alkaline conditions. Cement samples were measured by semiquantitative Inductively Coupled

Plasma Mass Spectrometry (ICP-MS) and quantitative Atomic Absorption (AA), Atomic

Emission (AE) and Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES). Prior

to analysis, powdered cement samples required dissolution, which was accomplished by

digesting the sample overnight in a mixture of HF and HClO4, followed by dissolution in diluted

HNO3. The authors discussed the criteria under which elements can be considered useful for

cement identification in concretes. The first criteria are the independency of elemental

concentrations of fluctuations in rotary kiln temperatures. This means that volatile elements

such as As, Bi, Cd, Cs, Ge, Sb, Se, Th should not be considered for identification purposes.

Second, elements must not be mobile in the highly alkaline pore solutions of concrete. Thus,

elements such as Cr, Mo, W and U are not useful for identification purposes due to the

formation of anions that are soluble under strongly alkaline conditions, as they are likely to be

extracted from the aggregate during cement hardening. Applying these criteria, the authors

concluded that substantial differences between cements in Sr, Ba and Mn concentrations are

considered most useful for identification purposes. Furthermore, it was concluded that Ca/Sr,

Ca/Ba and Ca/Mn ratios have the highest potential for identifying cements in hardened

concretes, with Ca/Sr being the most accurately quantifiable measurand. Due to the small

variation in Ca/Sr over a long period of time at a production site, this ratio is suitable for

discriminating between production sites. When it comes to rare earth elements (REE), their

distribution pattern had more limited identification potential.

The second part of their study [13] focused on chemical identification of OPCs in New Zealand

concretes. The potential of using 1 mol·L-1 HNO3 and 0.2 mol·L-1 NH4-EDTA for selective cement

leaching from hardened concrete samples was examined and compared. The authors stated

that any method used for cement identification in hardened concrete must consider

unwanted contributions from the aggregates. Thus, coarse aggregates were excluded by

crushing the concrete and selecting the fragments for leaching. However, a small portion of

fine aggregates could not be excluded. Although hardened cement is easily dissolved in

mineral acids such as dilute HNO3, which successfully extracts less from the aggregates than

picric acid/methanol mixtures [14], complete discrimination against the aggregate was not

achievable. [13] The authors concluded that extraction of concrete with 1 mol·L-1 HNO3 and

determination of Ca, Sr and Mn contents and Ca/Sr ratio could in principle be sufficient to

identify OPCs from four major cement plants in New Zealand.

However, leaching with 1 mol·L-1 HNO3 is not applicable for aggregates substantially consisting

of limestone. [13] Furthermore, leaching concrete with 1 mol·L-1 HNO3 results in significant

extraction of the lanthanides from four of the major aggregate types used as concrete

aggregates in New Zealand. The application of lanthanide pattern to provenance the cement

in a hardened concrete is limited by contamination of lanthanides released from aggregates

during leaching. On the other hand, a one-hour EDTA leaching process did not produce reliable

data for the Ca-Sr-Mn plot because Mn from hardened cement dissolved too slowly in alkaline

EDTA. [13]

Since the investigated chemical parameters for cement provenancing to-date might leave a

degree of ambiguity, together with Graham I.J, the authors investigated further potential

approaches for cement provenancing in hardened New Zealand concretes. [15]

They tested the suitability of Sr isotopes to provide additional evidence for cement

provenancing. Cement and aggregate samples were digested in a mixture of HF and HClO4 and

then taken up in 0.2 mol·L-1 HNO3. Concrete samples were dried at 105°C, crushed in a mortar,

sieved and a 0.5 – 1.0 mm fraction was collected. The sieved fraction was then shaken for one

hour at room temperature in the excess of 1 mol·L-1 HNO3 or 0.2 mol·L-1 NH4 EDTA at pH 7,

followed by centrifuging and decanting the clear liquid. Sr and Ca concentration analysis was

undertaken using ICP-AES. The extraction of strontium for isotopic analysis was performed

using standard cation exchange resin.

87Sr/86Sr was adjusted to 86Sr/88Sr = 0.1194 and results normalised to an SRM 987 value of

0.71025 (Conventional method). Determined conventional 87Sr/86Sr isotope ratios for the New

Zealand cements ranged from 0.7076 to 0.7120, with most being close to 0.7080. The authors

concluded that one important factor to be considered is gypsum, of which approximately 5

wt.% is added to the final clinker before crushing. Gypsum can have a significant effect on the

87Sr/86Sr isotope ratio of cement, particularly when coupled with low Sr content in the clinker.

Another conclusion was that the combined chemical and Sr isotopic analysis of commonly

used New Zealand cements has shown that they contain characteristic fingerprints, which may

be used to identify them in a concrete of unknown origin. However, in this study the

investigated OPC and concrete samples dated back to the 1980s or earlier, which might not

be comparable or analogous to modern cement. Furthermore, the number of cement plants

is rather low in New Zealand and neighbouring countries play no role in this context.

Additionally, even though cements majorly differ concerning their 87Sr/86Sr isotope ratios,

there are cements which have a very similar 87Sr/86Sr isotope ratio which makes distinguishing

between them based solely on this parameter rather complex.

These initial findings triggered other researchers to perform studies focused mostly on trace

element analysis in cements [16,9,17,18] and also, in clinkers [19-23,17,24-27,18,28] and

concrete [29] with the purpose of its provenance determination. The published literature

revealed that mostly clinkers were analysed, whereby cements were investigated far less.

These studies mostly relied on multivariate statistical analysis of trace elements´ distribution

in cements. However, such methods should not be used for evaluation of all elements. This is

because these elements either have no relation to the geographic origin or they are related

to additives. For example, elements coming from fuel used in clinker production (e.g., used

tires, heavy fuel oil, etc.) should not be used for provenance determination, since their origin

is not interlinked with the origin of the raw materials (limestone, clay, or the natural mix of

both) used for cement production. In addition, these constituents can also vary over time, so

that there is no uniform incorporation into the cement. Furthermore, in these studies the

resulting data interpretation required different visualisation methods (e.g., star-plots) for

comparison.

1.3. Isotopes in cement provenancing – The correlation

1.3.1. The background

Different isotopes of an element contain the same number of protons, thus the same atomic

number, yet a different atomic mass by virtue of having different numbers of neutrons. For

example, the hydrogen atoms within a sample of water will be comprised of some mixture of

hydrogen-1, hydrogen-2, and hydrogen-3 (i.e., 1H, 2H and 3H). Of these three naturally

occurring isotopes, 1H and 2H are stable and 3H has a half-life of 12.32 years. [30]

In contrast, a radiogenic isotope is a stable daughter isotope produced from the decay of a

radioactive parent (so-called mother) isotope. Differences in radiogenic isotopes can be

utilised for different purposes, such as: dating the time of formation of rocks and minerals

[31]; tracing the sources and transport of dissolved and detrital constituents in

biogeochemical, hydrologic, sedimentary cycles and ecosystems [32,33]; reconstructing

temporal changes in ocean water chemistry [34,35] and stratigraphic studies [36]; provenance

determination of different materials; as well as in the investigation of counterfeit products,

e.g., foodstuffs. [37]

Strontium, atomic number 38, is an element that contains four isotopes, namely those of 84,

86, 87, and 88, with the approximate proportions of 0.56: 9.87:7.04: 82.53 [35]. Geologically

short-lived nuclides, such as 90Sr, are not considered in this thesis. The above abundances are

however somewhat variable because 87Sr is a radiogenic isotope, generated by the emission

of a negative 𝛽𝛽− particle from 87Rb. This process is presented by the Eq. 1:

Rb →Sr

87 + 𝛽𝛽−+𝜈𝜈+𝑄𝑄 (1)

Where 𝛽𝛽−is the 𝛽𝛽−particle, 𝜈𝜈 is an antineutrino, and 𝑄𝑄 represents the decay energy of 0.275

MeV.[35]

Radioactive decay is the process of spontaneous disintegration of a radioactive parent isotope

to produce a radiogenic daughter isotope and a nuclear particle. The driving force of a decay

process is the instability of the radioactive parent isotope. This depends on factors such as

individual nucleus’ configuration of protons and neutrons, and its binding energy (per atomic

particle), which generally decreases with increasing atomic weight above mass 56. [38]

Radioactive decay is expressed by the following equation, where N is the number of

radioactive isotopes [39]:

𝑑𝑑𝑑𝑑

𝑑𝑑𝑑𝑑 =−𝜆𝜆𝜆𝜆 (1.1)

with 𝜆𝜆 being the decay constant. The half-life, T1/2 is related to 𝜆𝜆 as

Τ1/2 =ln 2𝜆𝜆

� = 0.693 / 𝜆𝜆 (1.2)

When Eq. (1.1) is integrated, the resulting Eq. is

Ν𝑑𝑑=Ν0𝑒𝑒−𝜆𝜆𝑑𝑑 (1.3)

Ν0 is the initial number of the radioactive isotope and t is the elapsed time from the start. The

number of the daughter isotope is represented as:

D𝑑𝑑= D0+ (Ν0−Ν𝑑𝑑) = D0+Ν0(1 −𝑒𝑒−𝜆𝜆𝑑𝑑) (1.4)

In geochemistry, T = 0 and T is the age of the sample. Here, Ν0=ΝP𝑒𝑒𝜆𝜆𝜆𝜆 from Eq. (1.4) and

D𝑑𝑑= DP, where the suffix “P” indicates “present”. Therefore, Eq. (1.4) changes to

DP= D0+Ν0 �1−𝑒𝑒−𝜆𝜆𝑑𝑑�= D0+ΝP�𝑒𝑒𝜆𝜆𝜆𝜆 −1� (1.5)

87Rb decays into 87Sr by beta emission. When we consider the decay of 87Rb to 87Sr, from Eq.

(1.4) the following Eq. can be obtained:

Sr =

PSr

8787 +Rb (𝑒𝑒𝜆𝜆𝜆𝜆 −1) (1.6)

Where 87RbP is the “present” amount of Rb and T is the age of the isotopic system. Since Sr

has a stable isotope (86Sr), whose number does not change over the time T, then (86Sr)P =

(86Sr)0:

�Sr

87Sr

86 �P=�Sr

87Sr

86 �0+�Rb

87Sr

86 �P�𝑒𝑒𝜆𝜆𝜆𝜆 −1� (1.7)

This equation represents the basic formula for age dating using Sr isotope ratios. In practice,

(87Sr/86Sr)p and (87Rb/86Sr)p are measured in several phases. If all phases became isotopic

equilibria at age T and did not suffer any disturbances from T to present, the data forms the

line which is known as isochron, as shown in Fig. 1.1. The slope corresponds to the age, and y-

intercept corresponds to the initial value of the (87Sr/86Sr)p at the T=0.

Fig.1.1. Illustration of the Rb-Sr isochron method of dating, adapted from [40].

Radioactive decay has thus been used for the development of techniques in the dating of rocks

and minerals by means of the radioactive isotopes of parent elements and the corresponding

stable isotopes of their daughter elements, e.g., Rb – Sr. The 87Sr/86Sr terms in Eq. (1.7) are

usually expressed directly as the ratios of the two isotopes.

There are several notations used in the scientific community for expressing Sr isotopic ratios

which must be differentiated. For example, the importance of differentiating between

87Sr/86Sr conventional isotope ratio (in further text 87Sr/86Sr isotope ratio) and absolute

isotope ratio (also known as isotope amount ratio). Dating can only be performed by using

isotope amount ratios n(87Sr)/n(86Sr) [41]. In addition, “delta notation”, e.g. 𝛿𝛿𝑆𝑆𝑆𝑆𝑑𝑑𝑁𝑁𝑁𝑁𝜆𝜆 𝑁𝑁𝑆𝑆𝑆𝑆 987

[42] is also used in geochemistry.

Concerning 87Sr/86Sr isotope ratio, the 87Sr/86Sr isotope ratio of NIST SRM 987 determined by

the conventional method is accepted as being 0.710 250 mol·mol-1 [43] in contrast to the

certified value n(87Sr)/n(86Sr) = (0.710 34 ± 0.000 26) mol·mol-1. [44] In addition, more than

1900 published results listed in the GeoReM database in 2019 for the 87Sr/86Sr isotope ratio of

NIST SRM 987 yield 0.710 250 mol·mol-1 as median with an expanded uncertainty of 0.000 001

mol·mol-1 with a coverage factor of k = 2. [45]

In addition to Sr, another element important to the provenancing discussion is Neodymium.

It contains five stable isotopes (142Nd, 143Nd, 145Nd, 146Nd, and 148Nd) and two radioisotopes

(144Nd and 150Nd) with half-lives of 2.29 x 1015 and 6.7 x 1018 years [46] respectively.

Due to their very long half-lives, 144Nd and 150Nd are regarded as stable isotopes. 147Sm decays

to 143Nd via α-decay and a half-life of 1.06 x 1011 years. [47] Furthermore, 146Sm decays to

142Nd via α-decay and a half-life estimated between 68 x 106 years [48] and 103 x 106 years

[49]. Since the alpha decay of 147Sm forms 143Nd, the Nd isotopic comparisons are expressed

in terms of the 143Nd/144Nd isotope ratio.

The decay of the radioactive samarium isotope 147Sm via alpha decay is represented by the

reaction:

Sm →Nd

143

147 +𝛼𝛼+𝑄𝑄 (1.8)

Where 𝛼𝛼 is an alpha particle, and 𝑄𝑄 represents the decay energy of 2.23 MeV. [50]

Isotopic variations in Nd result from the radiogenic growth of 143Nd in reservoirs with varying

Sm/Nd ratios. These isotopic variations are expressed relative to the stable, non-radiogenic

isotope 144Nd (143Nd/144Nd isotope ratio). The radiogenic growth of 143Nd is analogous to that

of 87Sr. [38]

For Sm – Nd isotopic system, the following Eq. 1.9 applies:

�Nd

143Nd

144 �P=�Nd

143Nd

144 �0+�Sm

147Nd

144 �P�𝑒𝑒𝜆𝜆𝜆𝜆 −1� (1.9)

These equations (1.7 and 1.9) can also be used to calculate Sr, and respectively Nd isotope

ratios in a sample at a certain age.

Sr and Nd also include stable isotopes with primordial origins (e.g., 86Sr) or radioisotopes with

very long half-lives. This is important, since relative abundances of Sr and Nd isotopes within

rocks and minerals are considered constant in environmental studies on time scales of <104

years but can vary significantly depending on their age of formation and the ratios of parent

and daughter elements (Rb/Sr, Sm/Nd). Unlike traditional stable isotopes of light elements (H,

C, N, O and S) which undergo isotope fractionation during chemical transformations and

changes of state, heavy elements such as Sr and Nd experience only minimal isotopic

fractionation. This is due to their small relative differences in the mass between their stable

isotopes. [43,51] The mass dependent isotope fractionation of Sr and Nd, both of which have

several non-radiogenic stable isotopes, can be compensated for during analysis, which is done

to some extent in the conventional isotope ratio method. [52]

These isotope systems are therefore used not only for dating rocks, but also for studies of

material origin and applications which include provenance determination of the raw

materials.

Due to the radiogenic in-growth of 87Sr and 143Nd by radioactive decay of 87Rb and 147Sm over

time, the isotopic compositions changed to higher values. The extent of radiogenic in-growth,

and thus today’s Sr and Nd isotopic compositions in the different earth reservoirs, is a function

of the half-lives of 87Rb and 147Sm, and both the age and parent to daughter ratio of the

individual geochemical reservoirs (e.g., earth mantle, continental crust and seawater) or rocks

(basalts, granites or sediments). [53]

Therefore, two factors that control the Sr and Nd stable isotope ratios of geologic materials

are: the age of the material; the initial ratio of parent to daughter elements, where higher

parent to daughter ratios lead to a higher abundance of radiogenic Sr and Nd. [53]

As an alkali element, Rb occurs as the Rb+ ion in an aqueous solution, whereas Sr, an alkaline

earth, occurs as the Sr2+ ion. Due to the important role that ionic size and charge plays in

governing an element’s geochemical behaviour, Rb and Sr will be incorporated into minerals

forming from solutions in different amounts. Thus, alkaline elements such as K and Rb show a

positive correlation with each other but a negative correlation with alkaline earth elements

such as Ca and Sr. Strontium preferentially substitutes for Ca, and Rb for K, in the crystal

lattices of natural minerals. In carbonate mineral lattices, Sr2+ substitutes at Ca2+ sites, whereas

Rb+ is comparatively excluded from Ca2+ sites. [38] The Earth’s crust and mantle have distinct

ranges of Rb/Sr ratios owing to their distinct mineral assemblages. Therefore, a sample with a

high Rb/Sr ratio will evolve to a higher present day 87Sr/86Sr value than a coeval sample with a

low Rb/Sr ratio.

The ancient granitic rocks, which are dominant in the continental crust, have elevated Rb/Sr

ratios and 87Sr/86Sr isotope ratios, whereas the mantle-derived basalts, dominant in the

oceanic crust, have low Rb/Sr ratios and 87Sr/86Sr isotope ratios. Volcanic rocks also have low

87Sr/86Sr isotope ratios, reflecting their derivation from the mantle material. Nd isotope ratios

generally negatively correlate with Sr isotope ratios since the ultramafic rocks within the

mantle exhibit higher Sm/Nd ratios than rocks within the continental crust. [43,54,52]

Isotope systems which contain radiogenic isotopes, such as 86Sr/88Sr and 143Nd/144Nd, show

wide geographical variations that reflect changes in the geological sources of these elements.

Therefore, these systems can be linked to their geographic origin, as they reflect the

mineralogical composition and geological history of the raw material. These elements have a

much smaller degree of isotope fractionation than the light stable isotopes, which enables the

mass-dependent fractionation of Sr and Nd to be compensated for. The mass dependent

effects which can occur are mostly within the noise of the measurement and surely much

smaller than the regional natural variations that are observed in provenancing applications.

Therefore, scientists may gain unique insight into the material's geographic source and/or

production history. [55,36]

In the past, either one of them or both Sr and Nd isotope fingerprints have been used for

studying and provenancing archaeological materials, such as gypsum mortars [56],

archaeological glass [57-60] and ceramics. [61,62] Thus, the differences in Sr and Nd isotopic

fingerprints in cements can be used for provenance determination.

1.3.2. Sr seawater curve

The ability to date and correlate marine geological materials using Sr isotope ratios, namely

87Sr/86Sr relies on the fact that the 87Sr/86Sr value of Sr dissolved in the world’s oceans has

varied over time. [35] Variations of 87Sr/86Sr through time are presented in a curve, well-

known as the “seawater curve”, which was plotted using 4119 data-pairs (Fig. 1.2).

Comparison of the measured 87Sr/86Sr in a marine mineral with this seawater curve can yield

a numerical age for the mineral. The use of the curve works only for marine minerals. It rests

on the assumption that the world’s oceans are and have always been homogeneous with

respect to 87Sr/86Sr. Such uniformity of 87Sr/86Sr is expected, due to the far longer residence

time of Sr in the oceans today (≈ 106 years) in comparison to the time that it takes currents to

mix the oceans (≈ 103 years). Thus, the oceans are thoroughly mixed on time scales that are

short relative to the rates of gain and loss of strontium. The degree to which this was true in

past times is unknown. [63]

Fig. 1.2. Variation of 87Sr/86Sr through time, adapted from [63].

The Phanerozoic Eon is the current geologic eon in the geologic time scale, covering

538.8 million years to the present. [64] The Phanerozoic seawater Sr isotopic curve was

originally derived from bulk carbonate samples [65], but subsequent studies have included

biogenic carbonates and phosphates, including foraminifera, ammonites, brachiopods, and

conodonts. Furthermore, newly updated seawater Sr isotopic curves were proposed based on

carbonates and fossils to several ages, as McArthur and co-authors [66,63] for 0 to 509 Ma,

Saltzman and co-authors [67] for Ordovician (485.4 – 443.8 Ma [64]), and Dudas and co-

authors [68] for the Permian-Triassic boundary. Phanerozoic seawater 87Sr/86Sr has varied

between the limits of crustal and mantle values, thereby reflecting the combined inputs from

these two sources to the world’s oceans. The curve has been age-calibrated for dating

purposes, which is the basis for Sr isotope stratigraphy.

Conversely, neodymium has a short residence time (≈ 500–2000 years), considerably shorter

than ocean mixing time, which is approximately 1000 years [69], therefore no seawater curve

can be established.

1.4. Chemical separation and purification of analytes

When it comes to selecting a suitable sample preparation procedure for isotope ratio analysis,

there are two important viewpoints to be considered. Prior to measurement of isotope ratios

using mass spectrometry, elements of interest must be extracted from cement and purified in

order to minimise mass spectral interferences. The elements of interests must be of high

purity when introduced into the mass spectrometer due to two main reasons: first, other

elements or compounds may have isotopes of similar mass that interfere with the mass

spectrum produced by the element of interest. Second, the interfering isotopes may inhibit or

alter the ionisation of the element of interest.

Thus, it is very important to properly extract the element of interest from the matrix and

perform elemental separation. Since elements of interest in this study are Sr and Nd in

cement, their purification requires dissolution of cement samples using carefully selected acid

or acid mixtures which will attack the intended phases (mineral assemblages) of cement. After

dissolution, elemental separation is accomplished by ion exchange chromatography, wherein

Sr and Nd are separated from the matrix by using acids and resin beads produced from

synthetic polymers. The ion exchange process is conducted by adding the dissolved cement

sample to a narrow column (made either of Teflon, PP, or glass) containing resin beads,

followed by adding acids of varying strength and type selected to partition the Sr and Nd.

Another important thing to be considered is the medium in which the elements are going to

be extracted. As already mentioned, those are usually acids or acid mixtures. The selection of

the acid is critical due to the nature of the material, meaning the mineralogical composition

and its complexity.

The importance of acid selection in Sr isotope ratios measurements was discussed in an article

by Wang and co-authors. [70] They investigated Sr isotopes in fly ashes, since they are utilised

in tracing the coal ash contaminants in impacted water resources. They found that the

87Sr/86Sr isotope ratios in the water leachates were significantly lower than that of the bulk fly

ash. It was concluded that distinction between the Sr isotope ratios of the bulk coal ash and

water-soluble Sr is important in understanding how to properly use strontium isotopes as a

forensic tracer for the identification of coal ash contamination in the environment.

Therefore, in the research conducted here, both Sr and Nd in cement samples had to be

prepared with care (see Chapter 2).

Furthermore, due to the possibility of contaminating the sample with extraneous material

introduced during the sample collection, preparation, pre-treatment, and chemical separation

procedures, the level of contamination (i.e., blank level) must be monitored and preferably,

kept low. For example, a certain amount of analyte (Sr and Nd) may be introduced to a cement

sample from airborne dust, saw blades, epoxy, polishing media, nitric acid, distilled water,

beakers, centrifuge tubes, contact with human hands, etc. [71]

By handling, purification of reagents, acid cleaning of labware, working in clean rooms, or

clean hoods with filtered air under laminar flow, use of Teflon and quartz glass labware, and

scaling down the number of materials used in these procedures, blanks can be maintained at

negligible levels, e.g., < 5 x 10-11 g for sample sizes of 5 x 10-8 g of Sr (e.g. [72]). This corresponds

to a sample size of 0.5 mg for calcite that contains 100 mg·kg-1 Sr. [38]

1.5. MC-TIMS Technique

Even though several instrumental techniques allow information on the isotopic composition

of elements of interest, mass spectrometry is undoubtedly the most versatile and powerful

technique. [73] Schematic representation of the essential parts of mass spectrometric

instrumentation are shown in Fig. 1.3.

For radiogenic isotope measurements while using a multi-collector thermal ionisation mass

spectrometer (MC-TIMS), the element of interest (e.g., Sr) is introduced into the ion source of

a thermal ionisation mass spectrometer as a solid salt loaded on a metal filament. There are

two possible filament geometries: single (evaporation and ionisation of the analyte is

accomplished with one filament) and double/triple filament geometry (evaporation and

ionisation occur on separate filaments placed parallel to each other). [55] The following

processes can be described in two steps: the first step involves the formation of atom ions on

a heated filament, while the following step involves the separation of these ions in a mass

spectrometer according to their mass/charge ratio. As a source, the surface material is in the

shape of a flat ribbon filament, and a liquid solution, previously separated from the matrix, is

deposited in the centre of the filament. Due to the applied current, the filament is heated, and

the analyte is evaporated and ionised. These ions are then accelerated via high voltage

through a magnetic field (in charge as mass separator), which deflects and separates ions of

different mass (and the same charge) into a mass spectrum (e.g., 84Sr, 86Sr, 87Sr, and 88Sr).

Lighter mass isotopes are more strongly deflected than heavier isotopes. The relative amounts

of the different masses are measured by the current produced when they strike ion detectors

(e.g., Faraday cups or secondary electron multipliers) positioned downstream from the

magnetic field.

Strontium isotope ratio measurements concern the simultaneous measurement of four stable

isotopes, namely 84Sr, 86Sr, 87Sr, and 88Sr. The only radiogenic isotope of Sr is 87Sr and it forms

due to the beta decay of 87Rb, as explained earlier. This means that the ratios of non-

radiogenic 84Sr, 86Sr, and 88Sr only change in case of isotope fractionation, with the isotopic

fractionation of 87Sr relative to 86Sr being almost half that of 87Sr relative to 88Sr. Setting the

86Sr/88Sr isotope ratio at a fixed value (0.1194) enables an approach that allows compensation

for changes in the 87Sr/86Sr isotope ratio due to isotope fractionation and the precise

determination of the abundance of radiogenic 87Sr in geological materials.

Neodymium isotopes are corrected in a similar fashion to the correction of 87Sr/86Sr

measurements by setting the 86Sr/88Sr isotope ratio to 0.1194, with Earth-derived materials

having 143Nd/144Nd isotope ratios that are corrected to a 146Nd/144Nd isotope ratio of 0.7219,

the chondritic uniform reservoir (CHUR) [74,75,54] value.

Fig. 1.3. Schematic representation of the essential parts of mass spectrometric

instrumentation, adapted from [73].

1.6. MC-ICP-MS Technique

While developing the most sophisticated instruments used to determine isotopic

composition, the goal is always to measure isotope ratios with the highest precision possible.

Building on the high precision achieved with MC mass spectrometers for thermal ionisation

ion sources, Walder and Freedman [76] described an instrument with an integrated ICP ion

source to an MC platform. In that way, both the versatility of an ICP source and the flexibility

of a variable MC array can be achieved. [77] Thus, multi-collector inductively coupled plasma

mass spectrometry (MC-ICP-MS) became an essential analytical technique in isotope

geochemistry. The initial intention for MC-ICP-MS, since the introduction of the first MC-ICP-

MS in 1992 (the Plasma 54 by VG Elemental), was that it would simplify measurements

normally made with MC-TIMS. [78] In addition, while MC-TIMS is a powerful ionisation

technique for various elements, it does have problems in ionising elements with high first

ionisation potentials (e.g., Hf). This challenge is overcome using ICP as ion source, since its

delivers high ionisation efficiency for almost all elements across the periodic table.

The principle of MC-ICP-MS is based on the formation of charged atomic ions in an inductively

coupled argon plasma at the temperature of almost 10 000 Kelvins. When sample aerosol

enters an ICP, the droplets undergo desolvation, molecules are broken down into the

constituting atoms and these are then ionised via electron impact and Penning ionisation,

which entails ionisation as a result of energy transfer from an excited Ar atom. These ions are

transferred from the plasma source at ambient pressure into a mass separator operated at

high vacuum via a set of cones. The next step is separation of these ions in a mass separator

according to their mass/charge ratio. In ICP-MS devices, argon plasma is under atmospheric

pressure, making handling very easy.

MC-ICP-MS shows several key features, including high-precision isotope ratio measurements,

simultaneous detection, and high sensitivity for all elements, among others. Over the past two

decades, the numbers of publications reporting on the use of MC-ICP-MS has grown

significantly, and continues to grow. [77]

MC-TIMS and MC-ICP-MS used as instrumental techniques for isotopic analysis in this research

are shown in Fig. 1.4.

Fig. 1.4. MC-TIMS Sector 54 (left, Micromass Ltd., Manchester, UK) and MC-ICP-MS Neptune

Plus (right, Thermo Scientific, Bremen, DE) at BAM, Berlin, DE.

1.7. Other techniques for cement provenancing

In order to analyse all important parameters which help in distinguishing cements and

determining their provenance, the application of varying techniques can be beneficial in

acquiring enough information about the samples themselves. Therefore, cement

characterisation sometimes requires a multidisciplinary approach, which combines

descriptive, analytical, and spatial information such as mapping. In principle, interpretation of

results depends on which technique was used for the cement provenancing.

Seeing as there are a variety of possible techniques and methods which can be employed

regarding cement provenance determination, the selection criteria presented herein, namely

of which available methods were the best suited and the proper techniques involved in those

methods, were based on literature research. Wherein analyses were carried out not only on

cement, but also on other materials, such as glass.

If the cement particles are incorporated in a sample of a different matrix, e.g., soil samples,

the first step should be examination of morphological features via microscope. Optical and

scanning electron microscopy can help in identifying the different components of forensic

samples, allowing recognition of the presence or absence of human-made particles, e.g.,

cements. [79]

The commonly used techniques for major, minor, and especially trace elements analysis are

ICP-OES and ICP-MS. These techniques are very useful and sufficiently precise for establishing

elemental patterns of various materials, including cements, as described in Section 1.2.

Another important technique to mention is Atomic Absorption Spectrometry (AAS), which

can be used in chemical analysis of the major and minor elements (Fig. 1.5), as described in

Galan and co-authors. [5] The authors used major and minor elements of building and quarry

sandstone samples for locating the original quarries used for constructing historical buildings

such as the Malaga Cathedral in Spain. This methodology helped them in identifying the

precise quarries from which the stones of a monument were taken.

Fig. 1.5. Major and minor elements of building and quarry limestone samples (average values

for each group). Adapted from [5].

The most important techniques with meritorious history for examining different sample types,

including those samples of irreplaceable and great cultural value, are X-ray diffraction (XRD)

and X-ray fluorescence (XRF) spectrometry. The usefulness of these techniques was proven in

various scientific fields, also in combination with other techniques, such as light and electron

microscopy. X-ray diffraction examinations are non-destructive, and therefore leave the

original specimen intact (e.g., for restauration purposes) and available for further analytical

study (e.g., in forensic science analysis). [80,81] For example, XRD technique revealed the type

of minerals present in the clay fraction of dust from a shipment of contraband ivory, enabling

them to determine the original location where the ivory was packed. [82]

The XRD technique can also assist in solving forensic cases including crime-scenes, such as that

of a double murder case in Australia, where it was necessary to identify soils present on a tool

and link them to a quarry. [83] Furthermore, XRD analysis was also successful in identifying

the provenance of industrial dust that had settling on parked vehicles. The mineralogy of these

dust particles enabled them to be identified as coming from cement plants. [84]

When it comes to the XRF spectrometry group of techniques, in combination with other

techniques, they provide valuable information about investigated samples. For example, μXRF

imaging was used for selecting those chemical elements enriched or depleted in the glaze

which serve as provenance and technological markers. [85] Portable XRF spectrometry has

undergone significant technological improvements in recent years and is now used in a wide

range of applications. [86] In the study conducted by Nakai and Shirataki [87] it was

demonstrated that the non-destructive energy dispersive XRF analysis can offer rich

information as to the characteristic chemical composition of archaeological glass and thus

help in locating the origin of said glass. However, the authors concluded that the sodium

content, which largely decreases due to surface weathering caused in turn by the long-term

underground burial conditions, impacts and affects the results of some other oxides. This was

also studied by Li and co-authors [88], and presents the limitation of the non-destructive

analysis of archaeological glass using the XRF technique.

In principle, both XRD and XRF could be used for determining provenance of cement and

concrete particles found at crime scenes, and their possible correlation with samples from a

suspected quarry.

Another technique which could give additional information about provenance is

cathodoluminescence analysis, which can help distinguish between several generations of

carbonate cements. [89]

Since alkali in cement is responsible for the Alkali–silica-reaction phenomenon that manifests

itself in the form of premature cracking in concrete structures such as bridge decks and

concrete pavements, information on cement alkali quantification can be found using Fourier

transform infrared spectroscopy (FTIR). [90]

Two other promising techniques include instrumental neutron activation analysis (INAA) and

laser ablation molecular isotopic spectrometry (LAMIS). Meyers and Vanzelts [91] used INAA

in their pilot study to distinguish between objects made of limestone from different sources.

When it comes to LAMIS, its application for performing optical isotopic analysis of solid

strontium-containing samples in ambient atmospheric air at normal pressure was explained

in a study published by Mao and co-authors [92], where measurement of strontium isotopes

showed promising results.

LA-ICP-MS proved to be a very useful tool to analyse trace elements present in 8th and 15th

century Islamic glass from the Middle East, in order to provide evidence for separate

production zones. [93] In the same study, Electron probe microanalysis (EPMA), along with

LA-ICP-MS, was carried out to provide more information about the glass samples.

However, for cements which contain a substantial amount of Rb, the use of LA-ICP-MS for

87Sr/86Sr isotope ratio analysis is not recommended. This is due to the 87Rb interference. In

fact, it must be removed prior to measurement, which is not possible for in-situ LA-ICP-MS.

Among all these listed methods, there is no one that, taken individually, can guarantee a

proper, unequivocal provenance result. Thus, it is very important to establish an approach

which allows low sample size and clearly developed means for Portland cement provenancing.

1.8. Research objectives

Portland cements are expected to carry a characteristic chemical imprint reflecting the

difference in geological environment of their raw materials. This characteristic chemical

imprint can be considered as a “fingerprint” for every cement sample stemming from an

individual quarry. Developing a method for provenance determination via the use of isotope

techniques could help in various scientific fields, such as solving liability issues when cement-

made structures show defect or even fail.

Therefore, one of the main objectives of this work was studying the potential use of

geochemical fingerprints of cements in provenance determination. Preliminary investigations

showed that, to determine Sr and Nd isotope fingerprints of cements, the development of a

sample preparation procedure for Sr isotope analysis would be necessary.

Thus, the first step in my research was reviewing relevant literature, in order to obtain a

deeper insight into the field of cement provenance studies and their corresponding research,

including methods investigated in the past.

Since commercial cements consist of multiple components, no detailed investigation into their

individual 87Sr/86Sr isotope ratios or their influence on the integral 87Sr/86Sr isotope ratio of

the resulting cement had been conducted previously. Thus, the second step was to investigate

four different methods of sample preparation procedures to find the best suitable approach

for cement sample preparation for the Sr isotope analysis.

Based on the results of the first two stages, a third part of the study was designed and included

Sr and Nd isotopic fingerprinting. In addition to Sr and Nd isotopic systems, carefully selected

elements, and their mass fractions, were used to form elemental ratios. Using these ratios,

together with the Sr and Nd isotope ratios, distances between cement samples were

computed. The next step was employing Divisive Analysis (DIANA) [94] to obtain a hierarchical

clustering, which increased the analytical resolution of the cement fingerprinting approach.

For this purpose, twenty-nine OPC samples from worldwide cement plants were investigated.

MC-TIMS was used to determine the isotope ratios in OPCs, and X-Ray Fluorescence (XRF) was

used in determining elemental mass fractions. Additionally, quality control was achieved by

organising and participating in the international interlaboratory study.

1.9. Research outline

The research outline is presented schematically in a Fig. 1.6.

Fig. 1.6. Schematic representation of research outline.

The experimental research presented can be found in three scientific publications, whereby

one is accepted, and two were published in peer-reviewed journals. Moreover, further work

which is submitted is also presented as part of this thesis.

Accordingly, this dissertation is structured in five chapters, as follows:

Chapter 1 contains general information and explains in detail the scientific background for the

use of Sr and Nd isotope ratios, as well as elemental ratios in cement provenancing.

Chapter 2 and 3 present the main body of the thesis where the peer-reviewed publications

and major work are included. Chapter 2 contains all scientific publications. The first peer-

reviewed publication is a literature overview encompassing the current research state and

future perspectives for cement provenancing using elemental analyses and isotope

techniques.

The published review article therefore provides an overview of cement provenance studies

and the main approaches commonly used. Moreover, the potential of isotope techniques in

regard to provenance studies in cement and concrete research was discussed in detail, proving

that a suitable approach could be found by using isotope ratios. The combined approach was

thereby proposed for determining the provenance of cements.

The second peer-reviewed publication is a research paper focusing on the development of a

sample preparation procedure for the Sr isotope analysis of Portland cements. Therefore, the

determination and comparison of the 87Sr/86Sr isotope ratios of a diverse set of Portland

cements and their corresponding Portland clinkers (the major component of these cements),

was made. The aim was to find the most appropriate sample preparation procedure for

cement provenancing, and the selection was realised by comparing the 87Sr/86Sr isotope ratios

of differently treated cements with those of the corresponding clinkers.

The third peer-reviewed publication is a research paper focusing on the development of an

analytical procedure for cement provenancing and applying this methodology to a set of

twenty-nine ordinary Portland cements. For this purpose, 87Sr/86Sr and 143Nd/144Nd isotope

systems and interelement ratios of Ca, Sr, K, Mn, Mg, and Ti were used as fingerprints. Therein,

the first database of Sr and Nd isotope systems was created through the investigated OPCs

stemming from twenty-nine cement plants located worldwide.

High quality isotope analysis can only be obtained when the analytical procedure has been

initially validated and the quality of further applications is continuously monitored via the use

of quality control samples. Thus, Chapter 3 presents the submitted work conducted on the

characterisation of conventional 87Sr/86Sr isotope ratios in cement, limestone and slate

reference materials based on an interlaboratory comparison study (ILC). The ILC study was

conducted as a part of this thesis. Thus, in this chapter, results from my individual research (as

organiser and as participant) are presented and compared.

In Chapter 4 all the results and achievements presented in this work are collected and

summarised. In addition to OPCs, the same approach was applied to recycled concrete,

seeking a comparison of applicability of the proposed approach to more sustainable materials

occurring in the construction industry. Chapter 5 presents the general conclusions drawn from

this thesis and recommendations for future research.

References

1. Penafiel K (2006) The Empire State Building: An Innovative Skyscraper.

https://www.buildings.com/feature/article/10193728/the-empire-state-building-an-

innovative-skyscraper. Accessed 25.09.2022

2. PJSC EP Architecture: Burj Khalifa Emmar. https://www.burjkhalifa.ae/en/the-

tower/design-construction/. Accessed 25.09.2022

3. (Ed.) V (2018) Activity Report 2015-2018

4. Goguel RL, Stjohn DA (1993) Chemical-Identification of Portland Cements in New-Zealand

Concretes .1. Characteristic Differences among New-Zealand Cements in Minor and Trace-

Element Chemistry. Cem Concr Res 23 (1):59-68. doi:https://doi.org/10.1016/0008-

8846(93)90136-W

5. Galan E, Carretero MI, Mayoral E (1999) A methodology for locating the original quarries

used for constructing historical buildings: application to Malaga Cathedral, Spain. Eng Geol 54

(3-4):287-298

6. Dreesen R, Dusar M (2004) Historical building stones in the province of Limburg (NE

Belgium): role of petrography in provenance and durability assessment. Mater Charact 53 (2-

4):273-287

7. Gomez-Heras M, Gonzalez RF (2004) Location of quarries of non traditional stony materials

in the architecture of Madrid: the Crypt of the Cathedral of Santa Maria la Real de la

Almudena. Mater Construcc 54 (274):33-49

8. Torok A, Prikryl R (2010) Current methods and future trends in testing, durability analyses

and provenance studies of natural stones used in historical monuments. Eng Geol 115 (3-

4):139-142

9. Schmidt-Döhl F, Koepke J, Schimrosczyk A (2005) Testing the identity of cementitious

materials by trace element analysis of belite. Paper presented at the 27th International

Conference on Cement Microscopy, Victoria, British Columbia, Canada April 24-April 28

10. Society TC Who would analyse a sample of concrete found at a crime scene?

http://www.concrete.org.uk/fingertips-nuggets.asp?cmd=display&id=1049. Accessed

04.03.2021

11. Becker JS (2008) Inorganic mass spectrometry : principles and applications. John Wiley &

Sons, Chichester, England; Hoboken, NJ

12. Murray RC (1992) Forensic geology / Raymond C. Murray, John C.F. Tedrow. vol Accessed

from https://nla.gov.au/nla.cat-vn3031098. Prentice Hall, Englewood Cliffs, N.J

13. Goguel RL, Stjohn DA (1993) Chemical-Identification of Portland Cements in New-Zealand

Concretes .2. The Ca-Sr-Mn Plot in Cement Identification and the Effect of Aggregates. Cem

Concr Res 23 (2):283-293. doi:https://doi.org/10.1016/0008-8846(93)90093-O

14. Milestone NB, John DAS, Baxter NI (1990) Investigation of methods for determining the

alkali content of old cements vol 22. Road Research Unit, National Roads Board, New Zealand

15. Graham IJ, Goguel RL, St John DA (2000) Use of strontium isotopes to determine the origin

of cement in concretes - Case examples from New Zealand. Cem Concr Res 30 (7):1105-1111.

doi:https://doi.org/10.1016/S0008-8846(00)00248-9

16. Potgieter JH (1997) Fingerprinting South African cements by chemical analysis. Paper

presented at the 10th International Congress on the Chemistry of Cement, Göteborg

[Sweden],

17. Potgieter JH, Potgieter SS, McCrindle RI, Tamas FD (2003) Fingerprinting of South African

cement clinkers and gypsum as a tool for cement identification purposes. Adv Cem Res 15

(2):45-50. doi:https://doi.org/10.1680/adcr.15.2.45.36725

18. Kalarus D (2008) Chemical identification of Portland cements produced in Poland on the

basis of the content of trace elements. Ph.D., Academia Gorniczo-Hutnicza, Krakow

19. Tamas FD (1996) Pattern recognition methods for the qualitative identification of

Hungarian clinkers. World Cement Research/ Development Section 27:75-79

20. Tamas FD, Patkai-Horvath M, Kristof-Mako E, Tritthart J Qualitative identification of

clinkers and cements - some results and possibilities. In: Justnes H (ed) 10th International

Congress on the Chemistry of Cement, Gothenburg, Sweden, June 2-6, 1997 1997.

