X-RAY CT-TECHNOLOGY REVEALING THE EFFECTS OF DENITRIFYING ON POROUS LIMESTONE

Laurenz Schröer Student number: 01202662

Promotor: Prof. Dr. Veerle Cnudde Copromotor: Dr. Tim De Kock

Jury: Prof. Dr. Ir. Nico Boon, Dr. Jan Dewanckele

Master’s dissertation submitted in partial fulfilment of the requirements for the degree of Master of Science in Geology

Academic year: 2016 - 2017

ACKNOWLEDGMENTS

I would like to express my gratitude for all the people who helped me to bring this thesis to a good end. First of all I would like to thank my promoters Tim De Kock and Professor Veerle Cnudde. Their input and expertise in µCT and porous media combined with their extreme creativity, were essential to bring this thesis to a higher level. Thank you to let me join to the Interpore conference and an extra very big thank you for Tim as you always were willing to help, when needed, to answer my many questions and to learn me to use the µCT scanner.

A lot of gratitude goes actually to everyone of Progress. Although Tim was my supervisor, I could ask everything to each one of you, and everybody was ready to help me: Jeroen Van Stappen with my setups, Géraldine Fiers with the microscopy, Maxim Deprez with mercury intrusion porosimetry, Redouane Meftah with the computer programs,… Thank you Stefanie Van Offenwert to share a desk together and to listen to all my thesis struggles and to motivate me afterwards.

As I almost did not know anything about bacteria, I would be literally completely lost with this thesis subject. Thank you Professor Nico Boon to guide me through the world of microbiology explaining it with normal terms and with your creativity, solving complex looking problems very easily. I also thank Jana De Bodt to help me in the laboratories of CMET and to prepare the bacteria for my experiments. This was certainly not an easy task as I was completely unfamiliar with all the techniques at CMET.

I thank furthermore Amelie De Muynck, Ivan Josipovic and all the people of UGCT from the Physics department to help me with the scanner and with the more fundamental questions about the scanner and radiation.

Besides all the professional help that I got, the thesis would also not have been possible without my great classmates, we had an amazing five years together. Although our university adventures has ended, it is certainly not a goodbye and we will stay always in touch.

To all my non-geological friends, thank you to let me forget my thesis now and then and to listen to all my geology facts. Hannes and Ilias thank you for letting me relax sometimes, to explain biology in detail and to travel together. It really refreshed my mind and let me work at my studies and thesis more efficiently. Also an extra thank you to Olav to explain statistics to me and motivate me constantly.

Most gratitude goes however to my parents and my grandparents, who supported me all the time. Thank you for preparing food, when I came home very lately, thank you to be silent and be capable to handle my grumpy mood, especially during exam periods and the last weeks of my thesis. But especially thank you for being there.

Table of Contents

1. INTRODUCTION ...... 1 1.1. State of Art ...... 1 1.2. Objectives ...... 2 2. BACTERIA WITHIN POROUS ROCKS: OCCURRENCE, INFLUENCE AND IMAGING ...... 3 2.1. Bacteria in the subsurface ...... 3 2.1.1. Microbial life and the geosphere ...... 3 2.1.2. Microscopic life within saturated porous media and its relationship to the pore distribution ...... 4 2.2. Bacterial colonization and its effects on porous rocks ...... 7 2.2.1. Bacterial adhesion and attachment ...... 7 2.2.2. Bacterial growth within porous media and their direct effects on the hydraulic conductivity ...... 9 2.2.3. Metabolites of denitrificans and its effects on porous media ...... 10 2.3. Imaging the biosphere on a microscale ...... 12 2.3.1. Bacteria and biofilms...... 12 2.3.2. Bacterial influence on a microscale – MICP, biodegradation and biogenic gas ...... 14 3. MATERIALS ...... 16 3.1. Savonnières limestone ...... 16 3.2. Tabaire stone...... 18 3.3. ...... 19 3.3.1. Basic characteristics ...... 19 3.3.2. Denitrification ...... 20 4. METHODS ...... 22 4.1. General overview about some of the most sophisticated used techniques ...... 22 4.1.1. Mercury intrusion porosimetry (MIP) ...... 22 4.1.2. Flow cytometry ...... 22 4.1.3. High resolution X-ray computed tomography (HRXCT) or micro-CT (µCT) ...... 23 4.2. Experimental setups and procedures...... 24 4.2.1. Characterization of the rocks ...... 24 4.2.2. Flow experiments – Estimation of bacterial adhesion within the porous rocks ...... 25 4.2.3. Radiation experiments ...... 26 4.2.4. Growth experiments ...... 26 4.3. Applied µCT reconstruction and data analysis protocol ...... 28 4.3.1. Reconstruction ...... 28 4.3.2. Data analysis ...... 28 4.3.3. Dataviewer and VGStudio ...... 29 5. RESULTS ...... 31 5.1. Characterization of the micro-environment – Porosity ...... 31 5.2. Flow experiments – Bacterial adhesion within porous limestone ...... 34

5.3. Radiation experiments ...... 35 5.4. Growth experiments ...... 36 5.4.1. Biological activity during the different cycles – Chemical analyses ...... 36 5.4.2. Reaction products – Gas ...... 37 5.4.3. Reaction products – Microbially induced calcite precipitation ...... 47 6. DISCUSSION ...... 50 6.1 Characterization of the micro-environment – Porosity ...... 50 6.2. Flow experiments – Bacterial adhesion in porous limestone ...... 50 6.3. Effects of X-ray radiation on Paracoccus denitrificans ...... 52 6.4. Growth experiments ...... 53 6.4.1. Biological activity – Chemical analyses ...... 53 6.4.2. Reaction products – Gas ...... 53 6.4.3 Reaction products – Microbially induced calcite precipitation ...... 56 7. FUTURE RESEARCH AND CONCLUSIONS ...... 59 8. REFERENCE LIST ...... 61 9.APPENDIX ...... 76 9.1. Bioclogging experiments ...... 76 9.2. Poster with preliminary results, presented at the Interpore Conference (8-11 May 2017, Rotterdam, The Netherlands) ...... 77

1. INTRODUCTION

1.1. State of Art

Bacteria are omnipresent and play a crucial role in the geosphere. Studying these organisms in porous materials became popular as they are vital for emerging subsurface remediation and engineering technologies, including in situ bioremediation of pollutants (e.g. Li et al., 2011), microbially induced calcite precipitation (MICP) to strengthen porous material, seal cracks and to control fluid flow (e.g. Ferris et al., 1997; Cuthbert al., 2012; Tobler et al., 2012; Phillips et al., 2013), soil stabilization (e.g. van Paassen et al., 2010), bioremediation of building stone (e.g. Warscheid and Braams, 2000; Webster and May, 2006) and to capture solid-phase pollutants (e.g. Fujita et al., 2010; Lauchnor et al., 2013).

Previous studies focused on the negative effects of microbial colonization like biodeterioration of natural materials (Danin and Caneva, 1990; Monte, 1993; Ortega-Calvo et al., 1995; Walsh, 2001; Caneva and Ceschin, 2009). Besides their effect on aesthetic appearance of a rock, biodeterioration is related to the development of biofilms (“bacterial colonies”) and the presence of extracellular polymeric substances (EPS). Their effect can however also be beneficial for the substrate. They have the capability to produce MICP, that binds the grains together as cement, potentially inducing bioclogging. This might hypothetically influence the rock strength as it reduces its hydraulic conductivity, that influences its durability. It can induce soil stabilization (van Paassen, 2010), but can also improve the strength and durability of concrete, mortar and natural stone (Van Der Star et al., 2009; Harkes et al., 2010; Erşan, 2016) or even seal cracks within buildings (De Muynck et al., 2008; Qian et al., 2010.; Achal et al., 2013) The bacteria might furthermore physically reduce the hydraulic conductivity by clogging the micropores (bioclogging) (Umar et al., 2016). Bioclogging can for this reason contribute to the rock durability and at is this moment not studied in detail at a pore scale level.

Bacillus sphaericus (Wang et al., 2016) and Bacillus pasteurii (da Silva et al., 2015) were the most prominent species to study MICP. They induce calcium carbonate precipitation through ureolysis. This is very efficient, but it contains some drawbacks, such as the production of the toxic ammonium and the need of oxygen. Anoxic conditions that are present in the deeper part of natural materials and the subsurface, will inhibit bacterial growth. For this reason another MICP pathway should be considered. Denitrification that oxidizes organic carbon by the reduction of nitrate, could be a good alternative: it does not produce toxic by-products and does not need any oxygen. This can be performed by denitrifies like Paracoccus denitrificans, the key species within this research (Erşan et al., 2015; Erşan 2016). Besides MICP it also generates nitrogen gas, that will allow to study the combined effect of MICP and gas within porous rocks. Its production and influence has not been studied yet in detail at this scale.

The interconnections between these organisms and the porous rocks, their habitat, are still poorly understood. A clear relationship between the location and growth of these biofilms together with the petrophysical properties remain till today unresolved at the pore scale. The colonization is influenced by the pore space structure: permeability, porosity, capillary kinetics, but also by the mineralogy and surface roughness. This is further influenced by the pH, climatic exposure, nutrient sources, water content,… (Warscheid and Braam, 2000) and complicated with the changes of all these characteristics over time (Miller et al., 2012). Some processes has been studied extensively in the past, such as MICP. The precise mechanisms however, the role of the organisms, as well as the link between petrophysical properties and microbial colonization remain unknown (e.g. Umar et al., 2016). The main reasons include the size of microorganisms, the difficulty to visualize them in their natural habitat and the huge variability of the microhabitats.

1

1.2. Objectives

The central hypothesis of this research is that the bacterial colonization, biofilm growth, MICP and microbial gas formation depends on the rock microstructure, more specifically on the chemical composition, pore wall roughness and pore network properties (due to e.g. retention and filtration of bacteria or nutrient transport through water in the pores). It further assumes that bacteria influence their environment and such the properties of the rocks. A possible example is a decreased water saturation, that decreases the flow of nutrients in the pores. This thesis won’t solve this big central hypothesis but the knowledge gathered here will be used as a starting point to explore the relationship between microbial colonization of porous sedimentary rocks on the one hand, and their petrophysical properties on the other hand. It will investigate how microbial colonization influences the rock’s permeability, capillarity and mechanical properties through the presence of biofilms and the bacterial production of mineral phases (e.g. calcium carbonate) and gasses (e.g. nitrogen gas). It will be a direct experimental observation that supplement the extensively used theoretical and numerical models to predict biofilm formation and MICP in porous media.

New developments such as high-resolution X-ray computed tomography (HRXCT) micro-CT (µCT) can partially solve the issue of the microscale. µCT is a high resolution non-destructive 3D imaging and analysis technique, that can visualize a huge variety of opaque materials. It can distinguish different phases down to a few micrometers based upon its density and atomic mass (Cnudde and Boone, 2013). It cannot easily resolve the at this moment but it should be possible thanks to some new developments (Iltis et al., 2011; Ivankovic et al., 2016). µCT is however able to resolve the bacterial influence like MCIP and biogenic gas formation.

This research is an introduction of a larger more detailed study, and introduces the complete colonization process of bacteria (Paracoccus denitrificans) within two porous limestones: Savonnières and Tabaire, each with their own unique pore structure. The limestones are a suitable environment for bacteria: the calcium carbonate is a source for inorganic carbon, calcium and has a high buffering capacity (Jones and Bennett, 2014). Colonization starts from bacterial adhesion, physical bioclogging, until their growth and the influence of their metabolism (MICP and biogenic gas). The research aims to detect the challenges and the potential of studying microbial research on a pore scale using 3D imaging techniques, combined with some more classical methods like mercury intrusion porosimetry (MIP), flow cytometry, optical density measurements etc.

Prior to any biological experiments, it is essential to know the microstructures of the two rocks and how this can relate to the bacteria. This is the first goal of this research that is followed by the objective to estimate the potential colonization of Paracoccus denitrificans within these rocks due to adhesion, using flow experiments. In a next stage bacterial bioclogging will be induced, to resolve it with µCT. µCT targets to locate the clogging, and its cause: the bacteria itself, MICP, biogenic gas,….

Studying microorganisms with µCT is contested as many believe the organisms are negatively influenced by radiation. A special radiation experiment intends to solve this question, at least for Paracoccus denitrificans as it will compare the growth of a radiated culture to a normal culture of bacteria. The last and main step of the process targets to provide a framework to resolve biogenic gas and MICP on a microscale with µCT. It will try to visualize the phases, to find the link with the pore structure and intends beyond this thesis to reveal the bacteria itself. Optical microscopy and Scanning Electron Microscopy (SEM) will verify at least the MICP and try to connect the thin sections in 2D and the 3D images. These experiments will test µCT on its limits, especially regarding the MICP. Additionally it will be an exercise to detect the influence of nutrients and pH on these reaction products and to control the biological activity on this scale.

2

2. BACTERIA WITHIN POROUS ROCKS: OCCURRENCE, INFLUENCE AND IMAGING

This chapter illustrates concisely the occurrence of bacteria and other microorganisms in the (deep) subsurface. It emphases on the difficulties to understand the microscopic processes in the underground and the complex relationship with microbial life. The beginning is general but later on it focuses on processes starting from bacterial attachment and colonization that have been studied directly with Paracoccus denitrificans whereas its denitrification and its effects on porous media are the core of this research. Chapter 2 ends with a short overview of previous studies that tried to visualize and image the microorganisms and some microbial processes on porous media.

2.1. Bacteria in the subsurface

2.1.1. Microbial life and the geosphere

Microorganisms inhabit almost every possible environment on the planet and extend the biosphere from the Earth’s atmosphere until several kilometres into crust. Their metabolic diversity and chemical reactivity makes them vital components of the global elemental cycles, ranging from aqueous redox processes to dissolution and precipitation reactions. Since its origin around 4 billion years ago, it shaped our planet and made it habitable for higher lifeforms (Konhauser, 2007). The number of prokaryotes are enormous and estimated to be 4-6 x 1030 with a combined total organic carbon of 60 to 100 % compared to land plants (Whitman et al., 1998).

Microorganisms can inhabit as only lifeform subsurface environments where pore spaces have only the size of a micrometre. The subsurface and other dark biospheres are the largest habitats on the planet, but the most poorly understood (Edwards et al., 2012). The penetration of the Earth by lifeforms is mainly limited by temperature (Pederson, 2000). Bacteria can survive temperatures up to 130°C, indicating that life could exist till about 5 km below the surface (Kashefi and Lovley, 2003; Akob and Küsel, 2011). Their metabolic pathways are similar with those utilized by surface species except the absence of light at depth, inhibiting photosynthesis and the rapid depletion of O2 (Konhauser, 2007). Most microorganisms are found attached to a rock surface scattered in individual cells, or in a group of cells or forming complex biofilm communities (Konhauser, 2007, Jones and Bennett, 2014). Prokaryote abundances decrease with depth. In soils they are the most diverse and reach abundances > 1010 cells/(soil g). Deeper in the subsurface in groundwater ecosystems, sediments and rocks, they range between 10² to 108 cells/(mL groundwater) and between 10² or 108 cells/(gram dry weight sediment or rock) (Akob and Küsel, 2011).

The microbial community is regarded as a dynamic and complex entity influenced almost exclusively by physiochemical (e.g. temperature, salinity, pH, etc.) and biochemical (e.g. nutrient or electron availability, etc.) factors (Madigan et al., 2009), while the mineral matrix is regarded as a simple, largely unreactive, homogeneous substratum. Rock surfaces are however not that simple or homogeneous as every rock type and mineral has its specific characteristics, including buffer capacity, trace element content, mineral surface, etc. These characteristics will each support a specific population on a substratum and research revealed that microorganisms are likely highly dependent on minerals (Jones and Bennett, 2014). Jones and Bennett (2014) assessed with laboratory reactors the impact of mineralogy on the microbial life. They created figure 2.1 that clearly indicates the differences in bacterial communities on different mineral surfaces.

3

Figure 2.1: Dendrogram illustrating the diversity of bacterial communities colonizing various solid surfaces. Dissimilarity increases with branch length in outward directions, the scale bar represents 5 % dissimilarity and each of the carbonate branches is < 2 % (Jones and Bennet, 2014).

Mineral surfaces and the local environment influence bacteria, however bacteria also impact their environment since geological times. They act as a main actor in geological processes, particularly in areas of biogeochemical cycling, element biotransformations, mineral and metal transformations, soil and sediment formation, bioweathering and decomposition (Ehrlich, 1996; Gadd, 2010). They play an important role in weathering as they accelerate mineral dissolution and oxidation reactions. Microbial weathering influences the Earth’s surface tremendously and act as important agent in soil formation. Microorganisms are on the other hand remarkably adapted to do the opposite and form mineral phases (Konhauser, 2007). Representing all the possible effects of microorganisms on the geological subsurface is out of the scope of this thesis. Some effects like Microbially induced calcite precipitation, gas production form the core of this thesis and will be discussed in detail.

Carbonates that are the main rock type for this thesis are very reactive and are influenced tremendously by microorganisms. A significant part is of biogenic origin, of which bacteria play an important role. Microorganisms attack on the other hand carbonates as well, usually by producing organic and inorganic acid (Bin et al., 2008; Gadd, 2010; Jones and Bennett, 2014). Karst environments, as we know today, would for example not exist without the biosphere and these microbe-mineral interactions (Phillips, 2016).

2.1.2. Microscopic life within saturated porous media and its relationship to the pore distribution

Microorganisms inhabit the pores of rocks. Processes on this scale are very complex and for this reason they are poorly understood and under-researched although it is expected to be fundamental in subsurface ecosystems. The research on bacteria within these environment is mainly based on unconsolidated sediments, e.g. alluvial aquifers as information is too patchy on other environments like crystalline aquifers or karst (Eisendle-Flöckner and Hilberg, 2015; Schmidt et al., 2017).

4

Life on a microscale involves magnitudes of less than one millimetre. It includes bacteria, fungi, archaea and protozoa (Schmidt et al., 2017). Microorganisms can be autochthonous or allochthonous. Autochthonous organisms thrives permanently inside the porous media, whereas allochthonous species originated from other environments (Goldscheider et al., 2006). Microorganisms and bacteria can furthermore be planktonic or benthic, whereas most organisms are not permanently the one or the other. The planktonic microorganisms can be free-floating or associated to suspended particles. Benthic microorganisms live attached to a surface e.g. a mineral grain. Figure 2.2 provides an illustration of potential bacterial occurrence in a microscale environment. Since Harvey et al., in 1984, researchers generally conclude that attached bacteria contain most of the biomass and activity in the subsurface, while planktonic cells are mostly inactive subsets of the attached organisms (Lehman et al., 2001; Goldscheider et al., 2006).

Figure 2.2: Schematic illustration of a heterogeneous sand and gravel aquifer with its macro- and microscale habitats for microorganisms (Goldscheider et al., 2006).

Although macroscale factors like geology, climate, soil properties determine the main drivers of the surbsurface ecosystems as described above, the microscale factors may not be neglected. While macro- and mesoscale processes determine the general scene, the timing, number of biochemical reactions and the range of types are influenced by the microscale. Figure 2.3 illustrates with two examples concisely how the different scales are related to each other and the importance of the microscale (Schmidt et al., 2017). The physical environment can be very different on a microscale, due to heterogeneity in grain size distributions, mineral composition, shape and patchy biological reactions, etc. It may result in so-called “multiporosity”, where the pore size distribution leads to preferential flow path of water on different scales. The preferential flow paths can also be created by a blockage due to a gas bubble or even cells. It induces different niches in which different biochemical reactions will take place. The liquid phase is less heterogeneous and some parameters may be nearly equally distributed in space, while others still show important heterogeneities on a microscale (Goldscheider et al., 2006; Schmidt et al., 2017). Figure 2.4 illustrates the different microniches within porous media. Once the microbial cells are thriving, they will influence their immediate surrounding up to a few micrometres by extruding metabolic products such as CO2 and using resources like dissolved oxygen. Some cell types might thrive while others are being outcompetes, making the majority of the groundwater microbial cells inactive, awaiting for improving conditions (e.g. Weaver et al., 2015) (Schmidt et al., 2017).

5

Figure 2.3: Illustration about the translation of microscale features to larger scales, with the example of a contamination within two different zones (oxic vs. anoxic), that might be different stages in degradation of the contamination. X-axis displays roughly the temporal scale and the y-axis the spatial scale (Schmidt et al., 2017).

Figure 2.4: Illustration of the micro pore scale variability and its hydraulic situation (Schmidt et al., 2017).

The microscale variability produces micro-niches for the organisms, but also constraints (Rebata-Landa and Santamarina, 2006; Schmidt et al., 2017). Small pores restrict bacteria movement and activity (Boivin-Jahns et al., 1996; Fredrickson et al., 1997). The diameter of the pore throat diameter should be twice the size of the bacterial cell to allow transport (Updegraff, 1982). Furthermore, in 1985, Jenneman et al., suggested that pore throat size and tortuosity (a measure for the interconnectivity of flow paths along the length of a core) are good indicators for bacterial penetration within rocks, better than permeability. Small pores reduce furthermore cell division over time and nutrient availability, that

6

becomes only possible through diffusion (Boivin-Jahns et al., 1996). It leads to reduced biodiversity and spatial isolation due to no pore connectivity. It also implies that all bacteria present in isolated pores are linear descendants of a bacterium entombed during deposition, which could be millions of years ago (Boivin-Jahns et al., 1996; Kieft et al., 1998).

2.2. Bacterial colonization and its effects on porous rocks

2.2.1. Bacterial adhesion and attachment

Colonization of substrate is a complex process and can be divided in different steps. Bacteria attaches rapidly to a surface that is being submerged in an aqueous environment. Organic and inorganic molecules absorb to the surface, where they form a conditioning film. This layer facilitates the accessibility of a surface to bacteria. It can alter the surface charge, tensions and potential positively and provides anchorage and extra nutrients for the growth of a bacterial community (Kumar et al., 1998; Garrett et al., 2008). The transport of bacteria and molecules to the surfaces involves sedimentation of the large particles, cells or aggregates, whereas the smaller cells or particles (< 1 mm) follows a certain diffusive transport due to Brownian motion. Convective transport occurs during fluid flow and is several orders faster. The bacteria can furthermore also move more freely during more quite conditions. They have some kind of mobility: some have a flagellum, while others can glide (van Loosdrecht et al., 1989; Konhauser, 2007; Palmer et al., 2007).

Bacterial adhesion and attachment is described as colloids retention by DLVO (Derjaguin, Verwey, Landau and Overbeek forces) and steric interactions. The DLVO theory describes the changes in Gibbs energy in function of the distance between the micro-organism and a flat surface (figure 2.5). The theory, describes in its simplest form where steric effects do not play a role, the total free Gibbs energy as the sum between mostly attractive van der Waals attractive and electrostatic forces. Colloids and bacteria can accumulate outward the repulsive barrier in the secondary minimum. Its depth determines the strength of adhesion. When the particles have enough energy to overcome the barrier (or when it is absent as in favourable conditions) they can enter the primarily minimum and attach physically (Hermansson, 1999; Rijnaarts et al., 1999; Tufenkji and Elimbech, 2005; Konhauser, 2007; Garrett et al., 2008; Molnar et al., 2015). The conditions are in general unfavourable as the electrostatic interaction are mostly repulsive as both the bacteria and the surface are negatively charged at neutral pH. The charge originates from the dissociation or protonation of carboxyl, amino or phosphate groups on the bacteria cell surface. This dissociation depends on the pH and the ionic concentration from the surrounding solution (Rijnaarts et al., 1995; 1999; van der Wal et al., 1997; Poortinga et al., 2002).