21. Tamas FD, Tagnit-Hamou A, Tritthart J (1998) Trace elements in clinker and their use as

"fingerprints" to facilitate their qualitative identification. In: M. Cohen SM, J.P. Skalny (ed)

Materials Science of Concrete Special Volume: The Sidney Diamond Symposium. Wiley, pp 57-

22. Abonyi J, Tamas FD, Tritthart J (2004) Exploratory data analysis of trace elements in clinker.

Adv Cem Res 16 (1):9-15. doi:https://doi.org/10.1680/adcr.2004.16.1.9

23. Abonyi J, Tamas FD, Potgieter S, Potgieter H (2003) Analysis of trace elements in South

African clinkers using latent variable model and clustering. South African Journal of Chemistry-

Suid-Afrikaanse Tydskrif Vir Chemie 56:15-20

24. Tamas FD, Abonyi J (2003) World map of clinkers - Visualization of trace element content

of clinkers by self-organizing map. Paper presented at the 11th International Congress on the

Chemistry of Cement, South Africa,

25. Tamas FD, Abonyi J (2002) Trace elements in clinker I. A graphical representation. Cem

Concr Res 32 (8):1319-1323. doi:https://doi.org/10.1016/S0008-8846(02)00794-9

26. Tamas FD, Abonyi J, Borszeki J, Halmos P (2002) Trace elements in clinker II. Qualitative

identification by fuzzy clustering. Cem Concr Res 32 (8):1325-1330.

doi:https://doi.org/10.1016/S0008-8846(02)00797-4

27. Tamas FD, Abonyi J, Puertas F (2001) Trace elements based classification on clinkers.

Application to Spanish clinkers. Mater Construcc 51 (263-64):85-96.

doi:https://doi.org/10.3989/mc.2001.v51.i263-264.355

28. Lin YH, Lee PC, Chang TP (2008) Applying gray relational analysis with gray numbers to

qualitative identification of the origin of hydraulic cement clinker by trace elements. J Mater

Civ Eng 20 (8):539-543. doi:https://doi.org/10.1061/(ASCE)0899-1561(2008)20:8(539)

29. Kasamatsu M, Igawa T, Suzuki S, Suzuki Y (2018) Forensic Discrimination of Concrete Pieces

by Elemental Analysis of Acid-soluble Component with Inductively Coupled Plasma-Mass

Spectrometry. Anal Sci 34 (6):729-733. doi:http://doi.org/10.2116/analsci.18SBN02

30. Kondev FG, Wang M, Huang WJ, Naimi S, Audi G (2021) The NUBASE2020 evaluation of

nuclear physics properties. Chinese Phys C 45 (3):030001. doi:https://doi.org/10.1088/1674-

1137/abddae

31. Schmitz MD, Singer BS, Rooney AD (2020) Chapter 6 - Radioisotope Geochronology. In:

Gradstein FM, Ogg JG, Schmitz MD, Ogg GM (eds) Geologic Time Scale 2020. Elsevier, pp 193-

209. doi:https://doi.org/10.1016/B978-0-12-824360-2.00006-1

32. Casse M, Montero-Serrano J-C, St-Onge G, Poirier A (2019) REE distribution and Nd isotope

composition of estuarine waters and bulk sediment leachates tracing lithogenic inputs in

eastern Canada. Mar Chem 211:117-130.

doi:https://doi.org/10.1016/j.marchem.2019.03.012

33. Abbott AN, Löhr SC, Payne A, Kumar H, Du J (2022) Widespread lithogenic control of

marine authigenic neodymium isotope records? Implications for paleoceanographic

reconstructions. Geochim Cosmochim Acta 319:318-336.

doi:https://doi.org/10.1016/j.gca.2021.11.021

34. Mcarthur JM, Kennedy WJ, Chen M, Thirlwall MF, Gale AS (1994) Strontium Isotope

Stratigraphy for Late Cretaceous Time - Direct Numerical Calibration of the Sr Isotope Curve

Based on the United-States Western Interior. Palaeogeogr Palaeocl 108 (1-2):95-119.

doi:https://doi.org/10.1016/0031-0182(94)90024-8

35. Veizer J (1989) Strontium Isotopes in Seawater through Time. Annu Rev Earth Pl Sc 17

(1):141-167. doi:https://doi.org/10.1146/annurev.ea.17.050189.001041

36. Hoogewerff JA, Reimann C, Ueckermann E, Frei R, Frei KM, van Aswegen T, Stirling C, Reid

M, Clayton A, Ladenberger A, Team GP (2019) Bioavailable Sr-87/Sr-86 in European soils: A

baseline for provenancing studies. Sci Total Environ 672:1033-1044.

doi:https://doi.org/10.1016/j.scitotenv.2019.03.387

37. Fortunato G, Mumic K, Wunderli S, Pillonel L, Bosset JO, Gremaud G (2004) Application of

strontium isotope abundance ratios measured by MC-ICP-MS for food authentication. J Anal

At Spectrom 19 (2):227-234. doi:https://doi.org/10.1039/B307068A

38. Banner JL (2004) Radiogenic isotopes: systematics and applications to earth surface

processes and chemical stratigraphy. Earth-Sci Rev 65 (3):141-194.

doi:https://doi.org/10.1016/S0012-8252(03)00086-2

39. Vanhaecke F, Degryse P (2012) The Isotopic Composition of the Elements. In: Vanhaecke

FK, Kurt (ed) Isotopic analysis : fundamentals and applications using ICP-MS. Wiley-VCH ; [John

Wiley [distributor], Weinheim; Chichester, pp 7 - 20

40. Elburg AM (2012) Geochronological Dating. In: Vanhaecke F, Degryse P (eds) Isotopic

analysis: Fundamentals and applications using ICP–MS. Wiley-VCH Verlag GmbH % Co. KGaA,

Weinham, pp 235-275

41. Irrgeher J, Prohaska T, Sturgeon RE, Mester Z, Yang L (2013) Determination of strontium

isotope amount ratios in biological tissues using MC-ICPMS. Anal Methods-Uk 5 (7):1687-

1694. doi:https://doi.org/10.1039/c3ay00028a

42. Brand WA, Coplen TB, Vogl J, Rosner M, Prohaska T (2014) Assessment of international

reference materials for isotope-ratio analysis (IUPAC Technical Report). Pure Appl Chem 86

(3):425-467

43. Faure G, Mensing TM (2013) Isotopes : principles and applications. 3rd edn. John Wiley &

Sons, Inc. , Hoboken, New Jersey

44. NIST (2007) Certificate of analysis SRM 987 isotopic standard for strontium. National

Institute of Standards and Technology. https://www-s.nist.gov/srmors/certificates/987.pdf.

Accessed 28.09.2022 2022

45. Jochum KP, Nohl U, Herwig K, Lammel E, Stoll B, Hofmann AW (2005) GeoReM: A New

Geochemical Database for Reference Materials and Isotopic Standards. Geostandards and

Geoanalytical Research 29 (3):333-338. doi:https://doi.org/10.1111/j.1751-

908X.2005.tb00904.x

46. Meija J, Coplen TB, Berglund M, Brand WA, Bièvre PD, Gröning M, Holden NE, Irrgeher J,

Loss RD, Walczyk T, Prohaska T (2016) Atomic weights of the elements 2013 (IUPAC Technical

Report). Pure Appl Chem 88 (3):265-291. doi:https://doi.org/10.1515/pac-2015-0305

47. Amakawa H, Tazoe H, Obata H, Gamo T, Sano Y, Shen C-C (2013) Neodymium isotopic

composition and concentration in the Southwest Pacific Ocean. Geochem J 47 (4):409-422.

doi:https://doi.org/10.2343/geochemj.2.0260

48. Kinoshita N, Paul M, Kashiv Y, Collon P, Deibel CM, DiGiovine B, Greene JP, Henderson DJ,

Jiang CL, Marley ST, Nakanishi T, Pardo RC, Rehm KE, Robertson D, Scott R, Schmitt C, Tang

XD, Vondrasek R, Yokoyama A (2012) A Shorter 146Sm Half-Life Measured and Implications for

146Sm-142Nd Chronology in the Solar System. Science 335 (6076):1614-1617.

doi:https://doi.org/10.1126/science.1215510

49. Meissner F, Schmidt-Ott W-D, Ziegeler L (1987) Half-life and α-ray energy of146Sm.

Zeitschrift für Physik A Atomic Nuclei 327:171-174. doi:https://doi.org/10.1007/BF01292406

50. Gupta MC, MacFarlane RD (1970) The natural alpha radioactivity of samarium. J Inorg Nucl

Chem 32 (11):3425-3432. doi:https://doi.org/10.1016/0022-1902(70)80149-X

51. Clark MJ, Brian LB, Francis A (2018) Geochemistry of Non-Traditional Stable Isotopes. De

Gruyter, Berlin, Boston. doi:https://doi.org/doi:10.1515/9781501509360

52. Nakano T (2016) Potential uses of stable isotope ratios of Sr, Nd, and Pb in geological

materials for environmental studies. Proceedings of the Japan Academy Series B-Physical and

Biological Sciences 92 (6):167-184. doi:https://doi.org/10.2183/pjab.92.167

53. Rosner M (2010) Geochemical and instrumental fundamentals for accurate and precise

strontium isotope data of food samples: Comment on "Determination of the strontium

isotope ratio by ICP-MS ginseng as a tracer of regional origin" (Choi et al., 2008). Food Chem

121 (3):918-921. doi:https://doi.org/10.1016/j.foodchem.2010.01.019

54. Whitehouse MJ (1996) Dickin A. P., 1995. Radiogenic Isotope Geology, xvi + 452 pp.

Cambridge, New York, Port Chester, Melbourne, Sydney: Cambridge University Press. Price

£65.00, US $89.95 (hard covers). ISBN 0 521 43151 4. Geological Magazine 133 (2):226-226.

doi:10.1017/S0016756800008852

55. Prohaska T (2015) Sector field mass spectrometry for elemental and isotopic analysis.

Royal Society of Chemistry, [Cambridge]

56. Dariz P, Wortmann UG, Vogl J, Schmid T (2022) Beautiful Pietàs in South Tyrol (Northern

Italy): local or imported works of art? Heritage Science 10 (1):37.

doi:https://doi.org/10.1186/s40494-022-00678-6

57. Henderson J, Evans J, Barkoudah Y (2009) The roots of provenance: glass, plants and

isotopes in the Islamic Middle East. Antiquity 83 (320):414-429.

doi:https://doi.org/10.1017/S0003598X00098525

58. Henderson J, Evans J, Bellintani P, Bietti-Sestieri AM (2015) Production, mixing and

provenance of Late Bronze Age mixed alkali glasses from northern Italy: an isotopic approach.

J Archaeol Sci 55:1-8. doi:https://doi.org/10.1016/j.jas.2014.12.006

59. Henderson J, Evans J, Nikita K (2010) Isotopic Evidence for the Primary Production,

Provenance and Trade of Late Bronze Age Glass in the Mediterranean. Mediterranean

Archaeology & Archaeometry 10 (1):1-24

60. Henderson J, Ma H, Evans J (2020) Glass production for the Silk Road? Provenance and

trade of islamic glasses using isotopic and chemical analyses in a geological context. J Archaeol

Sci 119:105164. doi:https://doi.org/10.1016/j.jas.2020.105164

61. Li BP, Zhao JX, Greig A, Collerson KD, Feng YX, Sun XM, Guo MS, Zhuo ZX (2006)

Characterisation of Chinese Tang sancal from Gongxian and Yaozhou kilns using ICP-MS trace

element and TIMS Sr-Nd isotopic analysis. J Archaeol Sci 33 (1):56-62

62. Li BP, Zhao JX, Greig A, Collerson KD, Zhuo ZX, Feng YX (2005) Potential of Sr isotopic

analysis in ceramic provenance studies: Characterisation of Chinese stonewares. Nucl Instrum

Meth B 240 (3):726-732

63. McArthur JM, Howarth RJ, Shields GA (2012) Chapter 7 - Strontium Isotope Stratigraphy.

In: Gradstein FM, Ogg JG, Schmitz MD, Ogg GM (eds) The Geologic Time Scale. Elsevier, Boston,

pp 127-144. doi:https://doi.org/10.1016/B978-0-444-59425-9.00007-X

64. Cohen KM, Finney, S.C., Gibbard, P.L. & Fan, J.-X. (2013) The ICS International

Chronostratigraphic Chart. http://www.stratigraphy.org/ICSchart/ChronostratChart2023-

04.pdf.

65. Burke WH, Denison RE, Hetherington EA, Koepnick RB, Nelson HF, Otto JB (1982) Variation

of seawater 87Sr/86Sr throughout Phanerozoic time. Geology 10 (10):516-519.

doi:https://doi.org/10.1130/0091-7613(1982)10<516:Vosstp>2.0.Co;2

66. McArthur JM, Howarth RJ, Bailey TR (2001) Strontium Isotope Stratigraphy: LOWESS

Version 3: Best Fit to the Marine Sr-Isotope Curve for 0–509 Ma and Accompanying Look-up

Table for Deriving Numerical Age. The Journal of Geology 109 (2):155-170.

doi:https://doi.org/10.1086/319243

67. Saltzman MR, Davidson JP, Holden P, Runnegar B, Lohmann KC (1995) Sea-level-driven

changes in ocean chemistry at an Upper Cambrian extinction horizon. Geology 23 (10):893-

896. doi:https://doi.org/10.1130/0091-7613(1995)023<0893:Sldcio>2.3.Co;2

68. Dudás FÖ, Yuan D-X, Shen S-Z, Bowring SA (2017) A conodont-based revision of the

87Sr/86Sr seawater curve across the Permian-Triassic boundary. Palaeogeography,

Palaeoclimatology, Palaeoecology 470:40-53.

doi:https://doi.org/10.1016/j.palaeo.2017.01.007

69. Mcarthur JM (1994) Recent Trends in Strontium Isotope Stratigraphy. Terra Nova 6

(4):331-358. doi:https://doi.org/10.1111/j.1365-3121.1994.tb00507.x

70. Wang Z, Coyte RM, Dwyer GS, Ruhl LS, Hsu-Kim H, Hower JC, Vengosh A (2020) Distinction

of strontium isotope ratios between water-soluble and bulk coal fly ash from the United

States. International Journal of Coal Geology 222:103464.

doi:https://doi.org/10.1016/j.coal.2020.103464

71. Makishima A, Wiley VCHVG, Co K (2016) Thermal Ionization Mass Spectrometry (TIMS)

Silicate Digestion, Separation, Measurement. Wiley-VCH, Weinheim

72. Musgrove M, Banner JL, Mack LE, Combs DM, James EW, Cheng H, Edwards RL (2001)

Geochronology of late Pleistocene to Holocene speleothemsfrom central Texas: Implications

for regional paleoclimate. GSA Bulletin 113 (12):1532-1543.

doi:https://doi.org/10.1130/0016-7606(2001)113<1532:Golpth>2.0.Co;2

73. Vanhaecke F (2012) Single-Collector Inductively Coupled Plasma Mass Spectrometry. In:

Vanhaecke F, Degryse P (eds) Isotopic analysis: Fundamentals and applications using ICP–MS.

Wiley-VCH Verlag GmbH & Co. KGaA, Weinham, pp 31-77

74. O'Nions RK, Hamilton PJ, Hooker PJ (1983) A Nd isotope investigation of sediments related

to crustal development in the British Isles. Earth Planet Sci Lett 63 (2):229-240.

doi:https://doi.org/10.1016/0012-821X(83)90039-0

75. Hamilton PJ, O’nions RK, Bridgwater D, Nutman AP (1983) Sm-Nd studies of Archaean

metasediments and metavolcanics from West Greenland and their implications for the Earth's

early history. Earth Planet Sci Lett 62:263-272. doi:https://doi.org/10.1016/0012-

821X(83)90089-4

76. Walder AJ, Freedman PA (1992) Isotopic Ratio Measurement Using a Double Focusing

Magnetic-Sector Mass Analyzer with an Inductively Coupled Plasma as an Ion-Source. J Anal

At Spectrom 7 (3):571-575. doi:https://doi.org/10.1039/ja9920700571

77. Wieser M, Swieters, J., Douthitt, C. (2012) Multi-Collector Inductively Coupled Plasma

Mass Spectrometry. In: Vanhaecke F, Degryse P (eds) Isotopic Analysis: Fundamentals and

Applications Using ICP-MS. 1 edn. Wiley-VCH Verlag GmbH & Co, Weinheim, Germany, pp 77-

78. Douthitt CB (2008) The evolution and applications of multicollector ICPMS (MC-ICPMS).

Anal Bioanal Chem 390 (2):437-440. doi:https://doi.org/10.1007/s00216-007-1660-x

79. Pirrie D, Pidduck AJ, Crean DE, Nicholls TM, Awbery RP (2019) Identification and analysis

of man-made geological product particles to aid forensic investigation of provenance in the

built environment. Forensic Sci Int 305:109974.

doi:https://doi.org/10.1016/j.forsciint.2019.109974

80. Kugler W (2003) X-Ray diffraction analysis in the forensic science: the last resort in many

criminal cases. Advances in X-Ray Analysis (46):1-16

81. Radosavljević-Mihajlović A The Use of X-ray Diffraction Analysis in Different Forensic

Disciplines. In: Lee WE, Gadow R, Mitic V, Obradovic N (eds) Proceedings of the III Advanced

Ceramics and Applications Conference, Paris, 2016// 2016. Atlantis Press, pp 353-359

82. Stoney D, Bowen A, Bryant V, Caven EA, Cimino M, Stoney PL (2019) Particle Combination

Analysis for Predictive Source Attribution: Tracing a Shipment of Contraband Ivory. Journal of

the American Society of Trace Evidence Examiners 2 (1):13-34

83. Fitzpatrick RW, Raven MD (2012) How Pedology and Mineralogy Helped Solve a Double

Murder Case: Using Forensics to Inspire Future Generations of Soil Scientists. Soil Horizons 53

(5):14-29. doi:https://doi.org/10.2136/sh12-05-0016

84. Fitzpatrick RW, Raven M (2005) Soil Mineralogy and Other Properties in Forensic

Investigations. Acta Crystallogr A 61:C14-C14

85. Gianoncelli A, Raneri S, Schoeder S, Okbinoglu T, Barone G, Santostefano A, Mazzoleni P

(2020) Synchrotron µ-XRF imaging and µ-XANES of black-glazed wares at the PUMA beamline:

Insights on technological markers for colonial productions. Microchem J 154:104629.

doi:https://doi.org/10.1016/j.microc.2020.104629

86. Vanhoof C, Bacon JR, Fittschen UEA, Vincze L (2021) Atomic spectrometry update – a

review of advances in X-ray fluorescence spectrometry and its special applications. J Anal At

Spectrom 36 (9):1797-1812. doi:https://doi.org/10.1039/D1JA90033A

87. Nakai I, Shirataki J (2016) Chemical Composition of Glass Beads Excavated from Kofun (ca.

AD 2nd to 7th c.) in Western Japan by Portable XRF Showing Glass Trade among Asian

Countries. In: Recent Advances in the Scientific Research on Ancient Glass and Glaze. pp 73-

94. doi:https://doi.org/10.1142/9789814630290_0004

88. Liu S, Li QH, Gan F, Zhang P, Lankton JW (2012) Silk Road glass in Xinjiang, China: chemical

compositional analysis and interpretation using a high-resolution portable XRF spectrometer.

J Archaeol Sci 39 (7):2128-2142

89. Pahl JK, Sikorska M (2004) Cathodoluminescence study of carbonate cements in the Upper

Cambrian conglomerates from the Wonominta Block, northwestern New South Wales.

Australian Journal of Earth Sciences 51 (2):247-259. doi:https://doi.org/10.1111/j.1440-

0952.2003.01055.x

90. Nasrazadani S, Springfield T (2014) Application of Fourier transform infrared spectroscopy

in cement Alkali quantification. Mater Struct 47 (10):1607-1615.

doi:https://doi.org/10.1617/s11527-013-0140-3

91. Meyers P, Vanzelst L (1977) Neutron-Activation Analysis of Limestone Objects - Pilot-

Study. Radiochimica Acta 24 (4):197-204

92. Mao XL, Bol'shakov AA, Choi I, McKay CP, Perry DL, Sorkhabi O, Russo RE (2011) Laser

Ablation Molecular Isotopic Spectrometry: Strontium and its isotopes. Spectrochim Acta B 66

(11-12):767-775

93. Henderson J, Chenery S, Kröger J, Faber EW (2016) Glass Provenance along the Silk Road:

The Use of Trace Element Analysis. In: Recent Advances in the Scientific Research on Ancient

Glass and Glaze. pp 17-42. doi:https://doi.org/10.1142/9789814630290_0002

94. Kaufman L, Rousseeuw JP (1990) Divisive Analysis (Program DIANA). In: Finding Groups in

Data: An Introduction to Cluster Analysis. Wiley Series in Probability and Statistics. John Wiley

& Sons, Inc., pp 253-279. doi:https://doi.org/10.1002/9780470316801.ch6

Chapter 2

Publications

2.1. Provenancing of cement using elemental analyses and isotope techniques –

the state-of-the-art and future perspectives

Publisher´s Version

Published in the journal “Journal of Analytical Atomic Spectrometry “

Volume 36, 03 September 2021, Pages 2030-2042

https://doi.org/10.1039/d1ja00144b

Authors: Anera Kazlagić, Jochen Vogl, Gregor J. G. Gluth, Dietmar Stephan

This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

Provenancing of cement using elemental analyses

and isotope techniques –the state-of-the-art and

future perspectives

Anera Kazlagi´

c, *

Jochen Vogl,

Gregor J. G. Gluth

and Dietmar Stephan

With the aim of identifying the origin and the manufacturer of a cement, a reliable procedure that provides

unambiguous results is needed. Such procedure could resolve practical issues in damage research, liability

issues and forensic investigations. A substantial number of attempts for ﬁngerprinting of building materials,

including cement, has already been carried out during the last decades. Most of them were based on

concentration analysis of the main elements/components. This review provides an overview of

provenance studies of cement and the main approaches commonly used. Provenance studies of cement

via isotope techniques are also presented and discussed as representatives of the state-of-the-art in the

ﬁeld. Due to the characteristic properties and the occurrence of carefully selected isotope ratios, unique

ﬁngerprints of diﬀerent kinds of materials can be provided by these methods. This property has largely

been explored in various scientiﬁcﬁelds such as geo- and cosmochemistry, food forensics, archaeology,

geochronology, biomedical studies, and climate change processes. However, the potential of isotope

techniques in cement and concrete research for provenance studies has barely been investigated.

Therefore, the review outlines a suitable approach using isotope ratios, which could lead to reliable

provenancing of cementitious materials in the future.

Anera Kazlagi´

c obtained

a master's degree in chemistry

(2018) from Faculty of Science,

University of Sarajevo (UNSA).

Anera received two silver badges

from UNSA, for being one of the

best students of 1

and 2

study

cycle. Aer graduation, she was

employed at UNSA as teaching

and research assistant. Since

December 2019 she is working

as a scientic researcher at the

division Inorganic Trace Anal-

ysis, Bundesanstalt f¨

ur Materialforschung und -pr¨

ufung (BAM),

Berlin, Germany. Anera is enrolled as PhD candidate at Depart-

ment of Civil Engineering, Building Materials and Construction

Chemistry at Technische Universit¨

at Berlin. Her current research

interests concern isotope research in cement, with the focus on

provenancing.

Dr Jochen Vogl studied chemistry

at the Universities of Regensburg

and Mainz in Germany with

a focus on speciation analysis

with on-line IDMS in his doctoral

thesis (1997). He then worked for

two years on reference measure-

ments applying IDMS at the

IRMM (now JRC Geel) in Bel-

gium. Since January 2000 he is in

charge of the eld Isotope Anal-

ysis at the Bundesanstalt f¨

Materialforschung und -pr¨

ufung

(BAM) in Berlin, Germany. His main working elds are the appli-

cation of IDMS for reference measurements, isotope ratio analysis,

isotope and matrix reference materials, elemental mass spectrom-

etry in general and metrology in chemistry.

Federal Institute for Materials Research and Testing, Division 1.1 Inorganic Trace

Analysis, Richard-Willst¨

ater-Straße 11, 12489 Berlin, Germany. E-mail: anera.

[email protected]

Federal Institute for Materials Research and Testing, Division 7.4 Technology of

Construction Materials, Unter den Eichen 87, 12205 Berlin, Germany

Technische Universit¨

at Berlin, Department of Civil Engineering, Building Materials

and Construction Chemistry, Gustav-Meyer-Allee 25, 13355 Berlin, Germany

Cite this: J. Anal. At. Spectrom.,2021,

36,2030

Received 28th April 2021

Accepted 2nd September 2021

DOI: 10.1039/d1ja00144b

rsc.li/jaas

2030 |J. Anal. At. Spectrom.,2021,36,2030–2042 This journal is © The Royal Society of Chemistry 2021

JAAS

CRITICAL REVIEW

Introduction

The built environment is one of the most signicant achieve-

ments of human civilisation, because with buildings it provides

protection against environmental inuences and the basis for

the production of goods, but with transport systems, it also

provides the basis for all forms of traﬃc. Concrete is the basis

for this in most cases, which is why concrete has become the

material that is articially produced in the largest quantity by

human hands during the last hundred years. The concrete

properties are dominated by the reactive component cement,

which is the synthetic chemical product produced in the largest

quantity. Due to the ageing of concrete as a building material

due to environmental inuences, but also due to application

errors caused by faulty constructions or wrong mix composi-

tions of concrete, various structural damages occur worldwide,

which cause high costs and, in severe cases, even endanger

human lives. Aer their occurrence, damages are being

preventively investigated and researched. Intending to mini-

mise the consequences of damages, it is crucial to identify the

source and the cause that lead to such events. Therefore, one of

the main reasons for cement provenancing is identifying the

causes of failure in concrete structures. Poor and substandard

building material, primarily steel reinforcement, structural

steel, and cement represent the main causes of building failure

and collapse.

Concrete roof collapses are occurring worldwide.

According to a recent review by Proske and Schmid,

the annual

collapse frequency of buildings is 2.4 10

7

per year in

industrialised countries, and 4.7 10

6

per year in developing

countries, in many cases caused by insuﬃcient quality of the

employed building materials. The authors also determined

mortality rate values of building collapses, ranging from 9.8 

8

to 1.5 10

6

per year, for industrialised and developing

countries, respectively, which is in agreement with the value

obtained by Blockley.

Furthermore, bridge collapse frequencies

worldwide have been reported to be on the order of magnitude

of 10

4

per year.

Many government agencies tried to develop a protocol and

a guideline to reduce and prevent the risk of such catastro-

phes. The type and origin of the construction and raw mate-

rials used, and the manufacturer of the components are of

great relevance not only for avoidance and damage assessment

but also for the resulting liability issues. This is also of interest

when considering the stability of concrete,

6,7

marble,

8,9

and

stone

10–15

in historic structures and investigating the origin of

the raw materials from which those structures were made. For

example, previous studies emphasized the investigation of raw

materials from which buildings in Portugal were con-

structed.

In most structural failure cases, analysis of the

building materials, including cement as the major compo-

nent, is one of the key issues for understanding the quality and

the strength of the structure. Testing institutes are also

interested in provenance studies, since they are sometimes

asked whether two cementitious samples are identical or are

ofthesameorigin.

Recycling of rubble may also require provenance studies in

some cases. Furthermore, the determination of rubble prove-

nance can contribute to expand the knowledge about this

material, and also about possible ways to return the material to

the construction processes.

Beyond this, cement provenancing is required in forensic

investigations. Given the variety of possible circumstances

that may be encountered on the crime scene, any building

material, such as cement, mortar, or concrete, may hold

evidentiary signicance. This includes cement particles or

dust found at a crime scene. Those particles can serve as

a physical evidence linking a suspect(s) to the victim or the

crime scene. Knowing the origin of the cement and concrete

Dr Gregor J. G. Gluth is a senior

researcher at the Bundesanstalt

f¨

ur Materialforschung und

-pr¨

ufung (BAM), Berlin, since

2010. Previously he had worked

at the institute for civil engi-

neering of the Technische Uni-

versit¨

at Berlin, where he also

obtained his PhD. He is active

member of several committees of

the International Union of

Laboratories and Experts in

Construction Materials, Systems

and Structures (RILEM) and the European Federation of Corrosion

(EFC). His research is focussed on concrete degradation mecha-

nisms, durability, and maintenance as well as special cements and

cements with reduced carbon footprint.

Prof. Dr Stephan Dietmar

studied chemistry at the

University of Siegen in Germany

with a focus on building chem-

istry (1996). He then worked for

three years as a research assis-

tant at the Institute for Building

Materials and Construction

Chemicals (University of Sie-

gen). Aer that, he was

employed as a trainee and

senior scientist at “Cement &

Quality, Heidelberg Cement”in

Heidelberg, Germany. For one year he worked at Heidelberger

Cement Group Technology Centre GmbH as a senior scientist. In

the period 2001–2006 he was working as a postdoc at TU Munich,

and aer that as academic senior councillor at the Department of

Construction Materials and Building Chemistry at University of

Kassel, Germany. Since March 2011 he is working as professor for

Building Materials and Construction Chemistry at TU Berlin.

This journal is © The Royal Society of Chemistry 2021 J. Anal. At. Spectrom.,2021,36,2030–2042 | 2031

Critical Review JAAS

particles can be benecial for locating the crime scene,

tracking back the particles to a packaging unit or providing

furtherevidenceforuseincourt.

Geographical provenancing of any kind of material can be

described as an act of identifying the place of origin or the

source of the material. Knowing the geographical origin of

a specic material has been very useful in elds such as

forensics and archaeology for unravelling residence histories of

suspects or trading routes between past cultures. The basis of

provenance studies is that there are specic, recurrent and

denable interdependences between raw materials and the

nal products derived from them.

Therefore, we can summarise three main cases:

(1) Finished product correlates to a raw material used in the

production, e.g. marble.

(2) Processed material, within which one component domi-

nates, either in terms of the mass fraction or in other terms, e.g.

curse tablets made of pure lead.

(3) Complex mixtures or a highly processed material, e.g.,

alloys, cement.

The provenancing of any material can be determined by

utilising specic properties of the material under investigation

and it's preceding (original) raw material.

Cementsforgeneralpurposesaredened in the standard

EN 197-1:2011,

which diﬀerentiates between ve main cate-

gories: CEM I Portland cement, CEM II Portland-composite

cement, CEM III Blast furnace cement, CEM IV Pozzolanic

cement, CEM V Composite cement. Portland cement clinker,

the main component of ordinary Portland cement (OPC; CEM

I according to EN 197-1:2011), is produced by sintering a nely

ground raw meal consisting mainly of limestone and clay or

the natural mixture marl from local materials. Other natural

raw materials, such as quartz sand, or secondary raw mate-

rials, such as foundry sands, slags, etc., can also be used to

adjust the composition. In the rotary kiln, the mixture is

heated to a temperature of approximately 1450 C. In addition

to hard coal and lignite, a complex mix of diﬀerent secondary

fuels is usually used in clinker production to achieve high

temperatures. The OPC clinker is mixed with a certain amount

of diﬀerent calcium sulphate minerals and then nely ground

to give OPC. The calcium-rich components of OPC that do not

originate from local sources are primarily the calcium

sulphates which either come from industrial by-products such

as gypsum from ue gas desulphurisation plants (FGD

gypsum

) or partly from natural sources such as natural

anhydrite and gypsum. Cements other than OPC contain

additional constituents, suchaslimestonepowder,ground

granulated blast furnace slag, natural pozzolana, yashor

burnt oil shale. Due to this broad range of components with

diﬀerent geographic origin, the provenancing of cement is not

straightforward.

This paper provides an overview of published provenance

studies of cement and concrete and lists the main approaches

usually adopted. Provenance studies of cement utilising isotope

techniques are also presented and discussed as representatives

of the state-of-the-art in the eld.

Previous attempts for cement

provenancing

Elemental ngerprinting

Trace elements analysis is being performed in diﬀerent scien-

ticelds and for provenance studies of various inorganic

materials. Trace elemental pattern was used for establishing the

provenance of pottery,

archaeological mortars,

22,23

as well as

for clinker and cement, which is described in this section.

For element to be used in cement provenance analysis, the

local raw materials must be the predominant source. It should

not originate from the ancillary components, the used fuel, or

the furnace. Furthermore, its concentration must be indepen-

dent of the temperature uctuations in the kiln. For the concept

to be transferred from cement to concrete, the elements must

be highly immobilised in the alkaline pore solution.

While

reviewing the literature, we noticed that many authors investi-

gated major, minor and trace elements in the studies. However,

trace element label in these studies partially diﬀer from the

denition put forward by IUPAC.

Therein, trace element is

considered as any element having an average mass fraction of

less than about 100 mgg

1

The rst paper on the provenancing of cement was pub-

lished in 1993 by Goguel and St John in two parts. In the rst

part,

major, minor and trace elements present in New Zealand

cements were investigated by semiquantitative Inductively

Coupled Plasma Mass Spectrometry (ICP-MS) and quantitative

Atomic Absorption (AA), Atomic Emission (AE) and Inductively

Coupled Plasma Optical Emission Spectrometry (ICP-OES).

Cement samples were prepared by digestion in a mixture of

HF and HClO

, followed by dissolution in diluted HNO

. The

authors concluded that Ca/Sr, Ca/Ba and Ca/Mn ratios have the

greatest potential for identifying cements in hardened

concretes and that the Ca/Sr ratio seems to be the most accu-

rately quantiable measurand. The variation of the Ca/Sr ratio

of cement from the same production site over a long period is

small, making this ratio suitable for discriminating between

production sites. Rare earth elements (REE) distribution

pattern, which were analysed as well, showed more limited

identication potential. In the second part,

the authors dis-

cussed Ca, Sr and Mn concentration leached from the concrete.

By performing selective acid cement leaching from hardened

New Zealand concretes, authors have tried to determine its

source of manufacture. Crushing the concrete and selecting the

fragments allowed them to exclude coarse aggregate from

leaching, while a small portion of ne aggregates could not be

excluded. The authors tested four types of aggregates to deter-

mine their contribution to leachates from concrete. It was

concluded that it is essential to assess the contributions from

the aggregates to the leaching solution. The application of REE

distribution pattern to provenance the cement in a hardened

concrete is limited by contamination of REEs released from

aggregates during leaching. These two articles triggered further

research in this eld. Tamas and co-authors noted that

cement's origin determination is both scientically and practi-

cally a challenge which could be solved by analytical

2032 |J. Anal. At. Spectrom.,2021,36,2030–2042 This journal is © The Royal Society of Chemistry 2021

JAAS Critical Review

determination of certain elements contained in cements, and

statistical processing of analytical data. In their study they

stated that a three-element approach based on Sr–Ba–Mn is not

fully adequate, especially in case of sulphate resistant clinkers

having high iron content.

Therefore, they added Mg, Ti and Zr, nally ending up with

a six-element approach, realised by ICP-OES measurements and

statistical processing of the data, using “pattern recognition”

methods.

In another project, Tamas and co-authors analysed minor

and trace elements in een clinker sorts to classify the clinkers

produced in diﬀerent factories in Spain. For the qualitative

identication of clinkers a rule base classier was designed -

trained by C4.5 algorithm.

In 2001 and 2002, Tamas and

Abonyi investigated the content of minor and trace elements in

OPC clinkers, and their use for provenancing. They determined

the Mg, Sr, Ba, Mn, Ti, Zr, Zn and V content in several hundred

clinkers. The maximum and minimum values, the standard

deviation of data, as well as the real and rounded averages were

used for plotting (Table 1). The authors concluded that the rst

six elements come from the local raw materials, while Zn and V

mainly come from fuel and are not useful for provenance

determination. They considered Ba, Sr and Mg as the most

relevant elements for clinker ngerprinting.

The sample

preparation was made by dissolving the clinker in HCl, followed

by SiO

precipitation, ltration and washing steps. The ltrate

was analysed by ICP-OES, statistically processed via MATLAB©

and the “star plots”were constructed (Fig. 1). Star plots are

a graphical method used to simplify and to visualize the minor

and trace element content in clinker, by comparing it with the

proposed standard. The area of the star in a star plot is

proportional to the total amount of the considered elements,

and the eight rays of the star represent the eight elements.

However, among other elements, Zn and V were still used for

constructing “star plots”. The trace elements were selected

based on the obtained clusters by the modied version of the

Fisher interclass separability method. For clinker identication

a specicsoware was developed

which is available on

internet.

The soware relies on unsupervised fuzzy clustering,

identied by a fuzzy classier using a control algorithm.

In 2004, Abonyi and co-authors presented a paper in which

data presentation box and quantile–quantile plots were

proposed to analyse the relationships between diﬀerent facto-

ries and diﬀerent minor and trace elements in clinkers.

The

authors concluded that exploratory data analysis can be useful

in the qualitative analysis of elemental content in clinker.

According to Lin and co-authors,

a standard pattern of each

origin must be established prior to the identication of the

clinker origin. Since each clinker origin shows a broad range of

trace element patterns, the uncertainty resulting from dening

a distinctive compound pattern from its various forms makes

these techniques diﬃcult for usage. Therefore, it is not easy to

dene a precise standard pattern of a clinker origin. Commonly

applied mathematical models for solving these uncertainty

issues are the random model, the fuzzy model

and the Gray

model.

Proposed by Deng in 1989, the Gray relational analysis is

a geometric method that can measure the approachability or

similarity between series.

Lin and co-authors' study integrated

Table 1 Maximum and minimum values, standard deviation, averages, and rounded averages (used for star plots) of investigated clinkers

(mg kg

1

) reprinted from F. D. Tamas and J. Abonyi, Cem. Concr. Res., 2002, 32, 1319–1323, Copyright (2002), with permission from Elsevier

Element Maximum value Minimum value Standard deviation Average Rounded average

Ba 442 47 122.74 192.68 200

Mn 6538 15 995.06 507.02 500

Sr 2972 19 677.63 560.28 500

Ti 1691 175 289.78 1189.13 1000

Zr 149 4 17.33 47.32 50

Mg 24 125 1751 6453.27 8976.93 8950

Zn 559 11 112.89 113.7 100

V 297 17 67.11 85.69 100

Fig. 1 Star plot of a Hungarian clinker (code number 5L5), a Spanish

one (code number 89ES1) and a South African one (code number

159SA17). Reprinted from F. D. Tamas and J. Abonyi, Cem. Concr. Res.,

2002, 32, 1319–1323, Copyright (2002), with permission from Elsevier.

This journal is © The Royal Society of Chemistry 2021 J. Anal. At. Spectrom.,2021,36,2030–2042 | 2033

Critical Review JAAS

both concepts of Gray number and Gray relational analysis to

set up a model for qualitative identication of clinker. The

results showed that proposed model was successful in identi-

fying the origin of clinker. The authors concluded that the

methodology of the model can be extended to various areas with

similar uncertain information.

In his doctoral thesis entitled “Chemical identication of

Portland cements produced in Poland on the basis of the

content of trace elements”, D. Kalarus determined Cr, Zn, Cd,

Pb, Co, Ni, Co, Sr, Ba, Mn, Ti and Mg. The analytical method

which the author used for the research was ICP-OES and scan-

ning electron microscope (SEM). Based on the concentrations of

the elements in clinkers, cements and raw materials, it was

concluded that Mg, Sr and Mn serve as “markers”for the

chemical identication of CEM I produced in Poland.

The

author concluded that the concentration of these elements in

cement depends almost exclusively on their content in the raw

materials, and not on the substitute fuels or waste materials

such as used car tires and municipal waste, which is in agree-

ment with the statement by Tamas and Abonyi. Other studies

on cement and clinker provenancing were mostly presented as

a conference proceedings or conference papers.

38–41

Sawaki and

co-authors estimated the chemical composition of cement in

hardened concrete via an electron probe microanalyser, but the

sensitivity turned out to be insuﬃcient for discrimination.

Kasamatsu and co-authors

examined elemental (Cu, Rb, Zn,

Sr, Zr, Ba, La, Ce, Nd and Pb) discrimination in nitric acid

soluble components of concrete. According to the authors, this

method can be used for forensic discrimination. However,

Goguel and co-authors

concluded that the nitric acid leaching

approach is not applicable to aggregates consisting substan-

tially of limestone. Therefore, nitric acid soluble component of

concrete cannot represent a cement sample, but rather a mix of

cement and calcareous additives. Moreover, dissolving concrete

in nitric acid results in signicant extraction of the lanthanides

from aggregates, which removes the potential of lanthanides for

cement identication in concrete.

In several papers, MATLAB© and the star plot method were

used to potentially pinpoint the origin of cement and, in some

cases, mortars.

44–46

A summary of the previous attempts for

clinker, cement and concrete provenancing is shown in Table 2,

where “elemental ngerprint”includes algorithm, pattern

recognition and geometric methods, considering that for its

realisation and interpretation, elemental analysis is a prerequi-

site. “Isotope ratio”implies the application of one isotopic ratio,

e.g.

Sr/

Sr. In summary, the elemental ngerprint approach

requires a large number of samples and the selection of appro-

priate (major, minor and trace) elements for establishing a stan-

dard pattern of clinker and later cement. Most of the previous

research relied on multivariate statistical analysis. Such methods

cannot be used for evaluation of all elements. This is because

these elements either have no relation to the geographic origin or

they are related to additives, whose origin is not connected to the

origin of the major components, such as fuel used in production

(used tires, heavy fuel oil etc.). Besides that, to have a meaningful

result, many samples and parameters should be considered.

Suﬃciently large numbers of the target elements and as well of

the samples are needed to achieve statistical signicance; other-

wise, the results are meaningless due to high standard errors. In

several studies, these requirements were met and the origin of

clinker for certain locations was determined via diﬀerent

approaches. Clinker is an intermediate product, which is avail-

able only at the production site and therefore, the practical rele-

vance is rather limited. The identication of clinker in cement

and hardened concrete, however, remains an open question.

Table 2 Summary of previous attempts for clinker, cement and concrete provenancing, “elemental ﬁngerprint”includes algorithm, pattern

recognition and geometric methods, considering that for its realisation and interpretation, elemental analysis is a prerequisite. “Isotope ratio”

implies application of one isotopic ratio

Clinker Cement Concrete Elemental ngerprint Isotope ratio Year Reference

x x 1993 24

x x x 1994 26

x x 1996 27

x x 1997 28

x x 1997 38

x 1998 39

x x x x 2000 48

x x 2001 29

x x 2002 30

x x 2002 35

x x 2003 44

x x 2003 45

x x 2003 40

x x 2004 33

x 2005 17

x x 2007 49

x x 2007 46

x x x 2007 37

x x 2008 34

x x 2018 43

2034 |J. Anal. At. Spectrom.,2021,36,2030–2042 This journal is © The Royal Society of Chemistry 2021

JAAS Critical Review

Isotope studies

One of the most suitable techniques which can be used in the

provenancing of building material is isotope ratio analysis of

metal elements by means of mass spectrometry. The basic

principle of all mass spectrometers is ionisation of a sample in

an ion source, separation of the ions via a mass separator

according to their m/z(mass/charge) ratio and nally, their

detection at a detection unit.

Thermal Ionisation Mass Spectrometry (TIMS) and Induc-

tively Coupled Plasma Mass Spectrometry (ICP-MS) are aer

Isotope-ratio mass spectrometry (IRMS), the most used tech-

niques for authentication purposes.

Besides TIMS, both single-collector (SC) ICP-MS

50–53

and MC-

ICP-MS

51,54

have been widely used for measuring isotope ratios.

Due to the high measurement precision, the large natural

isotope variation and the comparatively high contents in

diﬀerent sample types, strontium isotope ratios have found

wide application outside of geochemistry and have been

successfully used to determine the origin of inorganic mate-

rials, archaeological artefacts

55–57

and glass.

58,59

Therefore, Sr

isotope ratios can have high signicance in cement prove-

nancing. Since the main raw material for cement production is

limestone, and plenty of limestone deposits in Europe originate

from geological formations containing calcite or dolomite, the

diﬀerent geological age of limestones and the original mineral

composition combined with geological history are reected in

Sr/

Sr isotope ratio of cement.

Rb decays to

Sr with a half-

life of 4.8 10

years, while emitting a beta



particle.

Depending on the age, and the initial Rb/Sr ratio, diﬀerent

geochemical reservoirs and rocks therein developed a large

variation in today's

Sr/

Sr ratios, ranging from approximately

0.702 for the depleted mantle to >0.943 for old continental

crust.

For the light element isotope systems, such as S, H, N, O and

C, the mass-dependent eﬀects are signicant. Since their

isotope composition can be changed in individual process steps

(e.g. burning, annealing) by isotope fractionation, they have

limited suitability for determining the origin of cement and

concrete. Contrastingly, for the heavy element isotope systems,

such as Sr, Nd and Pb, the mass dependent eﬀects are mostly

within the noise of the measurement and surely much smaller

than the regional natural variations that are observed in prov-

enancing applications, therefore, scientists can get a unique

insight into the material's geographic source and/or production

history.

47,62

For the elements like Sr, which have more than one

non-radiogenic isotope, the relative abundance of the radio-

genic isotope (

Sr expressed as

Sr/

Sr) can be normalised to

the xed nominal abundance ratio of a pair of stable isotopes of

that element (

Sr/

Sr ¼0.1194 (ref. 63)). Therefore, for Sr, only

the radiogenic portion of a certain isotope is considered, and

this special internal calibration corrects for any stable isotope

fractionation (for example, burning) independently whether it

occurs in the mass spectrometer, during sample preparation or

even before in the history of the sample.

The rst idea for using Sr isotopes as tracer for the

geographic origin of cements dates to the year 2000 when

Graham and co-authors have published a case study from New

Zealand. They combined chemical and strontium isotopic

analysis, and showed that New Zealand cements carry

geochemical ngerprints from their raw materials to the nal

product, the concrete.

The authors demonstrated that there

are measurable diﬀerences in the

Sr/

Sr values of most of the

cements analysed, which can be used to pinpoint the

geographic origin. The number of cement plants, however, is

rather low in New Zealand and neighbouring countries play no

role in this context. Additionally, the investigated cement and

concrete samples date back to the 1980s and earlier, which

might not be comparable to modern cement.

Kosednar-Legenstein, in her PhD thesis,

and in studies

with co-authors

64,65

investigated the mineralogical, chemical

and isotopic composition of historic mortars and plasters (with

burnt lime as binder) from Austria. Among other results, the

authors reported

Sr/

Sr ratios of the mortars and plasters to

be between 0.7091 and 0.7115, whereas Sr/Ca ratios did not

exceed 0.003. In addition, burning and setting experiments

were conducted to investigate the behaviour of REE and the Sr

isotopes. The authors concluded that burning has a negligible

inuence on Sr/Ca,

Sr/

Sr and REE distribution, and that

Sr/Ca and

Sr/

Sr signatures can thus, in principle, be used to

identify natural deposits used for manufacturing of the mate-

rials. However, variations of the

Sr/

Sr ratios in some of the

investigated deposits as well as the possibility that limestone

from diﬀerent deposits was used for the binder and the aggre-

gates, respectively, made unequivocal provenancing diﬃcult.

All other publications that link isotope analysis and cement

research were focused on isotope applications other than

geographical origin determination. A study performed by

Pierkes and co-authors investigated the inuence of cement

hydration on the distribution of oxygen and hydrogen

isotopes.

Another example is the application of stable isotopes

(elements H, C, N, O, S) to investigate reactions during curing or

concrete damage.

67–70

Recently, Grengg and co-authors investi-

gated the deterioration mechanism of alkali-activated cements

in sulfuric acid, using oxygen-isotope signatures, among other

methods and tools.

However, if we take a look at the other examples of inorganic

material's provenancing, we would surely notice that isotope

techniques were frequently used, mostly for studying diﬀerent

categories of archaeological materials

72,73

and ceramics prove-

nancing.

74–76

In ceramics provenancing, with the aim of

assigning a production location to a specic ceramic artefact of

unknown origin, the rst step is the characterisation, followed

by identication of their chemical, physical or structural

“ngerprint”, and the last step is comparison. This comparison

is made by using collected ngerprints and advanced statistical

techniques to search for similarities and diﬀerences. This

procedure allows the collection of all ceramics with similar

features and diﬀerentiating groups of ceramics with diﬀerent

properties.

In archaeology, lead isotopes have been used as

tracers for provenancing of metal samples and artefacts since

its introduction by Brill and Wampler in 1967.

78–80

By combining

the isotopic information with additional information on the

sources and the artefacts being independent of the isotope data,

This journal is © The Royal Society of Chemistry 2021 J. Anal. At. Spectrom.,2021,36,2030–2042 | 2035

Critical Review JAAS

it is possible to overcome limitations such as overlapping lead

isotopic compositions of ore deposits, large spreads within one

ore deposit or a missing overlap with (un)known mining

locations.

57,81

A promising approach for cement provenancing is the

application of isotope ratios as described above for archaeo-

logical and ceramics provenance studies. In principle, the

transfer of these approaches to cement is possible, but works

best for approaches which have been established for glass and

ceramics due to the similarity of the materials.

The advantage of isotopic over elemental ngerprinting

Provenancing of building materials through isotope techniques

are based on the principle that every material, with a diﬀerent

geographical origin, has its unique isotopic ngerprint. The

major advantage of isotope techniques is that some isotope

systems which contain radiogenic isotope, such as

Sr/

Sr, can

be linked to the geographic origin, as they reect the mineral

composition and geological history of the raw material. More-

over, radiogenic isotope systems do not change, while the

elemental concentration are exposed to changes in environ-

ment as well as in technical processes. If strontium is consid-

ered as an element of interest in cement, the isotope ratio

depends on its origin from each of these raw materials,

predominantly in limestone or marl. The current literature

reveals that various methods can be employed for the prove-

nancing of cements and clinkers. Of particular interest is the

tendency of using elemental concentrations towards less

unequivocal data, and consequently, the need for a higher

number of features (appropriate elements) in the elemental

ngerprint. Elemental ngerprints seem easier to determine

but would, subsequently, require knowledge in multivariant

statistics and data analysis to be interpreted. However, isotope

ratio determination allows simpler isotopic ngerprints, which

are unique and unambiguous. Finally, isotope ngerprinting

approach oﬀers reduced number of measurands, leading to

a more straight forward statistical evaluation and interpretation

of data.

Other techniques

In addition to the above-mentioned analytical techniques, X-ray

diﬀraction (XRD) and X-ray uorescence (XRF) spectrometry are

important techniques for provenance studies. Their usefulness

was proven in various elds, e.g. the analysis of dust from

a shipment of contraband ivory to determine the original

location where the ivory was packed,

or solving a double

murder case in Australia,

where X-ray diﬀraction allowed to

identify soils on a tool and from a quarry, but also for solving

other forensic cases.

In principle, this capacity could be also

used for provenancing of concrete particles found on a crime

scene, and their possible agreement with samples from a sus-

pected quarry. For example, quartz gravels and sands, used as

aggregate for concrete, diﬀer as regards crystallinity,

which

can possibly be exploited for provenance studies. Besides XRD,

XRF was very oen used for characterising and provenancing of

diﬀerent types of building materials,

86,87

predominantly old

stones

8,15,88

and marbles

9,89

in buildings or artifacts with

historical signicance. Both XRD and XRF were used for

establishing the provenance of limestone.

With the purpose of

collecting both XRD and XRF data on the same spot of an object,

a portable XRD/XRF

instrument was developed. Another

technique which was used for limestone,

92,93

but also for

pottery

94–96

provenancing was Instrumental Neutron Activation

Analysis (INAA). Meyers and Vanzelts used INAA in their pilot-

study

to distinguish between objects made of limestone from

diﬀerent sources.

Another technique worth mentioning is laser ablation (LA)

ICP-MS. Van Ham-Meert and co-authors

124

published an article

about LA-ICP-MS for Pb and Sr isotopic determination in

archaeological glass. This technique was also used for prove-

nancing of ceramics,

98,99

as well as for cultural heritage

research.

100,101

The study of using portable LA for sampling solid

materials, with subsequent Sr and Nd measurements on TIMS

was reported by Knaf and co-authors.

102

An additional promising method for isotopic microanalysis

is laser ablation molecular isotopic spectrometry

103

(LAMIS). In

their article,

104

Bol'shakov and co-authors evaluated LAMIS for

rapid optical analysis of isotopes of diﬀerent elements. Appli-

cation of LAMIS for optical isotopic analysis of solid samples

was also investigated in a paper published by Mao and co-

authors,

105

where measurement of strontium isotopes showed

promising results. In forensic analysis, for an unknown soil

sample, scanning electron microscopy with energy dispersive X-

ray spectrometry (SEM-EDX) allows recognition of the presence

or absence of man-made particles, including concrete and

cement.

106

Therefore, automated mineralogy is amenable to

distinguish and recognise the particles derived from common

construction material classes. Indubitably, all these techniques

can provide valuable information for elemental and phase

identication and quantication in cement, which can be very

benecial for an overall picture in provenance study.

Since cement and concrete are both highly processed mate-

rials, IRMS is not covered within this review, since it relies on

analysis of light element isotope systems whose composition

can be strongly changed by isotope fractionation, as explained

in the section “Isotope studies”and thus tracing back cement to

the source materials is impossible.

The combined approach for

provenancing

The combined approach for provenancing, as proposed here,

incorporates the use of

Sr/

Sr and

143

Nd/

144

Nd isotope ratios

and of Ca/Sr elemental ratios to provide a unique ngerprint of

cement. The whole pathway for cement provenancing is shown

in Fig. 2.

Structure of the combined approach

First, a sample should be subsampled for mass fraction deter-

mination and isotope ratio analysis. For isotope ratio analysis,

a selection of the sample preparation procedure is very

important.

2036 |J. Anal. At. Spectrom.,2021,36,2030–2042 This journal is © The Royal Society of Chemistry 2021

JAAS Critical Review

For trace elements like Sr and Nd, prior to isotope ratio

analysis, mass fraction determination should take place pref-

erably on ICP-MS or ICP-OES. The following step is matrix

separation, to purify the small quantity of analytes. To obtain

accurate and precise isotope ratios, it is recommended to

perform the

Sr/

Sr and

143

Nd/

144

Nd measurements by TIMS,

on a single and on a double or triple lament geometry,

respectively. When both Sr and Nd isotopic ngerprints are not

capable of completely resolving the diﬀerent cement origins,

a third parameter should be introduced. Here, Ca/Sr concen-

tration ratio could be a promising measurand which could help

distinguishing between samples with very similar

Sr/

Sr–

143

Nd/

144

Nd isotopic ngerprint. This parameter is

promising and has already been studied by Goguel and co-

authors.

To measure total concentration, an element must be

fully dissolved. For this purposes, total digestion in HF and

HNO

are oen used. Due to the many reasons, such as pres-

ence of calcium as a major constituent in cement and the

formation of CaF

precipitate while dissolving cement in HF

and HNO

, numerous interferences on ICP-MS, as well as time-

consuming sample preparation, these measurements are rec-

ommended to be carried out using X-Ray Fluorescence (XRF) on

cement glass beads. Once prepared and properly stored, fused

(glass) beads are very stable and can be used several times.

Improvement by adding another isotope system

In the past mostly one isotope system (isotope ratios of one

chemical element) has been used for provenancing. For

complex materials such as alloys or glass, which consist of

several components of diﬀerent geographical origin, one

isotope system, however, oen resulted in large overlaps and

insuﬃcient resolution of the measurement space. To solve this,

additional isotope systems were introduced for glass prove-

nancing.

58,59,107,108

In 2009, Henderson and co-authors published

a paper on the provenancing of glass in the Islamic Middle East

by using oxygen, strontium, and neodymium isotope systems.

They concluded that the determination of Nd and Sr isotope

ratios for selected samples would be a suitable approach for

providing independent geographical information.

Elements such as Sr and Nd display variations in their

isotopic composition since one or more of their isotopes is the

nal product of the decay of naturally occurring and long-lived

radionuclides. Consequently, the isotopic composition of such

radiogenic isotopes is governed by the initial parent/daughter

ratio and the time these nuclides have spent together.

Therefore, the isotopic composition of these elements is not

only used for geochronological dating purposes but also for

tracing the geographical origin.

Neodymium is a classical isotope tracer in geochemistry and

has been used widely for dating and the study of petrogenesis,

alteration, and weathering processes.

109–111

Due to its aﬃnity to

silicate phases, the Nd isotope system has been applied in

addition to the Sr isotope system in provenance studies of

glass.

59,108,112–116

147

Sm decays to

143

Nd while emitting an alpha

particle with a half-life of 1.06 10

years.

117

Besides using this

decay system for dating purposes, it can be used as a natural

tracer since the isotopic composition of Nd as measured today is

time-integrated results of Sm/Nd ratio in a specicenviron-

ment.

118

The assumption is that

143

Nd/

144

Nd ratios are diﬀerent

in the raw materials from diﬀerent (geographically located)

plants used for clinker, and therefore also for cement production.

Sr is the tracer for the calcium carbonate-bearing raw material

(limestone or marl), and Nd for the silicate-bearing raw material

(clay). The addition of neodymium isotopes as a second param-

eter oﬀers the potential to achieve results that provide more

information about the geographical origin and empower the new

approach for provenancing. This is already demonstrated by

Henderson and co-authors, who applied both Sr and Nd isotope

ratios for glass provenance studies. When it comes to cement

a similar approach might be feasible, because cement and glass

have much in common. Both materials are complex mixtures of

a silicate and a carbonate component and contain a suite of

minor additives. Aer mixing, both materials are exposed to

temperatures above 1400 C hampering the application of clas-

sical stable isotopes of elements such as H, C and O. In Fig. 3 the

Sr/

Sr and

143

Nd/

144

Nd isotope measurements in Bronze Age

glass are plotted versus each other. This diagram clearly shows

that Mesopotamian glasses can be diﬀerentiated from Egyptian

Fig. 2 The pathway for cement provenancing.

Fig. 3 A plot of

Sr/

Sr vs.

143

Nd/

144

Nd isotope measurements in

Bronze Age glass from Mesopotamia (lozenges), Egypt (squares) and

Greece (triangles). Adapted from Henderson and co-authors, 2010.

107

This journal is © The Royal Society of Chemistry 2021 J. Anal. At. Spectrom.,2021,36,2030–2042 | 2037

Critical Review JAAS

and Greek glasses based on their combined

Sr/

Sr and

143

Nd/

144

Nd signatures, which is not possible based on the

Sr/

Sr signatures or the

143

Nd/

144

Nd signatures alone.

107

Samples which cannot be resolved by using one isotope system,

because the data overlap within the measurement uncertainty,

might be resolved by a second isotope system, adding a third or

fourth measurand, whether it is an isotope ratio, an interelement

ratio or an element concentration, will even increase the resolu-

tion, provided the measurand is correlated with the origin of one

of the raw material.

General limitations

The selection of a sample preparation procedure has an

important role in cement provenancing. If we consider prove-

nancing of CEM I, the total analyte's content and isotopic

ngerprint in minor additional constituent must be assessed.

Gypsum present in cement is exclusively from non-local sour-

ces, and Ca bearing minerals always carry a certain amount of

Sr, which may complicate origin determination. Besides

gypsum, CEM I can contain other constituents, included in

maximum 5% of minor additional constituents. Since they may

have an inuence on the Sr isotopic ngerprint, all the mineral

additions which are present in cement should be considered

while performing sample preparation. Furthermore, due to the

diﬀerent fuels used in rotary kilns for clinker production, the

possible contribution regarding the contamination from fuels

must be assessed. Since only clinker carries the local raw

materials correlated to the Sr ngerprint and therefore the

origin, one must ensure that while preparing the sample, only

the clinker fraction is obtained. Finally, the measured isotope

ratios should correspond to those of the clinker fraction. To

exclude everything but the analyte while performing the

measurement, it is necessary to check and aerwards validate

the right analytical procedure, including matrix separation.

Another limitation of isotope analysis is the duration, which is

mainly attributed to the sample preparation. To isolate the

analyte of interest (e.g. Sr), matrix separation via cation-

exchange or extraction chromatography must be performed.

To date, the column preparation, resin cleaning and analyte-

matrix separation is being performed manually at the

expenses of working time. A few years ago, however, a new way

of fully automated sample purication system has been oﬀered

as an option. The prepFAST MC® systems can overcome this

time-consuming issue in sample preparation.

119–123

An even

higher degree of automation might be allowed by laser ablation

coupled to ICP-MS when an automated sample changer and

corresponding soware packages are available.

Summary and outlook

Provenance studies of cement are of great importance for failure

research, damage assessment and resulting liability issues

related to concrete structures and in investigating ancient

building materials and in forensic science.

In this paper, the origin determination of cement has been

reviewed, and the main approaches from past and current

research have been presented. The majority of the published

work was related to trace elemental contents of cement and

clinkers, mostly combined with either the “star plot method”,

the “gray rational analysis”or other statistical and pattern

recognition methods. Even though the sample preparation and

the analysis presented in the publications were straightforward,

the data interpretation requires diﬀerent visualisation

methods, such as star graphs to intuitively compare the data. An

advanced statistical method like PCA and clustering algorithms,

which require solid statistical knowledge and computational

power, restrict the number of potential users. Besides this, two

(Zn, V) from a maximum of eight elements have proven useless

because they originate from a fuel deployed in the production of

the clinker. Finally, the published literature revealed that

mostly clinkers were analysed. Clinker, however, only occurs

within the production process before other components are

admixed. Therefore, the provenancing of clinker is of almost no

practical relevance. The application of isotope ratios in the

provenancing of cement is scarce and is limited to Graham and

co-authors

and Kosednar-Legenstein and co-authors,

64,65

who

showed that Sr isotopes, combined with elemental analysis, are

useful tools for provenancing of cement in a restricted area

concerning time and space. When expanding this approach to

a larger region such as Europe or even worldwide, a second

isotope system is required to increase the resolution of the

provenancing approach. Here, Nd being another radiogenic

isotope system, is selected in addition to Sr. Both elements, Sr

and Nd, have their main occurrence in diﬀerent components of

the raw mix of the cement. Thus, they act as tracers for diﬀerent

raw materials. By measuring both Sr and Nd isotope ratios,

a characteristic isotopic ngerprint of cement can be estab-

lished, which then can be used to identify concrete of unknown

origin. The addition of the interelement ratio of Ca/Sr to the

strontium and neodymium isotope ratios will increase the

analytical resolution of the approach and most likely lead to the

geographical origin of the material, and ideally to its manu-

facturer. The main advantage of the Sr–Nd isotope analysis with

or without the Ca/Sr ratio over the statistical and pattern

recognition approaches is a reduced number of measurands

leading to a more straight forward statistical evaluation, which

can be visualized in 2D or 3D plots. The measurands them-

selves, the Sr and the Nd isotope ratio and the Ca/Sr ratio, show

a stronger correlation with the origin of the raw materials, thus

enabling a more precise mapping of the cement origin.

However, sample preparation needs to be improved to avoid

interferences from mineral additions that do not originate from

the clinker but also in terms of time consumption. The use of

automated matrix separation facilitates the sample preparation

and reduces chemicals as well as time consumption. Finally, the

expansion of the analytical approach, e.g. addition of Ca/Sr

ratios, needs to be considered to enable the transfer of this

approach to the provenance determination of concrete samples.

Future research should focus on three main topics: rst,

optimisation and validation of the proposed wet-chemical

method for cement provenancing should take place. Second,

an in situ method for the determination of

Sr/

Sr and

143

Nd/

144

Nd isotope ratios based on LA-ICP-MS should be

2038 |J. Anal. At. Spectrom.,2021,36,2030–2042 This journal is © The Royal Society of Chemistry 2021

JAAS Critical Review

developed for concrete provenancing. The lateral resolution of

LA should allow the diﬀerentiation between diﬀerent mineral

phases of the concrete, i.e. the cementitious phase and the

aggregates. Together with the already determined isotope data

and the interelement ratios, the procedure for the determina-

tion of cement's origin should be transferable to concrete.

Third, a databank for isotopic ngerprints of cement should

be established. Besides answering individual questions

regarding the provenance of specic cement samples or

concrete structures, provenance studies also contribute to the

creation of cement datasets. Provenancing of cement allows the

complete documentation of the cement samples combining

isotope and concentration ratios along with their identied

provenance. As a result, new reference groups of various

cements of known origin are generated, while present ones

expand with new samples, leading to a reference databank.

Such a databank can be of great importance since it could

provide geochemists and other scientists with more knowledge

and could simplify future cement provenancing.

Abbreviations

AA Atomic absorption

AE Atomic emission

EDX Energy dispersive X-ray spectrometry

FGD Flue gas desulphurisation

ICP-MS Inductively coupled plasma mass spectrometry

ICP-

OES

Inductively coupled plasma optical emission

spectrometry

INAA Instrumental neutron activation analysis

IRMS Isotope ratio mass spectrometry

LA Laser ablation

LAMIS Laser ablation molecular isotopic spectrometry

MC Multi collector

OPC: ordinary Portland cement

PCA Principal component analysis

REE Rare earth elements

SC Single collector

SEM Scanning electron microscope

TIMS Multi collector thermal ionisation mass spectrometry

XRD X-ray diﬀraction

XRF X-ray uorescence

Author contributions

Anera Kazlagic: conceptualisation, methodology, project

administration, investigation, writing –original dra, writing –

review & editing, visualisation. Jochen Vogl: methodology,

supervision, funding acquisition, writing –original dra,

writing –review & editing. Gregor J. G. Gluth: methodology,

supervision, writing –review & editing. Dietmar Stephan:

methodology, supervision, writing –review & editing.

Conﬂicts of interest

There are no conicts to declare.

Acknowledgements

The authors gratefully acknowledge the Federal Institute for

Material Research and Testing (BAM) for providing funds under

the project number MIT1-2019-8.

References

1 M. Hamma-Adama, Am. J. Eng. Res., 2017, 6, 289–300.

2 G. Campione and G. Giambanco, J. Perform. Constr. Facil.,

2020, 34, 04020001.

3 D. Proske and M. Schmid, Bautechnik, 2021, 98, 423–432.

4 D. I. Blockley, The Nature of Structural Design and Safety,

Ellis Horwood Ltd, Chichester, England, 1980.

5 D. Proske, Bridge Collapse Frequencies versus Failure

Probabilities, Springer, Cham, 2019.

6 P. S. Reed, K. Schoonees and J. Salmond, Historic Concrete

Structures in New Zealand: Overview, Maintenance and

Management, Science & Technical Publishing, Department

of Conservation, New Zealand, 2008.

7 M. Bostenaru Dan, Materials, Technologies and Practice in

Historic Heritage Structures, Springer, Dordrecht, 2014.

8 K. Al-Bashaireh, F. Abudanah and M. Driessen, Archaeol.

Anthropol. Sci., 2020, 12, 26.

9 S. Columbu, F. Antonelli, M. Lezzerini, D. Miriello,

B. Adembri and A. Blanco, J. Archaeol. Sci., 2014, 49, 332–

342.

10 R. Fort, M. A. de Buergo, E. Perez-Monserrat and M. J. Varas,

Eng. Geol., 2010, 115, 149–157.

11 A. Torok and R. Prikryl, Eng. Geol., 2010, 115, 139–142.

12 E. Galan, M. I. Carretero and E. Mayoral, Eng. Geol., 1999,

54, 287–298.

13 R. Dreesen and M. Dusar, Mater. Charact., 2004, 53, 273–

287.

14 M. Gomez-Heras and R. F. Gonzalez, Mater. Constr., 2004,

54,33–49.

15 L. Carta, D. Calcaterra, P. Cappelletti, A. Langella and

M. de'Gennaro, J. Cult. Herit., 2005, 6, 277–286.

16 A. Murta, J. Pinto, H. Varum, J. Guedes, J. Lousada and

P. Tavares, Portugal Sb10: Sustainable Building Aﬀordable

to All –Low Cost Sustainable Solution,2010, pp. 589–596.

17 F. Schmidt-D¨

ohl, J. Koepke and A. Schimrosczyk, Presented

in Part at the 27th International Conference on Cement

Microscopy, Victoria, British Columbia, Canada, April 24

to 28, 2005.

18 A. Buko, World Archaeol., 1984, 15, 348–365.

19 M. A. Sanjuan and C. Argiz, Mater. Constr., 2012, 62, 425–

430.

20 F. W. Locher, Cement : Principles of Production and Use,

Bau+Technik Gmbh, D¨

usseldorf, 2006.

21 H. Mommsen, J. Radioanal. Nucl. Chem., 2001, 247, 657–

662.

22 E. D'Ambrosio, F. Marra, A. Cavallo, M. Gaeta and V. Guido,

J. Archaeol. Sci. Rep., 2015, 2, 186–203.

23 L. Ortega, M. Zuluaga, A. Alonso Olazabal, M. Insausti and

A. Iba˜

nez Etxeberria, Archaeometry, 2007, 50, 387–408.

This journal is © The Royal Society of Chemistry 2021 J. Anal. At. Spectrom.,2021,36,2030–2042 | 2039

Critical Review JAAS

24 R. L. Goguel and D. A. Stjohn, Cem. Concr. Res., 1993, 23,

59–68.

25 A. D. McNaught and A. Wilkinson, IUPAC Compendium of

chemical terminology, Oxford, 2nd edn, 1997.

26 R. L. Goguel and D. A. Stjohn, Cem. Concr. Res., 1993, 23,

283–293.

27 F. D. Tamas, World Cement Res. Dev. Sect., 1996, 27,75–79.

28 F. D. Tamas, M. Patkai-Horvath, E. Kristof-Mako and

J. Tritthart, Qualitative identication of clinkers and

cements - some results and possibilitiest, Proceedings of

the 10th International Congress on the Chemistry of Cement,

Gothenburg, 1997, vol. 3, p. 3v010.

29 F. D. Tamas, J. Abonyi and F. Puertas, Mater. Constr., 2001,

51,85–96.

30 F. D. Tamas and J. Abonyi, Cem. Concr. Res., 2002, 32, 1319–

1323.

31 J. Abonyi, Qualitative Identication of Clinkers by Fuzzy

Clustering, https://www.mathworks.com/matlabcentral/

leexchange/47162-qualitative-identication-of-clinkers-

by-fuzzy-clustering, accessed 18.03.2021.

32 J. Madar, J. Abonyi and F. Szeifert, Eng. Appl. Artif. Intell.,

2005, 18, 343–351.

33 J. Abonyi, F. D. Tamas and J. Tritthart, Adv. Cem. Res., 2004,

16,9–15.

34 Y. H. Lin, P. C. Lee and T. P. Chang, J. Mater. Civ. Eng., 2008,

20, 539–543.

35 F. D. Tamas, J. Abonyi, J. Borszeki and P. Halmos, Cem.

Concr. Res., 2002, 32, 1325–1330.

36 J. L. Deng, J. Grey Syst., 1989, 1,1–24.

37 D. Kalarus, PhD thesis, Academia Gorniczo-Hutnicza, 2008.

38 J. H. Potgieter, Presented in Part at the 10th International

Congress on the Chemistry of Cement,G

oteborg, Sweden,

1997.

39 F. D. Tamas, A. Tagnit-Hamou and J. Tritthart, in Materials

Science of Concrete Special Volume: the Sidney Diamond

Symposium, ed. S. M. M. Cohen and J. P. Skalny, Wiley,

1998, pp. 57–69.

40 F. D. Tamas and J. Abonyi, Presented in Part at the 11th

International Congress on the Chemistry of Cement, South

Africa, 2003.

41 H. Justnes, Proceedings of the 10th International Congress on

the Chemistry of Cement, Amarkai AB and Congrex,

Gothenburg, Sweden, June 2–6, 1997.

42 D. Sawaki, K. Kobayashi and E. Sakai, Bunseki Kagaku, 2010,

59, 1051–1064.

43 M. Kasamatsu, T. Igawa, S. Suzuki and Y. Suzuki, Anal. Sci.,

2018, 34, 729–733.

44 J. Abonyi, F. D. Tamas, S. Potgieter and H. Potgieter, S. Afr. J.

Chem., 2003, 56,15–20.

45 J. H. Potgieter, S. S. Potgieter, R. I. McCrindle and

F. D. Tamas, Adv. Cem. Res., 2003, 15,45–50.

46 S. S. Potgieter-Vermaak, J. H. Potgieter, K. Worobiec, R. van

Grieken, L. Majanovic and S. Moeketsi, Cem. Concr. Res.,

2007, 37, 834–843.

47 T. Prohaska, Sector Field Mass Spectrometry for Elemental

and Isotopic Analysis, Royal Society of Chemistry,

Cambridge, 2015.

48 I. J. Graham, R. L. Goguel and D. A. St John, Cem. Concr.

Res., 2000, 30, 1105–1111.

49 B. Kosednar-Legenstein, PhD thesis, Technische

Universitaet Graz, 2007.

50 S. Augagneur, B. Medina and F. Grousset, Fresenius' J. Anal.

Chem., 1997, 357, 1149–1152.

51 M. Barbaste, L. Halicz, A. Galy, B. Medina, H. Emteborg,

F. C. Adams and R. Lobinski, Talanta, 2001, 54, 307–317.

52 R. Larcher, G. Nicolini and P. Pangrazzi, J. Agric. Food

Chem., 2003, 51, 5956–5961.

53 B. Medina, S. Augagneur, M. Barbaste, F. E. Grouset and

P. Buat-Meard, Food Addit. Contam., 2000, 17, 435–445.

54 M. Barbaste, K. Robinson, S. Guilfoyle, B. Medina and

R. Lobinski, J. Anal. At. Spectrom., 2002, 17, 135–137.

55 H. Ma, J. Henderson and J. Evans, Archaeometry, 2016, 58,

68–80.

56 H. J. Ma, J. Henderson and J. Evans, J. Archaeol. Sci., 2014,

50, 551–558.

57 J. Vogl, M. Rosner, J. Curbera, U. Peltz and B. Peplinski,

Archaeol. Anthropol. Sci., 2018, 10, 1111–1127.

58 J. Henderson, J. Evans and Y. Barkoudah, Antiquity, 2009,

83, 414–429.

59 J. Henderson, J. Evans, P. Bellintani and A. M. Bietti-

Sestieri, J. Archaeol. Sci., 2015, 55,1–8.

60 R. Bowen, in Isotopes in the Earth Sciences, Springer

Netherlands, Dordrecht, 1994, pp. 162–200, DOI: 10.1007/

978-94-009-2611-0_4.

61 M. Rosner, Food Chem., 2010, 121, 918–921.

62 J. A. Hoogewerﬀ, C. Reimann, E. Ueckermann, R. Frei,

K. M. Frei, T. van Aswegen, C. Stirling, M. Reid,

A. Clayton, A. Ladenberger and G. P. Team, Sci. Total

Environ., 2019, 672, 1033–1044.

63 R. H. Steiger and E. J¨

ager, Earth Planet. Sci. Lett., 1977, 36,

359–362.

64 B. Kosednar-Legenstein, M. Dietzel, A. Leis, K. Stingl,

B. Wiegand and M. Baumgartner, Geochim. Cosmochim.

Acta, 2007, 71, A514.

65 B. Kosednar-Legenstein, M. Dietzel, A. Leis, B. Wiegand,

K. Stingl and M. Baumgartner, Presented in Part at the

European Geosciences Union General Assembly, Vienna,

Austria, 2006.

66 R. Pierkes, H. F¨

orstel and M. Boner, Inuence of Cement

Hydration on the Distribution of Oxygen and Hydrogen

Isotopes, Institut fuer Bauforschung Aachen, IBAC, Rheinisch-

Westfaelische Technische Hochschule Aachen, RWTH,

Aachen, 2005.

67 G. Macleod, A. E. Fallick and A. J. Hall, Chem. Geol., 1991,

86, 335–343.

68 F. Mittermayr, C. Bauer, D. Klammer, M. E. Bottcher,

A. Leis, P. Escher and M. Dietzel, Isot. Environ. Health

Stud., 2012, 48, 105–117.

69 K. Pye and N. Schiavon, Nature, 1989, 342, 663–664.

70 B. Kosednar-Legenstein, M. Dietzel, A. Leis and K. Stingl,

Appl. Geochem., 2008, 23, 2425–2437.

71 C. Grengg, G. J. G. Gluth, F. Mittermayr, N. Ukrainczyk,

M. Bertmer, A. Guilherme Buzanich, M. Radtke, A. Leis

and M. Dietzel, Cem. Concr. Res., 2021, 142, 106373.

2040 |J. Anal. At. Spectrom.,2021,36,2030–2042 This journal is © The Royal Society of Chemistry 2021

JAAS Critical Review

72 M. Murillo-Barroso, I. Montero-Ruiz, J. M. Nieto,

M. D. C. Massieu, D. M. Socas and M. Martinon-Torres, J.

Iber. Geol., 2019, 45, 585–608.

73 D. J. Killick, J. A. Stephens and T. R. Fenn, Archaeometry,

2020, 62,86–105.

74 M. Guzowska, I. Kuleﬀ, E. Pernicka and M. Satir, in Troia

and the Troad: Scientic Approaches, Springer, Heidelberg,

2003, pp. 233–249.

75 B. P. Li, J. X. Zhao, A. Greig, K. D. Collerson, Y. X. Feng,

X. M. Sun, M. S. Guo and Z. X. Zhuo, J. Archaeol. Sci.,

2006, 33,56–62.

76 B. P. Li, J. X. Zhao, A. Greig, K. D. Collerson, Z. X. Zhuo and

Y. X. Feng, Nucl. Instrum. Methods Phys. Res., Sect. B, 2005,

240, 726–732.

77 A. Sarris, Best Practices of Geoinformatic Technologies for the

Mapping of Archaeolandscapes, Archaeopress Publishing

Ltd., Oxford, 2015.

78 R. H. Brill and J. M. Wampler, Am. J. Archaeol., 1967, 71,63–

77.

79 Z. A. Stos-Gale and N. H. Gale, Archaeol. Anthropol. Sci.,

2009, 1, 195–213.

80 E. Niederschlag, E. Pernicka, T. Seifert and M. Bartelheim,

Archaeometry, 2003, 45,61–100.

81 J. Vogl, B. Paz and E. Volling, Archaeol. Anthropol. Sci., 2019,

11, 3267–3277.

82 D. Stoney, A. Bowen, V. Bryant, E. A. Caven, M. Cimino and

P. L. Stoney, J. Am. Soc. Trace Evid. Examiners, 2019, 2,13–

34.

83 R. W. Fitzpatrick and M. D. Raven, Soil Horiz., 2012, 53,14–

29.

84 R. W. Fitzpatrick and M. Raven, Acta Crystallogr., Sect. A:

Found. Adv., 2005, 61, C14.

85 N. Marinoni and M. A. T. M. Broekmans, Cem. Concr. Res.,

2013, 54,215–225.

86 D. Ergenc and R. Fort, Measurement, 2019, 147, 106876.

87 E. Delluniversita, I. M. Muntoni, I. Allegretta, M. Tarantini,

A. Monno, P. Maiorano, A. Girone, M. Morsilli, R. Terzano

and G. Eramo, Archaeol. Anthropol. Sci., 2019, 11, 6037–

6063.

88 J. Powers, D. M. Smilgies, E. C. Geil, K. Clinton,

N. Dimitrova, M. Peachin and R. E. Thorne, J. Archaeol.

Sci., 2009, 36, 343–350.

89 D. Magrini, D. Attanasio, S. Bracci, E. Cantisani and

W. Prochaska, Archaeol. Anthropol. Sci., 2018, 10, 1141–

1152.

90 L. Bianco, Rom. J. Phys., 2017, 62, 901.

91 P. Sarrazin, G. Chiari and M. Gailhanou, Adv. X-Ray Anal.,

2009, 52, 175–186.

92 J. Zhu, M. D. Glascock, C. S. Wang, X. J. Zhao and W. Lu, J.

Archaeol. Sci., 2012, 39, 2568–2573.

93 L. A. Dim, J. Adetunji, C. D. Okujeni, S. B. Elegba and

S. A. Agaja, J. Radioanal. Nucl. Chem., 1991, 148, 145–153.

94 L. S. Bohus, M. M. M. de Antczak, E. D. Greaves, A. Antczak,

J. Bermudez, Z. Kasztovszky, T. Poirier and A. Simonits, J.

Radioanal. Nucl. Chem., 2005, 265, 247–256.

95 P. S. Bedregal, P. A. Mendoza, M. S. Ubillus, W. Yepez,

J. Jennings and E. H. Montoya, J. Radioanal. Nucl. Chem.,

2015, 306, 729–736.

96 J. B. Tandoh, B. J. B. Nyarko, S. B. Dampare, Y. Bredwa-

Mensah, O. Gyampo and H. Ahiamadjie, J. Radioanal.

Nucl. Chem., 2010, 284, 567–573.

97 P. Meyers and L. Vanzelst, Radiochim. Acta, 1977, 24, 197–

204.

98 J. P´

erez-Arantegui, M. Resano, E. Garc´

ıa-Ruiz,

F. Vanhaecke, C. Rold´

an, J. Ferrero and J. Coll, Talanta,

2008, 74, 1271–1280.

99 M. Resano, P. Marzo, J. P´

erez-Arantegui, M. Aramend´

ıa,

C. Cloquet and F. Vanhaecke, J. Anal. At. Spectrom., 2008,

23, 1182–1191.

100 B. Giussani, D. Monticelli and L. Rampazzi, Anal. Chim.

Acta, 2009, 635,6–21.

101 M. Vannoorenberghe, T. Van Acker, J. Belza, D. Teetaert,

P. Crombe and F. Vanhaecke, J. Anal. At. Spectrom., 2020,

35, 2686–2696.

102 A. C. S. Knaf, J. M. Koornneef and G. R. Davies, J. Anal. At.

Spectrom., 2017, 32, 2210–2216.

103 A. A. Bol'shakov, X. L. Mao, J. J. Gonzalez and R. E. Russo, J.

Anal. At. Spectrom., 2016, 31, 119–134.

104 A. A. Bol'shakov, X. L. Mao, D. L. Perry and R. E. Russo,

Spectroscopy, 2014, 29,30–39.

105 X. L. Mao, A. A. Bol'shakov, I. Choi, C. P. McKay, D. L. Perry,

O. Sorkhabi and R. E. Russo, Spectrochim. Acta, Part B, 2011,

66, 767–775.

106 D. Pirrie, A. J. Pidduck, D. E. Crean, T. M. Nicholls and

R. P. Awbery, Forensic Sci. Int., 2019, 305, 109974.

107 J. Henderson, J. Evans and K. Nikita, Mediterr. Archaeol.

Archaeom., 2010, 10,1–24.

108 J. Henderson, H. Ma and J. Evans, J. Archaeol. Sci., 2020,

119, 105164.

109 J. Veizer, Annu. Rev. Earth Planet. Sci., 1989, 17, 141–167.

110 M. R. Walter, J. J. Veevers, C. R. Calver, P. Gorjan and

A. C. Hill, Precambrian Res.,2000, 100, 371–433.

111 D. J. DePaolo and E. E. Daley, Chem. Geol., 2000, 169, 157–

185.

112 P. Degryse, J. Henderson and G. Hodgins, Isotopes in

Vitreous Materials, Leuven University Press, Leuven, 2017.

113 A. Meek, J. Henderson and J. Evans, J. Anal. At. Spectrom.,

2012, 27, 786–795.

114 D. Brems, M. Ganio, K. Latruwe, L. Balcaen, M. Carremans,

D. Gimeno, A. Silvestri, F. Vanhaecke, P. Muchez and

P. Degryse, Archaeometry, 2013, 55, 449–464.

115 M. Ganio, M. Gulmini, K. Latruwe, F. Vanhaecke and

P. Degryse, J. Archaeol. Sci., 2013, 40, 4264–4270.

116 C. Boschetti, J. Henderson and J. Evans, J. Archaeol. Sci.

Rep., 2017, 11, 647–657.

117 M. C. Gupta and R. D. MacFarlane, J. Inorg. Nucl. Chem.,

1970, 32, 3425–3432.

118 P. Richard, N. Shimizu and C. J. All`

egre, Earth Planet. Sci.

Lett., 1976, 31, 269–278.

119 S. C. Metzger, B. W. Ticknor, K. T. Rogers, D. A. Bostick,

E. H. McBay and C. R. Hexel, Anal. Chem., 2018, 90, 9441–

9448.

This journal is © The Royal Society of Chemistry 2021 J. Anal. At. Spectrom.,2021,36,2030–2042 | 2041

Critical Review JAAS

120 A. Retzmann, T. Zimmermann, D. Profrock, T. Prohaska

and J. Irrgeher, Anal. Bioanal. Chem., 2017, 409, 5463–5480.

121 A. M. Weng, J. Arps, P. Blaser, C. Wienberg, D. Hebbeln

and N. Frank, Chem. Geol., 2017, 475, 140–148.

122 T. G. Enge, M. P. Field, D. F. Jolley, H. Ecroyd, M. H. Kim

and A. Dosseto, J. Anal. At. Spectrom., 2016, 31, 2023–2030.

123 S. J. Romaniello, M. P. Field, H. B. Smith, G. W. Gordon,

M. H. Kim and A. D. Anbar, J. Anal. At. Spectrom., 2015,

30, 1906–1912.

124 A. Van Ham-Meert, S. M. Chernonozhkin, S. J. M. Van

Malderen, T. Van Acker, F. Vanhaecke and P. Degryse,

Geostand. Geoanal. Res., 2018, 42(2), 223–238.