Figure 2.5: Classic DLVO interactions for favourable (red dashed line) and unfavourable conditions (solid black line) (Molnar et al., 2015).

7

Steric interactions, like cell-surface hydrophobicity can be important and are repulsive or attractive Steric repulsion occurs when the cell-surface molecules are hydrophilic and have no affinity to the substratum. When these molecules have on the other hand an affinity for the substratum, bridging might occur (Rijnaarts et al., 1995; 1999; Palmer et al., 2007). These hydrophobic/hydrophilic interactions together with the more or less negligible osmotic interactions are both included in the extended DLVO theory (van Oss, 1989; Hermansson, 1999).

A fraction of the bacteria reaching the conditional surface will adsorb reversibly. This process is affected by factors such as temperature, pressure conditions, available energy, surface functionality and bacterial orientation (Garrett et al., 2008). The bacteria will attach during this stage to the surface but without physical contact on a finite distance of 5-10 nm (Konhauser, 2007). In general this takes place in the secondary minimum (Van Loosdrecht et al., 1989, 1990). Secondary minimum deposition will be more important for larger cells as the energy barrier and the depth of the secondary minimum are directly proportional to the cell size (Hahn and O’Melia; Tufenkji and Elimelech, 2005).

Figure 2.6 illustrates the effect of the ionic strength of the solution for bacterial adhesion. The ionic strength is a critical factor within the DLVO theory. Its square is inversely proportional to the thickness of the double layer. An increase in electrolytic concentrations compresses the double layer and reduce the repulsive forces (van Loosdrecht et al., 1989). When the ionic strength increases, the energetic barrier decreases and the depth of secondary minimum increases. It brings the cells closer to the surface and enhances adhesion. Short-ranged force, like H-bounding will determine the adhesion strength, once the bacteria have eventually overcome the repulsion and become close to the surface (Van Loosdrecht et al., 1989; 1990; Hahn and O’Melia, 2004; Konhauser, 2007). The opposite is also valid as it is suggested that particles captured within the secondary minimum can be released when the ionic strength is reduced (Litton and Olson, 1996; Hahn and O’Melia, 2007; Tufenkji and Elimelech, 2005). Experimental studies (e.g. Jewett et al., 1995) support this as they indicated an increased attachment on mineral surface in high ionic strength conditions (Konhauser, 2007).

Figure 2.6: Illustration of the interaction between bacterial cell and negatively charged mineral surface in (a) low (≈ unfavourable conditions in figure 2.5) and (b) high ionic strength solutions (≈ favourable conditions in figure 2.5) (Konhauser, 2007).

Other important aspects for the initial adhesion includes the surface charge of both the bacteria and the mineral surface. This is characteristic for different mineral and bacterial species and is pH dependent. It can also be altered by the adsorption of inorganic and organic compounds. Yee et al., (2000) studied among others the effects of pH to bacterial adsorption (Konhauser, 2007). Surface roughness and topography also affects the retention of micro-organisms, as it in general promotes

8

bacterial attachment by increasing the contact area between surface and bacterial cell. The effects however vary tremendously with environmental factors such as cell shape, size and feature dimensions of the roughness (Whitehead and Verran, 2006; Palmer et al., 2007; Song et al., 2015).

A number of reversibly adsorbed cells will finally attach to the surface. Usually an extracellular polymeric substance (EPS) physically connects the cell and the solid, however some microorganisms can deploy specific structures such as fibrils, pili or holdfasts, to anchor themselves (De Weger et al., 1987; Konhauser, 2007). The attachment will occur in the primary minimum and is due to the high energy barrier, considered to be irreversible (Van loosdrecht, 1989; Hahn and O’Melia, 2004). Time is in direct correlation with the number of attached bacteria, as the number of bacterial collisions with the surface increases with time. It takes for some species in concentrated cells suspensions, only 10- 30 minutes to form a continuous monolayer (Characklis, 1973; Fletcher, 1977; Konhauser, 2007).

2.2.2. Bacterial growth within porous media and their direct effects on the hydraulic conductivity

After the attachment and adhesion, the bacteria will colonize the surface by growth and division. This is according to Fletcher (1980) the last step of the accumulation of microorganisms on a surface. Bacteria poorly occur as single isolated organisms, they tend to organize and form aggregates or biofilms. Inside these biofilms, each bacterium exists as an individual. All these cells are interconnected thanks to an extensive EPS network of highly hydrated exopolysaccharides (Costerton et al., 1995).

These microorganisms have a crucial influence on the porous media, in which they grow. Bioclogging is one of the most important effects: The bacteria reduce the hydraulic conductivity and porosity by filling/clogging the pore space (Vandevivere and Baveye, 1992a, c; Baveye et al., 1998). There are two central theories are related to it (Yarwood et al., 2006; Thullner, 2010). The first one states that the biofilms cover the pore walls uniformly (Cunningham et al., 1991), while the second theory suggests that the cells grow in interstitial space, forming large discrete aggregates (Vandevivere and Baveye, 1992c). Bacterial growth also alters the hydraulic properties of a porous medium by the production of capsules, gases, and induces changes in the chemical properties of the solid and liquid phases (Yarwood et al., 2006). Its effect on hydraulic properties is considerable as the general conclusion has been that bioclogging can reduce the hydraulic conductivity by 2 to 4 orders of magnitude (Cunningham et al., 1991; Vandevivere et al., 1992a, 1992b; Thullner, 2010; Iltis, 2013). Several researchers studied the interactions between biology and porous media, especially within soils (Yarwood et al., 2006). Yarwood et al., (2006) studied the effect of microbial growth on hydrological properties of unsaturated media. Others studied the effect of bacterial growth on hydraulic dispersion (Taylor and Jaffé, 1990; Bielefeldt et al., 2002) or addressed the transport of bacteria (Powelson and Mills, 1998; Jewett et al., 1999) and biodegradation of contaminants (Saranaj, 2013; Mathews et al., 2015). (Sand) column experiments are the most popular approach for the previous studies.

The exopolymers, or the EPS can also serve as key clogging agents. The idea that they could block or at least hinder, water flow was already suggested in the 1950s (McCalla, 1950). Vandevivere and Baveye (1992c) suggested that EPS production is not necessary to achieve significant clogging, however under specific conditions it give rise to large reductions of the hydraulic conductivity (Vandevivere and Baveye (1992a). The EPS includes tightly and loosely bound layers, also referred as “capsules” and “slime layers” (Vandevivere and Baveye, 1992a). The capsules are a discrete layer of polysaccharide that is firmly attached, while the amorphous slime is excreted more distantly (Flemming and Wingender, 2010). Xia et al., (2016) and Vandevivere and Baveye (1992a) stated that the loosely bound EPS have a larger contribution to reduce the hydraulic conductivity, while the tightly bound EPS have no significant effect. The “slime” could affect the conductivity by increasing the viscosity of the percolating fluid or enhancing cell adhesion and retention. Furthermore it can also lead to an increase in frictional resistance at the solid-liquid interfaces or it can decrease the effective porosity (Vandevivere and Baveye, 1992a).

9

Microorganisms itself also alter the density and viscosity of the percolating fluid. High cell concentrations may increase these parameters and alter the fluid flow. Bacterial growth reduces furthermore the surface tension or modifies the solid-liquid contact angle. This leads to a decreased liquid retention. The reduction of the surface tension relates to bacterial adhesion and secretions that can behave like surfactants on the gas-liquid interfaces leading to a reduction of the interfacial free energy. Sand grains on the other hand coated by biofilms, makes them more hydrophobic reducing the wettability compared to uncoated grains (Rockhold et al., 2002; Yarwood et al., 2006).

2.2.3. Metabolites of Paracoccus denitrificans and its effects on porous media

2.2.3.1. Microbially induced calcite precipitation (MICP)

Microbially induced calcite precipitation (MICP) is defined as: “The formation of calcium carbonate minerals from a solution due to the presence of microbial cells, biosynthetic products or metabolic activity” (Bosak, 2011, p. 223). It is regarded as “induced precipitation”, because the type of the produced mineral is largely dependent on the environmental conditions (Knorre and Krumbein, 2000; De Muynck et al., 2010a; Bosak, 2011). The carbonate precipitation is governed by: the calcium and dissolved inorganic carbon concentration (DIC), the pH and the availability of nucleation sites. Microorganisms influences the precipitation by altering one of those different precipitation parameters, mostly by increasing the alkalinity of their surroundings or by the secretion of substances that facilitate calcite nucleation. Precipitation increases with the pH, because this increases the concentration of carbonate ions. A lowering of the free energy barrier is required to form the initial nuclei. Bacterial surfaces can concentrate the Ca2+ ions around the negatively charged cell membrane (Knorre and Krumbein, 2000; Hammes and Verstraete, 2002; Bosak, 2011; Erşan, 2016). Bacteria can also precipitate actively by exchanging ions through their cell membrane (Castanier et al., 2000).

Besides the aforementioned parameters, MICP is also governed by environmental factors, like salinity and temperature. A temperature increase leads to a reduced solubility of calcium carbonate. Other important factors include the type of organisms, incubation time and the surrounding medium (Knorre and Krumbein, 2000; Cacchio et al., 2003; Rodriguez-Navarro et al., 2003; Zamarreňo et al., 2009). De Muynck et al., (2013) demonstrated that not only the amount of MICP is governed by environmental factors but also that the temperature and type of microorganisms affect the crystal morphology. Another governing factor, one that will be studied in this thesis, is the effect of porosity on MICP. As the pore structure affects fluid transport it also affects bacteria transport and the distribution of MICP. In 2011, De Muynck et al., found that the biogenic carbonate formation occurred more abundantly and at greater depths in a macroporous limestone.

MICP is a very common mechanism in nature and Boquet et al., already stated in 1979 that most bacteria are capable of producing calcium carbonate. There are several different metabolism pathways possible for MICP and to increase the pH, like sulphate reduction, degradation of amino-acids or calcium oxalate, denitrification, photosynthesis, degradation of urea or uric acid in aerobiosis, the utilization of organic acids and anaerobic methane oxidation coupled with sulphate reduction (De Muynck et al., 2010a; 2011; Bosak, 2011).

MICP has been studied extensively under natural environments and controlled laboratory conditions, but the precise mechanisms and the role of the organisms remains contentious (Umar et al., 2016). However, it is very applicable and studies are very diverse: soil consolidation (Whiffin et al., 2007; Karatas, 2008; Van Paassen et al., 2010), sandstone consolidation and sandstone production (Van Der Star, 2009; Harkes et al., 2010), enhanced oil recovery (Nemati and Voordouw, 2003), surface treatment for cementitious materials, healing of cracks in concrete (De Muynck et al., 2008; Erşan, 2016) and the removal of Ca2+ (Hammes et al., 2003a,b), heavy metals (Warren et al., 2001) or radionuclides (Fujita et al., 2000) from waste streams or ground water.

10

Until now, most studies looked at the urea hydrolysis pathway for MICP. Urea hydrolysis has the advantage that it produces a lot of carbonate in a short amount of time and it can be easily controlled (De Muynck et al., 2011; Erşan, 2016). However, this pathway has certain drawbacks. First of all, the bacteria require O2 as final electron acceptor to initiate and to maintain the microbial activity. However, oxygen is not always present and this will inhibit the microbial activity and the precipitation of calcium carbonate, causing problems in the deeper parts of for example soils and concrete cracks (Erşan et al., 2015; Erşan, 2016). It is also not very environmental friendly because it produces ammonium and ammonia which are toxic for aquatic life (Randall and Tsui, 2002) as these products will enter water sources through liquids and gasses (Erşan et al., 2015; Erşan, 2016). For this reason other pathways, like complete denitrification should be considered. It does not form any toxic products and does not need oxygen. The MICP rate is however 100 to 1000 times slower, but encouraging results have already been retrieved (Erşan et al., 2015; Erşan, 2016).

2.2.3.2. (Nitrogen) Gas production

Bacteria, like Paracoccus denitrificans will produce nitrogen gas (Amils, 2015b). Gas production is a measure of the bacteria growth. For this reason, it can be controlled by limiting bacterial growth factors such as nutrient availability and by environmental conditions like temperature (Sills and Gonzalez, 2001; Rebata-Landa and Santamarina, 2012) or geometrical limitations (Rebata-Landa and Santamarina, 2006).

Nitrogen gas is one of the most common gasses and has for this research several advantages as it is neither explosive, nor a greenhouse gas. Furthermore, its solubility in water is very low, meaning that the bubbles will remain undissolved for long periods (Rebata-Landa and Santamarina, 2012). This will make it possible to quantify the bacterial gas production with µCT very well.

There are three ways of spontaneous gas bubble formation. Firstly, it can form when the external pressure reduces below the vapour pressure of the pure liquid, potentially causing cavitation. Secondly, when the temperature increases to the boiling point, the vapour phase is more stable than pure liquid. Finally, when dissolved gas (if present) becomes supersaturated, due to changing environmental conditions (Lubetkin, 2003).

Bubbles nucleate spontaneously in a liquid phase when the supersaturation threshold is reached. This threshold is for homogeneous nucleation in function of the molecular interaction between the liquid and the dissolved phase. Solid-liquid interfaces, such as mineral surfaces affect heterogeneous nucleation additionally. Such surfaces tend to favour heterogeneous nucleation, which occurs at substantially lower supersaturations (Gerth and Hemmingsen, 1980; Rebata-Land and Santamarina, 2012). Microcavities, irregularities and impurities can act as nucleation centres in porous media (Dominguez et al., 2000). Bubbles form, once the supersaturation is reached, when the pore water pressure approaches the gas pressure (Rebata-Land and Santamarina, 2012).

(Nitrogen) gas bubbles can have a significant effect on the properties of porous media and the liquid. Small bubbles that do not distort the structure, can fit within the pores, where it will only change the compressibility of the pore fluid (Wheeler, 1988). Rebata-Land and Santamarina (2012) concluded that gas bubbles in soils reduces the pore fluid stiffness and the susceptibility to liquefaction. Gas production also might lower the permeability and change fluid flow paths (Van Paassen et al., 2010). Seki et al., (1998) noticed during column experiments a high gas retention as bubbles occluded the water pathways. Furthermore, the gas also altered arrangement of the soil particles. During flow experiments through sand columns, Van Paassen et al., (2010) noticed that the microbially produced bubbles are mobile and tend to flow upward where they are blocking the fluid flow downwards. The bubbles were agglomerating in the pores and when they formed a network, sudden upward gas bursts occurred, causing an irregular flow rate. However, the permeability stayed high enough to allow flow

11

at low pressures. On the other hand, the agglomerating bubbles also lead to pressure increase, causing some cracks in the upper part of the sand column. This did not affect the stability of the sand column. Another sand column experiment by Istok et al., (2007), also showed the migration of the gas upwards, leading to no significant decrease in hydraulic conductivity and only minor reductions in water saturation. These experiments are executed on coarse-grained prepared sediments and may not be valid for other sediment systems (Istok et al., 2007). Grain size and distribution is a key factor in gas retention and its effect on the hydraulic conductivity as fine particles may hinder bubble transport and trap them in the soil matrix (Rebata-Land and Santamarina, 2012).

Gas bubbles on a pore scale and its effects on for example the hydraulic conductivity has been popular in previous researches. However, those studies focused on mud rocks, particularly to study shale gas (e.g. Ambrose et al., 2010), peat (e.g. Kettridge and Binley, 2011; Kettridge et al., 2013) and sandstone (for among other tight gas, e.g. Ghanbarian et al., 2016) or (sand/soil) columns (Seki et al., 1998; Istok et al., 2007; van Paassen et al., 2010). Publications handling pore scale studies on gas and its transport within porous sedimentary (carbonate) rocks are, as far as we know, missing. We can relate our experiments to the ones with the sediments as we assume that the gas bubbles will not cause a lot of displacement when the grains are firmly connected. Furthermore, the porosity distribution, fraction open/closed porosity will be essential to retain the produced gas, as the fine particles can hinder bubble transport in sand columns as well.

2.3. Imaging the biosphere on a microscale

Microorganisms in porous media have been very popular research subjects in the past. Visualizing them in their habitat especially in three dimensions is challenging and research has been quite limited. Mathematical models gave us most information about the e.g. the architecture and distribution of bacteria and its related biofilms. These models needs to be verified and for this reason experimental data, including visualization of among others the biofilm distribution in porous media are essential (Davit et al., 2010; Iltis et al., 2011). This section will give a short overview of previous studies regarding the visualization of the bacteria itself, their biofilms and their influences.

2.3.1. Bacteria and biofilms

The traditional techniques to visualize and monitor biofilms and the bacteria are diverse but mainly focused on surface visualisation in 2D or quasi 2D porous systems. The most traditional methods are: light microscopy, confocal laser scanning microscopy, scanning electron microscopy (SEM), cryo-SEM and environmental-SEM, (Azeredo et al., 2016). Sometimes, these techniques are accompanied by other more sophisticated techniques such as optical fluometry (Beyenal et al., 2004) and spectrometry including attenuated total reflection/Fourier transform infrared spectroscopy (Nivens et al., 1993) and reflectance spectrometry (Wiggli et al., 1999) or even photoacoustic spectroscopy (Schmid et al., 2003). However, these techniques do not really visualise bacteria directly in-situ nor in three 3D, that allows characterisation of biofilm growth, surface architecture and the spatial distribution within porous media (Iltis et al., 2011). As techniques capable to visualise in 3D, Focused ion beam-SEM (FIB- SEM), atomic force and optical coherence tomography (Azeredo et al., 2016) can be mentioned. FIB- SEM can produce 3D reconstructions with the ‘slice and view’ process, where it mills away down to 10 nm slices from the surface, during which it records SEM images (Alhede et al., 2012). The atomic microscope is non-destructive and can reach resolutions till the nanometre and micrometre scale. It generates an image by recording the force between the sample surface and a very sharp probe (Dufrêne, 2002). Optical coherence tomography at least uses near infrared light (Xi et al., 2006) and high frequency ultrasound imaging, with sound waves as source (Shemesh et al., 2007).

12

These aforementioned imaging techniques all have important constraints and limitations (for instance, some of them are not applicable to visualize the pore structure of opaque materials, like rocks). Furthermore, these methods are not well suited to visualize large regions, larger than some mineral grains (Davit et al., 2011).

There are currently two techniques that are well suited to visualise bacteria within porous media: magnetic resonance microscopy (MRM) and µCT. MRM observes molecular dynamics in opaque systems. It observes velocity fields and can detect anomalous fluid flow due to microbial interactions (Potter et al., 1996; Seymour et al., 2004a; b, 2007). However, MRM is limited as it reaches only a resolution of 50-100 µm and needs very long acquisition times (Davit et al., 2010; Iltis et al., 2011).

At this moment, µCT is not capable to visualize the bacteria itself as their attenuation of X-rays is quite low and the attenuation signature is too similar with that of water (O’Donnell et al., 2007; Helliwell et al., 2013). Conventional contrast agents are insufficient for biofilms as they diffuse readily into the biofilms (Iltis et al., 2011). O’Donnell et al., (2007) proposed the possibility to use suitable biomarker stains that are able to attenuate. Microorganisms that naturally accumulate heavy metals might for instance attenuate enough X-rays. This technique is not demonstrated yet (Helliwell et al., 2013) but three different techniques have been proposed to distinguish between the different phases. The first technique involves visualisation of the interface between the biofilm and the surrounding pore water with a contrast agent that attaches to the biofilm surface. Iltis et al., (2011) demonstrated this strategy by resolving the biomass from the water by absorbing silver coated microspheres on the surface of the biofilm. These particles appeared then on synchrotron based CT-images as bright spots. Davit et al., (2010) demonstrated a second strategy, as they used barium sulphate as a contrast agent that do not diffuse or penetrate into the biofilm. This contrast agent has different X-ray attenuating properties and ideally shows the stained porous water and the unstained biofilm (= signal from unstained water). Du Roscoat et al., (2014) proposed a similar strategy but with 1-Choloronaphtalene as contrast agent. The third and last strategy involves an aqueous solution that diffuses and absorbs into the biofilm. Roberts (2009) and Klibert (2015) proposed and demonstrated this technique using a lugol’s iodine solution. Klibert (2015) verified this technique with synchrotron X-ray imaging and succeeded segmentation into low- and high-iodine phases. µCT seems to be used only for final visualization as Davit et al., 2010; Ilits et al., 2011 assume the X-ray radiation is lethal or retard future biofilm growth and silver is for example a biocide

O’Donnell et al., (2007) proposed on the other hand to replace X-rays with neutron radiation to establish tomographic images and visualize the microorganisms. Neutrons have the advantage to interact with the atomic nucleus instead of the electron shell, like X-rays. It enables visualization of the light, poorly detectable elements, that composes microorganisms (O’Donnell et al., 2007). Unfortunately this method does not reach such lower resolutions than 20 µm (Lehmann et al., 2007). Thieme et al., (2003) succeeded to visualize itself bacteria in a suspension with clay particles. They used very small clay flocks (6 µm diameter) and cryo X-ray nano tomography at a synchrotron setup reaching a resolution of about 45 nm, giving this method a high operational cost and important practical constraints (Cnudde and Boone, 2013).

13

2.3.2. Bacterial influence on a microscale – MICP, biodegradation and biogenic gas

Imaging the bacteria and the biofilms is very challenging, but imaging their influence is easier, although µCT remains often unexplored and lots of progress is still possible. Bacterial influence is so fundamental that it can even be monitored with geophysical techniques (Atekwana et al., 2006). The techniques to visualise bacterial influence on a microscale are similar as to visualize the bacteria itself and their biofilms. SEM or optical microscopy have been the most popular ones. Most of them, including SEM or optical microscopy are again only capable to visualize the surface, whereas the deeper part remains unknown. This section will summarize the three main techniques to visualize within opaque materials: MICP, biodegradation and biogenic gas. These three techniques have also been used for the experiments.

Thin sections microscopy and SEM are restricted to the surface but can provide information about the inner part. Unfortunately this requires a destructive sample preparation (Wang et al., 2014). Traditional optical microscopy can be used for MICP (figure 2.7) but also SEM is important as it can reach high resolutions (De Muynck et al., 2010; 2011). Optical microscopy can reveal a newly formed calcite layer from original calcite grains by a dark layer at the interface (figure 2.7) (De Muynck et al., 2010b; 2011). SEM images on the other hand can reveal the newly formed crystals and even the hole inside a crystal, that shows the place where the bacteria was located (figure 2.8) (De Muynck et al., 2007; Erşan et al., 2015). It can furthermore also reveal corrosion textures to study biodegradation (Jones and Bennet, 2014), while imaging gas production with these two techniques is not possible.

Figure 2.7: Thin section of Savonnières untreated (left) and biodeposition treated (right). Notice the large MICP crystals that are indicated by the arrows (De Muynck et al., 2011a; b).

Figure 2.8: SEM images showing Microbially induced calcite precipitation of calcite and aragonite crystals with microbial footprints “holes” (Erşan et al., 2015).