2042 |J. Anal. At. Spectrom.,2021,36,2030–2042 This journal is © The Royal Society of Chemistry 2021

JAAS Critical Review

Chapter 2

Publications

2.2. Development of a sample preparation procedure for Sr isotope analysis of

Portland cements

Publisher´s Version

Published in the journal “Analytical and Bioanalytical Chemistry “

Volume 414, 14 January 2022, Pages 4379–4389

https://doi.org/10.1007/s00216-021-03821-7

Authors: Anera Kazlagić, Francesco F. Russo, Jochen Vogl, Patrick Sturm, Dietmar

Stephan, Gregor J. G. Gluth

Creative Commons licence CC BY 4.0

Vol.:(0123456789)

1 3

Analytical and Bioanalytical Chemistry

https://doi.org/10.1007/s00216-021-03821-7

RESEARCH PAPER

Development ofasample preparation procedure forSr isotope

analysis ofPortland cements

AneraKazlagić1· Francesco F.Russo1· JochenVogl1· PatrickSturm2· DietmarStephan3· Gregor J.G.Gluth2

Received: 4 November 2021 / Accepted: 1 December 2021

Abstract

The 87Sr/86Sr isotope ratio can, in principle, be used for provenancing of cement. However, while commercial cements consist

of multiple components, no detailed investigation into their individual 87Sr/86Sr isotope ratios or their influence on the integral

87Sr/86Sr isotope ratio of the resulting cement was conducted previously. Therefore, the present study aimed at determin-

ing and comparing the conventional 87Sr/86Sr isotope ratios of a diverse set of Portland cements and their corresponding

Portland clinkers, the major component of these cements. Two approaches to remove the additives from the cements, i.e. to

measure the conventional 87Sr/86Sr isotopic fingerprint of the clinker only, were tested, namely, treatment with a potassium

hydroxide/sucrose solution and sieving on a 11-µm sieve. Dissolution in concentrated hydrochloric acid/nitric acid and in

diluted nitric acid was employed to determine the 87Sr/86Sr isotope ratios of the cements and the individual clinkers. The aim

was to find the most appropriate sample preparation procedure for cement provenancing, and the selection was realised by

comparing the 87Sr/86Sr isotope ratios of differently treated cements with those of the corresponding clinkers. None of the

methods to separate the clinkers from the cements proved to be satisfactory. However, it was found that the 87Sr/86Sr isotope

ratios of clinker and cement generally corresponded, meaning that the latter can be used as a proxy for the clinker 87Sr/86Sr

isotope ratio. Finally, the concentrated hydrochloric acid/nitric acid dissolution method was found to be the most suitable

sample preparation method for the cements; it is thus recommended for 87Sr/86Sr isotope analyses for cement provenancing.

Keywords Cement· Clinker· Sample preparation· Strontium isotopes· Provenancing

Abbreviations

CEM I Ordinary Portland cement as specified in EN

197–1

conc. acid Concentrated HNO3 and HCl 1:1 v/v

dil. acid 1 mol L−1 HNO3

FDG Flue gas desulfurisation

ICP-MS Inductively coupled plasma mass

spectrometry

KOSH Selective dissolution of clinker phases using

KOH/sucrose solution

MC-TIMS Multi collector thermal ionisation mass

spectrometry

Sieving Sieving on 11-μm sieve

XRD X-ray diffraction

Introduction

Concrete is the most important artificial material in the world

in terms of produced mass, and it is considered to be the basis

for our built environment [1]. Since the properties of concretes

are dominated by their key compound, cement [2], devising a

way to determine its origin, known as provenancing, is of great

importance. Provenance studies of concrete and cement are

required for failure research, damage assessment, and resulting

Published in the topical collection Analytical Methods and Applications

in the Materials and Life Sciences with guest editors Ute Resch-Genger,

Matthias Koch, Björn Meermann, and Michael G. Weller.

This manuscript is dedicated to the 150th anniversary of BAM.

*Anera Kazlagić

[email protected]

1 Division 1.1 Inorganic Trace Analysis, Federal

Institute forMaterials Research andTesting,

Richard-Willstäter-Straße 11, 12489Berlin, Germany

2 Division 7.4 Technology ofConstruction Materials, Federal

Institute forMaterials Research andTesting, Unter den

Eichen 87, 12205Berlin, Germany

3 Department ofCivil Engineering, Building Materials

andConstruction Chemistry, Technische Universität Berlin,

Gustav-Meyer-Allee 25, 13355Berlin, Germany

Kazlagić A.et al.

1 3

liability issues related to concrete structures, and they are also

important in forensic science [3–8].

To determine the origin of cement, it is important to under-

stand its composition and production process. At present, ordi-

nary Portland cement (OPC) is the most widely used cement

type, suitable for most purposes of concrete construction. As

specified in European Standard EN 197–1, Portland cement

(CEM I) contains at least 95 wt.% of Portland clinker (disre-

garding the calcium sulfate addition). The clinker is a complex

mixture of calcium silicates, calcium aluminates, and several

minor phases produced from limestone, chalk, marl, clays,

shale, and partly other minor components in a rotary kiln [2,

9]. The remaining 5 wt.% or less of the CEM I are minor addi-

tional constituents, as specified in EN 197–1; often, limestone

is used. To control the setting of the cement after addition of

water, calcium sulfate is added to the cement in small quan-

tities. Usually, natural gypsum and/or anhydrite, industrially

processed chemical gypsum, or flue gas desulfurisation (FDG)

gypsum is employed [10]. Each of these materials contains

strontium, an element whose isotopes provide information on

the geographical origin, and, therefore, is highly valuable for

provenance studies.

The potential of conventional 87Sr/86Sr isotope ratios,

hereafter referred to as 87Sr/86Sr isotope ratios, for unravel-

ling the origin of cement was first mentioned by Graham

and co-authors [11]. The advantage of using radiogenic iso-

tope systems such as the Sr isotope system is that the results

can be linked to the geographic origin of raw materials with

more confidence than the results of other approaches such as

elemental fingerprinting. This is partly due to the fact that

the elemental composition of a cement may be considerably

influenced by changes in the environment as well as changes

of the production process [3]. As ordinary Portland cement

consists mainly of Portland clinker, 87Sr/86Sr isotope ratios

of cement and clinker might be expected to be very similar.

However, since the additional cement constituents, as well as

the added calcium sulfates, can come from very different geo-

logical backgrounds or industrial processes, their effect on the

87Sr/86Sr isotope ratio of the resulting cement may be consider-

able. The magnitude and significance of this effect have not

been explored previously. Therefore, the present study com-

pared the 87Sr/86Sr isotope ratios of a number of cements and

clinkers of varying geographical origins and tested approaches

to separate the clinker from its parent cement to enable meas-

uring its 87Sr/86Sr isotope ratio for provenancing of the cement.

Materials andmethods

Cement andclinker samples

Fifteen cements (all ordinary Portland cement, CEM I

according to DIN EN 197–1) and the fifteen corresponding

Portland clinkers used to produce these cements were

obtained from the respective cement producers. All

cements contained the four major clinker phases alite

(Ca3SiO5), belite (Ca2SiO4), aluminate (Ca3Al2O6), and

ferrite (Ca2AlFeO5) in comparable amounts, as determined

by X-ray diffraction (XRD) analysis (see “X-ray diffrac-

tion measurements of non-treated and sieved samples” for

experimental conditions). However, the amount(s) of the

calcium sulfates, i.e. gypsum (CaSO4·2H2O), hemihydrate

(CaSO4·0.5H2O), and/or anhydrite (CaSO4), differed consid-

erably between the cements, as will be discussed in “XRD

results”. Nine cements and clinkers originate from Germany,

two from Serbia, one from Greece, one from Bosnia and

Herzegovina, one from Kosovo, and one from North Mac-

edonia. The rationale for the investigation of cements and

clinkers from different countries was to include materials

with a wide range of isotope ratios of the raw materials as

well as different CEM I production plants. The obtained

sample units (usually ~ 0.5–1kg) were divided into smaller

portions using standard procedures. The samples were stored

in PP beakers. The sample mass for 87Sr/86Sr isotope ratio

analysis was ≈ 100mg.

Chemicals

HNO3 (65–68% v/v) and HCl (≈ 30% v/v) were purchased

as pro analysis grade acids (Chemsolute®, Th. Geyer, Ber-

lin, DE) and were further purified by double sub-boiling

distillation. For the potassium hydroxide/sucrose treatment

(“KOSH treatment”), sucrose (SERVA Electrophoresis

GmbH, Heidelberg, DE, analytical grade, min. 99 wt.%

purity), KOH (Merck KGaA, Darmstadt, DE, pro analysi),

and Milli-Q water (Milli-Q Advantage A10 System, Merck

KGaA, Darmstadt, DE) were used. For Sr purification and

matrix separation, Sr·Spec™ resin (100–150µm, Eichrom

Technologies Inc, Lisle, IL, USA) was employed.

Sample preparation methods: overview

All procedures such as weighing, sample dissolution, and

analyte separation were performed in a metal-free clean

laboratory with ISO class 6 at the Federal Institute for

Materials Research and Testing (BAM). Crushing and grind-

ing of the samples were performed in a normal laboratory

environment.

Clinker samples were crushed and ground before diges-

tion in a 1:1 v/v mixture of concentrated HCl and HNO3 on

the hotplate (130°C) for 48h.

The cement samples were ground as well (except for

those being applied to the sieving procedure) and afterwards

prepared using four different procedures described below

(“Concentrated hydrochloric acid/nitric acid dissolution” to

“KOSH treatment”). Subsequently, the solutions containing

Development ofasample preparation procedure forSr isotope analysis ofPortland cements

1 3

the samples were evaporated until dryness, the precipitate

was redissolved in 2% w/w HNO3, and afterwards, an aliquot

was analysed by inductively coupled plasma mass spectrom-

etry (ICP-MS) to determine the Sr mass fraction. Then, a

subsample containing approximately 2µg Sr was used for

matrix separation by column chromatography using the Sr

resin. The isolated Sr fraction was then dried, redissolved,

and loaded on Re filaments, which were used for multi col-

lector thermal ionisation mass spectrometry (MC-TIMS)

measurements.

To evaluate the sample preparation procedures, the pre-

cipitates after each digestion treatment were analysed by

XRD and phase identification. In addition, to check the out-

comes of the sieving treatment (“Sieving”), the sieved and

non-treated cement samples were analysed by XRD as well.

Figure1 summarises the experimental programme.

Concentrated hydrochloric acid/nitric acid

dissolution

The concentrated hydrochloric and nitric acids were chosen

because of their strong acidic character and consequent abil-

ity to dissolve cement phases. After ball milling, each CEM

I sample was weighed (~ 100mg) in Savillex® beakers. The

next step was the addition of conc. HNO3 and conc. HCl

(2mL each). The beakers were closed, agitated by hand,

and sonicated (20min) and digestion was performed on a

hotplate (130°C) for 48h. The sample preparation using

concentrated HNO3 and HCl 1:1 v/v is abbreviated as “conc.

acid”.

Dilute nitric acid dissolution

The dissolution with 1mol·L−1 HNO3 was already used by

Graham and co-authors in their cement provenancing study

[11], where it has proven effective for the identification of

cement from hardened concrete samples. This approach was

also used by Kasamatsu and co-authors [7] for analysing

nitric acid-soluble components in the fragments of concrete

for forensic issues.

In the present study, the cement samples were first ground

in a planetary ball mill, weighed, and suspended in 4mL of

1mol·L−1 HNO3. The beakers were closed, agitated by hand,

sonicated (20min), and heated on a hotplate (130°C) for

48h. In the text below, the treatment by 1mol·L−1 HNO3 is

abbreviated as “dil. acid”.

Sieving

The rationale for the application of sieving was that the

clinker particles and the calcium sulfates in cements have

different particle size distributions due to their different

grindability. Portland clinker emerges from the kiln as

rounded granules or irregularly shaped lumps, in either

case with a dimension of about 3–30mm. These are subse-

quently blended and ground with the calcium sulfates. Due

to different hardnesses, the different cement components are

enriched in fractions with different particle sizes. The harder

the compound, the more is found in the larger particle size

fraction; gypsum and its dehydration products are concen-

trated in the finer fractions [2]. Theoretically, this property

should allow to separate the sulfates from the clinker by

sieving and subsequently collecting the fraction with parti-

cles greater than a specified size.

The samples were sieved on a 11-µm sieve (Atechnik

GmbH, Leinburg, DE) using a sieving machine (AS 200

control, Retsch GmbH, Haan, DE) and a setting of 2.52-

mm amplitude for 20min. Ideally, sieving should separate

the coarse particles from the fine particles, with the coarser

fraction remaining on the sieve being mainly composed of

Fig. 1 Schematic representation

of the four sample preparation

procedures for cement samples.

Conc. acid refers to dissolution

in concentrated hydrochloric

acid/nitric acid, KOSH refers to

selective dissolution of clinker

phases, dil. acid refers to dis-

solution in dilute nitric acid, and

sieving is dry sieving performed

on a 11-µm sieve. XRD refers

to X-ray diffraction and TIMS

refers to thermal ionisation

mass spectrometry

Kazlagić A.et al.

1 3

clinker. The so-obtained coarser fraction, of course, needs to

be digested for further Sr isotope analysis. Therefore, these

coarser fractions were dissolved in conc. acid as described

in “Concentrated hydrochloric acid/nitric acid dissolution”.

The whole applied procedure is abbreviated as “sieving”,

except for the XRD results, where “sieving” refers to sam-

ples analysed immediately after sieving.

KOSH treatment

A treatment with a potassium hydroxide/sucrose solution

(the so-called KOSH solution) to selectively dissolve alumi-

nate and ferrite phases from Portland cement, leaving alite

and belite phases undissolved, was described by Gutteridge

[12]. In the present context, the KOSH treatment was applied

to remove the additives from the cements, while the concom-

itant dissolution of aluminate and ferrite was unavoidable.

Before the KOSH treatment, the cements were ground

in a planetary ball mill (Pulverisette 5, Fritsch) to particle

sizes ≤ 65µm. The KOSH solution was prepared by weigh-

ing and mixing potassium hydroxide, sucrose, and water,

using the procedure described by Gutteridge [12], except

for slight modifications concerning smaller sample quanti-

ties while maintaining the same weight proportions as in the

original procedure. Instead of filtration, we used centrifu-

gation as the separation technique. The ball-milled CEM I

samples (≈ 1g) were mixed with KOSH solution (previously

prepared by mixing 37.5mL of Milli-Q H2O with 3.75g of

KOH and 3.75g of sucrose) at 95°C and stirred until the

solution turned pale yellow. The solution was then centri-

fuged; the residue was washed two times with water and

one time with methanol, with centrifuging steps in between.

The supernatant was removed by pipetting, and the subna-

tant was dried in the oven overnight (40°C). For the sub-

sequent determination of their 87Sr/86Sr isotope ratios, the

dry residues were then dissolved in conc. acid as described

in “Concentrated hydrochloric acid/nitric acid dissolution”.

The whole procedure is abbreviated as “KOSH”, except for

the XRD results, where “KOSH” refers to samples analysed

after KOSH treatment before digestion in conc. acid.

Analytical techniques

X‑ray diffraction measurements ofnon‑treated

andsieved samples

All samples of the non-treated and the sieved cements were

prepared for XRD measurements by filling the sample pow-

der into the cylindrical cavity of a standard polyvinylchlo-

ride sample holder and gently compressing it by using a

glass plate until the sample surface became aligned with the

sample holder surface.

The XRD measurements were performed on a

D8 ADVANCE diffractometer (Bruker AXS, DE) in

Bragg–Brentano geometry under the following conditions:

Cu Kα radiation (λ = 1.54187Å); X-ray tube: 40kV and

40mA; step size: 0.02° 2θ; scanning rate: 2.4° 2θ min−1;

LYNXEYE XE-T detector.

X‑ray diffraction measurements oftheresidues

aftertreatment

Five of fifteen cement and clinker samples (namely 3022,

3028, 3050, 3063, 3078) were chosen to analyse their resi-

dues by XRD after different preparation methods. The selec-

tion was based on a comparison of the phase assemblages

of all cements. The chosen samples contained different

amounts of calcium sulfates, i.e. the contents of gypsum

and anhydrite were either zero, close to the maximum of all

cements, or intermediate. Therefore, the XRD results of the

five cements are representative for the processes occurring

during treatments in all fifteen cement samples.

The KOSH, conc. acid, and dil. acid residues were ground

manually with mortar and pestle (agate) before the XRD

measurements. Since the available sample masses were low,

the resulting powders of the KOSH residues were dusted

on a flat (no cavity), single-crystal Si sample holder, cut to

give no reflections in the measurement range. The dil. acid

residues and the conc. acid residues had a jelly-like consist-

ency and were smeared on the centre region of the Si sample

holder. These latter samples had a hygroscopic character,

and as a result, the samples turned partially or completely

into viscous liquids during the XRD measurements.

The XRD patterns were recorded on an Ultima IV diffrac-

tometer (Rigaku, Japan) in Bragg–Brentano geometry under

the following conditions: Cu Kα radiation (λ = 1.54187Å);

X-ray tube: 40kV and 40mA; step size: 0.01° 2θ; scanning

rate: 0.2° 2θ min−1; scanning range: 5–65° 2θ; divergence

slit 1/2° (in plane), 10mm (axial); strip detector D/teX Ultra.

Conventional 87Sr/86Sr isotope analysis (MC‑TIMS

analysis)

Approximately 100mg of each cement and clinker sample

was weighted in a Savillex® beaker and prepared according

to the previously described methods. After determination of

the Sr mass fraction (by iCAP-Q ICP-MS, Thermo Fisher

Scientific, Bremen, DE), an aliquot of the sample was trans-

ferred to a new Savillex® beaker and dried on a hotplate.

After drying, the sample was redissolved in 3mol·L−1 HNO3

(1mL). An aliquot of this solution containing 2μg strontium

was taken to perform a strontium matrix separation using

the water suspension of Sr·Spec™ resin (350 µL) in poly-

vinylchloride columns (6mm inner diameter, 4cm long).

The resulting strontium fraction was evaporated to dryness

Development ofasample preparation procedure forSr isotope analysis ofPortland cements

1 3

and redissolved in nitric acid such that a final strontium

mass fraction of 100ng·μL−1 was obtained, which could be

directly used for loading 1 µL of the sample on Re filaments,

together with TaF5 activator for enhancing the ionisation.

Strontium isotope analyses were carried out by MC-TIMS

at BAM in Berlin using a Sector 54 instrument (Micromass

Ltd., Manchester, UK), in a dynamic multi-collection mode

via an automatic measurement procedure. The raw measured

data were corrected for interfering Rb and mass fractionation

(86Sr/88Sr = 0.1194) and finally were normalised to a NIST

SRM 987 87Sr/86Sr ratio of 0.71025 [13], which is also the

median of more than thousand published results listed in

the GeoRem database [14]. Furthermore, NASS-6 seawater

reference material was used as a control sample.

Results anddiscussion

XRD results

XRD results ofnon‑treated andsieved cements

The diffractograms of the cements before and after siev-

ing (see Supplementary Information (ESM), Figs.S1–S3)

reveal that the cements exhibited similar phase assemblages

as regards the major clinker phases alite (Powder Diffrac-

tion File [15] [PDF] # 01–073-0599), belite (PDF # 01–086-

0398), aluminate (PDF # 00–038-1429), and ferrite (PDF #

00–030-0226). In addition, in some of the cements, minor

amounts of quartz (SiO2; PDF # 00–046-1045) and calcite

(CaCO3; PDF # 01–086-0174) impurities were identified.

More significant differences existed regarding the calcium

sulfates in the cements, namely anhydrite (PDF # 01–072-

0916), hemihydrate (PDF # 01–083-0438), and gypsum

(PDF # 00–033-0311). Anhydrite exhibited the strongest

intensities, i.e. anhydrite was the main sulfate in most cases

(cements 3022–3029, 3032). The pertinent cements addi-

tionally contained minor amounts of hemihydrate, noticeable

as a shoulder at 14.74° 2θ in the diffractograms. Gypsum

was the main sulfate in cement 3063, which was the only

sample with no anhydrite signals. In all other samples, gyp-

sum occurred either as a minor component or this phase

was not present at all (cements no. 3022, 3029, 3030, 3078).

The differences between the cements result from different

additions of sulfates during cement production as well as

partial dehydration of the hydrated calcium sulfates during

cement milling. Their amounts are usually optimised and

adapted to the clinker composition by the cement producer

to achieve optimum setting and other properties of the final

material [2].

Comparison of the diffractograms of the non-treated

(un-sieved) and the sieved cements (see ESM, Figs.S1–S3)

shows that in no case sieving removed the calcium sulfates

from the cements. Instead, the sulfate contents of the

cements (as indicated by the heights of the reflections of

the sulfates, relative to the intensity of the reflection of alite

at 32.20° 2θ, i.e. d = 2.78Å) were comparable before and

after sieving, and in some cases, the fraction of anhydrite

or gypsum was even higher after sieving (e.g. cements 3022

and 3063). The present results clearly show that the applied

sieving procedure did not separate additives from clinker

in the cement. The underlying reasons could not be conclu-

sively clarified. However, a likely explanation is that the fine

calcium sulfate particles stuck to the clinker particles during

sieving due to physical bonding. Therefore, separation based

on a size differentiation with the proposed method is not

suitable to separate coarser clinker particles from the finer

particles, such as added calcium sulfates.

XRD results oftheresidues afterKOSH treatment

The KOSH treatment resulted in the selective dissolution of

the cements, as shown in Fig.2. The clinker phases alite and

belite remained unaffected by the KOSH treatment, while

aluminate largely dissolved, and ferrite completely disap-

peared. In addition, the sulfates were removed from the

cements. Reflections of quartz and calcite were detected in

the diffractograms of the residues, meaning that quartz had

remained undissolved and that calcite either had remained

undissolved or had reprecipitated during evaporation of the

solution after digestion, likely due to exposure to CO2 in

the air. The relative intensity of the peak at 32.68° 2θ was

increased compared to the untreated cements. This is the

position of the main peak of potassium hydroxide (KOH;

PDF # 01–089-7389); thus, the relative increase is possibly

related to precipitation of KOH from the KOSH solution

during treatment.

XRD results oftheresidues afterconcentrated hydrochloric

acid/nitric acid dissolution (conc. acid)

In contrast to the KOSH treatment, the conc. acid treat-

ment led to a virtually complete dissolution of the initial

cement phases (Fig.3). No alite-, belite-, aluminate-, and

ferrite-related peaks were detected in the diffractograms.

Most of the samples were fully X-ray amorphous. The

diffractogram of cement 3063 after the treatment exhib-

ited a reflection around 7.20° 2θ, which could be matched

with an LTA-type zeolite (PDF # 01–089-8015). Clinker

3050b had an additional peak at 9.64° 2θ (see Fig.S4,

ESM), which can be tentatively assigned to dealuminated

chabazite (PDF # 00–052-0784). The presence of these

newly formed silicates in the residues is an additional

indication that the clinker silicates, i.e. alite and belite,

had dissolved during the conc. acid treatment. In some

cases, signals with very low intensity were present around

Kazlagić A.et al.

1 3

25.4 and 29.6° 2θ, which were possibly caused by minor

amounts of remaining anhydrite and calcite, respectively.

Again, this would mean that these phases had either partly

remained undissolved or have reprecipitated during evapo-

ration after digestion of the samples.

XRD results oftheresidues afterdilute nitric acid

dissolution (dil. acid)

In general, the results for the cements treated with dil.

acid are comparable to those of the samples treated with

the conc. acid solution. Namely, the diffractograms indi-

cate that the initial cement phases had dissolved, and the

samples were virtually fully amorphous, though some of

them exhibited few distinct peaks of crystalline phases

(Fig.4). Cements 3022, 3028, 3050, 3063, and 3078 con-

tained minor amounts of gypsum, and samples 3028, 3063,

and 3050 additionally contained quartz, which was present

in the original cements, besides amorphous phase.

In samples 3022 and 3050, additional peaks were pre-

sent at 11.8° 2θ and 23.59° 2θ, which are assigned to tetra-

calcium monocarboaluminate (Ca4Al2(OH)12[CO3]·5H2O;

PDF # 01–087-0493). The carbonate ion in monocar-

boaluminate is readily replaced by nitrate [16]; thus, its

NO3-exchanged form (“NO3-AFm”) could have been

expected after treatment with nitric acid. The present

results, however, indicate that heating at 130°C for 48h

had effectively removed nitrate from the samples, so that

monocarboaluminate formed, with the required CO2 likely

provided by the atmosphere.

Fig. 2 X-ray diffractograms of the cement samples 3022, 3028, 3050,

3063, and 3078 after KOSH treatment: (a) alite, (b) belite, (d) alumi-

nate, (c) calcite, (q) quartz, (K) KOH

Fig. 3 X-ray diffractograms of the cement samples 3022, 3028, 3050,

3063, and 3078 after conc. acid treatment: (A) anhydrite, (c) calcite,

(q) quartz, (z1) zeolite-type phase

Fig. 4 X-ray diffractograms of the cement samples 3022, 3028, 3050,

3063 and 3078 after dil. acid treatment: (G) gypsum, (m) monocar-

boaluminate, (c) calcite, (q) quartz

Development ofasample preparation procedure forSr isotope analysis ofPortland cements

1 3

Results of87Sr/86Sr isotope analyses

Concentrated hydrochloric acid/nitric acid dissolution

(conc. acid) andsieving ofthecements

In Fig.5, the 87Sr/86Sr isotope ratios of the cements

digested by conc. acid treatment are compared with those

of the same cements after sieving and subsequent digestion

in conc. acid. The isotope ratios obtained with these two

methods excellently agree within the stated uncertainties

for almost all cements, in line with the XRD analyses,

which showed that sieving did not remove the calcium sul-

fate additions from the cements, nor significantly changed

their phase assemblages in other ways.

KOSH treatement

Sr isotope ratios for cement samples after KOSH method

and subsequent conc. acid digestion were compared to

the corresponding clinkers after conc. acid treatement, as

shown in Fig.6. For samples 3027, 3029, 3062, and 3078,

the KOSH treatment gave considerably different ratios for

the cements and the clinkers, the 87Sr/86Sr isotope ratios

of the cements being much higher than those obtained for

the corresponding clinker samples. In some cases, such

as 3026, 3030, and 3076, lower ratios were obtained for

the cements than for the corresponding clinker samples.

Therefore, there is no general trend discernable in the

data. However, a paired t-test applied to the data shows

that the difference between KOSH-treated cements and

the corresponding clinker samples is statistically signifi-

cant (t(14) = − 2.15, p = 0.05). The results of all calculated

t-tests can be found in TablesS4–S7 in the ESM.

Concentrated hydrochloric acid/nitric acid dissolution

(conc. acid)

When comparing the 87Sr/86Sr isotope ratios of the cement

samples after conc. acid treatment with the corresponding

clinker samples, the situation is different (Fig.7). The conc.

acid treatment gave a considerably higher ratio for only one

cement sample (3027); slightly higher ratios for the four

cement samples 3024, 3062, 3063, and 3078; and a lower

ratio for the other samples, especially for 3075 and 3076.

The results of the t-test showed no statistically significant

difference between the cement samples after the conc. acid

treatment and the clinker samples (t(14) = 0.26, p = 0.80).

A comparison was also performed between the cements

after sieving and subsequent conc. acid treatment and the

corresponding clinker samples after conc. acid digestion.

The results are shown in Fig.S5 in the ESM. Both sieving

Fig. 5 87Sr/86Sr isotope ratios of fifteen cement samples after conc.

acid treatement (orange) and after the sieving method (purple); error

bars represent expanded uncertainty (U, k = 2).

Fig. 6 Comparison of 87Sr/86Sr isotope ratios from fifteen cement

samples after KOSH treatment (red) and fifteen corresponding clinker

samples after conc. acid treatment (blue-green)

Fig. 7 Comparison of 87Sr/86Sr isotope ratios from fifteen cement

samples after conc. acid treatment (orange) and fifteen corresponding

clinker samples after conc. acid treatment (blue-green)

Kazlagić A.et al.

1 3

and conc. acid treatment of the cements yielded very similar

Sr isotope data (cf. Figure5). This confirms the results of

the XRD measurements.

Dilute nitric acid dissolution (dil. acid)

In Fig.8, the 87Sr/86Sr isotope ratios of the cements digested

by dil. acid treatment are compared with those of the corre-

sponding clinkers after digestion in conc. acid. The cements

after dil. acid treatment show the same trend as the cements

after conc. acid treatment. In fact, the dil. acid treatment

gave a much higher ratio for only one cement (3027), slightly

higher for the four cements 3024, 3062, 3063, and 3078,

and lower ratio for the other samples, especially for 3075

and 3076. There was no significant difference between the

samples after dil. acid treatment and the clinker samples

(t(14) = 0.45, p = 0.66).

Discussion

The selection of the most suitable sample preparation proce-

dure was realised by comparing the 87Sr/86Sr isotope ratios

of differently treated cement samples (processed cements)

with the corresponding clinker samples (see TablesS1 and

S2 in the ESM). The final goal was to find a procedure for

preparing cement in such a way that the 87Sr/86Sr isotope

ratio gives the closest value to the 87Sr/86Sr isotope ratio of

the corresponding clinker, but also to fulfil the requirements

regarding efficiency, e.g. time consumption.

To better compare the four different procedures applied to

fifteen pairs of samples, we reduced the data for each sam-

ple. The 87Sr/86Sr isotope ratio of each processed cement was

used to compute the absolute difference to the corresponding

clinker sample according to

The so-computed differences (see TableS3 in the ESM)

are shown in Fig.9 for all four methods applied to fifteen

processed cement samples.

For further data reduction, the average of Δabs was cal-

culated for each sample treatment. The results are shown

in Fig.10. The lower the average absolute difference is, the

closer are the 87Sr/86Sr isotope ratios determined for the

cements to those of the corresponding clinkers on average,

i.e. the better the method met the requirement to reflect the

87Sr/86Sr isotope ratio of the clinker. The average absolute

differences were 0.00021, 0.00018, 0.00018, and 0.00017,

for KOSH, sieving, dil. acid, and conc. acid respectively.

A possible explanation for the comparatively large aver-

age

Δabs

resulting from the KOSH treatment is contamina-

tion during the treatment. The measured procedural Sr blank

for the KOSH method was 75ng/g, resulting in 3.4µg of

Sr in a standard batch of KOSH solution (45g of KOSH

solution). Assuming that all Sr in the KOSH solution is

transferred to the cement, the result is 3.4µg of Sr contami-

nation per 1g of treated cement. This amount can have a

significant impact on the measured total 87Sr/86Sr isotope

ratios, because the Sr contents of the cements were in the

range of 100–1200µg/g, leading to a total Sr mass of 100 to

1200µg per sample treatment. However, an influence of the

87Sr/86Sr isotope ratio of the KOSH solution/ blank does not

fully explain the observed deviations, because if it had been

the sole influence, the contamination from the blank would

shift the samples isotope ratios towards the blank isotope

ratio. However, inspection of Fig.6 does not reveal a sys-

tematic trend, neither for a low, high, or seawater isotopic

(1)

abs =

|||||(

87Sr

86Sr

)

cement

−

(

87Sr

86Sr

)

clinker

|||||

Fig. 8 Comparison of 87Sr/86Sr isotope ratios from fifteen cement

samples after dil. acid treatment (fluorescent green) and fifteen cor-

responding clinker samples after conc. acid treatment (blue-green)

Fig. 9 Comparison of all four methods on fifteen processed cement

samples presented as the absolute difference between the 87Sr/86Sr

isotope ratio of the processed cement and the 87Sr/86Sr isotope ratio

of the corresponding clinker

Development ofasample preparation procedure forSr isotope analysis ofPortland cements

1 3

composition of the blank. Thus, evidently other factors play

a role here, which, however, could not be elucidated in the

present study.

To rank the usefulness of the four methods, we used five

parameters (Table1). The first parameter is the average

absolute difference (average Δabs) between the 87Sr/86Sr iso-

tope ratios of the processed cements and the corresponding

clinkers. This parameter was introduced because it reflects

extreme differences between the sample pairs (cement and

corresponding clinker). Conc. acid treatment exhibited

the lowest average Δabs, meaning this method yielded the

value closest to the target on average. The KOSH treatment

showed the highest average Δabs. For dil. acid treatment

and sieving, the average Δabs was approximately the same,

intermediate between the values of conc. acid treatment and

KOSH treatment.

The second parameter was the number (N1) of samples

(cement-clinker pairs) which gave a Δabs below a specified

threshold. Three times the expanded uncertainty (3 × U)

was selected as the threshold. Every sample pair (cement

and clinker) was evaluated within every sample preparation

procedure. The higher the number N1 of samples that fit the

criterion, the better the preparation procedure. The square

sum approach was used to assess the combined standard

uncertainty, uc, of the 87Sr/86Sr isotope ratio results. The fol-

lowing uncertainty components were identified: repeatability

of a single 87Sr/86Sr measurement in a sample, repeatabil-

ity of 87Sr/86Sr measurement in measured NIST SRM 987,

the bias to the reference value for 87Sr/86Sr measurement

in NIST SRM 987 (value published in GeoReM database

[14]), the experimental reproducibility of independently

processed samples, the bias of the measured 87Sr/86Sr ratio

in a processed AGV-2a sample versus the reference value

(value published in GeoReM database [14]), and the bias of

measured 87Sr/86Sr ratio in processed NASS-6 sample versus

literature value. The expanded uncertainty (U = k∙uc, with

k = 2) of the method for obtaining conventional 87Sr/86Sr iso-

tope ratios was calculated to be 0.000023. Based on these

calculations, N1 was 6, 5, 4, and 4 for KOSH, conc. acid,

sieving, and dil. acid treatment, respectively.

The other three parameters were evaluated by the analysts

and include the complexity of preparation, the time con-

sumption in hours, and the number N2 of prepared samples

per workday. Thus, these parameters are strictly applicable

only to the conditions of the present study, but they can serve

as an estimate for other laboratories too. Conc. acid and dil.

acid treatment were neither difficult nor time consuming,

and N2 was 15. Sieving and KOSH treatment were more

difficult to perform, particularly the KOSH treatment, due to

the involved compounds, and the need for repeating several

steps, e.g. dissolving, centrifugation, and drying. Further-

more, for KOSH treatment and sieving, the time consump-

tion were higher, and the number of samples which can be

handled per workday were lower. This especially applied

for sieving, since cleaning of the sieves after use was time

consuming, and the number of available sieves was a limit-

ing factor.

Summary andconclusions

The present study shows that with the employed setup

and conditions for the sieving method, satisfying results

could not be achieved. Sieving was not suitable to separate

coarser clinker particles from the finer calcium sulfate par-

ticles of cement, and therefore, the additives could not be

Fig. 10 Comparison of all methods: average absolute difference of

the 87Sr/86Sr isotope ratios of the processed cements and the corre-

sponding clinkers

Table 1 Comparison of

all methods regarding the

difficulty of preparation, time

consumption in hours and

number of prepared samples per

workday

*Including cleaning the sieves; **limited by the number of available sieves.

Method Average difference

from clinker N1 samples

below threshold Difficulty of

preparation Time consump-

tion in hours N2 samples

per work-

day

conc. acid 0.00017 5 Low 2 15

dil. acid 0.00018 4 Low 2 15

Sieving 11µm 0.00018 4 Medium 8* 3**

KOSH 0.00021 6 High 6 5

Kazlagić A.et al.

1 3

removed before Sr isotope analysis. This was confirmed

by Sr isotope analysis on MC-TIMS, as well as by XRD

analysis, which showed the calcium sulfates to be present

in the coarse fraction after sieving. Future work should

investigate whether air-jet sieving is more appropriate for

the present purpose.

XRD results confirmed that the KOSH method suc-

cessfully led to selective dissolution of clinker phases and

cement components, with the aluminate almost completely

dissolved, and the calcium sulfates and ferrite completely

removed. The clinker phases alite and belite remained largely

unaffected by the KOSH treatment. Thus, only KOSH has

proven effective out of the two employed approaches to

remove the calcium sulfate additives from the cements.

XRD showed that the conc. acid method led to a virtually

complete breakdown of the initial phases, where all alite-,

belite-, aluminate-, and ferrite-related peaks disappeared.

That means that all Sr-bearing compounds in the cement

were dissolved except for insoluble minor constitutents (e.g.

quartz). The dil. acid method yielded very similar results.

Despite the complete removal of the calcium sulfates

from the cements, due to reasons that could not be clari-

fied in the present study, the average deviation between

the 87Sr/86Sr isotope ratios of the cements treated with the

KOSH solution and the corresponding clinkers (Δabs) was

the largest in the present study, although the KOSH treat-

ment also returns the most pairs with a Δabs below the cho-

sen threshold (3 × U). This is somehow contradictory and

cannot be explained by blank issues only. The partial dis-

solution and reprecipitation during KOSH treatment might

dissolve varying Sr reservoirs depending on the individual

cement sample and thus leading to this behaviour.

The fact that the average Δabs was lowest for the conc.

acid treatment and that the t-test showed that the difference

between the 87Sr/86Sr isotope ratios of the cement samples

treated with conc. acid and the corresponding clinkers was

not statistically significant, while it was for cements after

KOSH treatment, strongly argues for the sample preparation

with conc. acid. Treatment with dil. acid led to slightly less

satisfactory results, i.e. a slightly higher average Δabs and a

lower number of samples with a 87Sr/86Sr ratios difference

between cement and clinker below the threshold.

It is thus concluded that dissolution in conc. acid (con-

centrated hydrochloric acid/nitric acid) yields satisfactory

results and is currently the most appropriate sample prepara-

tion method for determination the 87Sr/86Sr isotope ratios of

Portland cements (CEM I), compared to the other methods

employed in the present study. The obtained values can serve

as a proxy for the 87Sr/86Sr isotope ratio of the clinker in the

cement and can thus be used for cement provenance studies.

Supplementary Information The online version contains supplemen-

tary material available at https:// doi. or g/ 10. 1007/ s00216- 021- 03821-7.

Acknowledgements Dominik Al-Sabbagh (BAM) is thanked for his

assistance in performing the XRD measurements of the non-treated

and the sieved cements.

Author contribution Anera Kazlagić—conceptualisation, formal analy-

sis, data curation, investigation, methodology, project administration,

visualisation, writing—original draft, writing—review and editing.

Francesco F. Russo—formal analysis, data curation.

Jochen Vogl—conceptualisation, funding acquisition, supervision,

writing—review and editing.

Patrick Sturm—formal analysis, investigation, visualisation.

Dietmar Stephan—supervision, resources, writing—review and

editing.

Gregor J. G. Gluth—supervision, methodology, validation, writ-

ing—review and editing.

Funding Open Access funding enabled and organized by Projekt

DEAL. This work received funds under the project number MIT1-

2019–8 from the Federal Institute for Materials Research and Testing

(BAM).

Availability of data and material The data used to support the find-

ings of this study is available within the article and the associated

Supplementary Information file. The data is also available from the

corresponding author upon reasonable request.

Code availability Not applicable.

Declarations

Conflict of interest The authors declare no competing interests.

Open Access This article is licensed under a Creative Commons Attri-

bution 4.0 International License, which permits use, sharing, adapta-

tion, distribution and reproduction in any medium or format, as long

as you give appropriate credit to the original author(s) and the source,

provide a link to the Creative Commons licence, and indicate if changes

were made. The images or other third party material in this article are

included in the article's Creative Commons licence, unless indicated

otherwise in a credit line to the material. If material is not included in

the article's Creative Commons licence and your intended use is not

permitted by statutory regulation or exceeds the permitted use, you will

need to obtain permission directly from the copyright holder. To view a

copy of this licence, visit http:// creat iveco mmons. or g/ licen ses/ by/4. 0/.

References

1. Elhacham E, Ben-Uri L, Grozovski J, Bar-On YM, Milo

R. Global human-made mass exceeds all living biomass.

Nature. 2020;588(7838):442–4. https:// doi. org/ 10. 1038/

s41586- 020- 3010-5.

2. Taylor HFW. Cement chemistry. 2nd ed. London: Thomas Telford

Ltd.; 2009. p. 459.

3. Kazlagic A, Vogl J, Gluth GJG, Stephan D. Provenancing of

cement using elemental analyses and isotope techniques – the

state-of-the-art and future perspectives. J Anal At Spectrom.

2021;36:2030–42. https:// doi. org/ 10. 1039/ D1JA0 0144B.

4. Goguel RL, Stjohn DA. Chemical-identification of Portland

cements in New-Zealand concretes .1. Characteristic differ-

ences among New-Zealand cements in minor and trace-element

Development ofasample preparation procedure forSr isotope analysis ofPortland cements

1 3

chemistry. Cem Concr Res. 1993;23(1):59–68. https:// doi. org/ 10.

1016/ 0008- 8846(93) 90136-W.

5. Potgieter JH, Potgieter SS, McCrindle RI, Tamas FD. Fingerprint-

ing of South African cement clinkers and gypsum as a tool for

cement identification purposes. Adv Cem Res. 2003;15(2):45–50.

https:// doi. org/ 10. 1680/ adcr. 15.2. 45. 36725.

6. Potgieter-Vermaak SS, Potgieter JH, Worobiec K, van Grieken

R, Majanovic L, Moeketsi S. Fingerprinting of South African

ordinary Portland cements, cement blends and mortars for iden-

tification purposes - discrimination with starplots and PCA. Cem

Concr Res. 2007;37(6):834–43. https:// doi. org/ 10. 1016/j. cemco

nres. 2007. 02. 013.

7. Kasamatsu M, Igawa T, Suzuki S, Suzuki Y. Forensic discrimi-

nation of concrete pieces by elemental analysis of acid-soluble

component with inductively coupled plasma-Mass Spectrometry.

Anal Sci. 2018;34(6):729–33. https:// doi. org/ 10. 2116/ anals ci.

18SBN 02.

8. Pirrie D, Pidduck AJ, Crean DE, Nicholls TM, Awbery RP. Identi-

fication and analysis of man-made geological product particles to

aid forensic investigation of provenance in the built environment.

Forensic Sci Int. 2019;305:109974. https:// doi. org/ 10. 1016/j. forsc

iint. 2019. 109974.

9. Telschow S, Frandsen F, Theisen K, Dam-Johansen K. Cement

formation-a success Story in a black box: high temperature

phase formation of Portland cement clinker. Ind Eng Chem Res.

2012;51(34):10983–1004. https:// doi. org/ 10. 1021/ ie300 674j.

10. Locher FW. Cement : principles of production and use. Düssel-

dorf: Bau+Technik Gmbh; 2006. p. 535.

11. Graham IJ, Goguel RL, St John DA. Use of strontium isotopes to

determine the origin of cement in concretes - case examples from

New Zealand. Cem Concr Res. 2000;30(7):1105–11. https:// doi.

org/ 10. 1016/ S0008- 8846(00) 00248-9.

12. Gutteridge WA. Dissolution of the interstitial phases in Portland-

cement. Cem Concr Res. 1979;9(3):319–24. https:// doi. org/ 10.

1016/ 0008- 8846(79) 90124-8.

13. Faure G, Mensing TM. Isotopes : principles and applications. 3rd

ed. Hoboken: John Wiley & Sons, Inc.; 2013.

14. Jochum KP, Nohl U, Herwig K, Lammel E, Stoll B, Hofmann AW.

GeoReM: a new geochemical database for reference materials

and isotopic standards. Geostand Geoanal Res. 2005;29(3):333–8.

https:// doi. org/ 10. 1111/j. 1751- 908X. 2005. tb009 04.x.

15. Gates-Rector S, Blanton T. The Powder Diffraction File: a

quality materials characterization database. Powder Diffr.

2019;34(4):352–60. https:// doi. org/ 10. 1017/ S0885 71561 90008 12.

16. Balonis M, Medala M, Glasser FP. Influence of calcium nitrate

and nitrite on the constitution of AFm and AFt cement hydrates.

Adv Cem Res. 2011;23(3):129–43. https:// doi. org/ 10. 1680/ adcr.

10. 00002.

Publisher's note Springer Nature remains neutral with regard to

jurisdictional claims in published maps and institutional affiliations.

SUPPLEMENTARY INFORMATION FOR:

Development of a sample preparation procedure for Sr isotope analysis of

Portland cements

Anera Kazlagić1, Francesco F. Russo1, Jochen Vogl1, Patrick Sturm2, Dietmar Stephan3,

Gregor J. G. Gluth2

1 Federal Institute for Materials Research and Testing, Division 1.1 Inorganic Trace Analysis,

Richard-Willstäter-Straße 11, 12489 Berlin, Germany

2 Federal Institute for Materials Research and Testing, Division 7.4 Technology of Construction

Materials, Unter den Eichen 87, 12205 Berlin, Germany

3 Technische Universität Berlin, Department of Civil Engineering, Building Materials and

Construction Chemistry, Gustav-Meyer-Allee 25, 13355 Berlin, Germany

Correspondence to: anera.kazlagic@bam.de

Contents

Abbreviations .......................................................................................................................................... S-2

Equation .................................................................................................................................................. S-2

Fig. S1 ..................................................................................................................................................... S-3

Fig. S2 ..................................................................................................................................................... S-4

Fig. S3 ..................................................................................................................................................... S-5

Fig. S4 ..................................................................................................................................................... S-6

Fig. S5 ..................................................................................................................................................... S-7

Table S1 .................................................................................................................................................. S-8

Table S2 .................................................................................................................................................. S-9

Table S3 ................................................................................................................................................ S-10

Table S4 ................................................................................................................................................ S-11

Table S5 ................................................................................................................................................ S-12

Table S6 ................................................................................................................................................ S-13

Table S7 ................................................................................................................................................ S-14

Abbreviations:

Sieving – sieving on 11-µm sieve

KOSH - potassium hydroxide/sucrose solution

Conc. acid - Concentrated hydrochloric acid/nitric acid dissolution

Dil. acid - Dilute nitric acid dissolution

Equation:

Δ𝑎𝑎𝑎𝑎𝑎𝑎 =��𝑆𝑆𝑆𝑆

87𝑆𝑆𝑆𝑆

86 �𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑑𝑑 −�𝑆𝑆𝑆𝑆

87𝑆𝑆𝑆𝑆

86 �𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐�(1)

Legend:

Δabs - Absolute difference of processed cement from the clinker

(87Sr/86Sr)cement – Sr isotope ratio of processed cement

(87Sr/86Sr)clinker – Sr isotope ratio of the clinker

Fig. S1. X-ray diffractogram of non-treated cement (directly on top of the sample code) and

sieved cement (on top of the respective non-treated pattern) for samples 3022, 3024, 3026, 3027

and 3028. G = Gypsum, A = Anhydrite.

10 20 30 40 50

°2θ (Cu Kα)

3022

A3024

3026

3027

3028

Fig. S2. X-ray diffractogram of non-treated cement (directly on top of the sample code) and

sieved cement (on top of the respective non-treated pattern) for samples 3029, 3030,3032, 3050

and 3062. G = Gypsum, A = Anhydrite.

10 20 30 40 50

°2θ (Cu Kα)

3029

A3030

3032

3050

3062

Fig. S3. X- ray diffractogram of non-treated cement (directly on top of the sample code) and

sieved cement (on top of the respective non-treated pattern) for samples 3063, 3064, 3075, and

3078. G = Gypsum, A = Anhydrite.

10 20 30 40 50

°2θ (Cu Kα)

3063

3064

3075

3076

3078

Fig. S4. X-ray diffractograms of the clinker samples 3022b, 3028b, 3050b, 3063b and 3078b after

conc. acid treatment; A = anhydrite, z2 = zeolite-type phase.

Fig. S5. Comparison of 87Sr/86Sr isotope ratios from fifteen cement samples prepared with

sieving and subsequent conc. acid treatment (purple) and fifteen corresponding clinker samples

after conc. acid treatment (blue-green).

Table S1. 87Sr/86Sr isotope ratio of processed clinkers, and cements after the KOSH and sieving

treatment and clinker.1 2SE represents 2 x Standard Error (Standard Deviation divided by the

square root of number of measured isotope ratios).

Sample

87Sr/86Sr

(clinker)

2SE x 10-5

(clinker)

87Sr/86Sr

(KOSH)

2SE x 10-5

(KOSH)

87Sr/86Sr

(Sieving)

2SE x 10-5

(Sieving)

3022

0.70807

0.70810

0.70807

3024

0.70798

0.70816

0.70811

3026

0.70818

0.70812

0.70809

3027

0.70814

0.70888

0.70886

3028

0.70826

0.70827

0.70825

3029

0.70792

0.70816

0.70785

3030

0.70802

0.70795

0.70794

3032

0.70880

0.70899

0.70858

3050

0.70783

0.70782

0.70783

3062

0.70803

0.70838

0.70824

3063

0.70910

0.70927

0.70918

3064

0.70867

0.70874

0.70849

3075

0.70898

0.70897

0.70863

3076

0.70919

0.70896

0.70880

3078

0.70936

0.71012

0.70951

1 Each cement sample was loaded on a single filament, and the standard deviation was then

calculated from minimum 100, maximum 200 measured isotope ratios.

Table S2. 87Sr/86Sr isotope ratio of processed cements after the dil. acid and conc. acid treatment.

Sample

87Sr/86Sr (Dil.

Acid)

2SE x 10-5 (Dil.

Acid)

87Sr/86Sr (Conc.

Acid)

2SE x 10-5 (Conc.

Acid)

3022

0.70804

0.70803

3024

0.70811

0.70812

3026

0.70808

3027

0.70881

3028

0.70823

0.70824

3029

0.70783

0.70786

3030

0.70793

0.70795

3032

0.70856

0.70859

3050

0.70781

0.70783

3062

0.70821

3063

0.70917

3064

0.70848

3075

0.70860

0.70861

3076

0.70879

0.70881

3078

0.70946

0.70948

Table S3. Absolute difference Δabs between the 87Sr/86Sr isotope ratio of the cement and the

87Sr/86Sr isotope ratio of the corresponding clinker for the four investigated preparation methods.

Δabs was calculated using equation (1).

Sample

abs

(KOSH)

x 10-5

abs

(sieving)

x 10-5

abs

(dil.

acid) x 10-5

abs

(conc. acid)

x 10-5

3022

0.03

3024

3026

3027

3028

3029

3030

3032

3050

0.3

3062

3063

3064

3075

3076

3078

Table S4. Results of the paired t-test for 87Sr/86Sr isotope ratios of clinker and KOSH 87Sr/86Sr

isotope ratios of cements as calculated with Excel (Office 365).

t-Test: Paired Two Sample for Means

87Sr/86Sr Clinker

87Sr/86Sr KOSH

Mean

0.708436

0.708592

Variance

2.74E-07

3.8E-07

Observations

Pearson Correlation

0.891731

Hypothesized Mean Difference

t Stat

-2.15852

P(T<=t) one-tail

0.024369

t Critical one-tail

1.76131

P(T<=t) two-tail

0.048737

t Critical two-tail

2.144787

Table S5. Results of the paired t-test for clinker 87Sr/86Sr isotope ratios and sieving 87Sr/86Sr

isotope ratios of cements as calculated with Excel (Office 365).

t-Test: Paired Two Sample for Means

87Sr/86Sr Clinker

87Sr/86Sr Sieved

Mean

0.708436

0.708429

Variance

2.74E-07

2.48E-07

Observations

Pearson Correlation

0.865751

Hypothesized Mean Difference

t Stat

0.107146

P(T<=t) one-tail

0.458097

t Critical one-tail

1.76131

P(T<=t) two-tail

0.916194

t Critical two-tail

2.144787

Table S6. Results of the paired t-test for clinker 87Sr/86Sr isotope ratios and dil. acid 87Sr/86Sr

isotope ratios of cements as calculated with Excel (Office 365).

t-Test: Paired Two Sample for Means

87Sr/86Sr Clinker

87Sr/86Sr Dil

Mean

0.708436

0.708406

Variance

2.74E-07

2.41E-07

Observations

Pearson Correlation

0.873467

Hypothesized Mean Difference

t Stat

0.447992

P(T<=t) one-tail

0.330507

t Critical one-tail

1.76131

P(T<=t) two-tail

0.661014

t Critical two-tail

2.144787

Table S7. Results of the paired t-test for clinker 87Sr/86Sr isotope ratios and conc. acid 87Sr/86Sr

isotope ratios of cements as calculated with Excel (Office 365).

t-Test: Paired Two Sample for Means

87Sr/86Sr Clinker

87Sr/86Sr conc

Mean

0.708436

0.708419

Variance

2.74E-07

2.41E-07

Observations

Pearson Correlation

0.877963

Hypothesized Mean Difference

t Stat

0.259105

P(T<=t) one-tail

0.399662

t Critical one-tail

1.76131

P(T<=t) two-tail

0.799325

t Critical two-tail

2.144787

Chapter 2

Publications

2.3. Fingerprinting Portland Cements by means of 87Sr/86Sr and 143Nd/144Nd Isotope

Ratios and Geochemical Profiles

Preprint Version

Accepted in the journal “Advances in Cement Research “

Manuscript number: ADCR-2023-018

Submission date: 29.01.2023

Decision date: 25.05.2023

Authors: Anera Kazlagić, Dietmar Stephan, Markus Ostermann, Antonio Possolo, Jochen

Vogl

Crown Copyright Publication Agreement

Text was written in December 2022.

Fingerprinting Portland Cements by means of 87Sr/86Sr and 143Nd/144Nd Isotope

Ratios and Geochemical Profiles

Anera Kazlagić (1)*, Stephan Dietmar (2), Markus Ostermann (3), Antonio Possolo (4), Jochen Vogl (1)*

(1) Federal Institute for Materials Research and Testing, Division 1.1 Inorganic Trace Analysis, Richard-

Willstäter-Straße 11, 12489 Berlin, Germany.

(2) Department of Civil Engineering, Building Materials and Construction Chemistry, Technische

Universität Berlin, Gustav‑Meyer‑Allee 25, 13355 Berlin, Germany

(3) Federal Institute for Materials Research and Testing, Division 1.4 Process Analytical Technology,

Richard-Willstäter-Straße 11, 12489 Berlin, Germany

(4)National Institute of Standards & Technology, 100 Bureau Drive, Gaithersburg, MD 20899, USA

*Corresponding author: anera.kazlagic@bam.de

ORCID ID: Kazlagic, A. https://orcid.org/0000-0002-4428-3589; Stephan, D. https://orcid.org/0000-

0002-1893-6785; Ostermann, M. https://orcid.org/0000-0001-8116-5808; Possolo, A.

https://orcid.org/0000-0002-8691-4190; Vogl, J. https://orcid.org/0000-0002-0104-748X

Keywords

Portland cement, Statistical analysis, Geochemistry, Sr and Nd isotope analysis, elemental fingerprints

Abstract

This study uses conventional 87Sr/86Sr and 143Nd/144Nd isotope and interelement ratios of Ca, Sr, K, Mn,

Mg, and Ti as fingerprints for ordinary Portland cements (OPC) provenancing. Herein, the first database

of Sr and Nd isotope ratios investigated in OPCs stemming from twenty-nine cement plants located

worldwide was created. The results show that the Sr isotope ratios of OPCs are higher than those of

seawater from the observed geological period. The spread of 143Nd/144Nd in OPCs is not as large as the

spread for 87Sr/86Sr isotope ratios. However, the combination of both Sr and Nd isotope ratios provides

the potential for distinguishing between cements of different production sites. Most of the

investigated OPCs have measurable differences in their 87Sr/86Sr and 143Nd/144Nd isotope ratios, which

can be employed as a valuable analytical fingerprinting tool. In the case of equivocal results, Divisive

Hierarchical Clustering was employed to help overcome this issue. The construction of geochemical

profiles allowed us to compute suitably defined distances between cements and cluster them

according to their chemical similarity. By applying this methodology, successful fingerprinting was

achieved in 27 out of 29 analysed ordinary Portland cements.

1. Introduction1

Being globally and widely used, cement is considered an indispensable material in our built 2

environment. Due to the high cost of road transport by truck, cement is mainly delivered to local 3

markets. Production facilities of the German cement industry are spread over Germany according to 4

the appropriate mineral resources and located in the immediate vicinity of the respective limestone 5

deposits (VDZ, 2022a). In 2021, 21 companies produced about 35.0 million tons of cement in Germany 6

(VDZ, 2022b). At present, ordinary Portland cement (OPC) is still an important cement type, suitable 7

for most purposes of concrete construction (Telschow et al., 2012). OPC is produced by grinding 8

Portland cement clinker with anhydrite, bassanite, or gypsum. The Portland cement clinker is 9

manufactured in a rotary kiln at approx. 1450 °C, by sintering the raw materials and consists primarily 10

of alite (50–70 wt%), belite (15–30 wt%), aluminate (5–10 wt%), and ferrite (5–15 wt%), which all, in 11

addition to the main oxides, also contain foreign oxides (Taylor, 2009). Starting materials for Portland 12

cement production (CEM I, as described in European Standard EN 197–1) include limestone, clay, or a 13

naturally occurring mixture of the two, lime marl. Despite high quality standards of the building 14

materials employed, including cement as one of the essential components, buildings (Proske and 15

Schmid, 2021), bridges (Proske, 2019) and other concrete structures have to be subjected to intensive 16

testing again and again for various safety reasons, especially when they show defects or even fail. 17

Therefore, the determination of cement provenance has found its main purpose in solving liability 18

issues when damage occurs to concrete-made structures (Kazlagić et al., 2021) and for tracing the 19

origin of concrete as a waste product for further deposition of material. In these cases, interested 20

parties for cement provenancing include but are not restricted to, testing institutes, whose task is to 21

determine whether two cement samples are identical or are of the same origin (Schmidt-Döhl et al., 22

2005). Furthermore, finding the origin and comparing two cement samples can be significant in 23

forensic investigations. If cement dust is found at a crime scene, it can serve as evidence to link a 24

suspect(s) and/or a victim(s) or to determine the location of an incident (Kazlagić et al., 2021). Since 25

locally available raw materials (limestone and silica, or a natural mixture of both) form the basis of 26

cement production, this is an important property that can be used for linking cement with the 27

geographical origin, also known as ´provenancing´. 28

Over the last decades, numerous trials to fingerprint cement have been attempted. Most were based 29

on concentration analysis of the main or minor elements/components (Kazlagić et al., 2021). The use 30

of isotope ratios in cement provenancing is, however, scarce and is limited to Graham and co-authors 31

(Graham et al., 2000) and Kosednar-Legenstein (Kosednar-Legenstein, 2007) and his work with co-32

authors (Kosednar-Legenstein et al., 2007, Kosednar-Legenstein et al., 2006), who showed that Sr 33

isotopes, combined with elemental analysis, are useful tools for this purpose in a restricted area 34

concerning time and space. To expand this approach to a larger region, such as Europe or even 35

worldwide, a second isotope system is required to increase the resolution of the provenancing 36

approach (Kazlagić et al., 2021). Thus, Nd being another radiogenic isotope system is necessary in 37

addition to Sr. The isotope ratios of these elements remain unaltered in the high-temperature 38

processes (Gan et al., 2016) when conventional methods employ so-called internal normalisation 39

(Kosednar-Legenstein et al., 2006). Both elements, Sr and Nd, have their main occurrence in different 40

components of the raw mix of the cement, Sr being present in the limestone and Nd in the silicates. 41

Thus, they act as tracers for different raw materials. By measuring both Sr and Nd conventional isotope 42

ratios, a distinguishing isotopic fingerprint of the cement can be established (Kazlagić et al., 2021). 43

Interpretation of geochemical data based primarily on elemental concentrations often leads to 44

ambiguous results due to multiple potential influences, including mineral weathering, atmospheric 45

input, biological cycling, mineral precipitation, and exchange processes. The 87Sr/86Sr and 143Nd/144Nd 46

conventional ratios, hereafter referred as Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd) are, however, not 47

fractionated by these processes (Salifu et al., 2018, Hoogewerff et al., 2019), which gives these isotope 48

systems a great advantage over the other systems. To. To understand why Sr and Nd isotope systems 49

are selected for cement provenancing, it is important to explain why they act as tracers for different 50

raw materials. Factors that control the Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd) of geologic materials include 51

the age of the material and the ratio of parent to daughter elements (Rb/Sr, Sm/Nd), whereby higher 52

parent-to-daughter ratios lead to a higher abundance of radiogenic Sr and Nd isotopes. 53

The alkaline earth metal Sr is present in nature with four stable, naturally occurring isotopes: 88Sr, 87Sr, 54

86Sr and 84Sr, whereby the isotope composition varies according to the β-decay of 87Rb, which decays 55

into stable 87Sr with a half-life decay time of 4.7 · 1010 years. The Rb-Sr combination generates different 56

87Sr/86Sr, reflecting only the variations in the amount of radiogenic 87Sr present in the sample, 57

depending on the geological ages and the geochemical characteristics of the rock where they naturally 58

occur, namely their geographical origin (Bullen and Kendall, 1998). If Rb and Sr are integrated into a 59

mineral or rock at its formation, the amount of 87Sr rises over time as radioactive 87Rb decays to 87Sr. 60

The amounts of 84Sr, 86Sr, and 88Sr remain constant. Thus, older rocks have higher 87Sr/86Sr ratios than 61

younger ones with the same initial Rb/Sr ratio. Over geologic time, rocks of a given age composed of 62

minerals with a high Rb/Sr (e.g., granites) in the continental crust will develop a higher 87Sr/86Sr than 63

rocks with a lower Rb/Sr (e.g., oceanic basalt). Therefore, 87Sr/86Sr ratios in geologic materials indicate 64

age and geochemical origin (Capo et al., 1998). 65

Sr isotopes have been used to trace the provenance of numerous organic and inorganic modern and 66

archaeological materials (Vanhaecke and Degryse, 2012). The principle is based on the fact that the 67

87Sr/86Sr ratio of the investigated product is identical, within analytical uncertainties, with that of the 68

geological raw material or bedrock from which it was derived (Brill and Wampler, 1967). The 87Sr/86Sr 69

values in natural rocks differ from older granites, with 87Sr/86Sr ratios typically above 0.710 mol·mol-1 70

and as high as 0.740 mol·mol-1, to younger basalts, with lower 87Sr/86Sr ratios around 0.703 to 0.704 71

mol·mol-1. These differences, which occur in the percent range down to the sub per mil range, are 72

easily detected by highly precise analytical instruments. Thus, the possibility to discriminate different 73

geographical origins within small ranges of variation of the 87Sr/86Sr isotope ratios, requires high 74

precision measurement that can be achieved by using multi-collector thermal ionisation mass 75

spectrometry (MC-TIMS) and/or multi-collector inductively coupled plasma mass spectrometry (MC-76

ICPMS). 77

Phanerozoic (post-Cambrian) seawater 87Sr/86Sr has varied between the limits of crustal and mantle 78

values, reflecting the combined inputs from these two sources to the world’s oceans. Since the 79

residence time of Sr in the oceans today (≈ 106 years) is far longer than the time it takes currents to 80

mix the oceans (≈ 103 years), the world´s oceans are homogeneous with respect to 87Sr/86Sr ratios 81

(McArthur et al., 2012). This fact was used to construct the Sr seawater curve, which covers the 82

Phanerozoic eon and is constructed from analyses of marine carbonates and evaporites (Banner, 83

2004). The negligible Rb content of most marine carbonates means that they have not undergone 87Sr 84

enrichment, and their Sr isotope ratios reflect the conditions under which they were formed. This has 85

provided a means of determining the changes in the strontium isotope ratios over geological time, and 86

for global tectonic processes, mountain building and erosion activities (Dungworth, 2009). 87

Due to their large relative atomic mass, Sr isotopes in the first approximation retain the same 87Sr/86Sr 88

ratio as they pass through geochemical pathways. During ionisation and/or transmission through the 89

mass spectrometer, whilst measuring Sr isotope ratios in MC-TIMS, they may fractionate relative to 90

their value as a function of their mass. In the case of mass-dependent fractionation of Sr, it would be 91

corrected upon mass spectrometry measurement since Sr ratios are normalised to the constant value 92

of 86Sr/88Sr in natural rocks (Beard and Johnson, 2000). Thus, before being reported, measured 87Sr/86Sr 93

isotope ratios are normalised to a standard value of 0.1194 mol·mol-1 for 86Sr/88Sr (Steiger and Jäger, 94

1977). This procedure corrects for the large isotopic fractionations between 84Sr, 86Sr, 87Sr and 88Sr that 95

occur during mass-spectrometric measurement and, coincidentally, removes any natural fractionation 96

(McArthur et al., 2006). Additionally, the remaining bias caused, e.g., by differences in the detector 97

efficiencies, can be corrected by measuring NIST SRM 987 (Sr carbonate isotopic standard, NIST, 98

Maryland, USA), with accepted 87Sr/86Sr = 0.710 250 mol·mol-1 (convention value (Faure and Mensing, 99

2013, McArthur et al., 2006)). 100

In contrast, Nd isotope ratios (143Nd/144Nd) generally negatively correlate with Sr isotope ratios, 101

primarily as the ultramafic rocks within the mantle have higher Sm/Nd ratios than rocks within the 102

continental crust (Nakano, 2016). Neodymium, an element with seven natural isotopes, has been used 103

as an essential geochronometer and a key geochemical tracer for unraveling the compositional 104

evolution of terrestrial planets. 147Sm decays to 143Nd via α-decay with a half-life of 1.06 · 1011 year. 105

Furthermore, 146Sm decays to 142Nd via α-decay and a half-life estimated between 68 · 106 years 106

(Kinoshita et al., 2012) and 103 · 106 years (Meissner et al., 1987). Since Sm has a slightly lower ionic 107

radius, it is generally more compatible than Nd in the mantle of the Earth. These decays through time 108

have produced variations in Nd isotope ratios in terrestrial and extra-terrestrial materials. Since Sm is 109

more compatible than Nd, during processes such as melting and crystallisation, there are, over time, 110

significant differences in the Sm/Nd ratio and Nd isotopic composition of rocks and minerals on and 111

within the Earth (Bizimis and Scher, 2018). 112

Analogous to Sr, isotope fractionation correction for Nd isotope ratios must be applied. This 113

instrument-induced isotope fractionation is described by an exponential or power mass law 114

(Vanhaecke and Degryse, 2012) and can be corrected if one stable ratio is used. The ratio 146Nd/144Nd 115

= 0.7219 mol·mol-1 is conventionally used for this fractionation correction (Hamilton et al., 1983, 116

Whitehouse, 1996). Additionally, instrumental bias can be corrected by measuring a pure Nd standard, 117

such as JNdi-1 with accepted 143Nd/144Nd = 0.512 106 mol·mol-1, being the median of approx. 600 118

reported results in the GeoRem database (Jochum et al., 2005). 119

Variations and contrasts in the geological ages of the raw materials used to produce OPCs are reflected 120

in Sr and Nd isotope ratios. From a geological point of view, around 90 % of the quarried limestone in 121

Germany comes from the Mesozoic era and is thus between 65 and 250 million years old ((Ed.), 2018). 122

Whilst Sr isotopic composition is reflected in the limestone, Nd is geologically linked to silica and silicate 123

minerals used to produce cement. Moreover, 143Nd/144Nd ratios in cement reflect the formation age 124

of the silica and the concentration of samarium, the parent element. This enables discrimination 125

between sand derived from the erosion of old, metamorphic rock and sand derived from the erosion 126

of a younger rock, e.g., granite. 127

However, in this study, even though very important, the focus was not on finding deeper insight into 128

the geological background of raw materials (limestone and silicates, or a natural mixture of both) but 129

rather on developing an approach that could use variations of Sr and Nd isotope ratios in OPCs, as a 130

useful analytical tool for cement fingerprinting. In addition to Sr and Nd isotopic systems serving as 131

tracers for the geographical origin of cement, we used carefully selected elements and their mass 132

fractions to form elemental ratios. Using these ratios, together with Sr and Nd isotope ratios, distances 133

between cement samples were computed. The next step was employing Divisive Analysis (DIANA) 134

(Kaufman and Rousseeuw, 1990) to obtain a hierarchical clustering, which increased the analytical 135

resolution of the cement fingerprinting approach. Therefore, this study presents two approaches to 136

OPC provenancing: Sr and Nd isotopic fingerprinting and DIANA hierarchical clustering. For this 137

purpose, twenty-nine OPC samples from worldwide cement plants were used. MC-TIMS was used to 138

determine isotope ratios in OPCs, and X-Ray Fluorescence (XRF) was used to determine elemental mass 139

fractions. 140

2. Materials and Methods141

2.1. Ordinary Portland Cements 142

Twenty-nine cements (all ordinary Portland cement, CEM I according to DIN EN 197-1) were obtained 143

from the respective cement producers. The sample units, mostly weighing ≈ 0.5–1 kg, were divided 144

into smaller portions using standard procedures and stored in PP beakers. The sample mass for 145

87Sr/86Sr isotope ratio analysis was ≈ 100 mg, whereas for 143Nd/144Nd it was ≈ 200 mg. 146

Fifteen cements originated from Germany, three from Iraq, two from Serbia, two from Greece, two 147

from Austria, one from China, one from Italy, one from Bosnia and Herzegovina, one from Kosovo, and 148

one from North Macedonia. The rationale for investigating cements from different countries was to 149

include materials with a wide range of isotope ratios of the raw materials and different CEM I 150

production plants. OPC samples used in this study are presented in Table 1. The first two letters 151

represent the geological period of limestone, the raw material for cement production, meaning TE, CR, 152

JU, TR, DM, XX = Tertiary, Cretaceous, Jurassic, Triassic, Devonian and unknown, respectively. The two 153

letters after the hyphen represent the country of origin: AU, BH, CH, DE, GR, IQ, IT, KO, MC, SR = Austria, 154

Bosnia and Herzegovina, China, Germany, Greece, Iraq, Italy, Kosovo, North Macedonia, and Serbia. 155

respectively. The next number represents the sample number originating from the same country. Only 156

OPC samples from Germany, except for sample XX-DE05, have a clearly stated geological period. 157

158

Table 1. Ordinary Portland cement samples with respective ID 159

Sample

name

Sample

name

XX-IQ01

CR-DE11

CR-DE01

TR-DE12

XX-BH01

XX-IQ03

TR-DE02

TR-DE13

JU-DE03

XX-GR01

CR-DE04

TR-DE14

XX-DE05

XX-GR02

DM-DE06

TR-DE15

JU-DE07

XX-KO01

CR-DE08

XX-MC01

JU-DE09

XX-AU01

XX-SR01

XX-CH01

XX-SR02

XX-AU02

XX-IQ02

XX-IT01

TE-DE10

160

2.2. Chemicals 161

HNO3 (65% v/v) and HCl (30% v/v) were purchased as pro-analysis grade acids (Chemsolute®, Th. 162

Geyer, Berlin, DE) and were further purified by double sub-boiling distillation. HBF4 acid was purchased 163

as an ultra-pure (38% v/v) solution (Chemlab®, Zedelgem, BE). For all dilutions, Milli-Q water (Milli-Q 164

Advantage A10 System, Merck KGaA, Darmstadt, DE) was used. For Sr and Nd purification and matrix 165

separation, the following materials were employed: Sr·Spec™ resin (100–150 μm, Eichrom 166

Technologies Inc, Lisle, IL, USA) TRU·Spec™, Ln·Spec™ (100–150 μm, Eichrom Environment Bruz, FR). 167

Sample preparation for isotope analysis was performed in a metal-free clean laboratory with ISO class 168

6 at the Federal Institute for Materials Research and Testing (BAM). Crushing, grinding of the samples, 169

and sample preparation for X-Ray Fluorescence analysis was performed in a normal laboratory 170

environment. 171

172

2.3. Sample preparation for Sr analysis 173

For Sr isotope analysis, approximately 100 mg of each cement sample was weighed in a PFA beaker 174

and prepared by dissolving in the concentrated sub-boiled hydrochloric and nitric acids, a procedure 175

described in our previous study (Kazlagić et al., 2022). After determination of the Sr mass fraction (by 176

iCAP-Q ICP-MS, Thermo Fisher Scientific, Bremen, DE), an aliquot of the sample was transferred to a 177

new PFA beaker and dried on a hotplate. The next step was redissolving the sample in 2 mol·L−1 HNO3, 178

evaporating to dryness and redissolving using 1 mL of the 2 mol·L−1 HNO3. This solution contained 179

approx. 2 μg of Sr and was taken to perform a Sr matrix separation using the water suspension of 180

Sr·Spec™ resin (350 μL) in polyvinyl chloride columns with the following dimensions: 6 mm inner 181

diameter and 40 mm length. The procedure is shown as a scheme in Fig. S1, in the ESM. The resulting 182

Sr fraction was evaporated to dryness and redissolved in nitric acid such that a final Sr mass fraction of 183

100 ng·μL−1 was obtained, which could be directly used for loading 1 μL of the sample on previously 184

degassed single Re filaments, together with TaF5 as the enhancing ionisation activator. 185

186

2.4. Sample preparation for Nd analysis 187

Since Nd content in cement samples was 10-100 times lower than Sr content, approximately 200 mg 188

of each cement sample was weighted in PFA microwave beakers for Nd isotope analysis. The acid 189

mixture contained 5 mL of conc. sub-boiled HNO3, 2 mL of conc. sub-boiled HCl and 1 mL conc. HBF4, 190

which were added to the samples and after slight shaking, a MW oven (MLS Mikrowellen-Labor-191

Systeme GmbH, DE) was used to digest the samples. After digesting, samples were quantitatively 192

transferred to Falcon tubes by washing them with 2 mL of Milli-Q water. An aliquot of the sample 193

(2 mL - 4 mL) was transferred to a PFA beaker and dried on a hotplate. This solution contained approx. 194

1 µg to 2 μg of Nd and was taken to perform a Nd matrix separation using the Sr, TRU and Ln spec, as 195

described in (Mikova and Denkova, 2007), with a slight modification (see Fig. S2, ESM), using PFA 196

columns for Ln spec. The resulting Nd fraction was evaporated to dryness and redissolved in nitric acid 197

such that a final Nd mass fraction of 300 ng·μL−1 was obtained, which could be directly used for loading 198

1 μL of the sample on the previously degassed inner Re filaments, together with H3PO4 as an activator. 199

After final drying, the filament was heated up slowly until glowing dull red for about three seconds. 200

201

2.5. Sample preparation for elemental analysis (glass beads) 202

For elemental analysis, boron glass beads were prepared. The borate fusion technique involves mixing 203

a ground sample with a borate flux inside a 95 % Pt – 5 % Au crucible, heating it to 1200 °C with 204

agitation until the flux melts, and then dissolving the sample homogeneously in the flux. 205

As the first sample preparation step for borate fusion, 1.00 g of powdered cement samples are mixed 206

with 8.00 g of borate flux (consisting of Li-tetraborate: Li-metaborate in a 66:34 weight ratio) using 207

Fluxana Boramat® autosampler. Crucibles with cement samples are placed inside the automatic fusion 208

digestion unit (Fluxana Vitriox® Electic). At 1200 °C, the flux mixture dissolves oxidised solid cement 209

particles. The molten mixture is then poured into a mould and cooled to produce so-called “glass 210

beads” for XRF analysis. 211

212

3. Analytical techniques and clustering procedures213

3.1. Conventional 87Sr/86Sr and 143Nd/144Nd isotope analysis using MC-TIMS 214

Strontium isotope analyses were carried out by MC-TIMS at BAM in Berlin using a Sector 54 instrument 215

(Micromass Ltd., Manchester, UK) in a dynamic multi-collection mode via an automatic measurement 216

procedure. The raw measured data were corrected for interfering Rb and instrumental isotope 217

fractionation (86Sr/88Sr = 0.1194 mol·mol-1(Steiger and Jäger, 1977)) and finally were normalised to a 218

NIST SRM 987 87Sr/86Sr ratio of 0.710 250 mol·mol-1 (Faure and Mensing, 2013), which is also the 219

median of more than one thousand published results listed in the GeoReM database (Jochum et al., 220

2005). During measurements, a stable 88Sr signal of ≈ 3 V was obtained for 70-80 minutes, 221

corresponding to 150-200 individual signal scans. The measured procedural blank for Sr was <0.5 ng. 222

Furthermore, AGV2-a and NASS-6 seawater reference materials were used as control samples. The 223

total measurement time, including filament warm-up and actual measurement time was approx. 90 224

min per sample. For more details, see Kazlagić and co-authors (Kazlagić et al., 2022). 225

Neodymium isotope analyses were carried out on a Triton type MC-TIMS at FU in Berlin (Finnigan, 226

Thermo Scientific) in a static collection mode via a manual measurement procedure. The raw measured 227

data were corrected for interfering Sm and isotope fractionation (146Nd/144Nd = 0.7219 mol·mol-1, the 228

chondritic uniform reservoir ´CHUR´ value (Whitehouse, 1996, Hamilton et al., 1983)) and finally were 229

normalised to a JNd-1 146Nd/144Nd ratio of 0.512 106 mol·mol-1, which is the median of approx. 600 230

published results listed in the GeoReM database (Jochum et al., 2005). During the measurements, the 231

typical signal intensity was ≈3 V on 144Nd. AGV2-a and GSP-2 reference materials were used as control 232

samples. The measurement time was approx. 120 min per sample. 233

Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd) are metrologically traceable to the internationally agreed 234

convention method. 235

236

3.2. Elemental concentration analysis using X-Ray-Fluorescence 237

To determine the elemental mass fractions of fourteen elements in the OPC samples, previously 238

prepared and cooled cement glass beads were used for X-Ray Fluorescence (XRF) measurements. The 239

OPC samples were tested using two wavelengths dispersive XRF analysers manufactured by Panalytical 240

B.V. (MagiXPro and Zetium Ultimate). Both instruments are equipped with a 4 kW Rhodium X-Ray tube, 241

3 different collimator settings, 8 crystals and 4 different detectors. The instruments are calibrated 242

using a reference material set from Fluxana (Fluxana CEM). In this set, the reference material samples 243

are available as fused glasses. The measurement spot for the fused glass beads was 37 mm in diameter. 244

The measurement time was approx. 10 minutes for one sample. The fourteen analysed elements were 245

Al, Ca, Cr, Fe, K, Mg, Mn, Na, P, S, Si, Sr, Ti and Zn. However, only Ca, K, Mn, Mg, Sr, and Ti were selected 246

for cement clustering. This is because other measured elements either have no relation to the 247

geographic origin or are related to additives (such as fuel used in production, e.g., used tires, heavy 248

fuel oil, etc.), whose origin is not interlinked with the origin of the raw materials (limestone, clay or 249

natural mix of both) used for cement production. 250

Metrological traceability of the elemental mass fractions is to the International System of Units (SI). 251

3.3. Clustering procedures 252

Chemical composition can be used to quantify similarities between cements. In this contribution, we 253

use conventional isotope ratios for strontium and neodymium, and ratios of mass fractions between 254

selected elements to construct geochemical profiles that allow us to compute suitably defined 255

distances between cements, and then to cluster them according to their chemical similarity. The goal 256

is to refine the clustering based only on 87Sr/86Sr and 143Nd/144Nd. We have chosen ratios involving mass 257

fractions of Ca, K, Mn, Mg, Sr, and Ti, as well as 87Sr/86Sr and 143Nd/144Nd, to define the geochemical 258

profiles. 259

The first clustering method uses Euclidean distances between cement samples represented by points 260

in 5-dimensional space, whose coordinates are the ratios of mass fractions Ca/Sr, K/Mn, and Mg/Ti, 261

which were determined using XRF, and the isotope ratios 87Sr/86Sr and 146Nd/144Nd, with each ratio 262

standardised to have mean 0 and standard deviation 1. 263

The clustering technique used is Divisive Analysis (DIANA), as described in (Kaufman and Rousseeuw, 264

1990), which produces a hierarchical clustering. The top cluster comprises all n=29 cement samples. 265

Subsequently, each cluster is split into two clusters iteratively until each resulting cluster has the 266

smallest average dissimilarity. The process is repeated for a total of n-1 steps, and the results are 267

represented using a tree whose leaves are individual cement samples. 268

The second clustering method considers all 120 possible sets of ratios of two mass fractions that can 269

be formed using the mass fractions of Ca, K, Mn, Mg, Sr, and Ti, as well as 87Sr/86Sr and 143Nd/144Nd. 270

The set of ratios considered in the first method is only one of these 120 possible sets of ratios. In the 271

second clustering method, not only do we consider Ca/Sr, K/Mn, Mg/Ti, plus 87Sr/86Sr and 143Nd/144Nd, 272

but also Ca/Mg, Mn/Ti, K/Sr, plus 87Sr/86Sr and 143Nd/144Nd, as well as all the other possible 273

combinations of numerators and denominators of the fractions involving elemental mass fractions. 274

The idea here is to exploit the information conveyed by the mass fractions of Ca, K, Mn, Mg, Sr, and Ti 275

comprehensively without committing to a specific single set of ratios as in the first clustering method. 276

For this second clustering method, distances between cement samples are computed differently from 277

how they were computed for the first clustering method. However, once we have the corresponding 278

distance matrix, we will employ DIANA to obtain the hierarchical clustering corresponding to the new 279

set of distances. 280

The distance between cement sample i = 1,…,29 and cement sample j ≠ i that is used in the second 281

clustering method is computed by taking the following steps for each set s = 1,…,120 of different ratios 282

that can be formed using the six mass fractions aforementioned: 283

1) Compute the geochemical profile pi,s for sample i, and the geochemical profile pj,s for sample284

j, which correspond to set s of ratios of elemental mass fractions plus the two isotope ratios;285

2) Compute the Euclidean distance Ds(i,j) between pi,s and pj,s

286

3) Define D(i,j) as the median of D1(i,j), …, D120(i,j): that is, D(i,j) is the median of the distances287

between cement samples i and j over all 120 different geochemical profiles.288

These distances D(i,j) are symmetric (that is, D(i,j)=D(j,i)), but they do not satisfy the triangle inequality, 289

which requires that D(i,k) ≤ D(i,j) + D(j,k) for every triplet of cement samples i, j, and k. For this 290

particular collection of cement samples, the triangle inequality is satisfied for 96 % of the triplets (i,j,k). 291

In these circumstances, D is only a semi-metric, not a proper distance. Since the violation of the triangle 292

inequality is not very pervasive, D can be used as an “almost” proper distance metric and will not 293

preclude our using it as a dissimilarity matrix for clustering purposes. 294

3.4. Comparability of the data 295

Since no measurement is perfect, evaluating measurement uncertainty and factoring it in are 296

prerequisites for interpreting and using measurement results (Possolo and Meija, 2020). The 297

uncertainty evaluations were performed consistently with the Guide to the Expression of Uncertainty 298

in Measurement (GUM) ((JCGM), 2008). The following five uncertainty components were included in 299

the uncertainty for Sr isotope ratios calculation (eqn. 1): repeatability of a single 87Sr/86Sr measurement 300

in a sample, repeatability of 87Sr/86Sr measurement in NIST SRM 987, uncertainty of the reference value 301

for 87Sr/86Sr measurement in NIST SRM 987 (value published in GeoReM database) (Jochum et al., 302

2005), experimental reproducibility of independently processed homogeneous samples, and the bias 303

of the measured 87Sr/86Sr ratio in a processed control sample versus the reference value (value 304

published in GeoReM database) (Jochum et al., 2005). 305

306

eqn. 1 307

308

Similarly to Sr uncertainty calculation, the uncertainty for Nd ratios was calculated as follows: 309

repeatability of a single 143Nd/144Nd measurement in a sample, repeatability of 143Nd/144Nd 310

measurement in JNdi-1, uncertainty of the reference value for 143Nd/144Nd measurement in JNdi-1 311

(value published in GeoReM database) (Jochum et al., 2005), experimental reproducibility of 312

independently processed homogeneous samples, and the bias of the measured 143Nd/144Nd ratio in 313

the processed control sample (here, AGV-2a and GSP2) versus the reference value (value published in 314

GeoReM database) (Jochum et al., 2005). 315

316

𝑢𝑐,𝑓𝑖𝑛𝑎𝑙

𝑅(𝑟𝑎𝑡𝑖𝑜)𝑓𝑖𝑛𝑎𝑙=√(𝑢𝑐,1

𝑣𝑎𝑙𝑢𝑒1)2+( 𝑢𝑐,2

𝑣𝑎𝑙𝑢𝑒2)2+( 𝑢𝑐,3

𝑣𝑎𝑙𝑢𝑒3)2+( 𝑢𝑐,4

𝑣𝑎𝑙𝑢𝑒4)2+( 𝑢𝑐,5

𝑣𝑎𝑙𝑢𝑒5)2

4. Results and Discussion317

4.1. Results of conventional 87Sr/86Sr and 143Nd/144Nd isotope ratios in OPCs (Combined Approach) 318

The results of conventional 87Sr/86Sr and 143Nd/144Nd measured in twenty-nine OPCs are shown in Table 319