14

µCT is nondestructive and is a main tool for visualization of bacterial influence on porous media as it can image its habitat (the pore structure) in 3D with a high resolution (Cnudde and Boone, 2013; Schmidt et al., 2017). It is an upcoming technique in the study of MICP. As µCT lacks real chemical information, it can however be difficult to distinguish between MICP and matrix . During experiments it can be solved by differential images from before and after bacterial influence (Wang et al., 2014). However, it is possible that these X-rays influence the bacterial activity. µCT proved its potential in several engineering research fields e.g. self-healing concrete research, where it made it possible to reveal the spatial distribution of MICP in detail (figure 2.9) (Wang et al., 2014), but also in studies on natural building stones (De Muynck et al., 2011b) or to monitor biocementation of sandstone (Rong and Qian, 2013). Imaging techniques such as µCT are very high-tech and other more simple traditional are still important to verify µCT images. Besides MICP, µCT can also image and monitor biodegradation. De Graef et al., (2005) demonstrated the in depth changes of porosity of stones and concrete by bacterial degradation by using µCT combined with SEM studies. (De Graef et al., 2005). Gas production at least can be imaged by µCT as well, like for example Wilson et al., (2007) who used this technique to determine the gas distribution and void fraction inside gassy sediments. Kettridge and Binley (2011) determined on the other hand with µT the influence of peat structures on biogenic gas entrainment.

Figure 2.9: 3D rendered view created with µCT, illustrating the MICP (yellow) in three concrete samples after treatment with bacteria (Wang et al., 2014).

15

3. MATERIALS

3.1. Savonnières limestone

The Savonnières stone is a French layered oolitic limestone, which got exploited in a limited area near the village of Savonnières-en-Perthois. This village lies 15 km to the SE of Saint-Dizier in the department of La Meuse (Lothringian) (Camerman, 1957; Dessandier et al., 2000; Fronteau, 2000; Roels, 2000; Lorenz and Lehrberger 2013). It has a Late Jurassic, Tithonian age and belongs to the Oolithe Vacuolaire. This is a stratigraphic unit that also includes the Chévillon, Brauvillier, Combles stone and other variations and outcrops at the boundaries of the Champagne and Lorraine regions (Fronteau, 2000; Fronteau et al., 2010; Derluyn et al., 2014).

The stone is already known since Roman ages and was still popular during Merovian and Karolian times. Later however it lost its popularity until the end of the 19th century (Dusar and Dreesen, 2009), after which it regained its popularity in Belgium and France (Roels, 2000). This stone has also been used in countries such as Germany, The Netherlands, Austria and Czech Republic (Lorenz and Lehrberger, 2013). Originally, Savonnières limestone was firstly utilized in France close to the quarries and rivers which acted as transportation routes. Later on when water and rail transportation improved, the oolitic limestone from Lothringian spread across whole Central-Europe. This was particularly the case after the war of 1870/1871. It might indicate that a part of the restoration payments by France has been completed by stone deliveries (Lorenz and Lehrberger, 2013).

Some examples of the use of Savonnières limestone as building stone in France are the Notre Dame church and la cathédrale de Saint-Dizier and the prefecture of Nancy. Furthermore it was also popular as a replacement stone used for restorations (e.g. the cathedral and the Saint-Etienne church at Beauvais) (Dessandier et al., 2000; Lorenz and Lehrberger, 2013). In Belgium, the Railway station Nord(Brussels), the Leuven Railway station, and many others are constructed with Savonnières (Roels 2000).

Savonnières limestone (figure 3.1a) is a light beige limestone with a pitted texure that is considered as a grainstone according the Dunham (1962) classification and classified as a rounded oosparite according to Folk (1962). It consists almost entirely out of pure calcite (70 % sparite, 30 % micrite) and can contain dolomite as accessory mineral (Fronteau, 2000; Blows et al., 2003; Fronteau et al., 2010). It is a partially calcite cemented limestone, dominated by spherical to elliptical ooids with an average diameter of 500 µm and up to 6 mm. Some are intact, but most are poorly preserved. Bivalve fragments are furthermore also present, preserved their original shape and can get concentrated (more than 50 vol%) in zones (figure 3.1b). (Roels, 2000; Roels et al., 2000; Dewanckele et al., 2014; Lebedev et al., 2014). The stone is heterogeneous as some zones contain a lot of ooid fragments, while others exists solely out of ooids. The bivalves have a distinct orientation creating a certain degree of layering. The ooids are formed through calcium carbonate precipitation in concentric layers and/or radial crystal growth of calcium carbonate cement around a pre-existing grain (nuclei). Deposition occurred in warm shallow waters, in regions with an arid or semi-arid climate. This climate could induce supersaturation of calcium carbonate in the water (Flügel, 2004). The conditions in this case seemed to be rather moderate than high energetic. The nuclei include peloids and/or loosely lithified carbonate sediment. A large percentage of the nuclei has been leached out, leaving holes in the centre of the several ooids. However some intra-ooid pores have been partially to near-completely filled with micritic (carbonate mud) or fine peloidal sediment. Diagenesis has occurred in the following order: first precipitation of ooids, then micritisation, local compaction, cement formation and at least dissolution of aragonite (Lebedev et al., 2014). The cement is isopachous bladed sparite. Length of 100 µm are very common and it does not fill the pores completely, leaving a large and open interconnected pore structure (Dewanckele et al., 2014).

16

a b 2

1

3

4

Figure 3.1: Savonnières limestone (a) macro sample (cm-scale) (b) cross section under cross-polarized light (white bar = 500 µm) resolving the heterogeneous porosity with ooids (round) and bivalves (elongated), and bladed sparitic crystals: (1) intergranular micropores of the ooids, (2) intergranular micropores between the sparitic crystals, (3) intergranular macropores, (4) intragranular macropores.

The porosity is very complex and can be divided in primary and secondary porosity where the primary porosity refers to any porosity at the end of the depositional process. The secondary porosity developed as a result of diagenesis. The water saturation method using the European EN1936 norm determined a porosity between 30.88 and 37.75 with a mean of 32.94 %. This was based upon fourteen cylindrical samples of around 14 x 6 mm. Table 3.1. gives an overview of the porosity based on literature and reveals a certain degree of variability. Four types of porosity can be found in Savonnières limestone (figure 3.1b) whereas only the moldic porosity is the result of diagenesis. The first type is intragranular macroporosity of the ooid walls (main pore size 0.1 and 0.7 µm). Furthermore also intergranular microporosity occurs in between the sparitic calcite crystals (mean pore size of 12 µm) and triangular shaped macropores in between the sparitic cement (main pore size of 87 µm). At least also moldic porosity is present, obtained by the dissolution the ooids’ calcite (mean pore size of 87 µm) (Roels, 2000; Roels et al., 2001; 2003; Derluyn et al., 2014). These hollow cores mostly occurred in the facial pellets. They have a high Mg content and dissolves while the cortex (low Mg content) and sparite cement remained unaffected (Dewanckele et al., 2014). Figure 3.2 shows the wide range of pore size distribution and their contribution to the pore volume. The larger pores however are only connected via the smaller micropores. This makes from the macropores isolated regions, that are only accessible via the smaller pores (Roels et al., 2001; 2003; Derluyn et al., 2014).

Figure 3.2: Pore volume distribution of Savonnières limestone of the four different pore types, described in figure 3.1 (same legend) (Derluyn et al., 2014).

17

Table 3.1: Apparent density, porosity and capillary moisture content according to (1) Roels et al., 2000; (2) Pro Roc, 1998; (3) Noël, 1970; (4) WTCB, 1997; (5) values retrieved by water immersion porosimetry of fourteen cylindrical Savonnières samples of 14 mm x 6 mm. (1) (2) (3) (4) (5) Apparent density (kg/m³) 1660.5 1600-1800 1666-2078 1606-1717 1818 Porosity (%) 38.19 30-40 22.8-37.7 36.5-40.7 32.94 Capillary moisture content (%) 15.96 / / / /

3.2. Tabaire stone

The Tabaire stone is a calcarenite (table 3.2), retrieved from quarries 5.5 km west of Cartagena (Spain) in the vicinity of Canteras. It is formed during the latest of the Miocene and has been extensively used by the Carthaginians and Romans. A significant part of historical Cartagena was built with it, such as the Roman theatre (1st Century BC) and the Punic Wall (3rd Century BC). In modern times it has been utilized as well for example in the old hospital Marina (18th Century) and the main building of the Polytechnical University of Cartagena (Arana et al., 2003; Lanzón and Piñero, 2012; Lanzón et al., 2014).

Compared to the Savonnières limestone, Tabaire stone was poorly studied and literature is scarce. It is a yellowish calcarenite, that has a coarser texture (figure 3.3a). The stone microstructure consists out of closely packed grains of micrite (<4 µm) and sparite (4-30 µm) with quartz grains embedded in the stone matrix. It is fossil rich and contains many different foraminifera but also skeletal fragments of gastropods, bivalves and echinoid fossils. The macro pores have an angular yet regular form (figure 3.3b). µCT porosity determined by Lanzón et al., 2014 is 16.41 % of which 15.71 % open and only 0.70 % closed. The water saturation method that followed European EN1936 norm found however a porosity between 30.95 and 35.24 % based upon three 12 x 6 mm cylindrical samples. It indicates that around 50 % of the pores needs to be micropores that are not visible with µCT.

a b 3 4

2

5 1

Figure 3.3: Tabaire stone (a) macro sample (cm-scale) (b) cross section under cross-polarized light (white bar = 500 µm) resolving micritic calcite (1) and quartz (2) grains, (3) foraminifera, with (4) intergranular macroporosity and (5) microporosity.

Lanzón et al., 2014 executed also an X-ray fluorescence (XRF) and X-ray diffraction (XRD) analysis (table 3.21) that found besides Ca and C which indicates calcium carbonate, the presence of high concentration of Si, linked to the quartz and some Mg which often indicates dolomite. XRD confirms this but identified also muscovite, chlorite, ankerite and minor amounts of gypsum and halite (Lanzón and Piñero, 2012; Lanzón et al., 2014).

18

Table 3.2: Chemical composition of Tabaire stone (Lanzón et al., 2014).

3.3. Paracoccus denitrificans

3.3.1. Basic characteristics

Taxonomy: Bacteria; ; ; Rhodobacterales; ; Paracoccus; Paracoccus denitrificans Beijerinck and Minkman (1910) (Pujalte et al., 2014).

Paracoccus denitrificans is the type species of the genus Paracoccus. It got firstly described by Beijerinck and Minkman in 1910 as Micrococcus denitrificans, but got replaced in 1969 in the new genus of Paracoccus by Davis et al., It is a Gram-negative, non-spore forming, non-pigmented species, belonging to the order of the Alphaproteobacteria (Pujalte et al., 2014). They form short rods (coccobacilli) with average dimensions of 0.6 by 0.85 µm in stationary and 0.8 by 1.2 µm in exponential growth phase. They form white or cream coloured colonies and fissions transverse binary. Paracoccus denitrificans is immobile, containing no pili or flagella, but is covered with a fibrous material of largely extracellular polyanions and acidic mucosubstances (Baird-Parker, 1965; Kocur et al., 1968; Davis et al., 1969; Nokal and Mayer, 1979; Nokal and Schlegel, 1983; Ludwig et al., 1993, Kelly et al., 2006; Pujalte et al., 2014).

Paracoccus denitrificans is widely distributed and has been extracted from garden soil, arable soil, field soil, sewage sludge, meadow soil, horse manure, cow dung, canal and well mud (Nokhal and Schlegel, 1983; Kelly et al., 2006). It can grow within a temperature range of 8 to 40°C and pH values of 6 to 10. Optimum growth occurs in media with a pH of 7.6 and at a temperature of 36°C. Furthermore, it is also tolerant to oxygen concentrations up to 65 vol% and to about 7 % mass of NaCl (Nokal and Schlegel, 1983; Kelly et al., 2006).

The bacteria obtain their energy by the process of respiration, this is the process where the organism - oxidizes different substrates (Montero, 2015). Paracoccus denitrificans can use oxygen, nitrate (NO3 ), - nitrite (NO2 ) or nitrogen oxide (N2O or NO) as terminal electron accepter, making it capable for both aerobic or anaerobic respiration. Metabolism was furthermore never fermentative (Davis, 1969; Nokal and Schlegel, 1983; Ludwig et al., 1993; Kelly et al., 2006; Pujalte et al., 2014).

19

Anaerobic growth is supported by the respiratory reduction of nitrate, nitrite or nitrogen dioxide. This creates nitrogen gas as end product, however nitrogen oxide can also be produced as an intermediate and as an end product. Ammonium, nitrate, urea and glutamate can be used as nitrogen sources (Kelly et al., 2006). Aerobic denitrification however has also been observed (Davies et al., 1989).

Paracoccus denitrificans has versatile metabolic systems that can satisfy its energy demand in very different ways, illustrating the great adaptability of these bacteria. It is furthermore also a facultative chemolithoautroph: it obtains its energy from inorganic energy sources and by the oxidation of reduced inorganic compounds, found in the soil (Ludwig et al., 1993; Pujalte et al., 2014; Amils, 2015a). For Paracoccus denitrifcans these compounds can be thiosulfate, hydrogen (Robertson and Kuenen, 1983, Ludwig et al., 1993, Kelly et al., 2006) or iron (Kumaraswamy et al., 2006). Since Paracoccus is a facultative chemolithoautotroph, it can also obtain its energy heterotrophically, hence the word ‘facultative’. In this case they use reduced organic compounds as a carbon source and inorganic compounds as an energy source (Amils, 2015a). Paracoccus denitrificans is furthermore also a facultative methyloptroph, since it is able to oxidize reduced carbon compounds, containing one or more carbon atoms, but no carbon-carbon bonds. It is able to grow on methanol, methylamine, formate and formaldehyde as sole carbon and energy source (Bamforth and Quayle, 1978; Harms and Spanning, 1991, Kelly et al., 2006).

3.3.2. Denitrification

The main respiration pathway of Paracoccus denitrificans is denitrification, that will be important for the scope of this thesis. This is the process where nitrate reduces to nitrogen gas through intermediate products:

- - NO3  NO2  NO  N2O  N2

Denitrification completes the nitrogen cycle, by returning nitrogen to the atmosphere. This process is mostly performed by heterotrophic bacteria, however autotrophic denitrifiers have been identified (Amils, 2015b). This respiratory process requires four enzymes, which produces three intermediates prior to nitrogen gas. Each of the different steps, are shown in more detail in table 3.3 (Zumft, 1997; Karatas, 2008; Shapleigh, 2006, 2013, Amils 2015b). Every enzyme has a different localization, life time, regulatory mechanism, kinetics and sensitivity to inhibitory factors (Karatas, 2008).

The reduction of nitrate to nitrogen gas is a complete denitrification. However, this is less frequent than partial denitrification and generally several bacterial species are involved for a complete denitrification process. (Shapleigh, 2006, 2013; Amils, 2015b). Paracoccus denitrificans is a complete denitrifier and let accumulate only an insignificant amount of nitrite. It can denitrify efficiently and even up to 2 M of nitrate (Chakravarthy et al., 2011). This is beneficial because nitrite accumulation is undesired and can result in a reduction of conversion rates. Furthermore, it is also of environmental concern because increased concentrations of nitrite may lead to the production of nitrosamines, some of which are carcinogenic (Lijinski, 1977). Drinking and groundwater in the EU may only contain at maximum 0.50 mg/L nitrite, while this is 50 mg/L for nitrate (European Commission, 1991; Council of the European Union, 1998; European Union, 2006). Nitric oxide and nitrogen dioxide are on the other hand strong greenhouse gasses (Bouwman, 1989). The end products of this metabolism however, are nitrogen gas and carbon dioxide which are harmless (Karatas, 2008; Van Paassen et al., 2010).

Oxygen is the most widely used terminal oxidant for respiration, because it provides the most energy. However when Oxygen becomes limiting, the respiration of many bacteria will switch to other terminal electron acceptors, like nitrogen (Shapleigh, 2006, 2013). For a long time, it had been accepted that denitrification only occurred in anoxic conditions, but currently aerobic denitrification has been proven for some species, even for Paracoccus denitrificans (Davies, 1989; Robertson and Kuenen; 1984). This

20

does not change the fact that high dissolved oxygen concentrations have a negative effect on denitrification (Rajab et al., 2016). Denitrification is never an essential physiological trait as denitrifiers are also capable, with one or two exceptions, to respire oxygen gas (like Paracoccus denitrificans) (Shapleigh, 2006, 2013).

Table 3.3: Reduction of nitrate to nitrogen gas, showing the different enzymes and their cofactors (Karatas, 2008).

Several factors like pH, temperature, nitrate concentration and oxygen concentration (see above) affect denitrification and potential nitrite accumulation (Rajab et al., 2016). Studies indicated that the rate of denitrification continuously increased with the temperature as all the metabolic activity of microorganisms and participation of enzymes enhances at high temperature (Rajab et al., 2016). pH plays a major role in nitrate reduction and a pH outside the ideal pH range can lead to the accumulation of intermediates (Lee and Rittmann 2003). The literature is not that clear about it and gives different results probably depending on the applied set up (Rajab et al., 2016), however a low pH stimulates nitrate reduction because of an increase in proton accessibility (Di Capua et al., 2015). To give an idea about the absolute numbers: according to Lee and Rittmann, (2003) the optimal pH range for autotrophic denitrification is around 7.7 and 8.6 with a maximum efficiency at 8.4, while for heterotrophic denitrification this lies between 7 and 8. Paracoccus denitrificans is still a complete denitrifier at a pH of 7.4 (Chakravarthy et al., 2011), which is in the range of our experiments. Nitrate concentration has also an important effect, because as a high nitrate content (but also calcium overloading) can inhibit one of the reduction steps and can thus induce nitrite accumulation (Torrentó et al., 2010; van Paassen et al., 2010).

21

4. METHODS

Several methods have been used to determine the effects of bacteria on porous sedimentary rocks. µCT was by far the most important tool and most time went to the acquisition and processing of the images. This chapter will firstly describe the main methods to acquire the results. Afterwards it will give an overview of the three independent types of experiments, which we will call “flow experiments”, “radiation experiments” and “growth experiments”. It will describe the different setups and the acquisition of the data, followed by more information about the analyses. Every Savonnières and Tabaire sample is cylindrical of about 12 x 5.85 mm, these samples and all the data is stored at the Geology Department of the UGent (Krijgslaan 281 (building S8), 9000, Ghent, Belgium)

4.1. General overview about some of the most sophisticated used techniques

4.1.1. Mercury intrusion porosimetry (MIP)

MIP can estimate the pore throat size distributions of porous materials. Mercury needs, as it is nonwetting, an external pressure, to intrude capillaries. During MIP mercury intrudes a porous material due to a progressive applied pressure. The volume of mercury intruding the porous sample at each pressure increment determines its pore-size distribution, while the total intruded volume determines its total porosity. MIP detects pore (throat) radii between 1 mm and 2 nm depending on the applied pressure (Diamond, 2000; Cnudde et al., 2009; Berodier et al., 2016). The raw data delivers no direct information about the pore size distributions, implementing the necessity of a model, usually with the Washburn equation (Diamond, 2000; Cnudde et al., 2009; Berodier et al., 2016):

2훾 cos(휗) 푃 = − (Washburn, 1921) 푟

With: P = Pressure 훾 = Surface tension ( = 0.485 N/m at 25°C for Mercury) θ = Contact angle (usually = 140°C for mercury) r = Radius capillary

MIP and this model are straightforward; however several assumptions may not be forgotten. The model assumes a constant contact angle and surface tension for mercury. It assumes cylindrical pores which are equally accessible by the surrounding mercury (De Las Cuevas, 1997; Diamond, 2000; Berodier et al., 2016). Those assumptions are not valid for rocks as the pore space of natural rocks is more tortuous and consists of large pores interconnected with small throats (Andriani and Walsh, 2003). Larger pores are often only accessible through smaller ones, creating a wrong representation of the pore throats ("ink bottle effect") (Diamond, 2000; Berodier et al., 2016). It will especially problematic for Savonnières that has a very complex pore network (Roels, 2000; Roels et al., 2001; 2003; Derluyn et al., 2014). MIP can also not give information on closed pores, nor it can determine the pore connectivity in detail (Cnudde et al., 2009).

4.1.2. Flow cytometry

Flow cytometry depends on the basic laws of physics, especially on fluidics, electronics and optics (Watson, 1999). Flow cytometry determines the characteristics and phenotypes of cells as they move in a liquid through a laser. The system measures the light-scattering and colour-discriminated fluorescence of the particles, based upon the size, granularity of the cells and whether they carry fluorescent molecules (Brunsting and Mullaney, 1974; Macey, 2007). Staining cells may facilitate differentiation regarding cell types or identifying the presence of receptors and antigens, pH, membrane potential, DNA and enzyme activity (Macey, 2007).

22

Flow cytometers are capable to make several measurements of multiple parameters on every cell. Most commercial systems can do five or more, while specialised equipment can even reach to eleven parameters (Bigos et al., 1999, Macey, 2007). The flow cytometry consists out of three functional units: an optical system with one or more laser light sources, a sensing system comprising the sample/flow chamber and the optical assembly. A flow cytometer contains also a hydraulic system to control the flow of cells through the sensing system and a computer system to collect and analyse the data (Macey, 2007). Flow cytometry analyses every particle individually, but the interpretation happens collectively. A histogram represents the collective data with a single-, two- or three-parameters. Single-parameter histograms are 2D graphs with the parameter of interest on the x-axis, while the y-axis represents the number of events (Macey, 2007).

4.1.3. High resolution X-ray computed tomography (HRXCT) or micro-CT (µCT)

µCT is a widely used technique in geosciences. This section will give a summary of this method including the important elements for this research. Ketcham and Carlson, (2001), Cnudde et al., (2006) and Cnudde and Boone (2013) give a more detailed overview of this method and its applications.

4.1.3.1. Principles

µCT is a powerful non-destructive tool to analyze and visualize objects in three dimensions. It allows to investigate the internal structures of porous geomaterials. There are two different types of setups for µCT: a lab-based and a synchrotron setup (Cnudde and Boone, 2013). Figure 4.1 illustrates a lab- based setup with as main components: a microfocus X-ray tube, rotational stage with the sample and an X-ray detector.

Figure 4.1: Schematic illustration of a typical lab-based µCT setup, with a conical X-ray beam that allows geometrical magnification (Cnudde and Boone 2013).

An X-ray tube produces the X-rays. X-rays will penetrate the sample on the rotational stage in varying degree. They interact with material and attenuate (removal of X-rays by absorption or scattering). The attenuation depends on the material density and the mass attenuation coefficient, that is proportional to atomic number of the material. A detector measures the transmission of a narrow X-ray beam through a sample. The amount of X-rays that will reach the detector depends on the initial amount of X-rays, the attenuation and the thickness of the sample. Measuring the transmission of X-rays out of one directions will create a 2D radiograph. Multiple projections from different directions, makes it possible to reconstruct a 3D volume. (Ketcham and Carlson, 2001; Cnudde et al., 2006; Vlassenbroeck, 2010; Cnudde and Boone, 2013).

4.1.3.2. Image reconstructions and CT-analysis - Theory

Image reconstruction retrieves the linear attenuation coefficient at each point and using dedicated algorithms (Michael, 2001; Vlassenbroeck, 2010; Cnudde and Boone, 2013). Specialized rendering software such as VGStudio allows a visual inspection of the reconstructed 3D volume. A dedicated software package such as Octopus Analysis can furthermore obtain quantitative information when it

23

analyses the complete 3D volume. This information includes for geosciences the overall texture of a material, component volume fractions, grain and pore size parameters and morphology,… (Cnudde and Boone, 2013). The resolution of those acquired images reaches micrometre scale by a standard cone-beam µCT and is determined by the focal spot size, the resolving power of the detector and the magnification (Vlassenbroeck, 2010; Cnudde and Boone, 2013).

The reconstructed images are divided in voxels, the smallest feature that can be distinguished. Smaller volumes cannot distinguished but they contribute to the grey value of the voxel (partial volume effect). It is an image artefact that reduces the accuracy of the data analysis. It blurs the material boundaries and can make quantitative interpretation problematic. On the other hand, it gives us the opportunity to extract unexpectedly fine-scale details from µCT images (Ketcham and Carlson, 2001; Cnudde and Boone, 2013). µCT may be furthermore subjected to noise (Cnudde and Boone, 2013) and the operator dependency, as there no general accepted protocol to analyse all different kind of samples exists (Cnudde and Boone, 2013). The X-ray loses furthermore preferentially the lower part of its spectrum. It causes the edges of the object to appear brighter than the centre. An artefact called beam hardening (Ketcham and Carlson, 2001; Cnudde and Boone, 2013). Other artefacts include conical-beam artefacts (De Witte, 2010; Cnudde and Boone, 2013), ring artefacts, (Sijbers and Postnov, 2004), phase contrast (Wilkins et al., 1996), secondary radiation effects (Boone et al., 2012), artefacts produced by the instability or movement of the sample (Cnudde and Boone, 2013) and charge sharing effects (Firsching, 2009)

4.1.3.3. HECTOR (High Energy CT Optimized for Research)

All the scans for this thesis were acquired with HECTOR at the Centre for X-ray Tomography (UGCT) at UGent (figure 4.2). It is one of the latest developments by UGCT and X-Ray Engineering (XRE). The mechanical setup consists out of nine motorized axes and combines a large-flat panel detector with a microfocus X-ray source up to 240 kV. It can accommodate any type of samples up 1 m long, 80 cm in diameter and up to 80 kg, but it can also reach very high resolutions up to 3-4 µm (Masschaele et al., 2013).

1 3 2

Figure 4.2: Setup of HECTOR at UGCT, with: (1) X-ray source, (2) sample holder and (3) detector.

4.2. Experimental setups and procedures

4.2.1. Characterization of the rocks

Determining the microhabitat of the bacteria involved a concise description of thin sections by optical microscopy, combined with water immersion porosimetry according the European norm EN1936 and

24

literature (see Materials chapter). µCT revealed the µCT pore size distribution on ten samples of Savonnières and Tabaire. MIP measurements performed at the Magnel Laboratory for Concrete Research of UGent revealed the pore throat sizes on one sample of each rock.

4.2.2. Flow experiments – Estimation of bacterial adhesion within the porous rocks

Flow experiments have been executed in order to obtain an indication of the bacterial adhesion within one Tabaire and Savonnières sample. In a flow cell, a syringe flushed a solution with around 109 Paracoccus denitrificans cells/mL (=bacterial solution) through the samples. In order to make sure that the flow went through the samples and not around, a stent pump was connected to the flow cell creating a confining pressure.

The bacteria for all the experiments (including the growth experiments, described further) were prepared by Jana De Bodt (CMET, Faculty of Bioscience Engineering, UGent). The bacteria grew inside a Tryptic Soy Broth (TSB) medium (table 4.1) and were added at the end to a solution of Evian water with 10 g/L NaCl (= salt solution). This water is a standard within CMET, that will not enhance growth and the NaCl will allow bacterial adhesion during the experiments (cfr. figure 2.6b).

Table 4.1: Composition growth medium (Northeast Laboratory Services 2016). Compound Concentration (g/L) Tryptone (Pancreatic Digest of Casein) 17.0 Soytone (Peptic Digest of Soybean Meal) 3.0 Glucose (=Dextrose) 2.5 NaCl 5.0 K2HPO4 2.5 Ca(NO3)2 1.2

These experiments have been conducted manually by pressing a syringe with the help of Dr. Tim De Kock. The syringe pump was not available by then and only an estimation was necessary. After vacuum impregnation of the samples with demineralized. The samples were mounted in the flow cell after which a salt solution was flushed to create a suitable environment for bacterial adhesion. Afterwards, during the “impregnation phase“, the bacterial solution was flushed through the samples, while the solution leaving the flow cell was sampled: one sample for each ten droplets. After the 70th droplet, demineralized water flushed the bacteria out of the system (the so called “Elution phase”, cfr. figure 2.6a). The percolated solution was sampled again, but now for every five droplets till the 30th, after which a sample of 20 droplets was taken. The bacterial solution itself has also been sampled during the Tabaire experiment.

Obviously, it would have been better to use exact volume units instead of “droplets”, the volumes however were so small that the error would be very large. The flow did also not stop immediately after pressing the syringe. This made it very hard to sample the exact same amounts of volume.

Flow cytometry (BD Accuri C6) at CMET laboratories determined with the help of Jana De Bodt, the concentration of Paracoccus denitrificans within the collected liquid samples. A dilution of 10.000 times was necessary due to the high concentration of bacteria and SYBR Green coated the cells. This substance allows detection of the cells.

*This experiment would be followed by a bioclogging experiment. Unfortunately it was not possible to finish the setup in time and perform these experiments (See Appendix 9.1 for more information about this setup, its problems and its goal). It would be an upgraded version of this flow setup, that would also be used to redo these flow experiments with more detail.

25

4.2.3. Radiation experiments

To verify any potential negative the effect of X-ray radiation on Paracoccus denitrificans, a dense culture of Paracoccus denitrificans (109 cells/mL) in saline solution (Evian water with 10 g/L NaCl) was scanned in plastic containers by HECTOR. The acquisition parameters of the scan were the same as all the scans executed during the thesis (10W, 120kV, 1400 frames at 1s/frame). This solution and a control solution was subsampled to detect with flow cytometry the direct mortality, while optical density measurements followed the growth of radiated and non-radiated bacteria.

Flow cytometry counted the dead and living cells of three X-ray exposed samples only two hours after exposure and three non-exposed samples. This was possible after staining the bacteria (“life and death staining”) using SYBR Green 14 dye and propidium iodide. SYBR Green I will stain all the cells, while propidium iodide only penetrates non-viable cells (Williams et al., 1998; Barbesti et al., 2000). Propidium will fluorescence red under green light (Williams et al., 1998), while SYBR Green I will colour the cells green (Barbesti et al., 2000). The samples were stored for three days (two days in the fridge and one day in the incubator) after which flow cytometry counted again the viable and dead cells. This could show the robustness of the bacteria and any possible delayed effects. To verify the flow cytometry, extra subsamples have been taken and heated to 110°C aiming to kill the bacteria to reveal their properties on a flow cytometric diagram after life and dead staining.

An optical density meter (infinite 200Pro from Tecan) followed the recovery of the bacterial populations at 35°C during 48 hours. It establishes growth curves using a simple principle: microbial cells causes striking light to scatter. The amount of scattering is indirectly proportional to cell concentration and directly related to the biomass (Willey et al., 2009). The optical density was followed in triplicates of the radiated and non-radiated bacteria with the presence or absence of Ca(NO3)2 (See Table 4.2 for the compositions of the four different solutions). It allowed to verify the effects of radiation and the influence of Ca(NO3)2 within the growth medium.

Table 4.2: Overview of the four solutions (total 200 µL) with its components used to reveal the growth curves of Paracoccus denitrificans with optical density. Composition different samples for Optical density measurements Saline solution with Saline solution with TSB medium + 1.2 g/L TSB medium radiated-bacteria non-radiated bacteria Ca(NO3)2 20 µL 180 µL 20 µL 180 µL 20 µL 180 µL 20 µL 180 µL

4.2.4. Growth experiments

Three independent but similar growth experiments have been performed to study biogenic gas and MICP on limestone. One growth experiment, that will be named “a cycle”, lasted for one week. Every cycle simulated an exponential growth of Paracoccus denitrificans in and around the rock samples with identical conditions except one parameter: during cycle 1 it grew within a TSB medium with 1.2g/L Ca(NO3)2 at pH 7.2. At the start of cycle 2 Na(OH) was added to increase the initial pH of the medium to 8.5, to favour any potential MICP. During cycle 3 at least the TSB medium with 1.2 g/L Ca(NO3)2 was diluted nine times with Evian water (table 4.3). For each cycle, six samples underwent three scans using HECTOR: a reference scan, one scan one week after impregnation and the last one when the samples

26

were dried. Additionally two samples of cycle 1 and two blank samples with azide of cycle 2 underwent an extra scan directly after impregnation. Amelie De Muynck verified the amount of radiation by taking into account: the composition of all the materials, including the plastic containers, the dimensions, the scanning time, the porosity, energy of the X-ray source,.... The µCT data analyses will be described in further detail in section 4.3. The scanner settings were always similar, to allow comparison of the different scans: a beam voltage of 120 kV, 10 W and 1 mm Al shielding and a total of 1400 acquired images with an exposure time of 1 s. The resolution was 7.43 µm during the first cycle and 8.04 or 8.07 µm during the other cycles. Every sample of a cycle is indicated by a S or T, referring to Savonnières or Tabaire, followed by two numbers: the first one refers to the cycle it belongs to and the second acts a serial number. S2b and T2b refer to the two blank samples with azide of cycle 2 (table 4.3).

Table 4.3: Overview of the three cycles and its samples with its resolution. S10 and T10 have been scanned directly after impregnation, just as S2b and T2b but these are the blanks with azide. Cycle 3 (10 % TSB + 1.2g/L Cycle 1 (100 % TSB + 1.2g/L Cycle 2 (100 % TSB + 1.2g/L Ca(NO3)2, 90 % Ca(NO3)2 at pH 7.2) Ca(NO3)2 at pH 8.5) Evian at PH 7.2) Sample name µCT resolution Sample name µCT resolution Sample name µCT resolution S10 7.43 µm S2b 8.04 µm S30 8.07 µm S11 7.43 µm S20 8.04 µm S31 8.04 µm S12 7.43 µm S21 8.04 µm S32 8.04 µm S22 8.07 µm T10 7.43 µm T2b 8.07 µm T30 8.04 µm T11 7.43 µm T20 8.07 µm T31 8.04 µm T12 7.43 µm T21 8.07 µm T32 8.04 µm T22 8.07 µm

For each cycle, three samples of Savonnières and Tabaire stone were examined. The samples were placed in little plastic containers (figure 4.3). Before scanning, the mass of the samples in the container was acquired using a balance, with a sensitivity of 0.01 mg. After acquiring the reference scan, a growth medium with a high concentration of Paracoccus denitrificans (108 cells/mL) impregnated the samples under vacuum. Residual gas could be left behind, and to quantify and determine its distribution, one sample of each rock was scanned immediately after impregnation during the first cycle. This was conducted also during the second cycle, but this time in blank samples, with azide. This information will be used as a control to verify if all the gas after one week, has a biogenic origin.

Figure 4.3: Samples of cycle 2 within plastic pots, two days after impregnation.

Before the start of a cycle, the bacteria grew in a Tryptic Soy Broth (TSB) medium with 1.2 g/L Ca(NO3)2 (Table 4.1) till Paracoccus denitrificans reached a concentration of about 109 cell/mL. To make sure that the bacteria would still grow inside the rock samples, a new solution was prepared just before impregnation containing 90 % fresh TSB medium (or Evian with 10g/L NaCl during cycle 3) and 10 % of the TSB medium with the bacteria. The nitrate, served as electron acceptor for denitrification and as source for the nitrogen gas, while the calcium was necessary for MICP. After the impregnation and/or after the scans, the bacteria grew during one week in an incubator at 35°C, around the optimal temperature for Paracoccus denitrificans (Nokal and Schlegel, 1983; Kelly et al., 2006).

27

After one week in contact with the bacterial medium, all samples were scanned again using with the aim to detect biogenic gas bubbles. The solution was afterwards removed from the samples and frozen for further analysis. To resolve potential MICP, the samples were oven dried at 35°C after which they were scanned with HECTOR again. One thin section was produced after each cycle to study them with SEM and optical microscopy. This will complement and verify the µCT data.

After the three growth cycles, chemical analysis were performed on the frozen solutions. Electrodes firstly determined the pH after which a Volatile Fatty Acid (VFA) analysis was performed on all the samples with the Etheric extraction method (Greenberg et al., 1992); this to estimate the biological activity during the growth. 2 mL sample (diluted twice or four times as not enough solution was present) was consecutively added with 0.5 mL H2SO4 (50 %), +- 0.4 g NaCl, 400 µL internal standard and 2 mL diethylether. A test tube rotator mixed the sample for 2 minutes, after which it was centrifuged at 300 rpm for 3 minutes. Gras chromatography was at least performed on the diethyl phase to measure the acetate, propionate, isobutyrate, butyrate, isovalerate and valerate concentrations.

4.3. Applied µCT reconstruction and data analysis protocol

4.3.1. Reconstruction

Octopus Reconstruction, an in-house software package (Inside Matters) reconstructed all the scans (figure 4.4). It transformed the µCT projections, acquired from all directions into 2D slices. A spot and ring filter were applied together with a correction for beamharding. The reconstruction parameters were kept constant to allow comparison between the scans, especially to visualize potential MICP.

Figure 4.4: Illustration of Octopus Reconstruction.

4.3.2. Data analysis

Octopus Analysis can retrieve quantitative information from the reconstructed slices. It is a flexible 3D analysis in-house software package (Inside Matters) (figure 4.5). This program retrieved all the quantitative information regarding the gas phase and the µCT porosity. On the reference scans the porosity was determined, while the scans of the just impregnated samples gave information of the

28

residual gas phase. The scans of the wet samples after one week determined the final gas phase. For each data set a similar region of interest was selected, to compare the data carefully.

Figure 4.5: Illustration of Octopus reconstruction, during thresholding of gas bubbles within Tabaire.

In order to perform 3D analysis, a binary image of the pores or the gas bubbles needed to be retrieved. A single threshold was appropriate for the porosity, as only the obvious pores, were determined. This avoided the microporosity with sizes below the resolution for which MIP was used. The gas bubbles were more difficult to retrieve and a dual threshold was necessary, due to the low contrast between gas and liquid phase.

The binary images of the porosity, needed no further filtering. This was however not the case for the gas bubbles, because there was noise thresholded as well. Therefore a median filter and binary operations were applied to reduce this effect.

For the gas bubbles the image processing stopped here but the pores needed to be separated first, as many were interconnected with each other. The program retrieved then quantitative information about the total fraction of the different phases (including the open and closed fraction). It organised the bubbles or pores in classes according their equivalent diameter with a class width of two voxels. It retrieved the volume and quantity of each class allowing to reconstruct the (volume) distributions. The objects smaller than an equivalent diameter of four voxels, have been considered as noise and have not been incorporated in the distributions.

The pores or bubbles that were cut by the boundary were included in the analysis. It was not ideal, but removing these features would not be good as well, as the larger objects would have more chance to be removed with this action than the smaller ones.

4.3.3. Dataviewer and VGStudio

When MICP is deposited during the growth of the bacteria, it will potentially increase the volume of the samples and change the porosity. Dataviewer software (Bruker) (figure 4.6) matched the orientation of the scans and retrieved difference images before and after contact with bacteria. These images will reveal MICP if it is larger than the resolution of the scans.

29

Figure 4.6: Illustration of Dataviewer, while matching the two different scans.

The resulting difference image illustrated addition of material as a light phases and a removal as dark. Octopus Analysis could be used to retrieve quantitative information about these two phases in the same way as described above.

VGstudio (Volume graphics) was the last main program in the image analysis (Figure 4.7.). It is a tool that allowed to visualize the different phases in 3D. It enhanced the interpretation and aimed to find a correlation between different scans (e.g. between residual gas and the gas phase after one week). The rescaled images of Dataviewer made it possible to show these phases with no offset.

Figure 4.7: VGStudio display visualizing a Savonnières limestone with gas bubbles.

30

5. RESULTS

5.1. Characterization of the micro-environment – Porosity

Table 5.1. presents the µCT porosity for every sample of the growth experiments. Savonnières had an average µCT porosity of 16.03 % with more than 1/3 closed pores (at the µCT resolution level). Tabaire had a comparable µCT porosity with an average of 15.10 %. This rock contained however only a very low number of µCT closed pores (<1 %). 2D cross-sections like figure 5.1 showed many angular interconnected pores within Tabaire limestone, that started to form a network. The cross-sections resolved within Savonnières, especially rounded pores, that could be elongated but it could not resolve the pore connections. Figure 5.3 illustrates the pore volume distribution of the six samples from the first cycle (figure 5.3a,c) with their average distributions in (figure 5.3b,d). The volume-distribution was for both rocks lognormal. The pores of with an equivalent diameter between 200 and 300 µm for Savonnières and between 150 and 200 µm for Tabaire contributed the most significantly to the total volume.

Based on the µCT data: An important note, the µCT pore volume displayed in figure 5.3 ranges between 37.15 and 557.25 µm. Larger pores were present for both rock types, reaching an equivalent diameter of more than 1400 µm. They contributed significantly to the total volume. It reached sometimes 30 % within Savonnières of which one pore alone contributed 20 % of the total volume. Within Tabaire this was more limited, but could reach still 20 %.

Table 5.1: µCT porosity of the different samples, including the open and closed porosity. Sample Open porosity (%) Closed porosity (%) Total porosity (%) S10 10.42 6.72 17.13 S11 12.26 4.94 17.19 S12 9.91 5.14 15.05 S2b 8.09 6.36 14.44 S20 9.48 5.64 15.12 S21 9.86 5.34 15.20 S22 14.76 4.37 19.13 S30 12.94 5.24 18.18 S31 12.02 5.25 17.27 S32 4.42 7.18 11.60 Average 10.41 5.62 16.03

T10 14.23 0.80 15.03 T11 16.05 0.63 16.68 T12 11.10 0.93 12.03 T2b 13.12 0.85 13.97 T20 15.41 0.73 16.14 T21 13.61 0.72 14.33 T22 15.65 0.78 16.43 T30 15.20 0.69 15.89 T31 15.76 0.78 16.54 T32 13.16 0.81 13.97 Average 14.33 0.77 15.10

31

a b

Figure 5.1: 2D cross-section through (a) Savonnières and (b) Tabaire.

As shown in the materials section (3.1 and 3.2), both rocks contain similar water immersion porosity between ± 30 and ± 38 %. The water immersion porosity, combined with the µCT porosity (table 5.1) revealed that around 50 % of the pores cannot be resolved by µCT due to a lack in resolution (these pores will be defined here as micropores). MIP revealed the micropore throats with an equivalent diameter between 50 nm and 90 µm (figure 5.2). Within Tabaire 62 % of the micropores (access) were smaller than 2 µm, while for Savonnières it reached 84 %. The Tabaire sample contained an important number of pores (access) between 0.5 and 10 µm, while the pores (access) between 0.08 and 0.5 µm contributed significantly to the pore volume within the Savonnières. Most µCT closed pores were only isolated for µCT, but were actually open as water and mercury had access to it.

Relative pore volume distribution Savonnières and Tabaire: MIP 45 40 35 30 25

(vol/vol) 20 15 10

Incremential intrusionIncremential volume 5 Savonnières Tabaire 0 0,05 0,5 5 50 Pore size diameter (µm) Figure 5.2: Relative pore size distribution Savonnières and Tabaire retrieved by MIP.

32

a Pore volume distribution Savonnières and Tabaire b Average Pore volume distribution Savonnières and Tabaire 600 600 S10 S11 S12

500 T10 T11 T12 500 µm)

6 400

µm) 400 6

300 300

200 200

Volume Volume (x10 Pore volume volume Pore (x10 100 100 Savonnières Tabaire 0 0 0 100 200 300 400 500 0 100 200 300 400 500 Equivalent diameter (µm) Equivalent diameter (µm)

c Cumulative pore volume distribution Savonnières d Average cumulative pore volume distribution and Tabaire, between 37.15 and 551.25 µm Savonnières and Tabaire, between 37.15 and 557.25 µm 100 100 90 90 80 80 70 70 60 60 50 50 40 40 S10 S11 S12 30 30 20 T10 T11 T12 20

Pore volume volume Pore (cumulative %) 10 10 Pore volume volume Pore (cumulative %) Savonnières Tabaire 0 0 0 100 200 300 400 500 0 100 200 300 400 500 Equivalent diameter (µm) Equivalent diameter (µm) Figure 5.3: Pore volume distribution of three Savonnières and Tabaire samples (a) (zero values not displayed), with its cumulative (c) and their average (b) and cumulative distribution (d) .

33

5.2. Flow experiments – Bacterial adhesion within porous limestone

Figure 5.4 shows concentration of bacteria leaving the flow cell after percolating through the Savonnières and Tabaire samples. The results for both rocks were similar. During the “impregnation phase” (figure 5.4a,c) a plateau was rapidly reached. It contained some variation but its average value was for the Tabaire stone at least comparable with the initial concentration of Paracoccus denitrificans within the bacterial solution (538 x 107 bacteria/mL). The demineralized water during the “elution phase” (figure 5.4b,d) caused on the other hand for both rocks a similar exponential downward of the bacterial density.

a Concentration of Paracoccus denitrificans flushed b Concentration of Paracoccus denitrificans through Savonnières while flushing with bacterial solution coming through Savonnières while 1000 C flushing with demineralized water

/ mL / 1000

/ mL / 7

o7 *10 100 nc*10 100

10 10

y = 226,44e-0,129x R² = 0,9337 1 1

Paracoccusdenitrificans 0 20 40 60

Paracoccusdenitrificans 0 20 40 Bacterial solution (drops) Demineralized water (drops)

c Concentration of Paracoccus denitrificans flushed d Concentration of Paracoccus denitrificans through Tabaire while flushing with bacterial solution coming through Tabaire while flushing with

1000 demineralized water

/mL /mL

7 1000

7

*10 *10 100 100

10 10 y = 175,22e-0,103x R² = 0,9723

1 Paracoccusdenitrificans 1 Paracoccusdenitrificans 0 20 40 60 80 0 20 40 Bacterial solution (drops) Demineralized water (drops) Figure 5.4: (Left) “Impregnation phase”: Concentration of Paracoccus denitrificans flushed through Savonnières (a) and Tabaire (c). (Right) “Elution phase” Concentration of Paracoccus denitrificans flushed out of the same samples of Savonnières (b) and Tabaire (d) with demineralized water.

34

5.3. Radiation experiments

During one µCT scan of the growth experiments, the bacteria were exposed between 30.18 and 116.98 Gy for the Savonnières and between 30.31 and 117.40 Gy for the Tabaire samples. The exposed radiation for Paracoccus denitrificans was during this experiment slightly higher: in this case a pure dense culture of Paracoccus denitrificans (109 cells/mL) within plastic containers, was exposed to X-ray radiation. Flow cytometric data of the samples stained with SYBR Green 14 dye and propidium iodide (table 5.2) found no dead cells. After three days only a little amount (<10 %) of the bacteria was found dead. The difference with the heated sampled revealed clearly the dead bacteria on the flow cytometric plots (figure 5.5).

Table 5.2: Flow cytometric data comparing the concentration of living and dead Paracoccus denitrificans cells, whereas the X-ray samples were exposed to X-rays by HECTOR and the blank remained unexposed. (a) shows the results two hours after exposure to X-rays and (b) three days after exposure, compared with the results of samples that were heated in an oven to 100°C killing all the bacteria.

a: Situation two hours after exposure b: Situation three days after exposure Living cells Dead cells Living cells Dead cells Sample Sample (x107/mL) (x107/mL) (x108/mL) (x108/mL) X-ray A 365 0 X-ray 48 1 X-ray B 268 0 Blank 42 3 X-ray C 264 0 Blank A 378 0 Heated 1 0 25 Blank B 279 0 Heated 2 0 52 Blank C 328 0

a b

Dead Dead cells cells

Back- Back- ground Living ground Living noise cells noise cells

Figure 5.5: Plots of the flow cytometric measurements of a radiated (a) and a heated (dead) (b) sample three days after exposure by X-rays.