2. 320

Table 2. Results of conventional 87Sr/86Sr and 143Nd/144Nd in OPCs with expanded uncertainty being U 321

= 0.000 015 for Nd and 0.000 022 for Sr isotope ratios, k = 2. 322

Sample

name

Rcon(87Sr/86Sr) /mol·mol-1

U (k=2)

Rcon(143Nd/144Nd)/mol·mol-1

U (k=2)

XX-IQ01

0.708 177

0.000 022

0.512 396

0.000 015

CR-DE01

0.707 864

0.000 022

0.512 006

0.000 015

XX-BH01

0.708 482

0.000 022

0.512 248

0.000 015

TR-DE02

0.708 241

0.000 022

0.512 055

0.000 015

JU-DE03

0.708 594

0.000 022

0.512 112

0.000 015

CR-DE04

0.708 032

0.000 022

0.512 032

0.000 015

XX-DE05

0.708 081

0.000 022

0.512 154

0.000 015

DM-DE06

0.708 809

0.000 022

0.512 004

0.000 015

JU-DE07

0.707 953

0.000 022

0.512 099

0.000 015

CR-DE08

0.707 896

0.000 022

0.512 033

0.000 015

JU-DE09

0.708 122

0.000 022

0.512 103

0.000 015

XX-SR01

0.709 168

0.000 022

0.512 196

0.000 015

XX-SR02

0.708 209

0.000 022

0.512 148

0.000 015

XX-IQ02

0.708 214

0.000 022

0.512 408

0.000 015

TE-DE10

0.710 169

0.000 022

0.512 102

0.000 015

CR-DE11

0.707 718

0.000 022

0.512 038

0.000 015

TR-DE12

0.708 280

0.000 022

0.512 050

0.000 015

XX-IQ03

0.708 039

0.000 022

0.512 372

0.000 015

TR-DE13

0.708 052

0.000 022

0.512 059

0.000 015

XX-GR01

0.707 933

0.000 022

0.512 201

0.000 015

TR-DE14

0.708 245

0.000 022

0.512 047

0.000 015

XX-GR02

0.708 613

0.000 022

0.512 161

0.000 015

TR-DE15

0.708 970

0.000 022

0.512 112

0.000 015

XX-KO01

0.708 809

0.000 022

0.512 210

0.000 015

XX-MC01

0.709 481

0.000 022

0.512 240

0.000 015

XX-AU01

0.708 946

0.000 022

0.512 134

0.000 015

XX-CH01

0.710 464

0.000 022

0.511 857

0.000 015

XX-AU02

0.707 954

0.000 022

0.512 145

0.000 015

XX-IT01

0.707 689

0.000 022

0.512 119

0.000 015

323

The results for quality control materials that went through the full chemical procedures, including 324

sample preparation and analyte separation for Nd isotopes, are: AGV-2a = (0.512 779 ± 0.000 010) 325

mol·mol-1 and GSP-2 = (0.511 357 ± 0.000 013) mol·mol-1, and for Sr the results are: NASS-6 = (0.709 326

172 ± 0.000 027) mol·mol-1 and AGV-2a = (0.703 988 ± 0.000 025) mol·mol-1, all being in a good 327

agreement with the reference values from the GeoReM (Jochum et al., 2005) database and IAPSO 328

North Atlantic seawater value. The reference values for Nd are: AGV-2aGeoReM = (0.512 789 ± 0.000 329

011) mol·mol-1, GSP-2GeoReM = (0.511 368 ± 0.000 008) mol·mol-1, and for Sr: AGV-2aGeoReM = (0.703 987330

± 0.000 016) mol·mol-1, NASS-6IAPSO = (0.709 178 ± 0.000 007), where expanded measurement 331

uncertainties were obtained as MAD. 332

When we observe Sr isotope ratios in OPCs from the geologic time scale perspective, the following is 333

to be considered: the Cenozoic era covers the Tertiary period, the Mesozoic era covers the Triassic, 334

Jurassic and Cretaceous periods, and the Palaeozoic era covers the Devonian. The Tertiary period (TE) 335

dates from 66-2.6 Ma; the Cretaceous (CR) period dates from 145-66 Ma; the Jurassic period (JU) 336

covers 201-145 Ma; the Triassic (TR) covers 252-201 Ma; the Devonian (DM) covers 419-359 Ma (Geyh 337

and Schleicher, 1990). 338

Most of the samples from Germany have been produced from limestone, which dates to the Mesozoic 339

era (Jurassic, Cretaceous and Triassic), which is in agreement with the statement published by VDZ 340

((Ed.), 2018). In the investigated OPC sample set, only one sample presents Tertiary, and another one 341

Devonian origin. The isotope results plotted in Fig. 1 show that most OPCs are distinctive and can be 342

separated based on their geographical origin using Sr and Nd isotopic fingerprints. Using highly 343

sophisticated techniques such as MC-TIMS yields results with very low measurement uncertainties 344

(expanded uncertainties U = 0.000 015 for Nd and 0.000 022 for Sr isotope ratios, k = 2), which allows 345

the differentiation between different OPCs. Generally, as expected, 87Sr/86Sr ratios in OPCs show 48 346

times larger (range 0.7075-0.7105 mol·mol-1, the difference between lowest and highest measured 347

ratio being 0.029) spread in comparison to 143Nd/144Nd ratios (range 0.5118 – 0.5124 mol·mol-1, the 348

difference between measured lowest and highest ratio is 0.0006). 349

As depicted in Fig.1, samples originating from the same geological period are closer to each other, 350

meaning they have more similar 143Nd/144Nd isotope ratios (OPCs from the Jurassic period), and/or 351

87Sr/86Sr isotope ratios (Fig. 1, OPCs from Triassic). Samples depicted in purple (Fig. 1) date to the 352

Cretaceous period. They show higher 87Sr/86Sr isotope ratios than modelled 87Sr/86Sr values of the early 353

Cretaceous period, reflecting early Cretaceous seawater (McArthur et al., 2012) at the time of 354

limestone formation (Fig.3). 355

One sample from Germany of Tertiary origin (marked in red) presents a vastly different isotope 356

fingerprint than all the other OPCs, characterised by a high ratio of 87Sr/86Sr (0.71017 mol·mol-1). This 357

OPC, in particular, has 87Sr/86Sr isotope ratio much higher than Cenozoic limestone and seawater from 358

that era (Fig.3) (McArthur et al., 2012), most likely due to the presence of highly radiogenic clay 359

minerals in the bulk rock that incorporates Rb in concentrations of tens to hundreds of mg·kg-1. The 360

OPC sample originating from the Devonian period (DM-DE06; 0.708 809 mol·mol-1) presents not only 361

a different isotopic fingerprint than that of the other OPCs but also a higher value than those typical 362

of the Devonian period, which usually have a ratio in the range of 0.7080-0.7086 mol·mol-1 (McArthur 363

et al., 2012). However, the study reported by Wedepohl and Baumann reveals that 87Sr/86Sr isotope 364

ratios of six glasses (bottles and vessels) produced in the Eifel area of Germany in the second half of 365

the fourth century (Devonian) had mean value 0.70889 (Wedepohl and Baumann, 2000), which is also 366

higher than expected. The raw material for producing these glasses was marine molluscan shells, which 367

consist mainly of aragonite or a mixture of aragonite and calcite. Since the reported Sr ratios do not 368

agree with the seawater ratio for the Devonian period, Wedepohl and Baumann attribute this 369

difference to a slight diagenetic alteration of the molluscan shells in their unstable aragonite, which in 370

reaction with freshwater or another raw material, contributed to isotopically lighter Sr. 371

Even though one sample from the Triassic (TR-DE13) has Sr isotope ratios in agreement with the 372

seawater curve (Fig. 3), a general trend of higher 87Sr/86Sr isotope ratios of OPCs in comparison to the 373

seawater curve can be observed in Fig. 3. In principle, there are different possibilities for why the Sr 374

isotope ratios of OPCs in this study are higher than Sr isotope ratios derived from the seawater from 375

the observed geological period. One of the explanations is radiogenic strontium formation from 376

rubidium decay in raw material due to the presence of radiogenic clay. Furthermore, as explained by 377

Wedepohl and Baumann, diagenetic alteration could result in these differences. Another possibility is 378

the addition of an unknown additive during cement production, e.g., gypsum, that has a substantially 379

higher 87Sr/86Sr isotope ratio. This could, in turn, affect the ratio of the final product. 380

Nd isotope ratios in OPCs from Germany are in the range of 0.5120 – 0.5121 mol·mol-1. The spread of 381

these results is not insignificant but also not wide; thus, the differentiation between German OPCs 382

relies more on the differences in Sr isotope ratios. Nd isotope fingerprints in cements from countries 383

other than Germany, with China as an exception, range from 0.5121 – 0.5124 mol·mol-1. Therefore, 384

besides having a larger spread, these OPCs exhibit higher 143Nd/144Nd isotope ratios. Three samples 385

from Iraq contain the highest 143Nd/144Nd isotope ratio, indicating that Iraqi cement was made using 386

more radiogenic silica sources than those used in manufacturing European cements. 387

388

Fig. 1. A plot of 87Sr/86Sr vs. 143Nd/144Nd conventional isotope ratios with expanded uncertainties (U, 389

k=2) in ordinary Portland cements from different geographical origins and geological periods. The plot 390

shows that OPCs have different 87Sr/86Sr vs. 143Nd/144Nd signatures. Most of them, whose fingerprints 391

do not overlap, can be distinguished based on their specific Sr and Nd isotope fingerprints. 392

Three samples from the Triassic period show very similar Sr and Nd isotopic fingerprints within the 393

stated measurement uncertainties. This is possibly because these samples originate from cement 394

plants/limestone quarries dating back to the same geological period and most likely from the same 395

limestone or chalk deposit, known as “Muschelkalk”, which is a naturally occurring mixture of 396

carbonates and silicates. 397

Since the information about the geological age/period of the limestones is not available for all samples, 398

it was possible to investigate further in this direction only for samples originating from Germany. If, 399

however, the measured Sr and Nd isotope ratios are plotted in the same way, with the samples´ 400

countries of origin being marked instead of their geological age, a clear distinction can be observed, 401

see Fig. 2: OPCs originating from Germany have quite a large 87Sr/86Sr spread, from 0.7077 – 0.7102 402

mol·mol-1, with most being close to 0.7080 mol·mol-1 (Fig. 2), which is in agreement with the results 403

published in a study by Graham and co-authors (Graham et al., 2000). As expected, however, the 404

143Nd/144Nd spread is not significant (0.5120 – 0.5121 mol·mol-1). 405

Three OPCs originating from Iraq form a group (depicted in Fig. 2), easily distinguishable from the 406

others due to their high 143Nd/144Nd isotope ratios (0.51237-0.51241) mol·mol-1. Unlike any other 407

0.7075 0.7080 0.7085 0.7090 0.7095 0.7100 0.7105

0.5118

0.5119

0.5120

0.5121

0.5122

0.5123

0.5124

143Nd/144Nd

87Sr/86Sr

unknown

Jurassic

Cretaceus

Triassic

Tertiary

Devon

sample, the OPC from Shanghai, China, shows an extremely high 87Sr/86Sr (0.71046 mol·mol-1) and 408

extremely low 143Nd/144Nd isotope ratio (0.51186 mol·mol-1), making it easily distinguishable from 409

other groups and OPCs. The respective pairs of OPCs from Austria, Serbia and Greece did not form 410

groups, most likely due to the differences in intrinsic properties of raw materials from which their 411

cements were produced, and consequently, that affected Sr and Nd isotope ratios. Together with the 412

OPCs from Italy, North Macedonia, Kosovo and Bosnia and Herzegovina, they exhibited generally 413

higher 143Nd/144Nd ratios compared to the samples from Germany, with a relatively similar spread of 414

87Sr/86Sr ratios. Overall, there are measurable differences in the 87Sr/86Sr and 143Nd/144Nd of most of 415

the OPCs analysed, which can be used to fingerprint them. Three investigated OPCs from Germany (TR-416

DE02, TR-DE12 and TR-DE14) and two OPCs from Iraq (XX-IQ01 and XX-IQ02) have non-unique 87Sr/86Sr 417

and 143Nd/144Nd isotope ratios. Since they overlap within the stated uncertainties, they cannot be easily 418

distinguished from one another by applying this method. 419

420

Fig. 2. A plot of 87Sr/86Sr vs. 143Nd/144Nd conventional isotope ratios with expanded uncertainties (U, 421

k=2) in OPC stemming from ten different countries. Namely: Germany, Iraq, Bosnia and Herzegovina, 422

Serbia, Greece, Kosovo, North Macedonia, Austria, China, and Italy. 423

0.7075 0.7080 0.7085 0.7090 0.7095 0.7100 0.7105

0.5118

0.5119

0.5120

0.5121

0.5122

0.5123

0.5124

Greece

143Nd/144Nd

87Sr/86Sr

Iraq

China

Italy

Germany

Austria

Serbia

N. Macedonia

Greece

B&H

Kosovo

424

Fig. 3. 87Sr/86Sr isotope ratios of German OPC (grey rectangles) plotted against the Phanerozoic 425

seawater curve (data obtained from (McArthur et al., 2012)), from which the main constituent 426

limestones derive their Sr isotopic compositions. The information about the age of limestone queries 427

is derived from the official webpage of the Federal Institute for Geosciences and Natural Resources 428

((BGR)). 429

4.2. Geochemical Profiles and DIANA 430

The results of elemental concentrations measured by XRF are presented in Table S1, ESM. 431

In principle, if two samples cannot be distinguished using the combined approach discussed earlier, 432

then the second step would be the application of DIANA. In the first clustering method, the dissimilarity 433

is Euclidean distance between geochemical profiles comprising the ratios of elemental mass fractions 434

Ca/Sr, K/Mn and Mg/Ti plus 87Sr/86Sr and 143Nd/144Nd conventional isotope ratios. 435

The results of the first clustering method are presented in Fig. 4. The lower two samples (on the vertical 436

scale) are joined by a horizontal line segment, the more alike they are. Since the criterion of the cement 437

separation is the chemical similarity of the cements inside a particular group, the observed groups 438

suggest a definitive chemical correlation between the samples. Pairs of samples that are joined by a 439

horizontal segment close to the bottom of the scale on the left axis indicate the pairs of cement types 440

joined together based on their respective elemental concentrations and isotope ratios mentioned 441

earlier, and are thus very similar, such as samples with ID no. 2 & 6, 11 & 13. 442

Samples no. 15 & 27 are less similar than pairs like 2 & 6 or 11 & 13 because they are joined (by a 443

horizontal line segment) at a higher level of dissimilarity (almost 4) than 2 & 6 or 11 & 13, which are 444

joined by horizontal line segments at a level of dissimilarity less than 0.5. 445

The first clustering method considered only the selected ratios of mass fractions Ca/Sr, K/Mn, and 446

Mg/Ti, and the Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd). Thus, the geochemical profiles formed from this 447

clustering method depend strictly on these selected ratios. However, if instead all possible 120 448

combinations of numerators and denominators involving the mass fractions of Ca, K, Mg, Mn, Sr, and 449

Ti are considered as described in subsection 3.3, then the resulting profiles are more discriminating, 450

providing valuable information without committing to one specific set of ratios. Therefore, the second 451

clustering method presents the Divisive Hierarchical Clustering (DIANA) in twenty-nine OPCs 452

investigated in this study, where dissimilarity between each pair of cements is the median Euclidean 453

distance over all geochemical profiles with 120 combinations involving mass fraction ratios of six 454

selected elements, namely Ca, K, Mg, Mn, Sr, Ti and Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd) isotope ratios. 455

456

Fig. 4. Divisive Hierarchical Clustering (DIANA) – Dissimilarity is Euclidean distance between 457

geochemical profiles comprising the ratios of elemental mass fractions Ca/Sr, K/Mn and Mg/Ti plus the 458

Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd). 459

The results of the second clustering method are shown in Fig. 5. By measuring Euclidean distances 460

between cement samples using geochemical profiles based on DIANA, and cutting the tree horizontally 461

at dissimilarity level 25, four separated clusters (A, B, C and D branches) under the DIANA tree are 462

formed. However, the reasons behind forming these four clusters are partially uncertain and intriguing. 463

Nevertheless, we could say that most OPCs from the same geological period are located in the same 464

cluster. E.g., samples with ID no. 2 (CR-DE01), 6 (CR-DE04) and 16 (CR-DE11) (Cretaceous period) are 465

in cluster C. Samples no. 19 (TR-DE13), 4 (TR-DE02) and 21 (TR-DE14) are in cluster B (Triassic period), 466

and samples no. 5 (JU-DE03), 9 (JU-DE07) and 11 (JU-DE09) are in cluster D (Jurassic period). However, 467

it remains unclear why this is not applicable to all samples of known geological background, such as 468

samples with ID no. 10 (CR-DE08 from Cretaceous, but located in cluster A), ID no. 17 (TR-DE12) and 469

ID no. 23 (TR-DE15) (Triassic but located in clusters A and D, respectively). 470

Finally, samples with indistinguishable Sr and Nd isotopic fingerprints, namely samples no. 1 (XX-IQ01) 471

and 14 (XX-IQ02), can now be separated since they belong to different clusters in the DIANA tree. 472

The same case occurs with sample no. 17 (TR-DE12), which now can be separated from samples with 473

ID no. 21 (TR-DE14) and 4 (TR-DE02), and, therefore, distinguished. However, samples no. 21 (TR-DE14) 474

and no. 4 (TR-DE02) belong to the same cluster, so they cannot be distinguished. These three samples 475

are German cements made of limestone from the same geological period (Fig. 3). These samples 476

contain very similar elemental patterns besides having similar Sr and Nd isotopic fingerprints, making 477

it difficult to differentiate between them. These facts suggest that for the production of samples TR-478

DE14 and TR-DE02, raw materials (limestone and clay) that originated from a common geological area 479

were used, meaning they possess a common geological origin. 480

481

Fig. 5. Divisive Hierarchical Clustering (DIANA) – Dissimilarity is median Euclidean distance between all 482

geochemical profiles with ratios involving mass fractions of Ca, K, Mg, Mn, Sr, and Ti, plus the 483

Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd). 484

5. Conclusion and Outlook485

To compare and distinguish between two unknown Portland cement samples, it was necessary to 486

develop an analytical procedure that would allow cement fingerprinting. For this purpose, twenty-nine 487

ordinary Portland cement samples from cement plants located worldwide were investigated for their 488

elemental concentrations using XRF and Sr and Nd isotope ratios using MC-TIMS (so-called combined 489

approach). 87Sr/86Sr and 143Nd/144Nd conventional isotope systems in this study were selected because 490

they can be linked to their respective geographic origins, as they reflect the mineral composition and 491

geological history of the raw material used for cement production. 492

In Germany, many deposits contain limestone or dolomite in different geological formations. Different 493

geological ages of limestones, as well as the original mineral composition and geological history, are 494

reflected in the 87Sr/86Sr isotope ratio. This study reveals that the Sr isotope ratios of OPCs are higher 495

than that of seawater from the corresponding geological period. One of the explanations for this can 496

be a diagenetic alteration or the presence of radiogenic clay in the raw material. However, the other 497

explanation can be the addition of an unknown additive to the cement production process (e.g., 498

gypsum) which has a substantially higher 87Sr/86Sr isotope ratio, affecting the final product´s ratio. 499

The results indicate that most of the investigated OPCs possess measurable differences in Rcon(87Sr/86Sr) 500

and Rcon(143Nd/144Nd), which can be used as a valuable analytical tool to fingerprint them. Sr and Nd 501

isotope ratios were able to resolve all German OPCs, except for three chemically very similar German 502

OPCs. One of the biggest advantages of the combined isotope ratio approach is that it does not require 503

large sample sizes. Less than 0.5 g of sample is required for Sr and Nd isotope analysis, unlike other 504

techniques (XRD, XRF), which proves a useful criterion for forensic/crime scene investigations. 505

However, three samples in the combined approach showed limited provenancing potential due to 506

equivocal results. 507

In this case, Divisive Hierarchical Clustering (DIANA) can be employed, and geochemical profiles based 508

on Sr and Nd isotope ratios together with all ratios involving mass fractions of Ca, K, Mg, Mn, Sr, and 509

Ti can be determined. DIANA allowed further separation of individual samples by grouping them into 510

clusters based on their dissimilarity levels. This means that samples that were not separable based on 511

a combined approach could be successfully distinguished using geochemical profiles and DIANA. 512

However, when samples, besides having similar Sr and Nd isotopic fingerprints, also consist of similar 513

elemental patterns, they are grouped via DIANA into the same cluster, meaning they remain 514

indistinguishable. 515

There are different reasons for adopting this methodology for provenancing: 516

1. Fingerprinting of OPCs is feasible due to the measurable differences in the 87Sr/86Sr and 143Nd/144Nd517

isotope ratios of the investigated set of ordinary Portland cements. 518

2. Differentiation between cement samples is possible due to the small uncertainties provided by the519

high-quality measurement technique MC-TIMS. 520

3. Together with Sr and Nd isotope ratios, the ratios involving mass fractions of Ca, K, Mg, Mn, Sr, and521

Ti, plus the 87Sr/86Sr and 143Nd/144Nd conventional isotope ratios form the basis for Divisive Hierarchical 522

Clustering (DIANA) of cements. This helps differentiate cements that have similar Sr and Nd isotope 523

fingerprints and could not be differentiated using the combined approach. The results of this 524

study reveal that the proposed methodology is successful in distinguishing 27 out of 29 investigated 525

OPCs. 526

527

To achieve comparability, the outlook in expanding the database for Sr and Nd isotopic fingerprints in 528

cements and related materials in samples worldwide would have to occur. The database would acquire 529

complete documentation of cement samples combining chemical and isotopic data and their identified 530

provenance. Reporting chemically and isotopically defined cement will build on the existing database. 531

Concerning cement provenance determination, the amplification of a reference database would be 532

very important for the worldwide scientific community. 533

Abbreviations 534

Conc. – Concentrated 535

DIANA – Divisive Hierarchical Clustering 536

ICP-MS – Inductively Coupled Plasma Mass Spectrometry 537

MC – Multi Collector 538

OPC – ordinary Portland cement 539

PP – polypropylene 540

Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd) – Conventional 87Sr/86Sr and 143Nd/144Nd isotope ratios 541

TIMS –Thermal Ionisation Mass Spectrometry 542

Acknowledgements 543

The first author gratefully acknowledges Elis Hoffmann and Harry Becker for giving her the opportunity 544

to use FUB infrastructure for performing Nd measurements. Many thanks to Marcus Oelze for 545

constructive discussion and assistance concerning Fig. 3 formation. We wholeheartedly acknowledge 546

all cement producers who provided the cement samples. Owen Forbes is acknowledged for his help in 547

consensus clustering procedure, which will be reported in a future publication. 548

Author contribution 549

Anera Kazlagić: conceptualisation, methodology, project administration, investigation, formal analysis, 550

data curation, writing – original draft, writing – review & editing, visualisation. Dietmar Stephan: 551

supervision, writing – review & editing. Markus Ostermann: formal analysis. Antonio Possolo: data 552

curation, clustering, writing – review & editing. Jochen Vogl: supervision, conceptualisation & funding 553

acquisition, writing – review & editing. 554

Funding 555

This work received funding under project number MIT1-2019–8 from the Federal Institute for 556

Materials Research and Testing (BAM). 557

558

Availability of data and material 559

The data used to support the findings of this study is available within the article and the associated 560

Supplementary Information file. The data is also available from the corresponding author upon 561

request. 562

Conflict of interest 563

The authors declare no competing interests. 564

565

References 566

(BGR), B. F. G. U. R. Available: https://www.bgr.bund.de/DE/Themen/Sammlungen-567

Grundlagen/GG_geol_Info/Karten/Deutschland/GUEK250/guek250_node.html [Accessed 568

05.10.2022]. 569

(ED.), V. 2018. Activity Report 2015-2018. Düsseldorf. 570

(JCGM), J. C. F. G. I. M. 2008. Evaluation of Measurement Data - Guide to the Expression of Uncertainty 571

in Measurement. Sèvres, France: International Bureau of Weights and Measures (BIPM). 572

BANNER, J. L. 2004. Radiogenic isotopes: systematics and applications to earth surface processes and 573

chemical stratigraphy. Earth-Science Reviews, 65, 141-194. 574

BEARD, B. L. & JOHNSON, C. M. 2000. Strontium isotope composition of skeletal material can 575

determine the birth place and geographic mobility of humans and animals. Journal of Forensic 576

Sciences, 45, 1049-1061. 577

BIZIMIS, M. & SCHER, H. D. 2018. Neodymium Isotopes. In: WHITE, W. M. (ed.) Encyclopedia of 578

Geochemistry: A Comprehensive Reference Source on the Chemistry of the Earth. Cham: 579

Springer International Publishing. 580

BRILL, R. H. & WAMPLER, J. M. 1967. Isotope Studies of Ancient Lead. American Journal of Archaeology, 581

71, 63-77. 582

BULLEN, T. D. & KENDALL, C. 1998. Chapter 18 - Tracing of Weathering Reactions and Water Flowpaths: 583

A Multi-isotope Approach. In: KENDALL, C. & MCDONNELL, J. J. (eds.) Isotope Tracers in 584

Catchment Hydrology. Amsterdam: Elsevier. 585

CAPO, R. C., STEWART, B. W. & CHADWICK, O. A. 1998. Strontium isotopes as tracers of ecosystem 586

processes: theory and methods. Geoderma, 82, 197-225. 587

DUNGWORTH, D., DEGRYSE, P., SCHNEIDER, J. 2009. Kelp in historic glass: the application of strontium 588

isotope analysis. Isotopes in Vitreous Materials. Leuven, Belgium: Leuven University Press. 589

FAURE, G. & MENSING, T. M. 2013. Isotopes : principles and applications, Hoboken, New Jersey, John 590

Wiley & Sons, Inc. . 591

GAN, F., LI, Q. & HENDERSON, J. 2016. Recent Advances in the Scientific Research on Ancient Glass and 592

Glaze, Danvers, USA, World Century Publishing Corp. and World Scientific Oublishing Co. Pte. 593

Ltd. 594

GEYH, M. A. & SCHLEICHER, H. 1990. Phanerozoic Time-Scale. In: GEYH, M. A. & SCHLEICHER, H. (eds.) 595

Absolute Age Determination: Physical and Chemical Dating Methods and Their Application. 596

Berlin, Heidelberg: Springer Berlin Heidelberg. 597

GRAHAM, I. J., GOGUEL, R. L. & ST JOHN, D. A. 2000. Use of strontium isotopes to determine the origin 598

of cement in concretes - Case examples from New Zealand. Cement and Concrete Research, 599

30, 1105-1111. 600

HAMILTON, P. J., O’NIONS, R. K., BRIDGWATER, D. & NUTMAN, A. P. 1983. Sm-Nd studies of Archaean 601

metasediments and metavolcanics from West Greenland and their implications for the Earth's 602

early history. Earth and Planetary Science Letters, 62, 263-272. 603

HOOGEWERFF, J. A., REIMANN, C., UECKERMANN, E., FREI, R., FREI, K. M., VAN ASWEGEN, T., STIRLING, 604

C., REID, M., CLAYTON, A., LADENBERGER, A. & TEAM, G. P. 2019. Bioavailable Sr-87/Sr-86 in 605

European soils: A baseline for provenancing studies. Science of the Total Environment, 672, 606

1033-1044. 607

JOCHUM, K. P., NOHL, U., HERWIG, K., LAMMEL, E., STOLL, B. & HOFMANN, A. W. 2005. GeoReM: A 608

New Geochemical Database for Reference Materials and Isotopic Standards. Geostandards 609

and Geoanalytical Research, 29, 333-338. 610

KAUFMAN, L. & ROUSSEEUW, J. P. 1990. Divisive Analysis (Program DIANA). Finding Groups in Data: 611

An Introduction to Cluster Analysis. John Wiley & Sons, Inc. 612

KAZLAGIĆ, A., RUSSO, F. F., VOGL, J., STURM, P., STEPHAN, D. & GLUTH, G. J. G. 2022. Development of 613

a sample preparation procedure for Sr isotope analysis of Portland cements. Analytical and 614

Bioanalytical Chemistry, 414, 4379-4389. 615

KAZLAGIĆ, A., VOGL, J., GLUTH, G. J. G. & STEPHAN, D. 2021. Provenancing of cement using elemental 616

analyses and isotope techniques - the state-of-the-art and future perspectives. Journal of 617

Analytical Atomic Spectrometry, 36, 2030-2042. 618

KINOSHITA, N., PAUL, M., KASHIV, Y., COLLON, P., DEIBEL, C. M., DIGIOVINE, B., GREENE, J. P., 619

HENDERSON, D. J., JIANG, C. L., MARLEY, S. T., NAKANISHI, T., PARDO, R. C., REHM, K. E., 620

ROBERTSON, D., SCOTT, R., SCHMITT, C., TANG, X. D., VONDRASEK, R. & YOKOYAMA, A. 2012. 621

A Shorter 146Sm Half-Life Measured and Implications for 146Sm-142Nd Chronology in the Solar 622

System. Science, 335, 1614-1617. 623

KOSEDNAR-LEGENSTEIN, B. 2007. Historic mortars and plasters in Austria (Styria). Ph.D., Technische 624

Universitaet Graz. 625

KOSEDNAR-LEGENSTEIN, B., DIETZEL, M., LEIS, A., STINGL, K., WIEGAND, B. & BAUMGARTNER, M. 626

2007. Historical carbonate mortar and plaster - Proxies for ancient environments. Geochimica 627

Et Cosmochimica Acta, 71, A514-A514. 628

KOSEDNAR-LEGENSTEIN, B., DIETZEL, M., LEIS, A., WIEGAND, B., STINGL, K. & BAUMGARTNER, M. 629

2006. Historical Carbonate Mortar and Plaster-Isotopic and Chemical Signatures. European 630

Geosciences Union General Assembly. Vienna, Austria. 631

MCARTHUR, J. M., HOWARTH, R. J. & SHIELDS, G. A. 2012. Chapter 7 - Strontium Isotope Stratigraphy. 632

In: GRADSTEIN, F. M., OGG, J. G., SCHMITZ, M. D. & OGG, G. M. (eds.) The Geologic Time Scale. 633

Boston: Elsevier. 634

MCARTHUR, J. M., RIO, D., MASSARI, F., CASTRADORI, D., BAILEY, T. R., THIRLWALL, M. & HOUGHTON, 635

S. 2006. A revised Pliocene record for marine-87Sr/86Sr used to date an interglacial event636

recorded in the Cockburn Island Formation, Antarctic Peninsula. Palaeogeography,637

Palaeoclimatology, Palaeoecology, 242, 126-136.638

MEISSNER, F., SCHMIDT-OTT, W.-D. & ZIEGELER, L. 1987. Half-life and α-ray energy of146Sm. Zeitschrift 639

für Physik A Atomic Nuclei, 327, 171-174. 640

MIKOVA, J. & DENKOVA, P. 2007. Modified chromatographic separation scheme for Sr and Nd isotope 641

analysis in geological silicate samples. Journal of Geosciences, 52, 221-226. 642

NAKANO, T. 2016. Potential uses of stable isotope ratios of Sr, Nd, and Pb in geological materials for 643

environmental studies. Proceedings of the Japan Academy Series B-Physical and Biological 644

Sciences, 92, 167-184. 645

POSSOLO, A. & MEIJA, J. 2020. Measurement Uncertainty: A Reintroduction, Sistema Interamericano 646

de Metrologia, Inter American Metrology System. 647

PROSKE, D. 2019. Bridge collapse frequencies versus failure probabilities, Springer, Cham. 648

PROSKE, D. & SCHMID, M. 2021. Frequency and mortality of building collapses. Bautechnik, 98, 423-649

432. 650

SALIFU, M., AIGLSPERGER, T., HÄLLSTRÖM, L., MARTINSSON, O., BILLSTRÖM, K., INGRI, J., DOLD, B. & 651

ALAKANGAS, L. 2018. Strontium (87Sr/86Sr) isotopes: A tracer for geochemical processes in 652

mineralogically-complex mine wastes. Applied Geochemistry, 99, 42-54. 653

SCHMIDT-DÖHL, F., KOEPKE, J. & SCHIMROSCZYK, A. 2005. Testing the identity of cementitious 654

materials by trace element analysis of belite. 27th International Conference on Cement 655

Microscopy. Victoria, British Columbia, Canada. 656

STEIGER, R. H. & JÄGER, E. 1977. Subcommission on geochronology: Convention on the use of decay 657

constants in geo- and cosmochronology. Earth and Planetary Science Letters, 36, 359-362. 658

TAYLOR, H. F. W. 2009. Cement chemistry: 2nd edition, London, Thomas Telford Ltd. 659

TELSCHOW, S., FRANDSEN, F., THEISEN, K. & DAM-JOHANSEN, K. 2012. Cement Formation-A Success 660

Story in a Black Box: High Temperature Phase Formation of Portland Cement Clinker. Industrial 661

and Engineering Chemistry Research, 51, 10983-11004. 662

VANHAECKE, F. & DEGRYSE, P. 2012. The Isotopic Composition of the Elements. In: VANHAECKE, F. K., 663

KURT (ed.) Isotopic analysis : fundamentals and applications using ICP-MS. Weinheim; 664

Chichester: Wiley-VCH ; [John Wiley [distributor]. 665

VDZ, V. D. Z. E. V. 2022a. Environmental data of the German cement industry 2021. Düsseldorf. 666

VDZ, V. D. Z. E. V. 2022b. Zahlen und Daten – Zementindustrie in Deutschland 2022. Düsseldorf. 667

WEDEPOHL, K. H. & BAUMANN, A. 2000. The Use of Marine Molluskan Shells for Roman Glass and 668

Local Raw Glass Production in the Eifel Area (Western Germany). Naturwissenschaften, 87, 669

129-132.670

WHITEHOUSE, M. J. 1996. Radiogenic Isotope Geology. Geological Magazine, 133, 226-226. 671

672

100

SUPPLEMENTARY INFORMATION FOR:

Fingerprinting Portland Cements by means of 87Sr/86Sr and 143Nd/144Nd Isotope

Ratios and Geochemical Profiles

Anera Kazlagić (1)*, Stephan Dietmar (2), Markus Ostermann (3), Antonio Possolo (4), Jochen Vogl

(1)*

(1) Federal Institute for Materials Research and Testing, Division 1.1 Inorganic Trace Analysis,

Richard-Willstäter-Straße 11, 12489 Berlin, Germany

(2) Department of Civil Engineering, Building Materials and Construction Chemistry, Technische

Universität Berlin, Gustav

‑

Meyer

‑

Allee 25, 13355 Berlin, Germany

(3) Federal Institute for Materials Research and Testing, Division 1.4 Process Analytical Technology,

Richard-Willstäter-Straße 11, 12489 Berlin, Germany

(4)National Institute of Standards & Technology, 100 Bureau Drive, Gaithersburg, MD 20899, USA

*Corresponding author: anera.kazlagic@bam.de

Contents

Fig. S1 ............................................................................................................................................... S-2

Fig. S2 ............................................................................................................................................... S-3

Table S1 ............................................................................................................................................ S-4

References ......................................................................................................................................... S-5

101

Fig. S1. Strontium chromatographic separation scheme consists of four stages: preconditioning,

loading the sample solution, rinsing, and collecting the Sr fraction.

102

Fig. S2. Neodymium chromatographic separation scheme, whereby a – conditioning of columns, b –

coupling of Sr.spec and TRU.Spec columns and sample loading on Sr.Spec and TRU.Spec colum couple.

c – decoupling of columns and loading of LREE from TRU.Spec to Ln.Spec, e – decoupling of columns

and neodymium elution. Procedure scheme is adapted (with some changes in elution profile) from

Mikova and Denkova (Míková and Denková, 2012).

103

Table S1. The results of elemental concentrations (and elemental ratios) measured by XRF.

Sample Name ID BAM Sample Name

CaO Cr

OMgO Mn

SiO

SrO TiO

ZnO

Al Ca Cr Fe KMg Mn Na P S Si Sr Ti Zn

(%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%)

XX-IQ01 13065 5.25 63.99 0.06 3.62 0.73 3.19 0.14 0.29 0.08 2.52 20.75 0.07 0.29 0.01

CR-DE01 23029a 5.50 66.59 0.01 2.17 0.80 1.06 0.07 0.44 0.20 3.94 20.68 0.21 0.28 0.02

XX-BH01 33064a 6.74 64.21 0.02 3.60 0.66 1.13 0.16 0.10 0.11 2.76 21.84 0.05 0.26 0.02

TR-DE02 43028a 4.35 63.18 0.01 3.73 1.23 2.30 0.08 0.16 0.21 3.25 21.61 0.15 0.18 0.13

JU-DE03 53032a 4.82 66.03 0.01 2.98 0.63 1.28 0.10 0.15 0.44 3.20 20.23 0.03 0.32 0.02

CR-DE04 63022a 3.68 67.60 0.01 1.38 0.59 0.83 0.05 0.20 0.16 3.48 23.42 0.16 0.20 0.04

XX-DE05 73026a 5.30 64.47 0.01 3.07 0.90 3.04 0.11 0.35 0.14 3.67 20.04 0.17 0.28 0.03

DM-DE06 83027a 5.46 62.83 0.01 2.54 1.29 2.70 0.07 0.17 0.05 3.24 20.22 0.12 0.27 0.01

JU-DE07 93030a 5.99 63.12 0.01 2.21 0.97 1.29 0.07 0.21 0.22 3.02 20.65 0.25 0.32 0.02

CR-DE08 10 3050a 4.47 65.33 0.01 3.04 0.83 2.26 0.06 0.19 0.28 2.48 20.91 0.07 0.26 0.04

JU-DE09 11 3024a 5.81 65.26 0.01 2.78 0.51 1.50 0.05 0.08 0.19 3.71 20.86 0.06 0.31 0.01

XX-SR01 12 3063a 5.42 64.03 0.02 2.62 0.74 2.36 0.10 0.26 0.13 3.58 21.36 0.11 0.24 0.05

XX-SR02 13 3062a 4.28 62.94 0.01 2.97 0.57 1.31 0.08 0.10 0.05 2.56 19.67 0.05 0.21 0.02

XX-IQ02 14 3066 4.27 64.15 0.02 5.15 0.53 2.37 0.26 0.41 0.07 2.52 20.80 0.04 0.25 0.01

TE-DE10 15 3019a 5.09 64.57 0.01 3.14 0.88 1.93 0.28 0.12 0.23 2.96 20.93 0.09 0.23 0.04

CR-DE11 16 3020a 5.73 65.87 0.01 2.76 0.76 1.41 0.19 0.11 0.20 2.12 21.03 0.16 0.28 0.02

TR-DE12 17 3023a 5.29 64.47 0.02 3.25 0.67 2.98 0.06 0.20 0.42 3.90 20.04 0.15 0.31 0.03

XX-IQ03 18 3067 4.05 64.39 0.02 5.10 0.54 1.39 0.23 0.18 0.07 2.51 19.46 0.08 0.25 0.01

TR-DE13 19 3101 4.53 63.65 0.01 2.66 0.98 1.97 0.04 0.25 0.30 3.11 19.56 0.39 0.23 0.02

XX-GR01 20 3077a 4.51 64.84 0.01 4.31 0.57 1.01 0.13 0.18 0.08 2.59 19.87 0.10 0.28 0.07

TR-DE14 21 3123 5.10 62.09 0.01 2.86 1.00 1.64 0.05 0.20 0.06 3.46 18.40 0.16 0.26 0.03

XX-GR02 22 3075a 4.96 63.89 0.05 4.00 0.50 2.35 0.18 0.25 0.07 2.57 19.43 0.03 0.29 0.04

TR-DE15 23 3122 4.78 62.44 0.01 2.70 1.31 1.44 0.10 0.20 0.18 3.21 20.11 0.05 0.26 0.04

XX-KO01 24 3076a 5.71 62.87 0.02 3.48 0.61 2.22 0.08 0.19 0.07 2.64 20.71 0.04 0.31 0.02

XX-MC01 25 3078a 6.58 60.81 0.01 3.32 0.98 3.02 0.09 0.29 0.08 2.95 21.23 0.04 0.30 0.01

XX-AU01 26 3138 4.99 63.85 0.02 2.24 0.77 2.11 0.19 0.33 0.19 3.40 19.82 0.04 0.31 0.04

XX-CH01 27 3142 4.94 62.42 0.00 3.15 0.70 2.18 0.05 0.09 0.07 2.18 20.42 0.07 0.22 0.01

XX-AU02 28 3145 5.06 62.60 0.01 3.18 0.71 2.07 0.18 0.29 0.12 4.24 19.02 0.17 0.31 0.03

XX-IT01 29 3147 4.28 62.87 0.02 3.04 0.44 2.51 0.07 0.27 0.12 3.09 19.39 0.08 0.15 0.04

104

References

MÍKOVÁ, J. & DENKOVÁ, P. 2012. Modified chromatographic separation scheme for Sr and Nd

isotope analysis in geological silicate samples. Journal of GEOsciences, 221-226.

105

Chapter 3

Interlaboratory comparison on conventional 87Sr/86Sr isotope ratios as a

quality control indicator

Kazlagić Anera *

* Federal Institute for Materials Research and Testing, Richard-Willstäter-Straße 11, 12489

Berlin, Germany

Abstract

For performing isotope ratio measurements reported in previous publications (added in

Chapter 2), it was necessary to establish a quality control procedure.

Thus, as a part of this dissertation, an interlaboratory comparison (ILC) was organised to

characterise 87Sr/86Sr isotope ratios in geological and industrial reference materials by

applying the conventional method for 87Sr/86Sr isotope ratios. As reference material, four

cements (VDZ 100a, VDZ 200a, VDZ 300a, IAG OPC-1), one limestone (IAG/CGL ML-3), and one

slate (IAG OU-6) were selected, thus covering a wide range of Sr isotopic signatures. These six

powdered reference materials were sent to thirteen laboratories located worldwide. Together

with the materials, a detailed technical protocol was provided, prescribing the use of the

conventional method for determining 87Sr/86Sr isotope ratios, the evaluation of the

measurement uncertainty, and other requirements such as the minimum number of

digestions and the minimum test portion. Participants were asked to digest the reference

materials completely, e.g., with acid dissolution or alkaline fusion. Since both multi-collector

thermal ionisation mass spectrometry (MC-TIMS) and multi-collector inductively coupled

plasma mass spectrometry (MC-ICP-MS) instrumentation were used in this study, the first

objective of the ILC study was to test whether these two kinds of instruments yielded

significantly different results for 87Sr/86Sr isotope ratios. In other words, the objective was to

test for any differences in using either MC-TIMS or MC-ICP-MS. The second objective was to

assign reference values for 87Sr/86Sr isotope ratio measurements in the six materials. To meet

these goals, consensus values for the 87Sr/86Sr isotope ratios in these materials were estimated

by fitting a linear, Gaussian mixed effects model using R function “lmer” defined in

package “lme4”. [1] The ILC results are published in the scientific peer-reviewed journal

(Geostandards and Geoanalytical Research journal). [1] Besides being the pilot laboratory

and organising the ILC,

106

my own work as a participant was performed under the Laboratory ID 15 (in further text

Lab15). Thus, this chapter presents the results of Lab15 serving as a quality control indicator

for this thesis, explains in detail the selected sample preparation procedures, and elaborates

the individual results for 87Sr/86Sr isotope ratios in six reference materials.