Optical density measurements monitored the recovery of the exposed and non-exposed bacteria (figure 5.6). It revealed no big difference in growth between the two groups. The optical density of the radiated bacteria was only slightly lower. The presence or absence of Ca(NO3)2 affected the measurements tremendously: the bacteria reached with Ca(NO3)2 the stationary phase after fifteen hours, while the other samples lacking the Ca(NO3)2 were after 48 hours still approaching it. The absorbance increased also faster in the samples with the bacteria growing in the TSB medium with Ca(NO3)2. The absorbance increased more steadily when Ca(NO3)2 was missing, lacking a real exponential phase, however it surpassed the absorbance of the bacteria growing in the other medium.

35

a Growth Paracoccus denitrificans in TSB

medium without Ca(NO3)2 1,6

1,4

1,2 Stationary phase? 1

0,8 Exponential phase? 0,6

0,4 Lag X-ray A X-ray B

Optical densityat 600 nm phase X-ray C Bl A 0,2 Bl B Bl C 0 0 10 20 30 40 50 Time (hours)

b Growth Paracoccus denitrificans in TSB medium with Ca(NO3)2 1 0,9 0,8 Stationary phase 0,7 0,6 0,5 0,4 Lag Exponential phase 0,3 phase X-ray A X-ray B 0,2

Optical densityat 600 nm X-ray C Bl A 0,1 Bl B Bl C 0 0 10 20 30 40 50 Time (hours) Figure 5.6: Growth curve of Paracoccus denitrificans exposed to X-rays (red-orange) compared to non-radiated bacteria (greenish), within (a) TSB medium (b) TSB medium containing 1.2 g/L Ca(NO3)2.

5.4. Growth experiments

5.4.1. Biological activity during the different cycles – Chemical analyses

During cycle 1, the pH from around 7.3 to 7.69 for T12 and even 8.22 for S11 (table 5.3.). Cycle 3 followed this rising trend but less significantly reaching values between 7.47 and 7.76, except for S21 where a pH of 8,11 was achieved. The pH-values of cycle 2 were more variable with about the same range of cycle 1 and 3 combined. The trend was opposite as the pH decreased significantly from 8.5 (table 5.3). T22 was an exception with an anomalous pH of 8.71. The bacterial solution of T22 was also anomalous, as it was darker and contained a white substance. The pH values were for both rocks comparable with in general a slight higher pH for the Savonnières samples

The VFA concentrations varied during the cycle 1 between low values for Savonnières (120.46 – 185.88 mg/L) and slight higher values for Tabaire (284.79 – 606.29 mg/L). These concentrations were low after cycle 3 showing no big differences between the two rocks. Cycle 2 however reached ten times higher

36

concentrations (between 1636.27 and 6626.26 mg/L). Here there was a difference between the two rocks: Savonnières contained in general a higher concentration of the VFA. The acids were also present in blanks with a concentration comparable with the other samples of cycle 2 (table 5.3).

The VFA and the pH were inversely proportional. The pH increase during cycle 1 and 3 coincided with low to intermediate VFA concentrations, while the pH decrease during cycle 2 was accompanied with high VFA concentrations. There were however some exceptions like e.g. S22 (table 5.3).

Table 5.3: pH and VFA measurements of the growth media after one week within Savonnières and Tabaire. Every value of the VFA-analysis is expressed in mg/L. Propio- Isobutyl Buty- Isovale- Vale- Total Sample pH Acetate nate -rate rate rate rate VFA Cycle 1 (100 % TSB + 1.2g/L Ca(NO3)2 at pH 7.3) S10 8.22 123.58 45.98 0.00 0.00 0.00 0.00 169.56 S11 8.13 120.46 0.00 0.00 0.00 0.00 0.00 120.46 S12 8.06 185.88 0.00 0.00 0.00 0.00 0.00 185.88

T10 8.02 284.79 0.00 0.00 0.00 19.98 0.00 304.77 T11 8.03 468.01 48.81 43.86 0.00 77.22 0.00 637.89 T12 7.69 606.29 0.00 0.00 224.15 0.00 13.10 843.54

Cycle 2 (100 % TSB + 1.2g/L Ca(NO3)2 at pH 8.5) S2b 7.75 1509.11 618.88 91.69 0.00 224.27 0.00 2443.95 S20 7.71 3134.95 343.40 485.47 1432.63 1229.80 0.00 6626.26 S21 8.11 2294.01 180.19 15.78 64.28 52.37 0.00 2606.64 S22 8.01 1929.41 151.60 0.00 18.44 25.84 0.00 2125.30

T2b 7.53 1587.76 0.00 163.25 14.54 346.59 0.00 2112.14 T20 7.84 3132.58 163.25 208.13 634.28 605.62 0.00 4743.86 T21 7.93 1768.12 138.48 0.00 19.49 26.88 0.00 1952.97 T22 8.71 802.91 0.00 75.24 239.58 518.55 0.00 1636.27

Cycle 3 (10 % TSB + 1.2g/L Ca(NO3)2, 90 % Evian at pH 7.2) S30 7.54 118.15 0.00 0.00 0.00 0.00 0.00 118.15 S31 8.11 184.42 26.00 24.84 33.36 57.76 0.00 326.38 S32 7.76 204.76 0.00 0.00 23.23 57.11 0.00 285.10

T30 7.53 132.83 0.00 0.00 0.00 15.67 0.00 148.50 T31 7.47 106.18 0.00 0.00 0.00 0.00 0.00 106.18 T32 7.62 218.21 0.00 26.86 27.62 46.57 0.00 319.26

5.4.2. Reaction products – Gas

5.4.2.1. Gas saturations

The amount of gas increased significantly during a growth cycle (table 5.4). µCT at the start of the first two growth cycles determined the residual gas phase and uncovered gas trapped by the pores (table 5.4: residual gas). Its occurrence was low and the blanks indicated that most gas disappeared after one week. The gas saturations, retrieved by µCT, varied between the cycles. Gas saturations for Savonnières limestone varied between 13 and 23 %, which was significantly higher than the 1.5 -12 % in the Tabaire stone. The gas saturations for sample T10 and S10 differed with their respectively 12.85 and 1.57 % significantly compared to the other two rock samples of the same rock type.

37

The results of the second cycle were more variable. The overall gas saturations increased for Tabaire stone, reaching even 25 %. Sample T22 was an exception and contained almost no gas. The gas concentration was even lower compared to the blanks. Savonnières indicated a more complex behaviour with a very high saturation for S20 (23,42 %), while sample S21 and S22 contained less gas (respectively 14.92 % and 17.93 %) than in cycle 1.

The gas saturations did not differ significantly between the Tabaire and Savonnières stones during the last cycle. The result showed that in two out of three samples, almost no gas was present, while the other ones: S32 and T32 reached saturations of respectively 6.12 and 5.33 %.

Table 5.4: Percentages of the total volume of the gas phase, determined by µCT, directly after impregnation with Paracoccus denitrificans, as well as determined after one week. µCT porosity Residual gas Gas (%) after Gas saturation Sample (%) (%) one week (%)

Cycle 1 (100 % TSB + 1.2g/L Ca(NO3)2 at pH 7.3) S10 17.13 0.42 2.20 12.85 S11 17.19 / 3.92 22.80 S12 15.05 / 3.21 21.31

T10 15.03 0.19 0.24 1.57 T11 16.68 / 1.96 11.78 T22 12.03 / 1.28 10.62

Cycle 2 (100 % TSB + 1.2g/L Ca(NO3)2 at pH 8.5) S2b 14.44 0.94 0.06 0.43 S20 15.12 / 3.54 23.42 S21 15.20 / 2.27 14.92 S22 19.13 / 3.43 17.93

T2b 13.97 0.31 0.08 0.59 T20 16.14 / 3.54 21.92 T21 14.33 / 3.57 24.93 T22 16.43 / 0.01 0.08

Cycle 3 (10 % TSB + 1.2g/L Ca(NO3)2, 90 % Evian at pH 7.2) S30 18.18 / 0.02 0.13 S31 17.27 / 0.07 0.38 S32 11.60 / 0.71 6.12

T30 15.89 / 0.08 0.53 T31 16.54 / 0.01 0.04 T32 13.97 / 0.74 5.33

5.4.2.2. Residual gas phase

Figure 5.7 illustrates the gas (volume) distributions of the residual gas phases during the first two cycles. Figure 5.7a marks a lognormal distribution: it reached, especially for Savonnières limestone, a distinct peak around an equivalent diameter of 50 µm for S10 and 75 µm for S2b. Such a peak was present around the same values for Tabaire, but it was more obscure (figure 5.7, mode in table 5.5). Savonnières contained on average three times more residual gas bubbles than Tabaire and a higher gas volume. The differences between the two cycles were minor, except a small shift of the peak to higher equivalent diameters and a higher number of gas bubbles. The cumulative graph (figure 5.7b)

38

revealed only slight differences in the trends of the samples between the two growth cycles. The differences between the two stone types within the same cycle were even smaller or negligible.

The residual bubble volumes were more or less lognormal distributed (figure 5.7c,d). The gas bubbles with an equivalent diameter between more or less 100 and 150 µm accounted as the most important contributors for the whole residual gas phase. The cumulative distribution showed some variation, mainly between the first two cycles. The variation of the two rock types within the same cycle was more limited. In Table 5.5 some parameters of the distribution are given. They indicate the clear distinction in volume parameters between samples of the first and second cycle (e.g. average volume of a bubble was 0.32 and 0.35 x 106 µm³ at the start of the first cycle while it was 0.61 and 0.56 x 106 µm³during the second cycle).

Table 5.5: Parameters regarding residual gas (volume) distribution retrieved within one hour after impregnation. Average equivalent diameter + median are determined on the volume distribution and the mode on the bubble size distribution. Total Average Average Total # gas Median – Mode Samples volume volume equivalent bubbles (µm) (µm) (x106 µm³) (x106 µm³) diameter (µm) S10 1106 356.29 0.32 85.05 109.97 52.08 T10 381 133.53 0.35 87.48 115.66 52.08 S2b 1498 913.92 0.61 105.23 131.89 72.36 T2b 473 266.71 0.56 102.50 141.14 56.28

Measuring the blank samples, one week after impregnation, can give information about changes of the residual gas phase over time. Samples S30, S31, T22, T30 and T31 all contained gas levels of less than 1 % after one week (table 5.4). This assumes here no significant gas production, so these could serve as additional information about the evolution of the residual gas. The gas distributions after one week, was very chaotic and revealed no significant differences between the stones. The number of gas bubbles, the (average) volume and average equivalent diameter of the bubbles showed a spectacular variation (table 5.6). The differences reached even a factor of more than 100 between the average bubble volume of T2b and T31. The number of residual gas bubbles decreased spectacularly with around 99 % for S2b and T2b while the average volume and equivalent diameter increased (table 5.6). There was no data about the residual gas phase after one hour for the other samples, however by comparing table 5.5 to table 5.6 we can assume the same trend (except for T31, where the average volume and equivalent diameter were very low (table 5.6)).

Table 5.6: Parameters regarding the gas phase at the end of a cycle from the samples lacking significant gas production. The average equivalent diameter is determined on the volume distribution. Average Total volume (x106 Average volume Sample Total # gas bubbles equivalent µm³) (x106 µm³) diameter (µm) S2b 15 119.95 8.00 248.10 S30 6 29.60 4.93 211.22 S31 19 62.64 3.30 184.66

T2b 4 76.95 19.24 332.45 T22 3 11.71 3.90 195.37 T30 33 77.21 2.34 164.71 T31 36 5.45 0.15 66.11

39

a Residual gas phase b Cumulative Residual gas distribtuion

300 100 90 250 80 200 70 60 150 50 S10 S2b 40

# Gas Gas # bubbles 100 30 T10 T2b 20 50 S10 S2b T10 T2b Gas Gas bubbles (cumulative %) 10 0 0 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 Equivalent diameter (µm) Equivalent diameter (µm)

c Residual gas volume distribution d Cumulative residual gas volume distribution 120 100 90 100 80

80 70 µm³)

6 60 60 50 S10 S2b 40 40

T10 T2b 30 Volume Volume (x10 20 20 Volume (cummulative %) S10 S2b T10 T2b 10 0 0 0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350 Equivalent diameter (µm) Equivalent diameter (x10-6 m) Figure 5.7: (a) Residual gas bubble distribution, (b) cumulative distribution, (c) volume and (d) cumulative volume distribution retrieved within one hour after impregnation. Graphs do not display zero values.

40

5.4.2.3. Gas phase after one week

Figure 5.8, 5.9 and 5.10 illustrate the (volume) distribution of the gas phase after one week. It displays only the distributions of the samples which encountered a significant increase in gas saturation after impregnation with the bacteria (samples with gas saturation of > 1 % table 5.4). Only those results will be described in here. Savonnières retained again most gas bubbles, with a number that was on average three times higher than Tabaire (table 5.7). The samples of the second cycle contained most bubbles. Between the first two cycles the number of bubbles increased with several factors for Tabaire (e.g. 37 compared with 463), while the increase for Savonnières was more modest (e.g. 383 compared with 720). During the last cycle, when less nutrients were available only S32 and T32 contained a significant but lower gas concentration. There were also less bubbles but Savonnières retained here also the most, the difference with Tabaire however, was smaller (218 vs. 171).

The (volume) distribution of all the samples was initially more or less lognormal, with a peak in the number of bubbles between an equivalent diameter of ± 100 and 200 µm (see also mode table 5.7). It revealed no clear distinction between the different cycles and rocks. The trend of the volume distributions was similar, but here the peak lied generally between +-200 and 300 µm of equivalent diameter. Gas bubbles with an equivalent diameter starting at around 300 µm for Tabaire and 400 µm for Savonnières were rare. Tabaire contained however the largest bubbles. These large bubbles had if present, a significant impact, leading to an exponential increase of the volumes (c in figure 5.8, 5.9 and 5.10). The distribution, and parameters (figure 5.10 and table 5.7) were more moderated during cycle 3, with only minor differences between the two rocks.

Table 5.7: Parameters regarding the gas phase after one week. Average equivalent diameter, median are determined on the volume distribution. The mode is determined on the bubble size distribution (“number/number” indicates two modes and “/ “for T10 indicates no distinct mode). Samples Total # Total Average Average Median Mode (µm) gas volume volume equivalent (µm) bubbles (x106 (x106 µm³) diameter µm³) (µm)

Cycle 1 (100 % TSB + 1.2g/L Ca(NO3)2 at pH 7.3) S10 383 1382.60 3.61 190.33 238.97 185.75 S11 536 2870.00 5.35 217.06 295.45 111.45 S12 462 1940 .00 4.20 200 .16 265 .18 141 .17

T10 37 207.79 5.62 220.53 296.96 / T11 112 1339.04 11.96 283.70 485.91 185.75 T12 126 833.98 6.62 232.95 329.34 66.87/185.75

Cycle 2 (100 % TSB + 1.2g/L Ca(NO3)2 at pH 8.5) S20 1226 4262.18 3.48 187.95 257.60 120.6 S21 720 2730.68 3.79 193.48 275.62 136.68 S22 776 4114 .42 5.30 216 .35 289 .34 121 .05

T20 539 3322.84 6.16 227.50 357.96 121.05 T21 463 3405.43 7.36 241.29 395.94 121.05/88.77

Cycle 3 (10 % TSB + 1.2g/L Ca(NO3)2, 90 % Evian at pH 7.2) S32 218 673 .54 3.09 180 .70 227 .25 136 .68

T32 171 702.17 4.11 198.68 278.60 136.68/152.76

41

a Cycle 1 - Final gas distribution b Cycle 1 - Cumulative final gas distribution 60 100 90 50 80 70 40 60 30 50 S10 S11 S12 40 20 # Gas Gas # bubbles T10 T11 T12 30 S10 S11 S12 10 20 T10 T11 T12 Gas Gas bubbles (cumulative %) 10 0 0 0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800 Equivalent diameter (µm) Equivalent diameter (µm)

Cycle 1 - Cumulative final gas volume distribution b Cycle 1 - Final gas volume distribution d 250 100 90 200 80

µm³) 70 6 150 60 50 100 40 30

Gas Gas volume (x10 50 20 S10 S11 S12

S10 S11 S12 Gas bubbles (cumulative %) 10 T10 T11 T12 T10 T11 T12 0 0 0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800 Equivalent diameter (µm) Equivalent diameter (µm) Figure 5.8: (a) Size and cumulative (b) gas distribution within Savonnières and Tabaire at the end of cycle 1 with (c) and (d) illustrating respectively the volume and cumulative volume distribution. Graphs do not display zero values.

42

a Cycle 2 - Final gas distribution b Cycle 2 - Final gas distribution 140 100 120 90 80 100 70 80 60 50 60 40 # Gas Gas # bubbles 40 30 S20 S21 S22 T20 T21 20 20 S20 S21 S22 T20 T21

# Gas Gas # bubbles (cumulative %) 10 0 0 0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800 Equivalent diameter (µm) Equivalent diameter (µm)

c Cycle 2 - Final gas volume distribution d Cycle 2 - Cumulative final gas volume distribution 400 100 350 90 80 300

µm³) 70 6 250 60 200 50 150 40 30 100 Gas Gas volume (x10 S20 S21 20 S20 S21 S22

50 Gas volume (cumulative %) S22 T20 10 T21 T20 T21 0 0 0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800 Equivalent diameter (µm) Equivalent diameter (µm) Figure 5.9: (a) Size and cumulative (b) gas distribution within Savonnières and Tabaire at the end of cycle 2 with (c) and (d) illustrating respectively the volume and cumulative volume distribution. Graphs do not display zero values.

43

a Cycle 3 - Final gas distribution b Cycle 3 - Cumulative final gas distribution 25 100 90 20 80 70 15 60 50 10 40

# Gas Gas # bubbles S32 T32 30 S32 T32 5 20

# Gas Gas # bubbles(cumulative) 10 0 0 0 100 200 300 400 500 0 100 200 300 400 500 Equivalent diameter (µm) Equivalent diameter (µm)

c Cycle 3 - Final gas volume distribution d Cycle 3 - Cumulative final gas volume distribution

80 100 70 90

60 80 µm)

6 70 50 60 40 50 30 40 30 20

Gas Gas volume (x10 S32 T32 20 10 S32 T32 Gas Gas volume (cumulative %) 10 0 0 0 100 200 300 400 500 0 100 200 300 400 500 Equivalent diameter (µm) Equivalent diameter (µm) Figure 5.10: (a) Size and cumulative (b) gas distribution within Savonnières and Tabaire at the end of cycle 3 with (c) and (d) illustrating respectively the volume and cumulative volume distribution. Graphs do not display zero values.

44

The trends illustrated by the cumulative graphs (b) and (d) of figure 5.8, 5.9 and 5.10 were comparable for all the cycles and for the different rocks. In general, Tabaire contained more large gas bubbles. The cumulative volume distributions were more variable, especially between Savonnières and Tabaire. It revealed that the ten percent largest bubbles could contribute more than half of the total gas volume within the Tabaire samples. The volume of the largest bubbles within the Savonnières limestone was also a very important fraction, however here its influence was limited. The difference between the first and second cycle were negligible and the average volume, equivalent diameter and the median of the gas bubbles agreed with the cumulative trends (table 5.7). The gas volume distribution within Tabaire had a higher medium and was on average always larger then Savonnières (maximum of 217.06 µm for Savonnières vs. a minimum equivalent diameter of 220.53 µm for Tabaire). The average equivalent diameter for all the samples was 200.89 µm for Savonnières and 241.23 µm for Tabaire, or a difference of 16.72 %. This value did not include the third cycle, where the average diameter of gas bubbles was only 198.68 µm for T32, which was still larger than the average of 180.70 µm for S32 or a difference of only 9.05 %.

5.4.2.4. Gas distribution in two- and three dimensions

Revealing the gas in 2D or 3D showed in general a homogeneous distribution of the bubbles (fig 5.11). The gas bubbles were most constrained within the central cilinder of the rock, however they could reside in the outer part as well, certainly in the case of high gas saturations (figure 5.12e,f).

The residual gas bubbles (figure 5.12a,b) were in general evenly distributed as well. Those 3D images did not show a clear correlation between the two gas phases but it revealed again the small dimensions of the residual gas compared with the larger bubbles after one week. They were spherical, while the gas bubbles at the end of a cycle had a more complex shape, mimicking the pore shape. The gas bubbles in Savonnières (figure 5.12e) were mostly constrained to one pore, while taking its shape. The gas bubbles within Tabaire were less constrained to one pore and formed a network connecting several pores (figure 5.12f). Even 2D cross-section (figure 5.11) revealed the complex network. This was however also dependent on the number of gas. The gas bubbles within samples with a low saturation (figure 5.12a-d) were in general isolated and spherical.

a b

Figure 5.11: µCT cross-sections showing the gas at the end of growth the experiments within S20 (a) T21 (b), revealing isolated gas bubbles within Savonnières and more complex bubbles (see e.g. red arrow) within Tabaire.

45

a bb

c d

e f

Figure 5.12: 3D rendered volume of residual gas bubbles (red) and gas bubbles after one week (blue) in (a) S10 and (b) T10. Vertical cross-sections through (c) S32 and (d) T32 with a relative low number of gas after one week (blue). Horizontal cross-sections (e) and (f) through S12 and Tabaire T21 respectively, illustrating high gas saturations (blue). Diameter of the samples ≈ 5.85 µm.

46

5.4.3. Reaction products – Microbially induced calcite precipitation

While after one week there was a significant number of gas bubbles, the MICP was more obscure. Measuring the weight before and after a cycle revealed a small weight increase (table 5.8), except for sample T12. The weight increase was limited and significantly higher, for both rocks after cycle 2 (between 0.491 and 0.969 %) compared to other cycles. The increase during cycle 1 and 3 were similar, with intermediate and less variable values after cycle 3 (values ranged between -0.116 and 0.344 % for cycle 1 and between 0.170 and 0.229 % for cycle 2). The differences between Savonnières and Tabaire were small, and minor compared with the different growth cycles. Savonnières seemed however to experience a higher weight (average of 0.0102 g) increase in comparison with Tabaire (average of 0.0066 g).

The blanks of the second cycle experienced also a weight increase (0.407 and 0.387 %), it was the lowest of that cycle but still high compared to the other cycles.

Table 5.8: Weight difference between the start and the end of a cycle. Weight Weight Initial dry Final dry Sample difference difference weight (g) weight (g) (g) (%)

Cycle 1 (100 % TSB + 1.2g/L Ca(NO3)2 at pH 7.3) S10 2.4761 2.4747 0.0014 0.057 S11 2.4148 24.199 0.0051 0.211 S12 2.4689 24.774 0.0085 0.344

T10 2.7536 2.7569 0.0033 0.120 T11 2.6763 2.6778 0.0015 0.056 T12 2.5799 2.5769 -0.0030 -0.116

Cycle 2 (100 % TSB + 1.2g/L Ca(NO3)2 at pH 8.5) S2b 2.5784 2.5889 0.0105 0.407 S20 2.5804 2.5948 0.0144 0.558 S21 2.5741 2.5946 0.0205 0.796 S22 2.5381 2.5627 0.0246 0.969

T2b 2.5578 2.5677 0.0099 0.387 T20 2.5672 2.5798 0.0126 0.491 T21 2.5273 2.5398 0.0125 0.495 T22 2.5412 2.5559 0.0147 0.578

Cycle 3 (10 % TSB + 1.2g/L Ca(NO3)2, 90 % Evian at pH 7.2) S30 2.5754 2.5813 0.0059 0.229 S31 2.5344 2.5397 0.0053 0.209 S32 2.5800 2.5855 0.0055 0.213

T30 2.6469 2.6514 0.0045 0.170 T31 2.6365 2.6418 0.0053 0.201 T32 2.6606 2.6655 0.0049 0.184

47

a b c

d e f

Figure 5.13: Savonnières (S12): Initial (a), final reconstructed cross-section (b), and difference images (c); Tabaire (T20) Initial (d), final reconstructed cross-section (e), and difference images (f); Light colours of the difference images represent addition, while dark colours represent removal.

48

a b

Figure 5.14: Optical microscopy revealing potential MICP within Savonnières (a) S.10 and (b) S10. white bar = 100 µm.

a b

Figure 5.15: SEM images revealing (a) potential MICP within S10 and (b) a sharp edge (dashed line) at S32.

µCT (figure 5.13), optical microscopy (figure 5.14) and SEM (figure 5.15) tried to detect any possible MICP. Figure 5.13c,f show difference images of the initial and final state of a cycle. They contained in general more zones of addition compared to removal (or more light then dark colours). The images varied however and the difference images after the first cycle revealed addition, especially around the samples, which was very clear at their top. The images of the second cycle showed less addition and even some removal, while the results of cycle 3 were very variable. Internally there were also zones of addition and removal, especially after the first cycle (figure 5.13(a-c). These two zones laid sometimes at the opposite side of a grain. Optical microscopy (figure 5.14) revealed some rare distinct crystals that were deposited upon the ooids of Savonnières limestone. SEM images like figure 5.15a might also detect some MICP, however the edges around the samples were sharp (figure 5.15b).

49

6. DISCUSSION

6.1 Characterization of the micro-environment – Porosity

Characterizing the micro-environment where bacteria live, is essential to understand their effects on porous rocks. The minerals influence the chemical conditions, whereas the pore structure determines the geometry of the habitat. The chemical conditions between the different rocks used in this research are similar as they consist mainly out of calcium carbonate. This provides a buffering capacity and is a source for inorganic carbon and calcium. The quartz and the trace minerals in Tabaire can further add some trace elements that can be used for microbial life (Jones and Bennet, 2014). The pore structure on the other hand differed significantly between the two rocks, even though both rocks contained about the same porosity and a similar µCT pore volume distribution. Based on the µCT data, up to 30 % of the pores in the Savonnières and 20 % in the Tabaire consisted of pores bigger than 557 µm. These large pores can cause a significant amount of variability between the different samples. A variability that is too large to be representative according the volumes studied in here. These pores are also the main contributors to the variability within the total µCT porosity.

The water immersion porosity and µCT porosity values are representative and in agreement with the literature (Roels, 2000; Lanzón et al., 2014; Qajar et al., 2013). An important difference between the pore structures of the two stone types, is their connectivity. Because of limited spatial resolution (± 8 µm) in the µCT data, pore throats with a diameter equal or below ± 8 µm could not be detected. Therefore, a high percentage of so called “µCT closed porosity” was determined in Savonnières (table 5.1). However Mayo et al., (2015) determined with synchrotron µCT and xenon that Savonnières contains no isolated pores, meaning that the so called “closed pores” are all interconnected by micropores. This was less the case within Tabaire where µCT data and a visual inspection already show a better connectivity between the pores. The absence of closed pores within both rocks is in agreement with the high porosity for the water immersion method and the MIP results.

About 50 % of pores was smaller than the voxel resolution and interpreted as micropores. This was also the case for Tabaire, even though the high connectivity of the macropores. The amount of micropores is in agreement with the findings of Qajar et al., 2013 for Savonnières. The retrieved pore size distribution by MIP is not representative, especially for Savonnières due to the heterogeneity and its complex pore network (Diamond, 2000; Berodier et al., 2016). The MIP results give however a clear indication about the accessibility of the pores by Paracoccus denitrificans: An important amount of the micropores is not accessible for Paracoccus denitrificans as they need at least 2 µm (or twice their size (Updegraff, 1982)). Within Tabaire 62 % of the micropore (access) were smaller than 2 µm, while for Savonnières it reached even 84 %. It indicates that some larger pores are not accessible for the bacteria due to the pore throats. This will be especially the case for Savonnières, as also an important percentage of the macropores are disconnected. Within Tabaire this effect will be limited due to the high interconnectivity of the macropores.

6.2. Flow experiments – Bacterial adhesion in porous limestone

Paracoccus denitrificans is assumed to be negatively charged during all the experiments, as most bacteria (Rijnaarts et al., 1995; 1999; van der Wal et al., 1997; Poortinga et al., 2002). Tarada et al., (2005) measured even a negative ζ-potential for Paracoccus denitrificans at a pH of 7.2. Surface charge is pH dependent and can be positive or negative. A high pH will favour excess of negative species, while positive species will be favored at lower pH. Other important factors include the surfactant concentration and the ionic strength (Alotaibi et al., 2011). The surface charge of the calcite is complex mainly due to its solubility, governed by the chemical equilibrium and surface electrical charge (Rodríguez and Araujo, 2006). Literature, is quite confused and different charges have been assigned to calcite and limestone (Rodríguez and Araujo, 2006; Alotaibi et al., 2011). The isoelectric point of

50

calcite is 9.5 and some people believe that below that pH the charge of calcite is positive (as explained for quartz in Molnar et al., 2011). Rodriguez-Navarro et al., (2012) used this interpretation as an argument for better attachment of bacteria on calcite than on quartz. Studies looking specifically at ζ- potential in function of pH clearly indicate however the opposite. The charge of the surface with a pH lower than the isoelectric point is negative and vice versa (Alotaibi et al., 2011). Alotaibi et al., (2011) stated at pH 7 and 25°C that seawater generates positively charged (6.8 mV) limestone particles, deionized water negatively charged (-11.2 mV) limestone particles, just as NaCl solution (-7.9 mV) and aquifer water (-7.5 mV). Mahani et al., (2015) reported similar results. For this reason it is assumed that Savonnières and Tabaire are most likely also negatively charged within the salt solution of Evian with 10 g/L NaCl. See Alotaibi et al., (2011) and Mahani et al., (2015) for more information about the relation with surface charge, temperature, salinity, rock type, temperature,…

Figure 2.6 summarizes the interactions between the bacteria and the surface during the experiments. During the “impregnation phase”, a solution with a high ionic strength has been used. This enables adhesion due to a low energetic barrier and deep secondary minimum (figure 2.6a) (Konhauser, 2007). A plateau was rapidly achieved (figure 5.4a,c), indicating a limited number of bacteria retention sites. The plateau was reached at about the same concentrations of Paracoccus denitrificans for both rocks. It had for Tabaire (and most likely for Savonnières as well) the same magnitude as its initial bacterial solution. There was some variation but it is insignificant as many factors like the dilutions for the measurements caused errors. Some variation might also be related to ripening, when the bacteria serves as new attachment sites for bacteria retention (cfr. colloids Molnar et al., 2015) This is also the reason why a log-scale has been chosen for figure 5.4 as such experiments will never be exact, and of which only the main trends are important (Jarvis et al., 2007).

Demineralized water removed the cations, and inhibit bacterial adhesion because of electrostatic repulsion (figure 2.6b). It flushed the bacteria out of the system, causing an exponential decrease of bacteria in figure 5.4b,d. Excel describes this trend with an exponential function with a R-value > 0.93. Bacteria were still leaving the stone after 40 droplets and even with the assumed errors it shows that a significant of amount of bacteria retained within the rock samples: indicating that not everything was flushed out directly. The concentrations measured during the “elution phase” indicate that the bacteria do not attach physically to the surfaces, as they did not become immobilized. A slow release of the bacteria from the rock samples, referred in literature as tailing (Molnar et al., 2015), is also absent. Complete detachment can however not be proven as Tufenkji and Elimelech (2004, 2005), Bradford et al., (2007) showed that reducing the ionic strength of a solution does not necessarily release all the colloids. They assumed that other mechanisms are important as well such as pore space geometry. The pore space geometry creates hydrodynamically disconnected regions (“immobile regions”). Once bacteria adhere onto the surface of such a region, hydrodynamic constraints can impede its release when reducing the secondary minimum (Torkzaban et al., 2008). This effect has not been detected as the results for Savonnières and Tabaire were similar. A lack of spatial resolution is a main reason: the rock samples are very small and contain a pore volume of around 120 mm³ (= 0.12 mL), while only larger water volumes (minimal 0.5 mL) have been sampled, due to practical limitations.

There was furthermore a certain amount of “dead fluid” in the setup: when the solution was changed at the start of the “impregnation” and the “elution phase”, the flow cell still contained the fluid of the previous step. For this reason the first droplets belong to the previous step of the experiment. The difference in opacity between the bacterial solution and the demineralized water determines the “dead volume” at about two droplets. It might partly explain the high and low concentration of respectively the start of the “elution” and “impregnation” phase. Its influence is however limited and therefore did not affect the further signal.

51

These experiments reveal, although it is only an estimation, a rapid saturation with bacteria, after less than five droplets. It uncovers furthermore the adhesion of the bacteria in the porous rocks. This will allow in the future to study bioclogging using such a setup. It is a first step to study the colonization of these bacteria. The bioclogging experiments could not be executed due to ongoing methodological development of the setup and these experiments are foreseen to be recaptured in the near future (See Appendix 9.1 for more details).

6.3. Effects of X-ray radiation on Paracoccus denitrificans

In general the gas concentrations increased significantly during the first cycle compared to residual gas. There are however two exceptions that do not follow the trend: samples S10 and T10. The cause is unknown and can be natural variation. They have however one thing in common: they have both been scanned directly after impregnation to visualise the residual gas phase. This means that these samples, including the bacteria have been exposed to radioactive radiation at the start of the growth. Therefore, there were some doubts concerning the effect of X-ray radiation and if this could affect the bacteria population, as this is important to know for further studies with µCT.

The effect of X-ray radiation during µCT experiments on bacteria is contested as multiple experiments has been executed. Most experiments favour a limited effect of X-ray radiation. Bacteria have a very high tolerance for radiation: a dose of 20-70 kGy is for example necessary to sterilize soils (McNamara et al., 2003). The doses here were much lower: (30.18 and 116.98 Gy for Savonnières, and between 30.31 and 117.40 Gy for Tabaire). This relative low dose of ionizing radiation during µCT scans combined with the tolerance of bacteria for radiation, suggests a minimal impact on bacteria during µCT scans (Bouckaert et al., 2013). Bouckaert et al., (2013) concluded that µCT is very well compatible with biological soil experiments, as it caused only small effects. Zappala et al., (2013), Schmidt et al., (2015) and Kravchenko et al., (2014) confirm these results. Zappala et al., (2013) focused on the impact of total radiation on rhizospheres and microbial biomass and found only an insignificant effect of X-ray radiation. Schmidt et al., (2015) looked in detail to the influence of µCT on several microbial parameters. The conclusion here was similar, as it exhibited no significant differences among scanned and unscanned samples. Kravchenko et al., (2014) at least focused on dry soil samples and discovered only a minor influence. Fischer et al., (2013) disagrees however and concluded a strong impact of the ionizing radiation. It caused big differences in biological parameters between the scanned and control samples. It changed the microbial communities within the soils due to death of selected microbial groups. This study agrees with observations in Chernobyl where soils samples within the radioactive zone showed more radio-resistant bacteria (Zavilgelsky et al., 1998).

The aforementioned studies focused on bacterial populations within (wet) soil samples. By our knowledge the effect of radiation has not been conducted on bacteria in and around rock samples or on the effect on specific species like Paracoccus denitrificans. This species might be by coincidence less tolerant or the water within the samples during scanning might have played a role. Water might decrease the dose to harm Paracoccus as it increases (in wetter soils at least) the effects of indirect radiation: radiolysis of water enhance free radical formation. (Jackson et al., 1967; McNamara et al., 2003). The growth experiments took furthermore one week, less than the time lag of three weeks between µCT and microbiological analysis, that could be advisable according to Bouckaert et al., (2013). Fisher et al., (2013) observed furthermore a strong regeneration after one week of µCT exposure. The differences between the exposed and unexposed communities’ structures decreased after seven days.

For these reasons an experiment focused on Paracoccus denitrificans and HECTOR has been conducted. Our experiment indicates no effect of radioactive radiation, not a direct one and not after three days. The change of biological activity is negligible. The optical density curves illustrates (figure 5.6) a comparable growth of the radiated and non-exposed bacteria. The bacteria grew only faster in

52

the TSB medium with Ca(NO3)2. Likely, they grew so fast that they poisoned themselves. The other cultures with exposed and non-exposed bacteria in the medium lacking Ca(NO3)2 grew slower but reached higher optical density values, so higher cell densities. The radiation did furthermore not directly kill a significant part of Paracoccus denitrificans as the flow cytometry did not detect any dead cells. Even three days, of which two days in the fridge and one day in the incubator, did not change these results significantly (table 5.2). There is no problem with the differentiation between the viable and dead cells as the life and dead staining revealed clearly the dead bacteria, killed by heating (figure 5.5). This data clearly shows the robustness of Paracoccus denitrificans making it an appropriate species to study with µCT. It also means that there has to be another reason why the gas production for the samples S10 and T10 was much lower.

6.4. Growth experiments

6.4.1. Biological activity – Chemical analyses

The concentration of the VFA gives an estimation of the bacterial activity during a growth cycle as fatty acids are a constituent of lipids produced by the bacteria (Willey, 2009). The low VFA concentration during cycle 1 and especially during cycle 3, indicates a lower bacterial activity. This is caused by the low nutrient concentration during cycle 3 and verified by low gas saturations. It will have been higher during cycle 1 as revealed by the high gas saturations at the end. Cycle 2 experienced a very high biological activity, revealing a positive effect of the initial high pH on their growth. This is unexpected as a pH of 7.6 is regarded as the optimal growth condition for Paracoccus denitrificans (Nokal and Schlegel, 1983; Kelly et al., 2006). The blanks contained about the same amount of VFA’s, indicating that these samples did not remain sterile, and that something grew inside it. Most likely it will not have been Paracoccus denitrificans as no significant gas phase was present after one week (table 5.4.).

VFA’s are acids and will lower the pH values. The denitrification pathway explains the pH rise. Rust et al., (2000) measured for example a pH increase from 7 to beyond 8 within five days, because of denitrification. The final pH after one week is an interplay between these parameters (Drtil et al., 1995). The high concentration of VFA during cycle 2 lead to a decreasing pH trend, while it is not abundant enough to reverse the pH increase during cycle 1 and 3 by denitrification. All the interactions are buffered by the calcium carbonate within the sample, that prohibits large pH fluctuations. Savonnières contains more calcium carbonate, and this might explain the slight elevated pH values compared with Tabaire.

6.4.2. Reaction products – Gas

Now the micro-environment and the growth conditions during the experiments are known, this can be used to interpret some of mechanisms and parameters of the gas production, distribution and MICP.

6.4.2.1. Residual gas phase

Before we can interpret the gas after one week, we need to understand the residual gas phase (= starting point). Although impregnation occurred under vacuum, gas remained in every case within the pore network. The number of residual gas is small compared to the gas after one week. It “disappeared” mostly and the few bubbles present after one week were larger: the bubbles found their way out of the pores, coalesced, dissolved again in the water or were consumed by the bacteria.

53

The residual gas phase shows the influence of two different factors: the rock type with its specific pore structure and the cycle conditions. The data reveals that the pore structure is the main contributor to the number of trapped residual gas bubbles and the total gas volume. Savonnières with its complex pore structure and narrow pore throats can trap residual gas more effectively than Tabaire. The pore structures has no effect on the trends as shown by the cumulative graphs (figure 5.7b,d). The trends of the Savonnières and Tabaire are similar and alternate with each other. They can be divided into two groups: one for each growth cycle.

The average volume, equivalent diameter as displayed in table 5.5 do not differ significantly between the two rocks but they differ between the cycles. The small difference in resolution might explain the increased number of bubbles and the volume between the cycles (table 4.3). In all the µCT image- analysis the same number of µCT slices has been studied. It indicates that for the second growth cycle a volume of interest that is 7.5 % greater has been studied, leading automatically to an increase of the same percentage in absolute volumes and number of bubbles. The increases are however significantly larger and has no effect on the average volume and equivalent diameter of the bubbles.

It is assumed that that the bacteria do not produce biogenic gas immediately after impregnation. This is important to account for the residual gas, certainly during the first cycle. Gas production would result in a higher estimation of gas phase. The comparison with the results of the blank samples verifies this assumption: the residual gas concentration was higher within these blank samples. The increased gas concentrations of the second cycle, combined with an increased number of gas bubbles, (average) volume and average equivalent diameters can be linked to the impregnation process (or the small differences in solutions). The process was similar, however the samples of the second cycle stayed a few minutes under vacuum after impregnation. The diverse metabolism of Paracoccus denitrificans can explain the lower gas saturation during the first cycle: Paracoccus denitrificans can grow aerobically using O2 as terminal oxidant for respiration (Nokhal and Schlegel, 1983). This respiration pathway provides the most energy and is preferred above denitrification (Shapleigh, 2006, 2013). For this reason it is possible that Paracoccus denitrificans consumed already some of the residual gas bubbles before or during the µCT scan.

So in general these results indicate that the pore structure determines how easily gas bubbles got trapped, while the process of vacuuming, impregnation and the presence/absence of Paracoccus denitrificans has an effect on the bubbles their sizes, volume,… These results also seem to verify that the biogenic gas production did not occur immediately after impregnation. It proves that vacuum impregnation is a good method to study biogenic gas production as only a limited number stays behind, that mostly disappeared afterwards.

6.4.2.2. Final gas phase after one week

The gas concentrations increased significantly compared with the residual gas. This “extra” gas is biogenic and results from the denitrification of the Ca(NO3)2 by Paracoccus denitrificans. It proves that the bacteria can survive vacuum impregnation. We assume here, in analogy by Satik and Yortsos (1996) and Istok et al., 2007 the follow mechanism: Nitrogen gas forms initially microscopic bubbles around the pore walls. It grows radially within the pore body and coalescence to larger bubbles. Their form will be initially spherical and after filling the pore, and taking its shape, the bubble will start to invade a neighbouring pore throat. To do this it needs an high vapour pressure, which will be larger for smaller capillaries. During the pressurization the system is stable till the bubble eventually invades another pore when the vapour pressure exceeds the capillary barrier of the pore throat. This can occur with one or multiple pore throats a time. This process will eventually lead to gas saturated channels linked to the surface, allowing the flow of additional nitrogen gas out of the system.

54

Savonnières and Tabaire have a different pore network. Tabaire has an open pore network, allowing an easy access for the bacteria and gas bubbles to invade other pores. The complex pores and the narrow pore throats of Savonnières on will not all contain bacteria and only allow bubbles to invade other pores after a significant vapour pressure increase. VFA measurements determined the highest amount of biological activity during cycle 2, assuming most biogenic gas formation. The low nutrients during cycle 3 led to lower gas concentrations and shows a clear relation between nutrient availability and gas formation.

The first and second cycle are more complex as the trends differ for both rocks: gas saturations were for Tabaire significantly lower during cycle 1. A lower gas production can be a possible explanation. Savonnières on the other hand contains within these samples more gas. It is possible that the gas production within these samples was higher. Furthermore if the bacteria produce gas within the rocks, the gas has more chance within Savonnières to stay inside the pore network

Increasing the pH and the biological gas production during cycle 2 led to more gas bubbles. Its effect on the gas saturation is however more complicated. (Note due to the small difference in resolution the absolute results of table 5.7 should be corrected by 7.5 % like for the residual gas phase). The gas saturation increased for T20 and T21 to 25 % with clearly connected channel forming gas bubbles (figure 5.12f). It indicates that this saturation was around the maximum possible gas saturation within this stone. When more gas is produced, it will probably been flushed out of the rock, by the gas channels. This value is comparable to the 23 % measured by Istok et al., (2007) in a packed sedimentary column. Within Savonnières on the other hand, only one sample experienced an increase in gas saturation compared to cycle 1. An increase in gas production, can be an explanation. The increase was too low to lead to a significant pressure build up to invade other pores. Some larger pores contained no gas, potentially linked to the absence of bacteria by its inaccessibility to penetrate through all the small pores. The lower values for S21 and S22 are non-expected. Savonnières is a heterogeneous stone, with different pore structures (bivalves, ooids, dissolution features,…) so some structural or biological variability cause this. The enhanced biological activity and the weight increase during this cycle can affect the pore structure by MICP and bioclogging. This can clog the already narrow throats even more and increase the necessary vapour pressure to invade other pores. It is however not possible to verify these effects due to a lack of clear visualization of these phases.

In every cycle there are samples that experienced a variability which is so important that no structural property within the rocks, can cause this. For this reason it has to be a variation in the biological process, contamination,… . Within cycle 1 there are for example two samples with a significant lower gas saturation: S10 but especially T10 (table 5.4). A contamination is the most likely explanation as previous results exclude a negative impact of X-ray radiation on the growth of exposed Paracoccus denitrificans. Its effect on gas production is however not studied here. The absence of biogenic gas within sample T22 is another example where an with an abnormal gas saturation. In this case it is combined with an anomalous pH (table 5.3) and an odd white substance in the solution. A possible contamination is also for this sample the most plausible cause. Cycle 3 at least has also a strong variability. A threshold value between significant and no gas production can cause this variability where two out of three contained no gas. A contamination can however not be excluded.

The process of bubble formation and migration is dynamic. During this thesis the gas distribution is checked once after one week. Adding to this the uncertainty about the total biogenic gas production, the location of the bacteria, the extensive natural variability within biological processes, the study of only a limited number of samples with µCT, makes any interpretation with such varying results very difficult.

The gas distributions are more straightforward and seem to be related to the pore network and pore throats. They are rather independent of the amount of biogenic gas, as long as enough gas is present.

55

The number of gas bubbles was higher within Savonnières, like for the residual phase. The bubbles within the Savonnières were more confined to one pore compared with the Tabaire rock (figure 5.11 and 5.12). It induces a higher total number of gas bubbles, but a lower mean volume of each one, as confirmed in table 5.7. Tabaire contained some very large gas bubbles spread over multiple pores that could reach 20 % of the total gas volume, while such bubbles were absent in Savonnières.

The pore sizes determines the gas distribution of the smaller bubbles that are confined to one pore. This was similar for both rocks (figure 5.3) and results in comparable distributions (figure 5.8b, 5.9b, 5.10b) and similar modes (table 5.7). The average equivalent diameter of the bubbles, its (average) volume and the number of the bubbles are impacted by the occurrence of a large bubbles. This is especially the case for Tabaire. It is the main reason for the higher average volume and equivalent diameter within this rock. It explains for example the very high average volume and equivalent diameter of a bubble in T11. The formation of these large bubbles is directly linked to the connectivity and pore throat diameter. The open pores within Tabaire let some bubbles grow within multiples pores, while Savonnières kept in it confined within one pore. These lager bubbles impact the size distributions and the mode of the bubbles only slightly, as there are still enough small bubbles present, that stayed isolated.