3.1. Introduction

Monitoring the accuracy and precision of Sr isotope analyses according to the best practices

in metrology requires the use of characterised reference materials whose matrix is similar or

comparable to the matrix of the samples being analysed. These reference materials ensure

the highest metrological quality of the measurements for method validation, uncertainty

determination, and traceability of the measurement results. [2] Among different RMs,

certified isotope reference materials (iCRMs) for isotope ratio analysis have a wide range of

applications. They are required either for correction of mass fractionation/mass

discrimination, for method validation and quality control. [3,4] As a result, iCRMs are

indispensable for all isotope ratio measurements to obtain traceable and comparable isotope

results, and to ensure compliance with quality control protocols. [4] We distinguish two types

of iCRMs, namely primary iCRMs to be used for calibrating absolute isotope ratios and/or

anchoring delta scales and matrix-matched iCRMs. Besides the need for primary iCRMs

essential for correction and calibration of Instrumental Isotopic Fractionation (IIF), matrix

matched iCRMs are of increasing importance in validation of analytical procedures and quality

control due to the similarity of the sample matrix.

Due to the shortage of suitable matrix-matched iCRMs and at the same time a growing

necessity for them, several reference samples were selected by the community, measured,

and isotope ratio data was then reported. These data have been then compiled in specific

publications (e.g., Brand and co-authors [5]) or in databases such as GeoReM. [6] These data

are very beneficial since they provide a consensus value for a specific isotope ratio. However,

the reported data often come without an evaluation of possible measurement biases or

uncertainty components attributable to the measurement itself, material instability or

heterogeneity. In addition, with each new publication or new issue of the database, such data

are subject to change. [7]

One of the solutions for this problem is performing interlaboratory comparison studies. In this

way, it is possible to overcome the constant issues regarding the scarcity or unavailability of

107

iCRMs and the issues of changing values of data in databases. According to ISO/IEC

17043:2010, interlaboratory comparison (ILC) is defined as the organisation, performance,

and evaluation of measurements or tests on the same or similar items by two or more

laboratories or inspection bodies in accordance with predetermined conditions. [8] Therefore,

ILC studies in measurement science have the task to compare measurement results obtained

independently and to produce a consensus value for the common measurand that combines

the values measured by the participants. [9]

Isotope ratios of strontium (Sr) are key tracers used for solving various problems in different

scientific fields, including archaeology [10], food chemistry [11-13], forensic sciences and

provenancing [14-17], as well as geo- and environmental [18] chemistry. The application of Sr

in different fields is based on the principle that the Sr isotope ratios of natural materials reflect

the sources of Sr available during their formation.

Reports on reference values of 87Sr/86Sr isotope ratios in geological materials are limited to

several publications, mostly focused on 87Sr/86Sr isotope characterisation of silicate samples

and rock reference materials, [19-25] and in one case sediments. [26]

Although urgently needed to validate analytical procedures used for cement provenance

studies [27-29], matrix-matched reference materials with characterised 87Sr/86Sr are

unfortunately not available in the field of construction chemistry and cementitious materials.

In general, cements are produced from a carbonate (e.g., limestone) and a silicate (e.g., clay)

component and various additive materials (e.g., gypsum, anhydrite, flue gas desulphurisation

gypsum, limestone). [28] Marl contains both the carbonate and the silicate component and is

typically used if marl deposits are located close to the cement plants. Therefore, additional

87Sr/86Sr isotope characterisation of the raw materials used for cement production are of great

importance in cement provenancing. Due to the typically high Sr concentrations in the

carbonate raw materials (e.g., IAG/CGL ML-3) and the much lower Sr concentration in the

silicate raw materials (e.g., IAG OU-6) together with higher proportion of carbonate to silicate,

it is reasonable to assume that the 87Sr/86Sr isotope ratio of cements is dominated by the

carbonate raw material signature.

Furthermore, for conventional, so-called ´radiogenic´ 87Sr/86Sr isotope ratios, the number of

available reference materials with assigned reference values is limited.

108

Therefore, an interlaboratory comparison study was organised to characterise 87Sr/86Sr

isotope ratios in cements, limestone, and slate by applying the conventional method for

87Sr/86Sr isotope ratio determination. Four cements (VDZ 100a, VDZ 200a, VDZ 300a, IAG OPC-

1), one limestone (IAG/CGL ML-3), and one slate (IAG OU-6) were selected. The final goal was

to ensure quality control materials which are necessary for obtaining trustworthy results in

cement and other building materials whilst performing isotope analysis. Even though the

selected limestone and slate reference materials are not specifically used to produce the

cements considered in this ILC, they have very similar matrices compared to the raw material

used for cement production, and their 87Sr/86Sr isotopic characterisation is useful for quality

control assessment in geochemical as well as technical applications. In further sections, more

information about general requirements, RMs used, analytical procedures, and information

regarding Lab15, e.g., study layout, is provided.

3.2. Study design and 87Sr/86Sr isotope ratio

3.2.1. Study design

Based on their demonstrated ability to perform metrologically valid isotope ratio

measurements in general or based on their demonstrated experience in 87Sr/86Sr isotope ratio

determinations, twelve international analytical institutes (thirteen laboratories) were invited

to participate in the ILC study. All participants, in total thirteen labs, received a set of bottled

and packed reference materials, alongside a technical protocol with a reporting template. The

technical protocol contained detailed, mandatory guidelines to be followed. First, at least two

digestions per reference material needed to be carried out by the participants with a minimum

sample amount of 100 mg per digestion. Second, participants were asked to perform a

complete digestion of the reference material. Third, strontium was required to be isolated

from the matrix preferably via chromatographic means to remove interfering Rb and matrix

elements. Strontium recovery and procedural Sr blanks should be reported. Fourth, 87Sr/86Sr

isotope ratio measurements should be carried out using either MC-ICP-MS or MC-TIMS.

Additionally, one measurement of the certified isotope reference material NIST SRM 987 (Sr

carbonate isotopic standard, NIST, Maryland, USA) was requested per MC-ICP-MS sequence

or MC-TIMS turret, with a total of at least three measurements. Fifth, 87Sr/86Sr isotope ratio

measurements should be carried out following the 87Sr/86Sr isotope ratio method as described

below. Finally, participants were asked to report any quality control measures they had carried

109

out, such as 87Sr/86Sr isotope ratio measurement of international reference materials or

quality control samples. [1]

3.2.2. 87Sr/86Sr isotope ratio

Based on the definition for an operationally defined measurand in the ISO 17034 guidelines

[30,31], and also based on the definition in International Vocabulary [32], the commonly

applied conventional 87Sr/86Sr isotope ratio, also known as ‘radiogenic’ 87Sr/86Sr isotope ratio,

completely fulfils the requirements of an operationally defined measurand and is termed

‘conventional 87Sr/86Sr isotope ratio’ (in further text 87Sr/86Sr isotope ratio) with the quantity

symbol Rcon(87Sr/86Sr) and the unit mol·mol−1. Values of Rcon(87Sr/86Sr) are traceable to the

conventional 87Sr/86Sr isotope ratio method provided all steps of the method are followed and

an uncertainty budget is available. [1]

For obtaining and reporting 87Sr/86Sr isotope ratios, the internationally agreed-upon

guidelines had to be followed. The measurement method basically relies on the application of

mass spectrometry (in most cases MC-TIMS or MC-ICP-MS) and the internal (within the sample

measurement) correction of Instrumental Isotopic Fractionation, IIF (also known as ´mass

bias´) of the measured 87Sr/86Sr ion intensity ratio via the measured 86Sr/88Sr ion intensity

ratio, whereby the 86Sr/88Sr isotope ratio is set by convention to 0.1194. [33,34] Dividing the

convention value 86Sr/88Sr = 0.1194 by the measured 86Sr/88Sr ion intensity ratio and applying

it to the measured 87Sr/86Sr ion intensity ratio by using the exponential law or the power law,

the IIF is obtained. [35,36] This step makes the main difference of the 87Sr/86Sr isotope ratio

to the absolute isotope ratio n(87Sr)/n(86Sr), because stable isotopic variations are eliminated

by the IIF correction, and an insufficiently accurate value for the 86Sr/88Sr isotope ratio is used.

Moreover, the 87Sr/86Sr isotope ratio of NIST SRM 987 determined by the conventional

method is accepted to be 0.710 25 mol·mol−1 [37,38] in contrast to the certified value

n(87Sr)/n(86Sr) = (0.710 34 ± 0.000 26) mol·mol−1 [39]. In addition, more than 1900 published

results listed in the GeoReM database [6] up to 2019 for the 87Sr/86Sr isotope ratio of NIST

SRM 987 yield 0.710 250 mol·mol−1 as a median value with an expanded uncertainty of

0.000 001 mol·mol−1 with a coverage factor of k = 2. [6] Therefore, whenever a significant bias,

meaning bias beyond the analytical precision, between the measured value of NIST SRM 987

and the isotope ratio Rcon(87Sr/86Sr) = 0.710 250 mol·mol−1 was detected, it was recommended

to use this value for obtaining a final correction factor, which should have consequentially

110

been applied to all sample measurements. This kind of final correction factor may compensate

for remaining bias caused e.g., by differences in the detector efficiencies. By following the

exactly prescribed procedure and predefined method parameters described above, the

traceability to the internationally agreed convention method and comparability between all

measurement results obtained were ensured. [1]

3.3. Reference Materials

Six different powdered reference materials were used in this study, four cement materials and

two rock materials representing two potential raw materials of cement (limestone and

siliceous clay/marl, as listed in Table 3.1). These samples are reference materials designed for

use by laboratories undertaking the determination of the mass fractions of major and trace

elements therein, the assessment of measurement procedures, as well as for quality control.

Each participant received one sub-unit of each reference material, namely: VDZ 100a [40], VDZ

200a [41], VDZ 300a [42], IAG OPC-1 [43], IAG OU-6 [44], and IAG/CGL ML-3 [45]. The first four

samples (VDZ 100a, VDZ 200a, VDZ 300a, IAG OPC-1) are cement reference materials

delivering values of the mass fractions of the major and trace elements of different

compositions. Sample IAG OU-6 is a slate material and IAG/CGL ML-3 is a limestone material,

both being Certified Reference Materials, delivering certified values of the mass fractions of

the major and trace elements. More details and description of the materials are available

following the cited references.

Each of our sub-unit contains approx. 10 g of the corresponding reference material. Prior to

filling, the polypropylene (PP) bottles were precleaned by leaching in nitric acid (w = 10 % v/v,

Chemsolute®, Th. Geyer, Berlin, DE) and Milli-Q water (Milli-Q Advantage A10 System, Merck

KGaA, Darmstadt, DE). Reference materials were packed and dispatched in PP bottles sealed

in polyethylene-aluminium-composite foil bags (Fig. 3.1). Each reference material was labelled

according to its original name and unit number.

111

Table 3.1. Information on material, including supplier information, type of material (RM –

reference material, CRM Certified Reference Material) and Sr content indicated on

certificates.

Material name

Supplier

Type of

material

RM/CRM

Sr (mg·kg−1) ± U,

k=2

VDZ 100a

VDZ

Portland

cement

1107±50.4

VDZ 200a

VDZ

Portland-

composite

cement

1922±394

VDZ 300a

VDZ

Blast furnace

cement

811±49

IAG OPC-1

IAG

Portland

cement

118.2±2.1

IAG OU-6

IAG

Penrhyn Slate

CRM

132±3

IAG/CGL ML-3

IAG

Limestone

CRM

1018±30

Fig. 3.1. Left: six samples packed and prepared for shipment, right: labelled sample bottles

3.4. Analytical procedures – General Overview

Sample preparation procedures such as weighing, sample dissolution, and Sr isolation were

performed in clean laboratory environments in the respective participants´ laboratories.

Sample decomposition was realised by (i) acid digestion carried out in closed Teflon vials on a

112

hot plate and using various mineral acid mixtures, (ii) a microwave-assisted procedure, or (iii)

using a digestion bomb or by borate fusion. The amount of sample used for each independent

sample preparation ranged between 100 mg and 600 mg. Subsequent isolation of Sr was

carried out either by cation ion exchange resin (e.g. BioRad AG 50W X8) or by crown ether-

based resins (Triskem Sr resin). Both manual as well as automated procedures for Sr

separation were performed by the participants. Procedural blanks for Sr as reported by most

of the laboratories ranged from less than 2.7 ng down to 0.006 ng. [1] Schematic

representation of the analytical procedure applied in obtaining the 87Sr/86Sr isotope ratio is

shown on Fig. 3.2.

Fig. 3.2. Schematic representation of the analytical procedure applied in obtaining the

87Sr/86Sr isotope ratio

3.5. Study design by Lab15

3.5.1. Chemicals

HNO3 (65–68 % v/v) and HCl (≈ 30 % v/v) were purchased as pro analysis grade acids

(Chemsolute®, Th. Geyer, Berlin, DE) and were further purified by double sub-boiling

distillation. Milli-Q water (Milli-Q Advantage A10 System, Merck KGaA, Darmstadt, DE) was

used. Conc. HF (47-51 % v/v, VWR Leuven, BE) and conc. HClO4 (70 % v/v, J.T. Baker®, CAN)

were purchased as ultrapure grade acids and were used without further purification.

For Sr purification and matrix separation, Sr·Spec™ resin (1000-150 µm, SR-50B-S, Triskem Sr

Spec) was employed.

113

3.5.2. Sample preparation

Sample preparation for Sr isotope analysis was performed in a metal-free clean laboratory

with ISO class 6 at the Federal Institute for Materials Research and Testing (BAM).

Each sample was weighed (≈ 200 mg) in Savillex® beakers. The next step was the addition of

conc. HF (6 mL) and conc. HNO3 (2 mL). The beakers were closed, agitated by hand, sonicated

(20 minutes) and digestion was performed on a hotplate (120 °C) for 7 days. After 7 days,

beakers were opened, and the samples were dried at 100 °C until acid evaporation occurred.

Then, 1.5 mL of conc. HF and 0.5 mL of conc. HNO3 were added, the lid was closed and

tightened, and beakers were heated on a hotplate at 120 °C for 3 days. After 3 days, the

beakers were opened, and the liquid was evaporated. The last step was the addition of 3 mL

HNO3 (3 mol·L-1) and the beaker was left open until dry.

After drying, 5 mL of HNO3 (3 mol·L-1) was added, mixed, left to settle down and one portion

for matrix separation was taken. The Sr blank was measured on SC-ICP-MS and was less than

0.5 ng. The recovery of Sr was between 65-88 % for VDZ 100a, VDZ 200a, VDZ 300a, and

IAG/CGL ML-3. Since the Sr recovery for OU-6 and OPC-1 using the above described method

was only 25 % (low recoveries are possibly due to coprecipitation of Sr2+ with fluorides), these

samples were prepared again using a different, slightly modified method “C” by Yokoyama

and co-authors [46], whereby the amounts of HClO4 were increased two times. Therefore,

sample decomposition procedure included digestion with a mixture of HF and HClO4. This

mixture is known to be effective in removing insoluble fluorides and has been widely used by

many geochemists [47-49]. Following this procedure, 1 mL HClO4 (7 mol·L-1) and 1 ml of conc.

HF were added to the weighed sample. For complete decomposition, the tightly closed beaker

was heated overnight on a hot plate at 100 °C, in a fume hood used only for mixtures

containing HClO4. The decomposed sample was progressively evaporated at 120 °C for 12 h,

165 °C for 12 h and 195 °C until dry. The sample residue was treated with another portion (2

mL) of 7 mol·L-1 HClO4. The tightly closed beaker was again progressively evaporated at 120 °C

for 12 h, 165 °C for 12 h and 200 °C until dry. The next step was the addition of 0.5 mL HCl (6

mol·L-1) and evaporation to dryness at 110 °C. The dried residue was dissolved in 1 ml HNO3

(4 mol·L-1) and transferred to 5 ml concave bottom Savillex® Teflon beakers. These beakers

were then centrifuged for 15 min at 5000 rpm and the supernatant was transferred to another

clean beaker. Then, 0.5 ml HNO3 (4 mol·L-1) was added to the residue and agitated in an

114

ultrasonic bath to wash out any remaining elements. The beakers were then centrifuged, and

the supernatant was added to the primary supernatant. This washing procedure was carried

out twice. The next step was slowly drying the beakers with supernatant on 85 °C for about 2

days. After drying, 5 mL HNO3 (3 mol·L-1) was added.

An aliquot of this solution containing 2 μg strontium was taken to perform a strontium matrix

separation using the water suspension of Sr·Spec™ resin (350 µL) in polyvinylchloride columns

(6 mm inner diameter, 4 cm long). Columns were washed with Milli-Q H2O (6 · 350 μL), loaded

with resin suspension (350 μL, of resin suspended in milli-Q H2O) and conditioned with HNO3

(3 mol·L-1, 6 · 350 μL). The subsample for matrix separation was loaded onto the column and

the matrix was eluted with HNO3 (3 mol·L-1, 6 · 350 μL). The Sr fraction was eluted with milli-

Q H2O (6 · 350 μL) and collected.

After column chromatography, the resulting strontium fraction was evaporated to dryness

and re-dissolved in nitric acid such that a final strontium mass fraction of 100 ng·μL-1 was

obtained, which could be directly used for: (i) performing MC-TIMS measurements by loading

1 µL of the sample on Re filaments, together with TaF5 activator for enhancing the ionisation,

(ii) performing MC-ICP-MS measurements.

3.5.3. Measurements

Lab15 measured 87Sr/86Sr isotope ratios of RMs using both MC-TIMS and MC-ICP-MS.

Strontium isotope analyses were carried out by MC-TIMS at BAM in Berlin using a Sector 54

instrument (Micromass Ltd., Manchester, UK), in a dynamic multi-collection mode via both

manual and automatic measurement procedure. During measurements, a stable 88Sr signal of

≈ 3 V was obtained for 70-80 minutes, corresponding to 150-200 individual signal scans.

Furthermore, AGV2-a and NASS-6 seawater reference materials were used as control samples.

These control samples went through the complete sample treatment. The raw measured data

were corrected for interfering Rb and mass fractionation (86Sr/88Sr = 0.1194) and finally were

normalised to a NIST SRM 987 87Sr/86Sr isotope ratio of 0.71025 [37].

115

Strontium isotope analyses were also carried out by MC-ICP-MS (Thermo Scientific Neptune

Plus, Bremen, DE) at BAM in Berlin in a static multi-collection mode. The Standard-Sample-

Bracketing (SSB) approach was applied, with a blank measurement before each standard and

before each sample.

During measurements, a stable 88Sr signal of ≈ 10 V was obtained, with analysis time being

approximately 10 minutes per sample, corresponding to 70 individual signal scans. In addition

to Rb, Kr interference correction was applied. During the analysis, a PFA nebuliser with a 100

μL·min-1 flow was used. To minimise any bias by differences in sample and standard

composition, all samples and standards were diluted using the same nitric acid stock solution

(ω = 20 g·kg-1 HNO3) and all samples and standards were ‘concentration’-matched to within ±

10 % following a preceding screening of the Sr mass fractions using the MC-ICP-MS. Sr mass

fraction in measurement solution was ≈ 100 ng·g-1. Furthermore, AGV2-a was used as the

control sample. Each sample was measured with seventy cycles, and outliers were identified

based on the twofold standard deviation (2s) criterion and removed from the data. Blank

correction for the outlier-corrected intensities was performed by subtracting the mean

intensities from the preceding and the succeeding blank. Up to this point, calculations and

corrections were made within the measurement sequence by the Thermo Scientific Neptune

software. With an exception for blank corrections, all other calculations such as IIF correction,

interference correction and final correction using the tabulated value for NIST SRM 987 and

mean calculations were carried out in Excel.

3.6. Results and discussion – Lab15

3.6.1. 87Sr/86Sr isotope ratios and associated uncertainties

All participants used the provided template for reporting their data. Results of Lab15 obtained

on individual measurements of each sample dissolution of a material, and one combined, final

result (average of all measurements) for each material using MC-TIMS (Table 3.2, 3.3 and 3.4)

and MC-ICP-MS (Table 3.5, 3.6 and 3.7) were reported.

In the Tables 3.2-3.7 all individual measurement results, as well as final results are reported

together with the estimated uncertainty budget. In addition, measured NIST SRM 987 values

are reported, together with SD and number of measurements.

116

Table 3.2. Results for 2 RMs, namely IAG OPC-1 and CEM 100a using MC-TIMS reported by

Lab15.

Table 3.3. Results for CEM 200a and CEM 300a using MC-TIMS reported by Lab15.

Sample Name

Dissolution number

Measurement number 1 2 1 2 1 2 1 2 1 2 1 2

corrected conventional

Sr/

Sr value

0.726513 0.726509 0.726478 0.726474 0.726511 0.726492 0.708127 0.708138 0.708140 0.708127 0.708146 0.708140

Standard deviation SD 4.73018E-05 6.8E-05 5E-05 5.44E-05 8.7E-05 6.1E-05 9E-05 9.4E-05 8.2E-05 7.1E-05 8E-05 7.9E-05

Number of measured

isotope ratios

100 151 91 125 116 120 196 195 199 194 200 148

Standard deviation of the

mean SE

4.7E-06 5.6E-06 5.3E-06 6.5E-06 8.0E-06 5.6E-06 6.4E-06 6.8E-06 5.8E-06 5.4E-06 5.7E-06 6.5E-06

Final result (average of all

measurements)

Combined standard

uncertainty

Coverage factor

Expanded uncertainty

Measured NIST SRM 987

Sr/

Sr values

Standard deviation SD for

NIST SRM 987

Number of measurements

N for NIST SRM 987

0.726496

0.708136

3.85035E-05

1.32163E-05

2.64326E-05

1.9252E-05

III

OPC-1

CEM 100a

1.3E-05

1.6E-05

0.710246

Sample Name

Dissolution number

Measurement number 1 2 1 2 1 2 1 2 1 2 1 2

corrected conventional

Sr/

Sr value

0.708248 0.708247 0.708252 0.708254 0.708250 0.708248 0.709338 0.709339 0.709350 0.709355 0.709337 0.709348

Standard deviation SD 7.4E-05 7.3E-05 7.2E-05 8.9E-05 7.5E-05 8.2E-05 7.8E-05 9E-05 7.4E-05 9.1E-05 7.6E-05 8.1E-05

Number of measured

isotope ratios

178 199 173 87 193 200 199 168 200 168 180 189

Standard deviation of the

mean SE

5.9E-06 5.2E-06 5.5E-06 9.5E-06 5.4E-06 5.8E-06 5.6E-06 6.9E-06 5.3E-06 8.5E-06 5.6E-06 5.9E-06

Final result (average of all

measurements)

Combined standard

uncertainty

Coverage factor

Expanded uncertainty

Measured NIST SRM 987

Sr/

Sr values

Standard deviation SD for

NIST SRM 987

Number of measurements

N for NIST SRM 987

1.15415E-05

2.30829E-05

2.14691E-05

0.708250

1.07345E-05

0.709344

III

CEM 300a

CEM 200a

III

1.6E-05

1.3E-05

0.710246

0.710243

117

Table 3.4. Results for IAG OU-6 and IAG/CGL ML-3 using MC-TIMS reported by Lab15.

Table 3.5. Results for IAG OPC-1 and CEM 100a using MC-ICP-MS reported by Lab15.

Sample Name

Dissolution number

Measurement number 1 2 1 2 1 2 1 2 1 2 1 2

corrected conventional

Sr/

Sr value

0.729798 0.729776 0.729790 0.729796 0.729798 0.729781 0.708251 0.708259 0.708261 0.708255 0.708264 0.708253

Standard deviation SD 7.7E-05 7.6E-05 8E-05 7.1E-05 6.8E-05 6.2E-05 6.5E-05 7.3E-05 6.3E-05 7.9E-05 8.4E-05 7.9E-05

Number of measured

isotope ratios

199 190 134 190 180 95 199 190 178 190 180 200

Standard deviation of the

mean SE

5.5E-06 5.5E-06 6.9E-06 5.1E-06 5.5E-06 6.4E-06 4.7E-06 5.3E-06 4.7E-06 5.7E-06 6.2E-06 5.6E-06

Final result (average of all

measurements)

Combined standard

uncertainty

Coverage factor

Expanded uncertainty

Measured NIST SRM 987

Sr/

Sr values

Standard deviation SD for

NIST SRM 987

Number of measurements

N for NIST SRM 987

2.03403E-05

1.2989E-05

2.5978E-05

III

0.708257

0.729790

1.01702E-05

III

OU-6

IAG/CGL 020 ML-3

0.710243

1.3E-05

0.710246

Sample Name

Dissolution number

Measurement number 1 2 3 1 2 3 1 2 3 1 2 1 2 3 1 2 3

corrected conventional

Sr/

Sr value

0.726490 0.726481 0.726459 0.726470 0.726462 0.726477 0.726489 0.726481 0.708122 0.708126 0.708137 0.708133 0.708125 0.708135 0.708133

Standard deviation SD 5.2E-05 5.39E-05 4.9E-05 5.3E-05 5.5E-05 5.6E-05 5.4E-05 6.1E-05 7.2E-05 7.4E-05 5.5E-05 5.8E-05 7.5E-05 6.7E-05 5.5E-05

Number of measured

isotope ratios

62 56 61 65 50 70 70 66 62 70 65 66 70 70 66

Standard deviation of the

mean SE

6.6E-06 7.2E-06 6.3E-06 6.5E-06 7.7E-06 6.7E-06 6.4E-06 7.5E-06 9.2E-06 8.8E-06 6.8E-06 7.1E-06 8.9E-06 8.1E-06 6.7E-06

Final result (average of all

measurements)

Combined standard

uncertainty

Coverage factor

Expanded uncertainty

Measured NIST SRM 987

Sr/

Sr values

Standard deviation SD for

NIST SRM 987

Number of measurements

N for NIST SRM 987

3.2E-05

1.0E-05

0.710300

0.710253

3.4E-05

2.2E-05

1.71521E-05

0.726476

0.708130

III

OPC-1

CEM 100a

1.11732E-05

118

Table 3.6. Results for CEM 200a and CEM 300a using MC-ICP-MS reported by Lab15.

Table 3.7. Results for IAG OU-6 and IAG/CGL ML-3 using MC-ICP-MS reported by Lab15.

Sample Name

Dissolution number

Measurement number 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3

corrected conventional

Sr/

Sr value

0.708245 0.708241 0.708246 0.708250 0.708239 0.708236 0.708244 0.708233 0.708246 0.709325 0.709325 0.709322 0.709326 0.709322 0.709308 0.709334 0.709327 0.709302

Standard deviation SD 7E-05 6E-05 5.2E-05 6.9E-05 5.7E-05 7E-05 6.1E-05 5.3E-05 6.2E-05 5.7E-05 7.1E-05 6.9E-05 6.4E-05 6.3E-05 6E-05 5.6E-05 6.4E-05 5.4E-05

Number of measured

isotope ratios

70 70 70 70 70 70 70 70 70 70 70 70 70 70 70 70 70 70

Standard deviation of the

mean SE

8.3E-06 7.1E-06 6.2E-06 8.3E-06 6.8E-06 8.3E-06 7.3E-06 6.4E-06 7.4E-06 6.9E-06 8.5E-06 8.3E-06 7.6E-06 7.6E-06 7.2E-06 6.7E-06 7.7E-06 6.5E-06

Final result (average of all

measurements)

Combined standard

uncertainty

Coverage factor

Expanded uncertainty

Measured NIST SRM 987

Sr/

Sr values

Standard deviation SD for

NIST SRM 987

Number of measurements

N for NIST SRM 987

1.0E-05

0.710253

2.1E-05

2.5E-05

0.708242

0.709321

III

CEM 200a

CEM 300a

1.03512E-05

1.24624E-05

Sample Name

Dissolution number

Measurement number 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3

corrected conventional

87Sr/86Sr value

0.729786 0.729775 0.729764 0.729786 0.729790 0.729792 0.729695 0.729698 0.729696 0.708233 0.708236 0.708252 0.708236 0.708236 0.708255 0.708240 0.708239 0.708250

Standard deviation SD 4.98249E-05 5.48E-05 5.7E-05 4.28E-05 4.736E-05 5.43E-05 5.4E-05 5.616E-05 5.78E-05 5.8E-05 5.9E-05 6.1E-05 5.9E-05 5.3E-05 5.8E-05 5.7E-05 6.1E-05 4.8E-05

Number of measured

isotope ratios

70 70 60 70 70 51 70 70 40 70 64 70 70 57 70 70 70 70

Standard deviation of the

mean SE

6.0E-06 6.5E-06 7.4E-06 5.1E-06 5.7E-06 7.6E-06 6.5E-06 6.7E-06 9.1E-06 6.9E-06 7.4E-06 7.3E-06 7.0E-06 7.0E-06 6.9E-06 6.8E-06 7.3E-06 5.8E-06

Final result (average of all

measurements)

Combined standard

uncertainty

Coverage factor

Expanded uncertainty

Measured NIST SRM 987

87Sr/86Sr values

Standard deviation SD for

NIST SRM 987

Number of measurements

N for NIST SRM 987

1.0E-05

3.2E-05

0.710253

0.710300

9.02E-05

0.729754

IAG/CGL 020 ML-3

OU-6

2.2E-05

III

0.708242

1.11655E-05

4.50976E-05

119

Figures 3.3 and 3.4 depict the final 87Sr/86Sr isotope ratios of all RMs and associated expanded

uncertainties U, k = 2 vs. technique used (MC-ICP-MS and MC-TIMS) reported by Lab15.

Fig. 3.3. 87Sr/86Sr isotope ratios of four investigated cement RMs and associated expanded

uncertainties U, k = 2 vs. technique used, namely MC-ICP-MS and MC-TIMS. The presented

results are from Lab15.

120

When comparing the measurement results (Tables 3.2 –3.7), a larger spread for IAG OPC-1

(Table 3.2 and 3.5) and IAG OU-6 (Table 3.4 and 3.7) compared to the other reference

materials can be observed. This might be attributed to material heterogeneity but also to

increased difficulties with sample digestion. In this regard, it is also possible that the test

portion/sample size taken as a minimum sample size (100 mg) should be larger for this

purpose, since it suggests sample heterogeneity concerning the 87Sr/86Sr isotope ratio.

Fig. 3.4. 87Sr/86Sr isotope ratios of slate IAG OU-6 and limestone IAG/CGL ML-3 CRMs and

associated expanded uncertainties U, k = 2 vs. technique used, namely MC-ICP-MS and MC-

TIMS. The presented results are from Lab15.

Lab15 provided a straightforward and kind of top-down approach for calculating

measurement uncertainties for 87Sr/86Sr isotope ratio measurements which is based on Eq.

3.1:

𝑢𝑢𝑐𝑐,𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓

𝑆𝑆

𝑐𝑐𝑐𝑐𝑓𝑓(𝑁𝑁𝑐𝑐

87 /𝑁𝑁𝑐𝑐

86 ) 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 =��𝑢𝑢𝑐𝑐,1

𝑣𝑣𝑎𝑎𝑐𝑐𝑢𝑢𝑐𝑐1�2+�𝑢𝑢𝑐𝑐,2

𝑣𝑣𝑎𝑎𝑐𝑐𝑢𝑢𝑐𝑐2�2+�𝑢𝑢𝑐𝑐,3

𝑣𝑣𝑎𝑎𝑐𝑐𝑢𝑢𝑐𝑐3�2+�𝑢𝑢𝑐𝑐,4

𝑣𝑣𝑎𝑎𝑐𝑐𝑢𝑢𝑐𝑐4�2+�𝑢𝑢𝑐𝑐,5

𝑣𝑣𝑎𝑎𝑐𝑐𝑢𝑢𝑐𝑐5�2

(3.1)

121

This uncertainty evaluations were based on the principles as laid down in the Guide to the

Expression of Uncertainty in Measurement. [50] The following five uncertainty components

were included in the calculation below: repeatability of a single 87Sr/86Sr measurement in the

sample, repeatability of 87Sr/86Sr measurement in NIST SRM 987, the bias of the measured

87Sr/86Sr isotope ratio in NIST SRM 987 against the reference value as obtained from the

literature. [37,6] Rcon(87Sr/86Sr) (Sr, NIST SRM 987) = (0.710 250 ± 0.000 001) mol·mol−1, the

experimental reproducibility of independently processed homogeneous samples, and the bias

of the measured 87Sr/86Sr isotope ratio in the processed control sample against the reference

value (e.g., value published in GeoRem database [6]).

3.6.2. MC-ICP-MS and MC-TIMS comparison

Welch's unequal variances t-test [51] was applied to compare MC-TIMS and MC-ICP-MS

separately for each material, using the Welch-Satterthwaite formula to compute the effective

number of degrees of freedom associated with the standard error of each difference.

Effective numbers of degrees of freedom for standard uncertainty of difference between MC-

TIMS and MC-ICP-MS [52] are presented in Table 3.8. After that, test statistics (the

standardized differences between the values measured using MC-TIMS and MC-ICP-MS) and

p-values were calculated and are also shown in Table 3.8.

Table 3.8. Effective numbers of degrees of freedom (d.f.) for standard uncertainty of

difference between MC-TIMS and MC-ICP-MS, test statistics (t-test) and p-values from the

two-tailed tests of the hypotheses of no differences for OPC-1, CEM 100a, 200a and 300a,

IAG/CGL ML-3 and IAG OU-6.

OPC-1

CEM 100a

CEM 200a

CEM 300a

IAG/CGL

ML-3

IAG

OU-6

d.f.

11.1

10.3

12.1

12.7

9.3

12.7

t-test

0.767

0.354

0.541

1.360

0.775

1.000

p-value

0.46

0.73

0.60

0.20

0.46

0.33

The reference distributions for the six tests were Student's t distributions [51] with numbers

of degrees of freedom equal to the effective numbers of degrees of freedom above.

122

3.7. Conclusion

Sr isotope ratios determined for IAG ML3 Rcon(87Sr/86Sr) of 0.708 257 mol·mol-1 (MC-TIMS),

0.708 242 mol·mol-1 (MC-ICP-MS) and the three cement RMs stemming from Germany,

namely VDZ-100a, VDZ-200a, VDZ-300a (0.708 136 (MC-TIMS) and 0.708 130 (MC-ICP-MS),

0.708 250 (MC-TIMS) and 0.708 242 (MC-ICP-MS), 0.709 344 (MC-TIMS) and 0.709 321 (MC-

ICP-MS)) mol·mol-1 respectively, are very close to Phanerozoic seawater signatures (0.709 180

to 0.706 900 mol·mol-1;[10]). In contrast, the IAG OU-6 and the cement RM stemming from

South Africa IAG OPC-1 have high, very radiogenic Rcon(87Sr/86Sr) of 0.729 790 mol·mol-1 (MC-

TIMS), 0.729 754 (MC-ICP-MS) and 0.726 496 mol·mol-1 (MC-TIMS), 0.726 476 (MC-ICP-MS)

respectively.

In order to test the differences in the application of MC-ICP-MS and MC-TIMS for

Rcon(87Sr/86Sr), the Welch's unequal variances t-test [51] was applied. Since the p-values are all

large, no adjustment for multiple testing is needed, and none of the hypotheses of no-

difference between MC-TIMS and MC-ICP-MS is rejected. In other words, no statistically

significant effects attributable to differences between both instrumental techniques for

measuring the 87Sr/86Sr isotope ratios in these RMs could be observed.

All the work which was published before was monitored by quality control materials (QC). QC

enables traceability of the measurement results and helps in verifying the performance of the

instruments, including sample treatment. Moreover, within the ILC, the measurement

trueness was monitored, and estimation of the measurement uncertainties was performed.

Hence, the measurement results of Lab15 proved to be in accordance with the above-

mentioned quality attributes, whilst also being compliant.

123

References

1. Kazlagić A, Rosner M, Cipriani A, Frick DA, Glodny J, Hoffmann EJ, Hora JM, Irrgeher J, Lugli

F, Magna T, Meisel TC, Meixner A, Possolo A, Pramann A, Pribil MJ, Prohaska T, Retzmann A,

Rienitz O, Rutherford D, Paula-Santos GM, Tatzel M, Widhalm S, Willbold M, Zuliani T, Vogl J

(2023) Characterisation of conventional 87Sr/86Sr isotope ratios in cement, limestone and

slate reference materials based on an interlaboratory comparison study. Geostandards and

Geoanalytical Research. doi: https://doi.org/10.1111/ggr.12517

2. Prohaska T (2015) Sector field mass spectrometry for elemental and isotopic analysis. Royal

Society of Chemistry, [Cambridge]

3. Vogl J, Pritzkow W (2010) Isotope reference materials for present and future isotope

research. J Anal At Spectrom 25 (7):923-932

4. Vogl J, Rosner M, Pritzkow W (2013) The need for new isotope reference materials. Anal

Bioanal Chem 405 (9):2763-2770

5. Brand WA, Coplen TB, Vogl J, Rosner M, Prohaska T (2014) Assessment of international

reference materials for isotope-ratio analysis (IUPAC Technical Report). Pure Appl Chem 86

(3):425-467

6. Jochum KP, Nohl U, Herwig K, Lammel E, Stoll B, Hofmann AW (2005) GeoReM: A New

Geochemical Database for Reference Materials and Isotopic Standards. Geostandards and

Geoanalytical Research 29 (3):333-338. doi:https://doi.org/10.1111/

j.1751-908X.2005.tb00904.x

7. Vogl J, Yim YH, Lee KS, Goenaga-Infante H, Malinovskiy D, Hill S, Ren TX, Wang J, Vocke RD,

Murphy KE, Nonose N, Rienitz O, Noordmann J (2019) Certification of ERM-EB400, the

First Matrix Reference Material for Lead Isotope Amount Ratios, and ERM-AE142, a Lead

Solution Providing a Lead Isotopic Composition at the Edge of Natural Variation.

Geostandards and Geoanalytical Research 43 (1):23-37

8. ISO/IEC (2010) ISO/IEC 17043:2010. Conformity assessment — General requirements for

proficiency testing.

9. Koepke A, Lafarge T, Possolo A, Toman B (2017) Consensus building for interlaboratory

studies, key comparisons, and meta-analysis. Metrologia 54 (3):534-562.

doi:https://doi.org/10.1088/1681-7575/aa6c0e

124

10. Bentley RA (2006) Strontium isotopes from the earth to the archaeological skeleton: A

review. J Archaeol Method Th 13 (3):135-187. doi:https://doi.org/10.1007/s10816-006-9009-

11. Rosner M (2010) Geochemical and instrumental fundamentals for accurate and precise

strontium isotope data of food samples: Comment on "Determination of the strontium

isotope ratio by ICP-MS ginseng as a tracer of regional origin" (Choi et al., 2008). Food Chem

121 (3):918-921. doi:https://doi.org/10.1016/j.foodchem.2010.01.019

12. Trincherini PR, Baffi C, Barbero P, Pizzoglio E, Spalla S (2014) Precise determination of

strontium isotope ratios by TIMS to authenticate tomato geographical origin. Food Chem

145:349-355. doi:https://doi.org/10.1016/j.foodchem.2013.08.030

13. Voerkelius S, Lorenz GD, Rummel S, Quetel CR, Heiss G, Baxter M, Brach-Papa C, Deters-

Itzelsberger P, Hoelzl S, Hoogewerff J, Ponzevera E, Van Bocxstaele M, Ueckermann H (2010)

Strontium isotopic signatures of natural mineral waters, the reference to a simple geological

map and its potential for authentication of food. Food Chem 118 (4):933-940.

doi:https://doi.org/10.1016/j.foodchem.2009.04.125

14. Boschetti C, Henderson J, Evans J (2017) Mosaic tesserae from Italy and the production of

Mediterranean coloured glass (4th century BCE–4th century CE). Part II: Isotopic provenance.

Journal of Archaeological Science: Reports 11:647-657.

doi:https://doi.org/10.1016/j.jasrep.2016.12.032

15. Brilli M, Cavazzini G, Turi B (2005) New data of Sr-87/Sr-86 ratio in classical marble: an

initial database for marble provenance determination. J Archaeol Sci 32 (10):1543-1551.

doi:https://doi.org/10.1016/j.jas.2005.04.007

16. Hoogewerff JA, Reimann C, Ueckermann E, Frei R, Frei KM, van Aswegen T, Stirling C, Reid

M, Clayton A, Ladenberger A, Team GP (2019) Bioavailable Sr-87/Sr-86 in European soils: A

baseline for provenancing studies. Sci Total Environ 672:1033-1044.

doi:https://doi.org/10.1016/j.scitotenv.2019.03.387

17. Henderson J, Evans J, Bellintani P, Bietti-Sestieri AM (2015) Production, mixing and

provenance of Late Bronze Age mixed alkali glasses from northern Italy: an isotopic approach.

J Archaeol Sci 55:1-8. doi:https://doi.org/10.1016/j.jas.2014.12.006

18. Nakano T (2016) Potential uses of stable isotope ratios of Sr, Nd, and Pb in geological

materials for environmental studies. Proceedings of the Japan Academy Series B-Physical and

Biological Sciences 92 (6):167-184. doi:https://doi.org/10.2183/pjab.92.167

125

19. Balcaen L, De Schrijver I, Moens L, Vanhaecke F (2005) Determination of the Sr-87/Sr-86

isotope ratio in USGS silicate reference materials by multi-collector ICP-mass spectrometry.