The difference between Tabaire and Savonnières decreases as less gas was produced (cycle 3). In these samples the bubbles were spherical and less influenced by the pore throats as most of them remained significantly smaller than a pore. The connectivity between the pores plays here a smaller role and there were less large bubbles to impact the average volume of a bubble,… etc. (analogous with the residual gas phase, which revealed no differences in average bubble size,… between the two rocks).

3D images shows (figure 5.12) that gas bubbles reside preferentially in the centre of the rocks, especially during low gas saturations. The gas at the centre has the most chance to get trapped. The edge of the top part of the rock contained more bubbles, and does not follow the previous interpretation. A possible mechanism here can be the buoyancy of the bubbles, bringing more bubbles to the upper regions of the samples. This sets an interplay between the loss of gas bubbles by buoyancy and diffusion out of the rock and their supply not only by local biogenic production but also by the same buoyancy.

6.4.3 Reaction products – Microbially induced calcite precipitation

As MICP is composed out of the same material as the rock samples, it is very hard to detect it. The experiments do not detect clear MICP precipitation. It should be present as denitrification is a pathway to MICP for where Paracoccus denitrificans has been considered for concrete applications (Erşan et al., 2015, Erşan, 2016). The weight increase, zones of addition on the difference images and the rare distinct crystals detected with optical microscopy assumes MICP. The µCT difference images are hard to use, as the quality of the scans is not always the same: the reference scans (figure 5.13a,d) are sharper than the scans at a later stage of the cycle (figure 5.13b,e). This causes some artefacts on the difference images (figure 5.13c,e). Many possibilities can cause this, including the use of different reconstruction techniques, other filters, movement of the samples, differences in X-ray flux during scanning,… All these possibilities have been checked, making the cause still unclear. The vague scans make the edges of every grain visible on the difference images. There is also some displacement visible in these difference images: some grains had at one side a white border and on the other a black one. It makes any quantitative estimation impossible. The zones of addition around the sample are most significantly at the top of the samples during cycle 1. This follows the expectations as the bacteria leading to the MICP also floated above it. Unfortunately this is not detected during the other cycles.

56

The weight increase is very low, and the solution, or the bacteria itself can partially cause this. After removal of the medium with bacteria, some fluid with dissolved particles stayed behind within the rock samples. This will cause crystallization within the samples before the final weighting. Table 6.1 illustrates an estimation of the contribution of the bacterial cells itself and the solutes. These are minimum values as possibly a significant amount of the solution stayed behind adhered to the rock and walls of the box. It illustrates that the weight increase for the first cycle is not significant, in a lesser amount also for the third growth cycle. The values of the second cycle are low but still considerably higher than the estimated extra weight by the bacteria itself and the growth medium. This increase is expected as the initial pH here was increased to 8.5. Those values are however small compared to the weight increase reported by Willem De Muynck et al., (2011). They conducted a biodeposition treatment with ureolytic pathway, and measured a weight increase of about 4.51 % for Savonnières after the treatment. The lowest values were reported for Massangis stone but was with its 1,49 % still considerable larger than during these experiments. The weighting results disagrees with the difference images, as a lower amount of addition is detected during the second cycle including the absence of the white border on the difference images (figure 5.13c,f).

The blank samples of the second cycle also experienced a considerable weight increase. The main reasons here has to be precipitation due to the initial pH of 8.5, the crystallization after evaporation of the liquid that was left behind and maybe some precipitation as the samples did not stay sterile. The weight increase of T22, is not related to MICP by Paracoccus denitrificans as no significant gas production has been detected here. Precipitation by the solution (mark the final high pH of this sample, enhancing precipitation) and a possible contamination, might here be the main cause. Dissolution can occur as well, however besides the minor weight decrease in T12 and some zones of removal on the difference images, it has not been detected. The pH stayed for all samples also above 7.5, alkaline enough to avoid significant dissolution. It is important to note that the measured pH gives an indication of the pH within the rock samples, however locally it could be different due to local bacterial growth and specific environmental conditions.

Biological activity can explain the variation of the weight differences between the different samples. An overall decreased biological activity during cycle 3, induced less variation. The larger variation on the other hand during the first cycle, can be an artefact due to its very low weight increase. The different growth conditions within the cycles, caused the differing weight increases. The initial elevated pH increased the precipitation during the second cycle. The low nutrient conditions and biological activity during the last cycle decreased the potential of MICP. The subtle dissimilarities between the two rocks, can be linked to the composition, as Savonnières induced a slight higher pH (table 5.3) after one week. The pore network will also have an important impact. MICP is expected to be less widespread within the interior of Savonnières as Paracoccus denitrificans cannot access all the pores. The results however do not let us verify this assumption.

The optical microscopy assumes some MICP that it is not clear again. The assumed MICP crystals are also significantly smaller then reported in literature by De Muynck et al., 2008, where some crystals reached around 100 µm (figure 2.7). The MICP detected over there was widespread across the surfaces, which is here certainly not the case. There are however some indications as the crystals in figure 5.14b, lay on a cut ooid. The SEM images verifies these results with the small crystals in figure 5.15a but large crystal as in figure 2.8, with a small hole, has not been found. The edges of the samples were at least also very sharp like in figure 5.15b indicating that nothing precipitated upon it.

57

Optical microscopy and SEM images verifies that µCT is not able to detect clear MICP in this case as the crystals of MICP, are maybe below the resolution. There are multiple reasons why MICP during the experiments could be that low. The denitrification pathway is less appropriate to study MICP, as it achieves MICP rates that are 100 to 1000 times slower than using the ureolysis pathway (Erşan, 2016). The VFA formation is another problem as it counteracts the pH increase induced by denitrification.

Table 6.1: Estimation extra weight dueo bacteria and growth medium based on an average pore volume of 0.12 mL per sample for the three growth cycles. With (1) Northeast Laboratory Services (2013), (2) Neidhardt (1996), (3) Evian (2014). Estimation extra weight due to bacteria and growth medium based on an average pore volume of 0.12 mL per sample Extra weight (g) Extra weight (g) per Concentration per sample - Cycle sample - Cycle 3 (10 % g/L 1+ 2 (100 % TSB) TSB - 90 % Evian) TSB medium (1)

Enzymatic Digest of Casein 17.0 2.04E-03 2.04E-04 Enzymatic Digest of 3.0 3.60E-04 3.60E-05 Soybean Meal NaCl 5.0 6.00E-04 6.00E-05 Dextrose 2.5 3.00E-04 3.00E-05

K2HPO4 2.5 3.00E-04 3.00E-05 +

Ca(NO3)2 1.2 1.44E-04 1.44E-05

SUM 3.74E-03 3.74E-04

Bacteria Bacteria in solution (#/mL) 3.00E+09 3.60E+08 bacteria 3.60E+07 bacteria Dry weightTSB one) bacteria bacteria/mL2.80E-13 g Dry weight(g) bacteria(2) (g) TSB 1.01E-04 1.01E-05

Evian – Total dissolved 0.5 / 6.00E-05 solids (3)

Total extra weight 0.00384 0.00044

58

7. FUTURE RESEARCH AND CONCLUSIONS

This preliminary research uncovered the complex interactions between microorganisms and rocks. It introduced the effects starting from bacterial adhesion to growth in porous limestone, of which only bioclogging could not be studied directly. This research was unique as it brought the expertise of three different departments together: Geology, Bioscience Engineering and Physics. The results are satisfying and give a clear overview of the challenges and the potential for future research. The main conclusions of the thesis are the challenging nature of working with microorganisms in a not-fully controlled environment, such as rocks that each have their unique signature and heterogeneity. The organisms made it difficult as they have the power to change their environment and create other growth conditions, then anticipated. The data revealed however the potential of µCT in geological microbial research as it visualized the pore network, natural gas production and assumed MICP in 3D.

Water immersion porosimetry, combined with µCT and MIP described successfully the general physical habitat in both rocks types. It revealed an open versus a closed µCT pore network. The large pore throats within Tabaire made every pore accessible for Paracoccus denitrificans while many pores were impassable within Savonnières.

Bacterial adhesion is the first step of bacterial colonization. Counting Paracoccus denitrificans flushed through rock samples with the flow cytometer proved that the bacteria do adhere easily to porous rocks. This reveals its potential for further bioclogging experiments. The setup and the small rock samples did not allow any reliable quantitative information. Saturation was reached very rapidly and the discrete samples for our measurements were too big compared to the pore volume. A new setup comparable with the one developed for the bioclogging experiments, including the syringe pump and manometers, could control the experiments better and lead to more quantitative information. Continuous optical density measurements can supplement the direct measurements with a flow cytometer in the future. This will increase the resolution and will potentially uncover differences between the different types of porous materials.

Unfortunately it has not been possible to study bioclogging, due to some technical issues with the setup. The setup has already been upgraded during this thesis and the last problems have been documented precisely. This includes a leak in the flow cells, that will be repaired during upcoming months. Hereafter it should be possible to induce clogging and uncover it with µCT.

The next step in this thesis involved the study of the growth of microorganisms within porous limestone , using µCT. This technique proved to be very promising, as the radiation of one scan did not affect the growth of Paracoccus denitrificans. This means that µCT can likely follow the growth of this bacteria within the porous limestone without influencing it due to X-rays. This will be studied in more detail in the near future, so we can exclude this effect completely.

µCT resolved bacterial growth and detected the gas bubbles within the rocks with the utmost detail. The bacteria produced a significant amount of biogenic gas during one week. The gas production is related to the biological activity. The biological activity was positively influenced by a higher pH of 8.5 and a high amount of nutrients. The results revealed however a natural variability, that impacted the results and interpretations significantly. Unfortunately, some samples have also been contaminated. These variations illustrated that we cannot fully control the growth of Paracoccus denitrificans within porous sedimentary rocks at this moment. To control the different parameters, future research should start from sterile samples that will exclude any potential contamination. It should also begin with homogeneous samples, with a fixed pore structure and composition, such as glass beads. This all will allow to unravel and document the biological variation. µCT was furthermore limited during this project as it can only study the biological variability on a few samples. For this reason it needs to be supplemented with some extra methods, that are capable to study many samples such as VFA

59

extractions. These extractions proved e.g. to be a good indicator for the biological activity. A new setup should also be created that can collect all the biogenic gas, as a lot of the biogenic gas escaped during the experiments. A bigger sample could be impregnated and placed with the solution inside a box connected on top with a balloon. The nitrogen gas trapped inside the rock sample will cause a rise in the water level, while the other bubbles will fill the balloon. Measuring the air pressure above the water level could be another possibility.

µCT imaged successfully the gas bubbles and resolved the effects of pore size distribution on the biogenic gas bubbles. These distributions were not affected by the biological gas production, as long as enough gas was present. Their sizes were affected by the pores, while the pore throats and the connectivity determined the presence of large bubbles and gas saturated channels. Savonnières and Tabaire both contained a similar gas distribution. Large bubbles formed within Tabaire due to the open porosity, while most bubbles stayed isolated within Savonnières. These large bubbles affected the gas volume distribution significantly by incorporating an important percentage of the total gas volume.

The residual gas at the start of the growth cycle, was insignificant and disappeared mostly after one week. It distribution was for both rocks linked to the impregnation process, or the presence/absence of bacteria. The pore structure affected only the number of trapped gas bubbles.

The goal to describe the presence of MICP by Paracoccus denitrificans was not succeeded as less precipitation formed than anticipated. Theoretically it should be present, and it is assumed by the weight increase, µCT, SEM and optical microscopy. Clear evidence has however not been found. The VFA production countered partially the beneficial conditions of denitrification on MICP. Paracoccus denitrificans produced more acids during higher biological activity, that were abundant enough to shift an initial pH increase to a significant decrease. This species or the denitrification pathway in general might not be the best way to study MICP. In the future we might solve this issue by simulating multiple cycles within the same sample or by shifting to ureolytic bacteria.

During this research we only looked at the initial situation and the situation after one week. Microbial processes are however continuous in time. To understand this better, continuous visualization or visualization within shorter time intervals should occur in future research. µCT should be the perfect tool, since the bacteria probably will not be affected by the multiple exposure to X-ray radiation.

Future experiments could also be directed towards finding the locations of bacteria attachment and biofilm growth within porous rocks. In 3D dimensions this would be possible with e.g. contrasting agents. Those experiments can include 2D visualization methods with specialized microscopes such as the cryo-SEM techniques developed at the MAPS spinoff of RWTH Aachen. The microbial habitat will be investigated in more details including other properties of the porous rocks such as the roughness. 4D µCT, like EMCT at UGent could monitor directly how the gas bubbles, the precipitation alter the flow paths end evolve. It could also reveal the locations with significant bioclogging even without direct visualization.

This research is only a starting point of a bigger research project that will be developed in the upcoming years. The results give a starting point and revealed the potential and challenges of the different methods and experiments.

60

8. REFERENCE LIST

Achal, V., Mukerjee, A., and Reddy, M. S. (2013). Biogenic treatment improves the durability and remediates the cracks of concrete structures. Construction and Building Materials, 48, 1-5. Akob, D. M. and Küsel, K. (2011). Where microorganisms meet rocks in the Earth's Critical Zone. Biogeosciences, 8(12), 3531-3543. Alhede, M., Qvortrup, K., Liebrechts, R., Høiby, N., Givskov, M. and Bjarnsholt, T. (2012). Combination of microscopic techniques reveals a comprehensive visual impression of biofilm structure and composition. FEMS Immunology and Medical Microbiology, 65(2), 335-342. Alotaibi, M. B., Nasr-El-Din, H. A., and Fletcher, J. J. (2011). Electrokinetics of limestone and dolomite rock particles. SPE Reservoir Evaluation & Engineering, 14(05), 594-603. Ambrose, R. J., Hartman, R. C., Diaz Campos, M., Akkutlu, I. Y. and Sondergeld, C. (2010). New pore- scale considerations for shale gas in place calculations. In SPE Unconventional Gas Conference, 23-25 February, Pittsburgh, Pennsylvania, USA. Amils, R. (2015a). Chemotroph. In M. Gargaud, W. M. Irvine, R. Amils, H. J. Cleaves II, D. L. Pinti, J. C. Quintanilla, … M. Viso (Eds.), Encyclopedia of Astrobiology (pp. 438–439). Berlin, Heidelberg: Springer. Amils, R. (2015b). Denitrification. In M. Gargaud, W. M. Irvine, R. Amils, H. J. Cleaves II, D. L. Pinti, J. C. Quintanilla, … M. Viso (Eds.), Encyclopedia of Astrobiology (pp. 624-625). Berlin, Heidelberg: Springer. Andriani, G., and Walsh, N. (2003). Fabric, porosity and water permeability of calcarenites from Apulia (SE Italy) used as building and ornamental stone. Bulletin of Engineering Geology and the Environment, 62(1), 77-84. Arana, R., Mancheño, M. A., Martínezz, J.I. M., Estrellaz, T. R., Martinez-Conde, J. A. R. and Serrano, F. (2003). Las canteras de" roca tabaire" de canteras (Cartagena, Murcia). Contexto geológico e importancia como patrimonio geológico y minero. In I. Rábano, I. Manteca, and C. Garcia, (eds.) Patrimonio geológico y minero y desarrollo regional, Publicaciones del instituto geológico y minero de España, Madrid, Spain, (2), 75-85. Atekwana E.A., Werkema, D.D., Atekwana, E.A. (2006) Biogeophysics: The effects of microbial processes on geophysical properties of the shallow subsurface. In: H. Vereecken, A. Binley, G. Cassiani, A. Revil. K., Titov (Eds.) (p.p. 161–193) Applied Hydrogeophysics, Dordrecht: Springer Netherlands. Azeredo, J., Azevedo, N. F., Briandet, R., Cerca, N., Coenye, T., Costa, A. R., ... and Kačániová, M. (2016). Critical review on biofilm methods. Critical Reviews in Microbiology, 1-39. Baird-Parker, A. C. (1965). The Classification of Staphylococci and Micrococci from World-wide Sources. Journal of General Microbiology, 38(3), 363–387. Bamforth, C. W. and Quayle, J. R. (1978). Aerobic and anaerobic growth of Paracoccus denitrificans on methanol. Archives of Microbiology, 119(91), 91–97. Barbesti, S., Citterio, S., Labra, M., Baroni, M. D., Neri, M. G., and Sgorbati, S. (2000). Two and three‐ color fluorescence flow cytometric analysis of immunoidentified viable bacteria. Cytometry, 40(3), 214- 218. Baveye, P., Vandevivere, P., Hoyle, B. L., DeLeo, P. C. and de Lozada, D. S. (1998). Environmental impact and mechanisms of the biological clogging of saturated soils and aquifer materials. Critical reviews in environmental science & technology, 28(2), 123-191. Beijerinck, M. W. and Minkman, D. C. J. (1910). Bildung und Verbrauch von Stickoxydul durch Bakterien. Zentbl Bakteriol Parasitenkd Infektionskr Hyg Abt II25, 30-63.

61

Berodier, E., Bizzozero, J. and Muller, A.C.A. (2016). Mercury intrusion porosimetry, In K. Scrivener, Snellings and B. A. Lothenbach (eds.) practical guide to microstructural analysis of cementitious materials. (p.p. 419-444), Boca Raton: CRC Press. Beyenal, H., Yakymyshyn, C., Hyungnak, J., Davis, C. C. and Lewandowski, Z. (2004). An optical microsensor to measure fluorescent light intensity in biofilms. Journal of microbiological methods, 58(3), 367-374. Bielefeldt, A. R., McEachern, C. and Illangasekare, T. (2002). Hydrodynamic changes in sand due to biogrowth on naphthalene and decane. Journal of environmental engineering, 128(1), 51-59. Bigos, M., Baumgarth, N., Jager, G. C., Herman, O.C., Nozaki, T., Stovel, R.T., Parks, D.R. and Herzenberg, L. A. (1999). Nine color eleven parameter immunophenotyping using three laser flow cytometry. Cytometry Part A, 36(1), 36-45. Bin, L., Ye, C., Lijun, Z. and Ruidong, Y. (2008). Effect of microbial weathering on carbonate rocks. Earth Science Frontiers, 15(6), 90-99. Blows, J. F., Carey, P. J. and Poole, A. B. (2003). Preliminary investigations into Caen Stone in the UK, its use, weathering and comparison with repair stone. Building and Environment, 38(9), 1143-1149. Boivin-Jahns, V., Ruimy, R., Bianchi, A., Daumas, S. and Christen, R. (1996). Bacterial diversity in a deep- subsurface clay environment. Applied and Environmental Microbiology, 62(9), 3405-3412. Boone, M.N., Vlassenbroeck, J., Peetermans, S., Van Loo, D., Dierick, M. and Van Hoorebeke, L., (2012). Secondary radiation in transmission-type X-ray tubes: simulation, practical issues and solution in the context of X-ray microtomography. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 661(1), 7–12. Boquet, E., Boronat, A., Ramos-Cormenzana, A., (1973). Production of calcite (calcium carbonate) crystals by soil bacteria is a general phenomenon. Nature 246 (5434), 527–529. Bosak, T. (2011). Calcite Precipitation, Microbially Induced. In R. J. and V. Theil (Eds.), Encyclopedia of Geobiology (pp. 223–227). Dordrecht: Springer Netherlands. Bouckaert, L., Van Loo, D., Ameloot, N., Buchan, D., Van Hoorebeke, L., and Sleutel, S. (2013). Compatibility of X-ray micro-Computed Tomography with soil biological experiments. Soil Biology and Biochemistry, 56, 10-12. Bouwman, A. F. (1989). The role of soils and land use in the greenhouse effect. Netherlands Journal of Agricultural Science, 37, 13–19. Bradford, S. A., Torkzaban, S., and Walker, S. L. (2007). Coupling of physical and chemical mechanisms of colloid straining in saturated porous media. Water Research, 41(13), 3012-3024. Brunsting, A., and Mullaney, P. F. (1974). Differential light scattering from spherical mammalian cells. Biophysical journal, 14(6), 439-453. Cacchio, P., Ercole, C., Cappuccio, G. and Lepidi, A. (2003). Calcium carbonate precipitation by bacterial strains isolated from a limestone cave and from a loamy soil. Geomicrobiology Journal, 20(2), 85-98. Camerman, C., 1957. Description et emploi en Belgique et aux Pays-Bas des Pierres Blanches Françaises. Brussels: Hayez, 92 p.p. Caneva, C. and Ceschin, S. (2009). Ecology of biodeterioration. In G. Caneva, M. P., Nugari and O. Salvadori (Eds.), Plant biology for cultural heritage. (p.p. 31–39.), Los Angeles: Getty Publications. Castanier, S., Le Métayer-Levrel, Perthuisot, J.-P. (2000). Bacterial Roles in the Precipitation of Carbonate Minerals. In R. E. Riding and S. M. Awramik (Eds.), Microbial Sediments. (p.p. 32-39). Berlin, Heidelberg: Springer.

62

Chakravarthy, S. S., Pande, S., Kapoor, A. and Nerurkar, A. S. (2011). Comparison of Denitrification Between Paracoccus sp. and Diaphorobacter sp. Applied Biochemistry and Biotechnology, 165(1), 260– 269. Characklis, W. G. (1973). Attached microbial growths—I. Attachment and growth. Water Research, 7(8), 1113-1127. Costerton, J. W., Lewandowski, Z., Caldwell, D. E., Korber, D. R. and Lappin-Scott, H. M. (1995). Microbial biofilms. Annual Reviews in Microbiology, 49(1), 711-745. Cnudde, V., Masschaele, B., Dierick, M., Vlassenbroeck, J., Van Hoorebeke, L., and Jacobs, P. (2006). Recent progress in X-ray CT as a geosciences tool. Applied Geochemistry, 21(5), 826-832. Cnudde, V., Cwirzen, A., Masschaele, B., and Jacobs, P. J. S. (2009). Porosity and microstructure characterization of building stones and concretes. Engineering geology, 103(3), 76-83. Cnudde, V., and Boone, M. N. (2013). High-resolution X-ray computed tomography in geosciences: A review of the current technology and applications. Earth-Science Reviews, 123, 1-17. Council of the European Union. (1998). Council Directive 98/83/EC of 3 November 1998 on the quality of water intended for human consumption. Official Journal of the European Communities, L330, 32– 54. Cunningham, A. B., Characklis, W. G., Abedeen, F. and Crawford, D. (1991). Influence of biofilm accumulation on porous media hydrodynamics. Environmental science & technology, 25(7), 1305- 1311. Cuthbert, M. O., Riley, M. S., Handley-Sidhu, S., Renshaw, J. C., Tobler, D. J., Phoenix, V. R., and Mackay, R. (2012). Controls on the rate of ureolysis and the morphology of carbonate precipitated by S. Pasteurii biofilms and limits due to bacterial encapsulation. Ecological Engineering, 41, 32-40. Danin, A., and Caneva, G. (1990). Deterioration of limestone walls in Jerusalem and marble monuments in Rome caused by cyanobacteria and cyanophilous lichens. International Biodeterioration, 26(6), 397- 417. da Silva, F. B., De Belie, N., Boon, N., and Verstraete, W. (2015). Production of non-axenic ureolytic spores for self-healing concrete applications. Construction and Building Materials, 93, 1034-1041. Davies, K. J. P., Lloyd, D. and Boddy, L. (1989). The Effect of Oxygen on Denitrification in Paracoccus denitrificans and Pseudomonas aeruginosa. Microbiology, 135(9), 2445–2451. Davis, D. H., Doudoroff, M., Stanier, R. Y. and Mandel, M. (1969). Proposal to reject the genus Hydrogenomonas: Taxonomic implications. International Journal of Systematic Bacteriology, 19(4), 375–390. Davit, Y., Iltis, G., Debenest, G., VERAN‐TISSOIRES, S., Wildenschild, D., Gérino, M. and Quintard, M. (2011). Imaging biofilm in porous media using X‐ray computed microtomography. Journal of microscopy, 242(1), 15-25. De Graef, B., Cnudde, V., Dick, J., De Belie, N., Jacobs, P. and Verstraete, W. (2005). A sensitivity study for the visualisation of bacterial weathering of concrete and stone with computerised X-ray microtomography. Science of the total environment, 341(1), 173-183. De Las Cuevas, C. (1997). Pore structure characterization in rock salt. Engineering geology, 47(1-2), 17- 30. De Muynck, W., De Belie, N. and Verstraete, W. (2007). Improvement of concrete durability with the aid of bacteria. Proceedings of the first international conference on self healing materials. 18-20 April 2007, Noordwijk aan Zee, The Netherlands.