Int J Mass spectrom 242 (2-3):251-255

20. Fourny A, Weis D, Scoates JS (2016) Comprehensive Pb-Sr-Nd-Hf isotopic, trace element,

and mineralogical characterization of mafic to ultramafic rock reference materials. Geochem

Geophy Geosy 17 (3):739-773

21. Jo HJ, Lee HM, Kim GE, Choi WM, Kim T (2021) Determination of Sr-Nd-Pb Isotopic Ratios

of Rock Reference Materials Using Column Separation Techniques and TIMS. Separations 8

(11)

22. Weis D, Kieffer B, Maerschalk C, Barling J, de Jong J, Williams GA, Hanano D, Pretorius W,

Mattielli N, Scoates JS, Goolaerts A, Friedman RM, Mahoney JB (2006) High-precision isotopic

characterization of USGS reference materials by TIMS and MC-ICP-MS. Geochem Geophy

Geosy 7

23. Weis D, Kieffer B, Maerschalk C, Pretorius W, Barling J (2005) High-precision Pb-Sr-Nd-Hf

isotopic characterization of USGS BHVO-1 and BHVO-2 reference materials. Geochem Geophy

Geosy 6

24. Raczek I, Jochum KP, Hofmann AW (2003) Neodymium and strontium isotope data for

USGS reference materials BCR-1, BCR-2, BHVO-1, BHVO-2, AGV-1, AGV-2F GSP-1, GSP-2 and

eight MPI-DING reference glasses. Geostandard Newslett 27 (2):173-179

25. Yang YH, Zhang HF, Chu ZY, Xie LW, Wu FY (2010) Combined chemical separation of Lu, Hf,

Rb, Sr, Sm and Nd from a single rock digest and precise and accurate isotope determinations

of Lu-Hf, Rb-Sr and Sm-Nd isotope systems using Multi-Collector ICP-MS and TIMS. Int J Mass

spectrom 290 (2-3):120-126

26. Chauvel C, Bureau S, Poggi C (2011) Comprehensive Chemical and Isotopic Analyses of

Basalt and Sediment Reference Materials. Geostandards and Geoanalytical Research 35

(1):125-143

27. Kazlagić A, Vogl J, Gluth GJG, Stephan D (2021) Provenancing of cement using elemental

analyses and isotope techniques - the state-of-the-art and future perspectives. J Anal At

Spectrom 36 (10):2030-2042. doi:https://doi.org/10.1039/D1JA00144B

28. Kazlagić A, Russo FF, Vogl J, Sturm P, Stephan D, Gluth GJG (2022) Development of a sample

preparation procedure for Sr isotope analysis of Portland cements. Anal Bioanal Chem 414

(15):4379-4389. doi:https://doi.org/10.1007/s00216-021-03821-7

126

29. Graham IJ, Goguel RL, St John DA (2000) Use of strontium isotopes to determine the origin

of cement in concretes - Case examples from New Zealand. Cem Concr Res 30 (7):1105-1111.

doi:https://doi.org/10.1016/S0008-8846(00)00248-9

30. ISO (2017) General requirements for the competence of reference material producers (ISO

17034:2016). International Organization for Standardization Geneva

31. Trapmann S, Botha A, Linsinger TPJ, Mac Curtain S, Emons H (2017) The new International

Standard ISO 17034: general requirements for the competence of reference material

producers. Accredit Qual Assur 22 (6):381-387. doi:https://doi.org/10.1007/s00769-017-

1285-5

32. JCGM (2012) International vocabulary of metrology – Basic and general concepts and

associated terms (VIM).

33. Nier AO (1938) The isotopic constitution of strontium, barium, bismuth, thallium and

mercury. Phys Rev 54 (4):275-278

34. Steiger RH, Jager E (1977) Subcommission on Geochronology - Convention on Use of Decay

Constants in Geochronology and Cosmochronology. Earth Planet Sci Lett 36 (3):359-362.

doi:10.1016/0012-821x(77)90060-7

35. Habfast K (1998) Fractionation correction and multiple collectors in thermal ionization

isotope ratio mass spectrometry. Int J Mass spectrom 176 (1-2):133-148

36. Hart SR, Zindler A (1989) Isotope Fractionation Laws - a Test Using Calcium. Int J Mass

Spectrom Ion Processes 89 (2-3):287-301

37. Faure G, Mensing TM (2013) Isotopes : principles and applications. 3rd edn. John Wiley &

Sons, Inc. , Hoboken, New Jersey

38. McArthur JM, Rio D, Massari F, Castradori D, Bailey TR, Thirlwall M, Houghton S (2006) A

revised Pliocene record for marine-87Sr/86Sr used to date an interglacial event recorded in

the Cockburn Island Formation, Antarctic Peninsula. Palaeogeography, Palaeoclimatology,

Palaeoecology 242 (1):126-136. doi:https://doi.org/10.1016/j.palaeo.2006.06.004

39. NIST (2007) Certificate of analysis SRM 987 isotopic standard for strontium. National

Institute of Standards and Technology. https://www-s.nist.gov/srmors/certificates/987.pdf.

Accessed 28.09.2022 2022

40. VDZ VDZeV (2020) Data sheet of reference cement VDZ 100a. https://www.vdz-

online.de/en/services/laboratory-services/reference-materials Accessed 28.09.2022

127

41. VDZ VDZeV (2020) Data sheet of reference cement VDZ 200a. https://www.vdz-

online.de/en/services/laboratory-services/reference-materials. Accessed 28.09.2022

42. VDZ VDZeV (2020) Data sheet of reference cement VDZ 300a.

43. IAG (2015) Reference Material Data Sheet. OPC-1 Ordinary Portland Cement.

http://iageo.com/wp-content/uploads/2017/11/OPC-1-RM-information-sheet-21st-Dec-

2015.pdf Accessed 28.09.2022

44. IAG (2020) Certificate of Analysis. IAG CRM 1 - OU-6 (Penrhyn Slate). http://iageo.com/wp-

content/uploads/2020/05/IAG-OU-6-revised-May2020v2.pdf. Accessed 28.09.2022

45. IAG (2016) Certificate of Analysis IAG / CGL 020 – ML-3 (Limestone). http://iageo.com/wp-

content/uploads/2020/08/ML-3-certificate.pdf Accessed 28.09.2022

46. Yokoyama T, Makishima A, Nakamura E (1999) Evaluation of the coprecipitation of

incompatible trace elements with fluoride during silicate rock dissolution by acid digestion.

Chem Geol 157 (3-4):175-187. doi:https://doi.org/10.1016/S0009-2541(98)00206-X

47. Burnett DS, Wasserburg GJ (1967) 87Rb-87Sr Ages of Silicate Inclusions in Iron Meteorites.

Earth Planet Sci Lett 2 (5):397-+. doi:https://doi.org/10.1016/0012-821x(67)90180-X

48. Croudace IW (1980) A Possible Error Source in Silicate Wet-Chemistry Caused by Insoluble

Fluorides. Chem Geol 31 (1-2):153-155. doi:https://doi.org/10.1016/0009-2541(80)90074-1

49. Manhes G, Minster JF, Allegre CJ (1978) Comparative Uranium-Thorium-Lead and

Rubidium-Strontium Study of Saint-Severin Amphoterite - Consequences for Early Solar-

System Chronology. Earth Planet Sci Lett 39 (1):14-24. doi:https://doi.org/10.1016/0012-

821x(78)90137-1

50. (JCGM) JCfGiM (2008) Evaluation of Measurement Data - Guide to the Expression of

Uncertainty in Measurement

51. Welch BL (1947) The generalization of ‘Student's’ problem when several different

population varlances are involved. Biometrika 34 (1-2):28-35.

doi:https://doi.org/10.1093/biomet/34.1-2.28

52. Miller RG (1986) Beyond ANOVA, Basics of Applied Statistics. John Wiley & Sons, New York,

128

Chapter 4

Main results and discussion

This Chapter discusses the principal results presented in Chapter 2 and Chapter 3 and aims to

summarise major findings of the conducted research.

4.1. Literature review and method proposal

The first step of the study included a comprehensive investigation of previous research and

published literature pertinent to the field. The origin determination of cement and its related

materials has thus already been reviewed and the need for a reliable cement provenancing

technique was shown to be necessary. Reviewed research ranging from the time period of

1993-2018, indicated a focus on the use of trace elemental contents of cements and clinkers,

mostly combined with either the “star plot method”, the “gray rational analysis” or other

statistical and pattern recognition methods.[1] Thus, to compare data, an interpretation

requires different visualisation methods such as star graphs (Chapter 2, section 2.1, Fig. 1.1

and Table 1.1, respectively). [1] One of the main disadvantages of these techniques is that

solid statistical knowledge and computational power restrict the number of potential users.

Another disadvantage was that two (Zn, V) of a maximum of eight analysed elements were

proven useless because they originated from a fuel used in clinker production. Additionally,

the published articles mostly analysed clinkers (materials which occur within the production

process before other components are admixed) and not cements. (Chapter 2, section 2.1,

Table 2.)

In contrast, the use of isotope ratios for cement provenancing is scarce and limited to Graham

and co-authors [2] as well as Kosednar and co-authors [3-5], who showed that Sr isotopes,

combined with elemental analysis, can be useful tools for cement provenancing in a restricted

area concerning time and space. It was proven that 87Sr/86Sr isotope ratios do not change

during the high temperature processes which occur in the kiln during clinker formation. [3]

However, sample preparation for Sr isotope determination had to be improved to avoid

interference from mineral additions that did not originate from the clinker, but also in terms

of reducing the time consumption.

129

In the past, mainly one isotope system (isotope ratios of one chemical element) had been used

for provenancing. However, for complex materials (e.g., alloys, glass, cement) which consist

of several components/raw materials of different geographical origin, one isotope system

frequently resulted in large overlaps and insufficient resolution of the measurement space.[1]

Thus, in the previously published review article, a new approach for cement provenancing was

proposed, which in addition to 87Sr/86Sr isotope ratios, also incorporates 143Nd/144Nd isotope

and Ca/Sr elemental ratios. The purpose of the new approach is to provide an unambiguous

provenancing technique, which also helps in resolving such samples which cannot be resolved

using only one isotope system (e.g., if data overlap within the measurement uncertainty).

Thus, the idea behind the proposed approach is that these samples can be resolved by adding

a second isotope system, alongside a third or fourth measurand, whether that be an isotope

ratio, an interelement ratio, or an elemental concentration. The pathway for cement

provenancing was thus presented in form of a scheme (Chapter 2, section 2.1, Fig 2.)

After reviewing the literature research and proposing a new approach for cement

provenancing, the next step was to perform preliminary experiments and develop a suitable

sample preparation method for determination of Sr isotopes in cement.

4.2. Preliminary study

Ordinary Portland cement (OPC) mostly consists of Portland cement clinker. Theoretically, the

87Sr/86Sr isotope ratios of cement and of clinker are expected to be nearly the same. However,

the experimental results showed differing fingerprints. Since cements produced worldwide

contain different additives, the partial influence of these additives can cause variance in the

final Sr isotopic signature of cement. Preliminary analysis of three pairs of clinker and CEM I,

showed that there is a difference in 87Sr/86Sr isotopic fingerprints of clinker and those of the

corresponding CEM I, and that this difference varies for each clinker-CEM I pair (see Fig. 4.1).

When the only additives in cement production are calcium sulphates (CS), their addition,

typically in small quantities (< 5 %) can positively or negatively bias the final product’s ratio,

depending on the isotope ratio and the Sr mass fractions in the clinker and in the CS.

Another possible explanation for obtaining different fingerprints even from the same

production site is the variation of signatures within a specific limestone deposit, also discussed

by Kosednar-Legenstein and co-authors [3], which affects the ratio shown by the difference in

the ten thousandths place after the decimal (0.7075 and 0.7076). Such differences, however,

130

might also be explained by the addition of an unknown compound contributing to the clinker

ratio (for example carbonate aggregate, especially if not from the same limestone deposit,

which is albeit very unlikely).

Fig. 4.1. 87Sr/86Sr isotopic fingerprint of CEM I, clinker, and CS (calcium sulphates such as

gypsum and/or anhydrite) from three different sample sets, namely 3019, 3020, and 3023.

After observing said differences in the isotope fingerprints of the clinker, the cement, and

variations of these differences, it was necessary to develop a suitable sample preparation

procedure for cement provenancing. An ideal procedure needs to fulfil several requirements.

Namely, the procedure must be fast, effective, and low-cost, while also able to remove any

additives in the cement. In first approximation it would return the 87Sr/86Sr isotopic fingerprint

of the clinker only.

4.3. The sample preparation method for Sr isotope ratios in OPC

The outcome of the preliminary study was to develop a sample preparation method for

determination of Sr isotopes in cement.

Therefore, the aim was to determine and compare the 87Sr/86Sr isotope ratios of a diverse set

of Portland cements and their corresponding Portland clinkers depending on the sample

preparation method. For this purpose, four approaches were tested with two approaches

131

employed in order to remove additives from the cements (approach 1 and 2), i.e., to measure

the 87Sr/86Sr isotopic fingerprint solely of the clinker. Namely: 1) treatment with a potassium

hydroxide/sucrose solution and 2) sieving through a 11-μm sieve. 3) Dissolution in

concentrated hydrochloric acid/nitric acid and in 4) diluted nitric acid was employed to

determine the 87Sr/86Sr isotope ratios of the cements and the individual clinkers. The aim was

to find the most appropriate sample preparation procedure for cement provenancing in

regard to Sr isotope fingerprints and the selection was realised by comparing the 87Sr/86Sr

isotope ratios of differently treated cements with those of the corresponding clinkers.

(Chapter 2, section 2.2, Fig.1.)[6]

The results revealed that sieving was not suitable to separate coarser clinker particles from

the finer calcium sulfate particles of cement and therefore the additives could not be removed

before Sr isotope analysis. This was confirmed by Sr isotope analysis on MC-TIMS, as well as

by XRD analysis (Chapter 2, section 2.2, Fig.5 and Supplementary Information, Figs. S1–S3

respectively), which showed that calcium sulfates were still present in the coarse fraction after

sieving.

XRD results confirmed that the KOSH method successfully led to selective dissolution of clinker

phases and cement components (Chapter 2, section 2.2, Fig. 2). The aluminate was almost

completely dissolved, and the calcium sulfates and ferrite were completely removed.

Nevertheless, the average deviation Δabs between the 87Sr/86Sr isotope ratios of the cements

treated with the KOSH solution and the corresponding clinkers (Chapter 2, section 2.2, eq.1)

were the largest of all investigated approaches. (Chapter 2, section 2.2, Fig. 6, and Table 1.)

When it comes to the concentrated acid method, XRD results showed that it led to a virtually

complete breakdown of the initial phases, where all alite-, belite-, aluminate- and ferrite-

related peaks disappeared. Thus, all Sr-bearing compounds in the cement were dissolved

except for insoluble minor constituents (e.g., quartz). Similar results were achieved with the

diluted acid method (Chapter 2, section 2.2, Fig. 3, and Fig. 4 respectively). This method

however was deemed less satisfactory due to a slightly higher average Δabs and a lower

number of samples with an 87Sr/86Sr isotope ratios difference between cement and clinker

below the threshold (Chapter 2, section 2.2, Table 1.)

132

The comparison of all employed methods was presented as the average absolute difference

of the 87Sr/86Sr isotope ratios of the processed cements and the corresponding clinkers in

Chapter 2, section 2.2, Fig. 10.

The average Δabs was lowest for the concentrated acid treatment and the t-test showed that

the difference between the 87Sr/86Sr isotope ratios of the cement samples treated with

concentrated acid and the corresponding clinkers were not statistically significant, while it was

for cements after KOSH treatment (Chapter 2, section 2.2, Supplementary Information, Figs.

S4–S7). The sample preparation with concentrated acid (consisting of concentrated

hydrochloric acid/nitric acid) was therefore deemed the most appropriate sample preparation

method for determination of the 87Sr/86Sr isotope ratios of Portland cement. This method was

then used for the further analyses in cement provenancing presented here.

4.4. The sample preparation method for Nd isotope ratios in OPC

Conversely to Sr, most of the Nd content in cements stems from silicate sources, such as shale

or clay. The sample preparation method for the determination of Nd isotope ratios in cement

comprised a simple and straight forward approach of fully digesting the samples.

In the past, in order to dissolve the silicate matrix and refractory oxides, most digestion

protocols were based on the use of HF. Nevertheless, loss of specific analytes due to fluoride

co-precipitation have been reported, making it difficult to establish multi-elemental extraction

procedures for specific analytes. Furthermore, the acute toxicity of HF requires careful sample

handling, and its use is restricted in many labs causing additional sample preparation steps

like evaporation and re-dissolution. The Nd digestion protocol used in this research utilises

the less dangerous and non-acute toxicity source of fluoride - HBF4. Thus, the risks of handling

HF-containing solutions were greatly decreased, since HF was being generated in situ and

excess fluoride ions were directly complexed by in situ generated H3BO3. In other words, free

HF during digestion either reacted with the sample or was being neutralised by H3BO3 to form

HBF4 again. [7]

After digestion, Nd was purified from the matrix by using column chemistry, as described in

Supplementary Information (Chapter 2, section 2.3, Fig.S2). The Nd column chemistry was

adapted from [8], the columns were calibrated and samples were prepared for measurement.

(see Fig. S2, ESM).

133

4.5. Fingerprinting Portland Cements

To make cement provenance determination possible, a reliable procedure which provides

consistent and unambiguous results had to be established. The aim of developing such a

procedure was to resolve practical issues in damages research, liability issues, and forensic

investigations (Fig. 4.2). To compare and distinguish between two unknown Portland cement

samples, it was therefore necessary to develop an analytical procedure that would allow

cement fingerprinting.

Fig. 4.2. Schematic presentation of cement provenancing via the use of isotope techniques.

For this purpose, as evidenced in Chapter 2, section 2.3, 87Sr/86Sr and 143Nd/144Nd isotope

systems (Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd)) and interelement ratios of Ca, Sr, K, Mn, Mg,

and Ti were used as fingerprints for ordinary Portland cement (OPC) provenancing. Twenty-

nine ordinary Portland cement samples (Chapter 2, section 2.3, Table 1) from cement plants

located worldwide were investigated to determine their elemental mass fractions using XRF

(Chapter 2, section 2.3, Table S1) and Sr and Nd isotope ratios using MC-TIMS (Chapter 2,

section 2.3. Table 2).

134

The study presented in Chapter 2.3 reveals that the Sr isotope ratios of OPCs are higher than

that of seawater from the corresponding geological period. [9] (Chapter 2, section 2.3, Fig.3)

There are different possibilities and therefore explanations why the Sr isotope ratios of OPCs

in this study are higher than Sr isotope ratios derived from seawater of the observed geological

period. One of the possible reasons is radiogenic strontium formation from rubidium decay

occurring in the raw materials due to the presence of radiogenic clay. Furthermore, as

explained by Wedepohl [10], diagenetic alteration could result in these differences. Another

possibility is the addition of an unknown additive during cement production, e.g., gypsum,

that has a substantially higher 87Sr/86Sr isotope ratio, which could in turn affect the ratio of

the final product.

OPCs originating from Germany have quite a large 87Sr/86Sr spread, from 0.7077 – 0.7102

mol·mol-1, with most being close to 0.7080 mol·mol-1 (Chapter 2, section 2.3, Fig. 2), which is

in agreement with the results published in a study by Graham and co-authors [2]. The spread

of 143Nd/144Nd in German OPCs is not significant (0.5120 – 0.5121 mol·mol-1) but improves

resolution of the approach. Nevertheless, the combination of both Sr and Nd isotopic

fingerprints provides the potential for distinguishing between cements of different production

sites.

Three OPCs originating from Iraq form a group (depicted in Fig. 2, Chapter 2, section 2.3),

easily distinguishable from the others due to their high 143Nd/144Nd isotope ratios (0.51237-

0.51241) mol·mol-1. Unlike any other sample, the OPC from Shanghai, China, shows an

extremely high 87Sr/86Sr (0.71046 mol·mol-1) and extremely low 143Nd/144Nd isotope ratio

(0.51186 mol·mol-1), making it easily distinguishable from other groups and OPCs. The

respective pairs of OPCs from Austria, Serbia and Greece did not form groups, most likely due

to the differences in intrinsic properties of raw materials from which their cements were

produced, and consequently, this affected their Sr and Nd isotope ratios. Together with the

OPCs from Italy, North Macedonia, Kosovo, and Bosnia and Herzegovina, they exhibited

generally higher 143Nd/144Nd isotope ratios compared to the samples from Germany, with a

relatively similar spread of 87Sr/86Sr isotope ratios.

Most of the investigated OPCs possess measurable differences in Rcon(87Sr/86Sr) and

Rcon(143Nd/144Nd), which can be used as a valuable analytical tool for establishing their

fingerprints. One of the biggest advantages of the combined isotope ratio approach is that it

135

does not require large sample sizes, unlike other techniques (XRD, XRF). Thus, less than 500

mg of the sample is sufficient for Sr and Nd isotope analysis, which is very useful in case of

forensic investigations where trace cement particles are found at a crime scene. In addition to

the small sample size required for MC-TIMS analyses, another advantage of the provenancing

approach via the use of isotope techniques are the low measurement uncertainties. This

means, it is possible to differentiate between cement samples due to the small uncertainties

provided by the high-quality measurement technique MC-TIMS or MC-ICP-MS.

However, three investigated OPCs from Germany (namely TR-DE02, TR-DE12 and TR-DE14)

and two OPCs from Iraq (XX-IQ01 and XX-IQ02) have non-unique 87Sr/86Sr and 143Nd/144Nd

isotope ratios. Since they overlap within the stated uncertainties, they were not easily

distinguishable from one another by applying this method.

For this case, Divisive Hierarchical Clustering (DIANA)[11] was employed, and geochemical

profiles based on Sr and Nd isotope ratios together with all ratios involving mass fractions of

Ca, K, Mg, Mn, Sr, and Ti were determined. In the applied clustering technique, the

dissimilarity between each pair of cements is the median Euclidean distance over all

geochemical profiles, with 120 combinations involving mass fraction ratios of six selected

elements, namely Ca, K, Mg, Mn, Sr, Ti and Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd) isotope ratios.

The results of this clustering method are shown in Chapter 2., section 2.3., Fig. 5.

By measuring Euclidean distances between cement samples using geochemical profiles based

on DIANA, and cutting the tree horizontally at dissimilarity level 25, four separated clusters (A,

B, C and D branches) under the DIANA tree are formed. However, the reason behind the

formation of these four clusters is partially uncertain and intriguing. Nevertheless, we could

say that most OPCs from the same geological period fall in the same cluster. Samples with ID

no. 2 (CR-DE01), 6 (CR-DE04) and 16 (CR-DE11) (Cretaceous period) are in cluster C. Samples

no. 19 (TR-DE13), 4 (TR-DE02) and 21 (TR-DE14) are in cluster B (Triassic period), and samples

no. 5 (JU-DE03), 9 (JU-DE07) and 11 (JU-DE09) are in cluster D (Jurassic period). However, it

remains unclear why this is not applicable to all samples of known geological background, such

as samples with ID no. 10 (CR-DE08 from Cretaceous, but located in cluster A), ID no. 17 (TR-

DE12) and ID no. 23 (TR-DE15) (Triassic but located in clusters A and D, respectively).

136

When it comes to samples with indistinguishable Sr and Nd isotopic fingerprints, namely

samples no. 1 (XX-IQ01) and 14 (XX-IQ02), they were separated using this approach due to

their belonging in different clusters within the DIANA tree.

The same case occurs with sample no. 17 (TR-DE12), which now can be separated from

samples with ID no. 21 (TR-DE14) and 4 (TR-DE02), and, therefore, distinguished. However,

samples no. 21 (TR-DE14) and no. 4 (TR-DE02) belong to the same cluster, and thus they

cannot be distinguished using this method. These three samples are German OPCs made of

limestone stemming from the same geological period (Chapter 2, section 2.3, Fig. 3). The

samples contained very similar elemental patterns besides having similar Sr and Nd isotopic

fingerprints, making it difficult to differentiate between them. These facts suggest that to

produce samples TR-DE14 and TR-DE02, raw materials that originated from a common

geological area were used, meaning they possess a common geological origin.

4.6. Recycled concrete– An example for fingerprinting future building materials

In view of sustainable development within the construction industry, comprising a vision of

lower CO2 emission and lessening energy consumption, recycling cementitious waste

materials is gaining more attention. Recycling of cementitious materials could be a sustainable

way forward if recycled concrete can produce similar physical properties as that of

ordinary Portland cement. [12] In addition to the twenty-nine OPCs in Chapter 2, section

2.3., the same approach was applied to recycled concrete, seeking a comparison of

applicability of this approach to more sustainable materials occurring within the

construction industry. Therefore, one additional sample made of recycled concrete

(source TU Berlin) was investigated. The measured isotope ratios in this sample were:

87Sr/86Sr = (0.709 009 ± 0.000022) mol·mol-1, 143Nd/144Nd = (0.512 033 ± 0.000015)

mol·mol-1. These results show that the proposed methodology works using cement-

related materials, such as recycled concrete. This is a critical property, because the utilisation

of recycled materials in concretes and as raw materials in cement kilns is expected to

increase.

137

4.7. Quality Control via performing an Interlaboratory Comparison Study

To ensure and maintain the high quality of isotope ratio results, quality control was

performed. This was done by organising and participating in an interlaboratory comparison

(ILC) to characterise Rcon(87Sr/86Sr) isotope ratios in geological and technical reference

materials (RMs) by applying the conventional method for 87Sr/86Sr isotope ratios.

Six different powdered reference materials were used in this study, four cement materials (3

VDZ cements – CEM 100a, CEM 200a, CEM 300a and one Portland cement from South Africa

IAG OPC-1) and two rock materials (limestone IAG ML-3 and slate IAG OU-6) representing two

potential raw materials of cement. Both MC-TIMS and MC-ICP-MS were used to determine

87Sr/86Sr isotope ratios in RMs.

Rcon(87Sr/86Sr) determined for IAG ML3 Rcon(87Sr/86Sr) of 0.708 257 mol·mol-1 (MC-TIMS), 0.708

242 mol·mol-1 (MC-ICP-MS) and the three cement RMs stemming from Germany, namely VDZ-

100a, VDZ-200a, VDZ-300a (0.708 136 (MC-TIMS) and 0.708 130 (MC-ICP-MS), 0.708 250 (MC-

TIMS) and 0.708 242 (MC-ICP-MS), 0.709 344 (MC-TIMS) and 0.709 321 (MC-ICP-MS)) mol·mol-

1 respectively are very close to Phanerozoic seawater signatures (0.709 180 to 0.706 900

mol·mol-1;[13]). In contrast, the IAG OU-6 and the cement RM stemming from South Africa

IAG OPC-1 have high, very radiogenic Rcon(87Sr/86Sr) of 0.729 790 mol·mol-1 (MC-TIMS), 0.729

754 (MC-ICP-MS) and 0.726 496 mol·mol-1 (MC-TIMS), 0.726 476 (MC-ICP-MS) respectively.

In order to compare MC-TIMS and MC-ICP-MS separately for each material, Welch's unequal

variances t-test [14] was applied, using the Welch-Satterthwaite formula to compute the

effective number of degrees of freedom associated with the standard error of each difference.

Since the p-values are all large, no adjustment for multiple testing is needed, and none of the

hypotheses of no-difference between MC-TIMS and MC-ICP-MS must be rejected (Chapter 3,

section 3.6.2, Table 3.8). In other words, no statistically significant effects attributable to

differences between both instrumental techniques for measuring the 87Sr/86Sr isotope ratios

in these RMs could be observed.

When comparing the measurement results (Chapter 3, section 3.6., Tables 3.2 – 3.7),

Rcon(87Sr/86Sr) overlap within the estimated uncertainties for the 3 VDZ cements and slate RM

(IAG ML3). The larger spread for IAG OPC-1 (Chapter 3, section 3.6., Table 3.2 and 3.5) and IAG

OU-6 (Chapter 3, section 3.6., Table 3.4 and 3.7) compared to the other reference materials

138

can be observed. This might be attributed to material heterogeneity but also to increased

difficulties in sample digestion. Therefore, to remove possibility of sample heterogeneity

issues, it is recommended that the test portion/sample size taken as the minimum, should be

larger than 100 mg.

Results within this study were used to verify the performance of instruments, validate

analytical procedures (including sample digestion and Sr separation), calculate measurement

uncertainties, as well as to monitor measurement trueness and assess the quality of Sr isotope

ratio measurements. Furthermore, the results of this lab (Lab15) in comparison to the results

from other participants in this ILC [15] proved that the analytical method performs well and is

fit for its intended purpose.

139

References

1. Kazlagić A, Vogl J, Gluth GJG, Stephan D (2021) Provenancing of cement using elemental

analyses and isotope techniques - the state-of-the-art and future perspectives. J Anal At

Spectrom 36 (10):2030-2042. doi:https://doi.org/10.1039/D1JA00144B

2. Graham IJ, Goguel RL, St John DA (2000) Use of strontium isotopes to determine the origin

of cement in concretes - Case examples from New Zealand. Cem Concr Res 30 (7):1105-1111.

doi:https://doi.org/10.1016/S0008-8846(00)00248-9

3. Kosednar-Legenstein B (2007) Historic mortars and plasters in Austria (Styria). Technische

Universitaet Graz, Graz

4. Kosednar-Legenstein B, Dietzel M, Leis A, Wiegand B, Stingl K, Baumgartner M (2006)

Historical Carbonate Mortar and Plaster-Isotopic and Chemical Signatures. Paper presented at

the European Geosciences Union General Assembly, Vienna, Austria

5. Kosednar-Legenstein B, Dietzel M, Leis A, Stingl K, Wiegand B, Baumgartner M (2007)

Historical carbonate mortar and plaster - Proxies for ancient environments. Geochim

Cosmochim Acta 71 (15):A514-A514

6. Kazlagić A, Russo FF, Vogl J, Sturm P, Stephan D, Gluth GJG (2022) Development of a sample

preparation procedure for Sr isotope analysis of Portland cements. Anal Bioanal Chem 414

(15):4379-4389. doi:https://doi.org/10.1007/s00216-021-03821-7

7. Zimmermann T, von der Au M, Reese A, Klein O, Hildebrandt L, Pröfrock D (2020)

Substituting HF by HBF4 – an optimized digestion method for multi-elemental sediment

analysis via ICP-MS/MS. vol 2611. doi:https://doi.org/10.1039/D0AY01049A

8. Mikova J, Denkova P (2007) Modified chromatographic separation scheme for Sr and Nd

isotope analysis in geological silicate samples. Journal of Geosciences 52 (3-4):221-226.

doi:http://doi.org/10.3190/jgeosci.015

9. McArthur JM, Howarth RJ, Shields GA (2012) Chapter 7 - Strontium Isotope Stratigraphy. In:

Gradstein FM, Ogg JG, Schmitz MD, Ogg GM (eds) The Geologic Time Scale. Elsevier, Boston,

pp 127-144. doi:https://doi.org/10.1016/B978-0-444-59425-9.00007-X

10. Wedepohl KH, Baumann A (2000) The Use of Marine Molluskan Shells for Roman Glass

and Local Raw Glass Production in the Eifel Area (Western Germany). Naturwissenschaften 87

(3):129-132. doi:https://doi.org/10.1007/s001140050690

140

11. Kaufman L, Rousseeuw JP (1990) Divisive Analysis (Program DIANA). In: Finding Groups in

Data: An Introduction to Cluster Analysis. Wiley Series in Probability and Statistics. John Wiley

& Sons, Inc., pp 253-279. doi:https://doi.org/10.1002/9780470316801.ch6

12. Wang J, Mu M, Liu Y (2018) Recycled cement. Construction and Building Materials

190:1124-1132. doi:https://doi.org/10.1016/j.conbuildmat.2018.09.181

13. Bentley RA (2006) Strontium isotopes from the earth to the archaeological skeleton: A

review. J Archaeol Method Th 13 (3):135-187. doi:https://doi.org/10.1007/s10816-006-9009-

14. Welch BL (1947) The generalization of ‘Student's’ problem when several different

population varlances are involved. Biometrika 34 (1-2):28-35.

doi:https://doi.org/10.1093/biomet/34.1-2.28

15. Kazlagić A, Rosner M, Cipriani A, Frick DA, Glodny J, Hoffmann EJ, Hora JM, Irrgeher J,

Lugli F, Magna T, Meisel TC, Meixner A, Possolo A, Pramann A, Pribil MJ, Prohaska T,

Retzmann A, Rienitz O, Rutherford D, Paula-Santos GM, Tatzel M, Widhalm S, Willbold M,

Zuliani T, Vogl J (2023) Characterisation of conventional 87Sr/86Sr isotope ratios in cement,

limestone and slate reference materials based on an interlaboratory comparison study.

Geostandards and Geoanalytical Research. doi: https://doi.org/10.1111/ggr.12517

141

Chapter 5

Conclusions and future recommendations

Around 90 % of the quarried limestone in Germany comes from the Mesozoic era and is

between 65 and 250 million years old. [1] Variations and contrasts in the geological ages of

the raw materials used to produce Portland cements are reflected in 87Sr/86Sr and 143Nd/144Nd

isotope ratios. These isotope ratios can be linked to their respective geographic origins, as

they reflect the mineral composition and geological history of the raw material used for

cement production. [2] Whilst Sr isotopic composition is reflected in the limestone, Nd is

geologically linked to silica and silicate minerals that were used to produce cement. Thus,

143Nd/144Nd in cement are a reflection of the formation age of the silicate material and the

concentration of Sm, which is the parent element. This makes discrimination between sand

derived from the erosion of old, metamorphic rocks and one derived from the erosion of a

younger rock, e.g., granite, possible.

Therefore, the important step for cement provenancing was to develop an analytical

procedure which would enable straight-forward cement fingerprinting. Sr and Nd have their

main occurrence in different components of the raw mix of the cement (Sr – carbonate and

Nd – silicate component). Thus, 87Sr/86Sr and 143Nd/144Nd isotope systems (Rcon(87Sr/86Sr) and

Rcon(143Nd/144Nd)) act as tracers for different raw materials. By measuring both Sr and Nd

isotope ratios, a characteristic isotopic fingerprint of cement can be established. For this

reason, twenty-nine ordinary Portland cement samples from cement plants located

worldwide were investigated for their 87Sr/86Sr and 143Nd/144Nd isotope ratios using MC-TIMS.

Prior to their investigation, sample preparation method had to be developed. For the

determination of 87Sr/86Sr isotope ratios in Portland cement, the sample preparation with

concentrated hydrochloric and nitric acid mixture performed in a clean bench on a hotplate

showed as the most appropriate sample preparation method. On the other hand, for the

determination of 143Nd/144Nd isotope ratios in Portland cement, sample preparation with HBF4

using microwave digestion showed as satisfactory method. Therefore, these methods were

then selected to be used for continued research regarding cement provenancing.

The results revealed that Sr isotope ratios of OPCs are higher than those of seawater from the

observed geological period. The spread of Rcon(143Nd/144Nd) in cements is not as large as the

142

spread of Rcon(87Sr/86Sr). However, most of the investigated OPCs possess measurable

differences in Rcon(87Sr/86Sr) and Rcon(143Nd/144Nd), which serves as a valuable analytical tool

for fingerprinting them. With this, almost all German OPCs had been resolved, except for three

chemically very similar German OPCs. To increase the analytical resolution of the approach,

elemental ratios of OPCs consisting of Ca, Sr, K, Mn, Mg, and Ti were used as additional

fingerprints. Mass fractions of these elements were measured using XRF. Divisive Hierarchical

Clustering (DIANA) was employed, and geochemical profiles based on Sr and Nd isotope ratios

together with all ratios involving mass fractions of Ca, K, Mg, Mn, Sr, and Ti were determined.

DIANA allowed further separation of individual samples. This was done by grouping them into

clusters based on their dissimilarity levels. In other words, samples that were not previously

distinguishable could be successfully distinguished using geochemical profiles and DIANA.

However, when samples, besides having similar Sr and Nd isotopic fingerprints, also consisted

of similar elemental patterns, they were grouped via DIANA into the same cluster, meaning

they remained indistinguishable. Using this methodology, successful fingerprinting was

achieved in 27 out of 29 analysed ordinary Portland cements. Unlike other techniques (e.g.,

XRD, XRF), less than 500 mg of a sample is required for Sr and Nd isotope analysis. Thus, the

developed approach does not demand large sample sizes, which is advantageous for

forensic/crime scene investigations (e.g., in determining origin of cement dust found at a

crime scene). In addition, the differentiation between cement samples is possible due to the

small uncertainties provided with the high-quality measurement technique MC-TIMS. When

it comes to the use of recycled materials, such as recycled concrete, the results showed

distinguishable fingerprint which did not overlap with other fingerprints. Therefore, the use

of the “combined approach” presented in this study can be applied for cements other than

OPCs and cement-related materials, such as recycled concrete. These results are not only

interesting but also of high importance, as the utilisation of recycled materials in concretes

and raw materials in cement kilns is only expected to increase.

When discussing applicability of cement provenancing approach to concrete,

several parameters must be considered. First, when evaluating a similarity between

fragmented concrete samples or simply investigating a single concrete fragment, a

morphological observation, such as colour and texture, should be conducted. Second, to

answer concrete provenance-related questions, the use of different techniques (e.g.,

XRD; XRF etc.) in combination with proposed provenancing approach can be very

beneficial.

143

The selection of analytical techniques varies from case-to-case scenario (e.g., sample size

limitations) and depends on the purpose of concrete investigation itself. Third, a separation

of coarse particles from cement in concrete should be performed with carefully selected wet-

chemistry method. Other possibility is the use of in-situ LA-ICP-MS, whereby the wet-

chemistry sample preparation step is completely excluded. However, for samples which

contain a substantial amount of Rb and Sm, the use of LA-ICP-MS for 87Sr/86Sr and 143Nd/144Nd

isotope ratio analysis is not recommended due to the interferences.

Quality control was achieved by organising and participating in the international

interlaboratory study. Another important objective of the study was the evaluation of

differences in the application of MC-TIMS and MC-ICP-MS. No statistically significant effects

attributable to differences between the instrumental techniques employed for measuring

87Sr/86Sr isotope ratios were observed. The Sr isotope ratios of investigated reference

materials are ranging from low, so-called ‘non-radiogenic’ signatures, to high, very ‘radiogenic’

signatures. Associated measurement uncertainties were calculated using a top-down

approach. All measured 87Sr/86Sr isotope ratios are traceable to the conventional method.

Even though the purpose of the ILC was not to monitor the performance of its participating

laboratories, the results obtained by Lab15 (meaning, the work performed by myself, Anera

Kazlagic) within this study were valid and satisfactory, and thereby demonstrated competency

for obtaining and reporting 87Sr/86Sr isotope ratios.

Based on the obtained research results, the following recommendations for future work are

proposed:

The enlargement of the database for Sr and Nd isotopic fingerprints in cements and related

materials in samples worldwide is necessary. In other words, more chemically and isotopically

defined cements are needed to support provenance determination. Reporting of chemically

and isotopically defined cement is needed to build on the existing database and thus amplify

it. Being very important tools of 21st century, data mining, artificial intelligence and machine

learning can be used in the future to establish new data-driven models. By leveraging these

datasets provided in databases with data-driven models, data analysis can be guided by

automatically solving some implicit patterns and extracting valuable information.

Another future recommendation would be organising and performing interlaboratory

comparison for Nd isotopes in cements.

144

References

1. (Ed.) V (2018) Activity Report 2015-2018

2. Kazlagić A, Vogl J, Gluth GJG, Stephan D (2021) Provenancing of cement using elemental

analyses and isotope techniques - the state-of-the-art and future perspectives. J Anal At

Spectrom 36 (10):2030-2042. doi:https://doi.org/10.1039/D1JA00144B

145

Bibliographic information

•Anera Kazlagić, Jochen Vogl, Gregor J. G. Gluth, Dietmar Stephan (2021): Provenancing

of cement using elemental analyses and isotope techniques – the state-of-the-art and

future perspectives. In Journal of Analytical Atomic Spectrometry, 36, 2030-2042,

Publisher’s version added in chapter 2, section 2.1. DOI:

https://doi.org/10.1039/d1ja00144b

•Anera Kazlagić, Francesco F. Russo, Jochen Vogl, Patrick Sturm, Dietmar Stephan,

Gregor J. G. Gluth (2022): Development of a sample preparation procedure for Sr

isotope analysis of Portland cements. In Analytical and Bioanalytical Chemistry, 414,

4379–4389, Publisher’s version added in chapter 2, section 2.2. DOI:

https://doi.org/10.1007/s00216-021-03821-7

•Anera Kazlagić, Dietmar Stephan, Markus Ostermann, Antonio Possolo, Jochen Vogl

(2023): Fingerprinting Portland Cements by means of 87Sr/86Sr and 143Nd/144Nd Isotope

Ratios and Geochemical Profiles. In Advances in Cement Research, Manuscript

number: ADCR-2023-018. Accepted version added in chapter 2, section 2.3.

146

List of abbreviations

AA Atomic absorption

AAS Atomic absorption spectrometry

AE Atomic emission

BAM Federal Institute for Materials Research and Testing

CEM I Ordinary Portland cement as specified in EN 197–1

CHUR Chondritic uniform reservoir

Conc. acid Concentrated HNO3 and HCl 1:1 v/v

Conc. Concentrated

CS Calcium sulphates

DIANA Divisive Hierarchical Clustering

dil. acid 1 mol·L−1 HNO3

EDTA Ethylenediaminetetraacetic acid

EDX Energy dispersive X-ray spectrometry

EPMA Electron probe microanalysis

ESM Electronic supplementary material

FGD Flue gas desulphurisation

FTIR Fourier transform infrared spectroscopy

ICP Inductively coupled plasma

ICP-MS Inductively coupled plasma mass spectrometry

ICP-OES Inductively coupled plasma mass spectrometry optical emission spectrometry

iCRMs Certified isotope reference materials

IIF Instrumental isotopic fractionation

ILC Interlaboratory comparison

147

INAA Instrumental neutron activation analysis

IRMS Isotope ratio mass spectrometry

KOSH Selective dissolution of clinker phases using KOH/sucrose solution

LA Laser ablation

Lab15 Laboratory ID 15

LAMIS Laser ablation molecular isotopic spectrometry

MC Multi collector

NIST National Institute of Standards and Technology

OPC Ordinary Portland cement

PCA Principal component analysis

PP Polypropylene

Rcon(143Nd/144Nd) Conventional 143Nd/144Nd isotope ratios

Rcon(87Sr/86Sr) Conventional 87Sr/86Sr isotope ratios

REE Rare earth elements

RM Reference Material

SC Single collector

SD Standard deviation

SE Standard deviation of the mean

SEM Scanning electron microscope

Sieving Sieving on 11-μm sieve

SSB Standard-sample-bracketing

TIMS Multi collector thermal ionisation mass spectrometry

XRD X-ray diffraction

XRF X-ray fluorescence

148