63

De Muynck, W., Cox, K., Belie, N. De and Verstraete, W. (2008). Bacterial carbonate precipitation as an alternative surface treatment for concrete. Construction and Building Materials, 22(5), 875–885. De Muynck, W., De Belie, N. and Verstraete, W. (2010a). Microbial carbonate precipitation in construction materials: A review. Ecological Engineering, 36(2), 118–136. De Muynck, W., Verbeken, K., De Belie, N. and Verstraete, W. (2010b). Influence of urea and calcium dosage on the effectiveness of bacterially induced carbonate precipitation on limestone. Ecological Engineering, 36(2), 99-111 De Muynck, W., Leuridan, S., Van Loo, D., Verbeken, K., Cnudde, V., De Belie, N. and Verstraete, W. (2011). Influence of Pore Structure on the Effectiveness of a Biogenic Carbonate Surface Treatment for Limestone Conservation. Applied and Environmental Microbiology, 77(19), 6808–6820. De Muynck, W., Verbeken, K., De Belie, N. and Verstraete, W. (2013). Influence of temperature on the effectiveness of a biogenic carbonate surface treatment for limestone conservation. Applied microbiology and biotechnology, 97(3), 1335-1347. Derluyn, H., Dewanckele, J., Boone, M. N., Cnudde, V., Derome, D. and Carmeliet, J. (2014). Crystallization of hydrated and anhydrous salts in porous limestone resolved by synchrotron X-ray microtomography. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, 324, 102-112. Dessandier, D., Auger, P., Haas, H. and Hugues, G. (2000). Guide méthodologique de sélection des pierres des monuments en termes de durabilîté et compatibilité. BRGM/RP-50137.FR, 76 p.p. Dewanckele, J., De Kock, T., Fronteau, G., Derluyn, H., Vontobel, P., Dierick, M., ... and Cnudde, V. (2014). Neutron radiography and X-ray computed tomography for quantifying weathering and water uptake processes inside porous limestone used as building material. Materials Characterization, 88, 86-99. De Weger, L. A., Van der Vlugt, C. I., Wijfjes, A. H. M., Bakker, P. A., Schippers, B. and Lugtenberg, B. (1987). Flagella of a plant-growth-stimulating Pseudomonas fluorescens strain are required for colonization of potato roots. Journal of bacteriology, 169(6), 2769-2773. De Witte, Y., 2010. Improved and Practically Feasible Reconstruction Methods for High Resolution X- ray Tomography. PhD Thesis, Ghent University, Ghent. 270 p.p. Diamond, S. (2000). Mercury porosimetry: an inappropriate method for the measurement of pore size distributions in cement-based materials. Cement and concrete research, 30(10), 1517-1525. Di Capua, F., Papirio, S., Lens, P. N. L. and Esposito, G. (2015). Chemolithotrophic denitrification in biofilm reactors. Chemical Engineering Journal, 280, 643–657. Dominguez, A., Bories, S. and Prat, M. (2000). Gas cluster growth by solute diffusion in porous media. Experiments and automaton simulation on pore network. International Journal of Multiphase Flow, 26(12), 1951-1979. Drtil, M., Németh, P., Kucman, K., Bodík, I., and Kasperek,̆ V. (1995). Acidobasic balances in the course of heterotrophic denitrification. Water Research, 29(5), 1353-1360. Dufrêne, Y. F. (2002). Atomic force microscopy, a powerful tool in microbiology. Journal of bacteriology, 184(19), 5205-5213. Dunham, R. J., 1962, Classification of carbonate rocks according to depositional texture. In W.E. Ham. (ed.), Classification of carbonate rocks: American Association of Petroleum Geologists Memoir, (p.p. 108-121). Tulsa: PLS.

64

du Roscoat, S. R., Martins, J. M. F., Séchet, P., Vince, E., Latil, P. and Geindreau, C. (2014). Application of synchrotron X-ray microtomography for visualizing bacterial biofilms 3D microstructure in porous media. Biotechnology and bioengineering, 111(6), 1265-1271. Dusar, M. and Dreesen, R. (2009) Geodiversiteit weerspielged in historische monumenten: Vlaamse natuursteenlandschappen als geotoeristische trekpleister. Geological survey of Belgium professional paper, 1(305), 79-100. Edwards, K. J., Becker, K. and Colwell, F. (2012). The deep, dark energy biosphere: intraterrestrial life on earth. Annual review of earth and planetary sciences, 40, 551-568. Ehrlich, H. L. (1996). How microbes influence mineral growth and dissolution. Chemical geology, 132(1- 4), 5-9. Eisendle‐Flöckner, U. and Hilberg, S. (2015). Hard rock aquifers and free‐living nematodes–an interdisciplinary approach based on two widely neglected components in groundwater research. Ecohydrology, 8(3), 368-377. Erşan, Y. Ç., Belie, N. de and Boon, N. (2015). Microbially induced CaCO3 precipitation through denitrification: An optimization study in minimal nutrient environment. Biochemical Engineering Journal, 101, 108–118. Erşan, Y. Ç. (2016). Microbial nitrate reduction induced autonomous self-healing in concrete. PhD thesis. Ghent University, Ghent, Belgium. 215 p.p. European Commission. (1991). First Council Directive 91/676/EEC of 12 December 1991 concerning the protection of waters agains pollution caused by nitrates from agricultural sources. Official Journal of the European Communities, L 375(1), 1–8. European Union. (2006). Directive 2006/118/EC of the European Parliament and of the council of 12 December 2006 on the protection of groundwater against pollution and deterioration. Official Journal of the European Union, L372(19), 19–31. Evian. (2014). Evian natural spring water – Annual water quality report. 10 p.p. Ferris, F. G., Stehmeier, L. G., Kantzas, A., and Mourits, F. M. (1997). Bacteriogenic mineral plugging. Journal of Canadian Petroleum Technology, 36(09). Firsching, M., (2009). Material Reconstruction in X-ray Imaging. PhD thesis, Friedrich-Alexander- Universität Erlangen-Nürnberg, Erlangen-Nürnberg, Germany, 103 p.p. Fischer, D., Pagenkemper, S., Nellesen, J., Peth, S., Horn, R., and Schloter, M. (2013). Influence of non- invasive X-ray computed tomography (XRCT) on the microbial community structure and function in soil. Journal of microbiological methods, 93(2), 121-123. Flemming, H. C. and Wingender, J. (2010). The biofilm matrix. Nature Reviews Microbiology, 8(9), 623- 633. Fletcher, M. (1980). Adherence of marine micro-organisms to smooth surfaces. In E. H. Beachey. Bacterial adherence (pp. 345-374). Dordrecht: Springer Netherlands. Flügel (2004). Microfacies of carbonate rocks, Analysis, interpretation and application. Berlin Heidelberg: Springer. 976 p.p. Folk, R.L. (1962). Spectral subdivision of limestone types. in W.E. Ham. (ed.) Classification of carbonate Rocks-A Symposium: American Association of Petroleum Geologists Memoir 1, (p.p. 62-84). Tulsa: PLS. Fredrickson, J. K., McKinley, J. P., Bjornstad, B. N., Long, P. E., Ringelberg, D. B., White, D. C.,... and Onstott, T. C. (1997). Pore‐size constraints on the activity and survival of subsurface bacteria in a late

65

cretaceous shale‐sandstone sequence, northwestern New Mexico. Geomicrobiology Journal, 14(3), 183-202. Fronteau, G. (2000). Comportements télogénétiques des principaux calcaires de Champagne-Ardenne en relation avec leur faciès de dépôt et leur séquençage diagénétique. PHD thesis, l’Université de Reims Champagne-Ardenne, Reims, France. 304 p.p. Fronteau, G., Schneider-Thomachot, C., Chopin, E., Barbin, V., Mouze, D. and Pascal, A. (2010). Black- crust growth and interaction with underlying limestone microfacies. In R. Přikryl and Á´. Török (eds.) ,Natural Stone Resources for Historical Monuments (p.p. 25 -34). Geological Society, London, Special Publications, 333. Fujita, Y., Ferris, F. G., Lawson, R. D., Colwell, F. S. and Smith, R. W. (2000). Subscribed Content Calcium Carbonate Precipitation by Ureolytic Subsurface Bacteria. Geomicrobiology Journal, 17(4), 305–318. Fujita, Y., Taylor, J. L., Wendt, L. M., Reed, D. W., and Smith, R. W. (2010). Evaluating the potential of native ureolytic microbes to remediate a 90Sr contaminated environment. Environmental science & technology, 44(19), 7652-7658. Gadd, G. M. (2010). Metals, minerals and microbes: geomicrobiology and bioremediation. Microbiology, 156(3), 609-643. Garrett, T. R., Bhakoo, M. and Zhang, Z. (2008). Bacterial adhesion and biofilms on surfaces. Progress in Natural Science, 18(9), 1049-1056. Gerth, W. A. and Hemmingsen, E. A. (1980). Heterogeneous nucleation of bubbles at solid surfaces in gas-supersaturated aqueous solutions. Journal of Colloid and Interface Science, 74(1), 80-89. Ghanbarian, B., Torres-Verdin, C. and Skaggs, T. H. (2016). Quantifying tight-gas sandstone permeability via critical path analysis. Advances in Water Resources, 92, 316-322. Goldscheider, N., Hunkeler, D. and Rossi, P. (2006). Review: microbial biocenoses in pristine aquifers and an assessment of investigative methods. Hydrogeology Journal, 14(6), 926-941. Greenberg, A. E., Clesceri, L. S., and Eaton, A. D. (1992). Standard methods for the examination of water and wastewater. American Public Health Association. Hahn, M. W. and O'Melia, C. R. (2004). Deposition and reentrainment of Brownian particles in porous media under unfavorable chemical conditions: Some concepts and applications. Environmental science & technology, 38(1), 210-220. Hammes, F. and Verstraete, W. (2002). Key roles of pH and calcium metabolism in microbial carbonate precipitation. Reviews in Environmental Science & Bio/Technology, 1(1), 3–7. Hammes, F., Seka, A., de Knijf, S. and Verstraete, W. (2003a). A novel approach to calcium removal from calcium-rich industrial wastewater. Water Research, 37(3), 699–704. Hammes, F., Seka, A., Van Hege, K., Van de Wiele, T., Vanderdeelen, J., Siciliano, S. D. and Verstraete, W. (2003b). Calcium removal from industrial wastewater by bio-catalytic CaCO3 precipitation. Journal of Chemical Technology and Biotechnology, 78(6), 670–677. Harkes, M. P., van Paassen, L. A., Booster, J. L., Whiffin, V. S. and van Loosdrecht, M. C. M. (2010). Fixation and distribution of bacterial activity in sand to induce carbonate precipitation for ground reinforcement. Ecological Engineering, 36(2), 112–117. Harms, N. and van Spanning, R. J. M. (1991). C1 metabolism in Paracoccus denitrificans: Genetics of Paracoccus denitrificans. Journal of Bioenergetics and Biomembranes, 23(2), 187–210.

66

Harvey, R. W., Smith, R. L. and George, L. (1984). Effect of organic contamination upon microbial distributions and heterotrophic uptake in a Cape Cod, Mass., aquifer. Applied and Environmental Microbiology, 48(6), 1197-1202. Helliwell, J. R., Sturrock, C. J., Grayling, K. M., Tracy, S. R., Flavel, R. J., Young, I. M.,… and Mooney, S. J. (2013). Applications of X‐ray computed tomography for examining biophysical interactions and structural development in soil systems: a review. European Journal of Soil Science, 64(3), 279-297. Hermansson, M. (1999). The DLVO theory in microbial adhesion. Colloids and Surfaces B: Biointerfaces, 14(1), 105-119. Iltis, G. C., Armstrong, R. T., Jansik, D. P., Wood, B. D. and Wildenschild, D. (2011). Imaging biofilm architecture within porous media using synchrotron‐based X‐ray computed microtomography. Water Resources Research, 47(2). Iltis, G. C. (2013). Visualization and characterization of biofilm spatial distribution in porous media using x-ray computed microtomography. PhD thesis, Oregeon State University, Corvallis, Oregon, USA. 209 p.p. Istok, J. D., Park, M. M., Peacock, A. D., Oostrom, M. and Wietsma, T. W. (2007). An experimental investigation of nitrogen gas produced during denitrification. Ground Water, 45(4), 461-467. Ivankovic, T., Rolland du Roscoat, S., Geindreau, C., Séchet, P., Huang, Z., and Martins, J. M. (2016). Development and evaluation of an experimental protocol for 3-D visualization and characterization of the structure of bacterial biofilms in porous media using laboratory X-ray tomography. Biofouling, 32(10), 1235-1244. Jackson, N. E., Corey, J. C., Frederick, L. R., and Picken, J. C. (1967). Gamma irradiation and the microbial population of soils at two water contents. Soil Science Society of America Journal, 31(4), 491-494. Jarvis, B., Hedges, A. J., and Corry, J. E. (2007). Assessment of measurement uncertainty for quantitative methods of analysis: Comparative assessment of the precision (uncertainty) of bacterial colony counts. International journal of food microbiology, 116(1), 44-51. Jenneman, G. E., McInerney, M. J. and Knapp, R. M. (1985). Microbial penetration through nutrient- saturated Berea sandstone. Applied and Environmental Microbiology, 50(2), 383-391. Jewett, D. G., Hilbert, T. A., Logan, B. E., Arnold, R. G. and Bales, R. C. (1995). Bacterial transport in laboratory columns and filters: influence of ionic strength and pH on collision efficiency. Water Research, 29(7), 1673-1680. Jewett, D. G., Logan, B. E., Arnold, R. G. and Bales, R. C. (1999). Transport of Pseudomonas fluorescens strain P17 through quartz sand columns as a function of water content. Journal of Contaminant Hydrology, 36(1), 73-89. Jones, A. A. and Bennett, P. C. (2014). Mineral microniches control the diversity of subsurface microbial populations. Geomicrobiology Journal, 31(3), 246-261. Karatas, I. (2008). Microbiological improvement of the physical properties of soil. PhD thesis, Arizona State University, Tempe, Arizona, USA. 200 p.p. Kashefi, K. and Lovley, D. R. (2003). Extending the upper temperature limit for life. Science, 301(5635), 934-934. Kelly, D. P., Rainey, F. a. and Wood, A. P. (2006). The Genus Paracoccus. In M. Dworkin, S. Falkow, E. Rosenberg, K.-H. Schleifer and E. Stackebrandt (Eds.), The Prokaryotes (3rd ed., pp. 232–249). New York, New York: Springer New York.

67

Ketcham, R. A., and Carlson, W. D. (2001). Acquisition, optimization and interpretation of X-ray computed tomographic imagery: applications to the geosciences. Computers & Geosciences, 27(4), 381-400. Kettridge, N. and Binley, A. (2011). Characterization of peat structure using X‐ray computed tomography and its control on the ebullition of biogenic gas bubbles. Journal of Geophysical Research: Biogeosciences, 116(G1). Kettridge, N., Kellner, E., Price, J. S. and Waddington, J. M. (2013). Peat deformation and biogenic gas bubbles control seasonal variations in peat hydraulic conductivity. Hydrological Processes, 27(22), 3208-3216. Kieft, T. L., Murphy, E. M., Haldeman, D. L., Amy, P. S., Bjornstad, B. N., McDonald, E. V.,… and Boone, D. R. (1998). Microbial transport, survival and succession in a sequence of buried sediments. Microbial Ecology, 36(3), 336-348. Klibert, C. (2015). Method for Synchrotron X-Ray Computed Tomographic Imaging of Biofilms in Porous Media. Master thesis, Department of Civil and Environmental Engineering, BA, Washington University, St. Louis, Missouri, USA, 74 p.p. Knorre, H. v. and Krumbein, W. E. (2000). Bacterial calcification. In R. E. Riding and S. M. Awramik (Eds.), Microbial Sediments (p.p. 25-32). Berlin, Heidelberg: Springer. Kocur, M., Martinec, T. and Mazanec, K. (1968). Fine structure of Micrococcus denitrificans and M. halodenitrificans in relation to their . Antonie van Leeuwenhoek, 34(1), 19–26. Konhauser, K. (2007) Introduction to geomicrobiology. Malden: Blackwell Publishing, 425 p.p. Kravchenko, A. N., Negassa, W. C., Guber, A. K., Hildebrandt, B., Marsh, T. L., and Rivers, M. L. (2014). Intra-aggregate pore structure influences phylogenetic composition of bacterial community in macroaggregates. Soil Science Society of America Journal, 78(6), 1924-1939. Kumar, C. G. and Anand, S. K. (1998). Significance of microbial biofilms in food industry: a review. International journal of food microbiology, 42(1), 9-27. Kumaraswamy, R., Sjollema, K., Kuenen, G., van Loosdrecht, M. and Muyzer, G. (2006). Nitrate- dependent [Fe(II)EDTA]2− oxidation by Paracoccus ferrooxidans sp. nov., isolated from a denitrifying bioreactor. Systematic and Applied Microbiology, 29(4), 276–286. Lauchnor, E. G., Schultz, L. N., Bugni, S., Mitchell, A. C., Cunningham, A. B., and Gerlach, R. (2013). Bacterially induced calcium carbonate precipitation and strontium coprecipitation in a porous media flow system. Environmental science & technology, 47(3), 1557-1564. Lanzón M. and Piñero, A. (2012). Caracterización químico-física de la piedra Tabaire y eficacia del tratamiento deconsolidación mediante hidróxido cálcico. In: XI CONGRESSO Internacional de Reabilitação do Património Arquitectónico e Edificado = XI Congreso Internacional de Rehabilitación del Patrimonio Arquitectónico y Edificación (p.p. 401-410), Cascais: ICES, CICOP, 12-14 of July 2012. Lanzón, M., Cnudde, V., De Kock, T., Dewanckele, J. and Piñero, A. (2014). X-ray tomography and chemical–physical study of a calcarenite extracted from a Roman quarry in Cartagena (Spain). Engineering Geology, 171, 21-30. Lebedev, M., Wilson, M. E. and Mikhaltsevitch, V. (2014). An experimental study of solid matrix weakening in water‐saturated Savonnières limestone. Geophysical Prospecting, 62(6), 1253-1265. Lee, K.-C. and Rittmann, B. E. (2003). Effects of pH and precipitation on autohydrogenotrophic denitrification using the hollow-fiber membrane-biofilm reactor. Water Research, 37(7), 1551–1556.

68

Lehmann, E. H., Frei, G., Kühne, G., and Boillat, P. (2007). The micro-setup for neutron imaging: A major step forward to improve the spatial resolution. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 576(2), 389-396. Lehman, R. M., Colwell, F. S. and Bala, G. A. (2001). Attached and unattached microbial communities in a simulated basalt aquifer under fracture-and porous-flow conditions. Applied and Environmental Microbiology, 67(6), 2799-2809. Li, M., Mahmudov, R., and Huang, C. P. (2011). Hazardous waste treatment technologies. Water Environment Research, 83(10), 1598-1632. Lijinski, W. (1977). "How nitrosamines cause cancer." New Science, 27, 216-217. Litton, G. M. and Olson, T. M. (1996). Particle size effects on colloid deposition kinetics: Evidence of secondary minimum deposition. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 107, 273-283. Lorenz, H. and Lehrberger, G. Savonnières, Morley and Co.: Oolithische Kalksteine aus Lothringen (Frankreich) als Bau-und Denkmalgesteine in Mitteleuropa. Tagung für Ingenieurgeologie mit Forum für junge Ingenieurgeologen München 2013, 359-366. Lubetkin, S. D. (2003). Why is it much easier to nucleate gas bubbles than theory predicts?. Langmuir, 19(7), 2575-2587. Ludwig, W., Mittenhuber, G. and Friedrich, C. G. (1993). Transfer of Thiosphaera pantotropha to Paracoccus denitrificans. International Journal of Systematic Bacteriology, 43(2), 363–367. Macey M. G. (2007) Principles of flow cytometry. In M. G. Macey (Ed.) Flow cytometry, Principles and applications. (p.p. 1-17). Totowa: Humana Press Inc. Madigan, M., Martinko, J.M., Dunlap, P.V. and Clark, D.P. (2009) Brock Biology of microorganisms. 12th edition. San Francisco: Pearson Benjamin Cummings. 1061 p.p. Mahani, H., Keya, A. L., Berg, S., and Nasralla, R. (2015). The Effect of Salinity, Rock Type and pH on the Electrokinetics of Carbonate-Brine Interface and Surface Complexation Modeling. In SPE Reservoir Characterisation and Simulation Conference and Exhibition. Society of Petroleum Engineers, 14-16 September, Abu Dhabi. Masschaele, B., Dierick, M., Van Loo, D., Boone, M. N., Brabant, L., Pauwels, E.,… and Van Hoorebeke, L., (2013). HECTOR: A 240kV micro-CT setup optimized for research. Journal of Physics: Conference Series 463, 12012. Mathews, S. L., Pawlak, J. and Grunden, A. M. (2015). Bacterial biodegradation and bioconversion of industrial lignocellulosic streams. Applied microbiology and biotechnology, 99(7), 2939-2954. Mayo, S., Josh, M., Nesterets, Y., Esteban, L., Pervukhina, M., Clennell, M. B., ... and Hall, C. (2015). Quantitative micro-porosity characterization using synchrotron micro-CT and xenon K-edge subtraction in sandstones, carbonates, shales and coal. Fuel, 154, 167-173. McCalla, T. M. 1950. Studies on the effect of microorganisms on rate of percolation of water through soils. Soil Science Society of America, Proceedings. 15, 182-186. McNamara, N. P., Black, H. I. J., Beresford, N. A., and Parekh, N. R. (2003). Effects of acute gamma irradiation on chemical, physical and biological properties of soils. Applied Soil Ecology, 24(2), 117-132. Michael, G. (2001). X-ray computed tomography. Physics Education, 36(6), 442-451. Miller, A. Z., Sanmartín, P., Pereira-Pardo, L., Dionísio, A., Sáiz-Jiménez, C., Macedo, M. F., and Prieto, B. (2012). Bioreceptivity of building stones: a review. Science of the total environment, 426, 1-12.

69

Molnar, I. L., O'Carroll, D. M., and Gerhard, J. I. (2011). Impact of surfactant-induced wettability alterations on DNAPL invasion in quartz and iron oxide-coated sand systems. Journal of contaminant hydrology, 119(1), 1-12. Molnar, I. L., Johnson, W. P., Gerhard, J. I., Willson, C. S. and O'Carroll, D. M. (2015). Predicting colloid transport through saturated porous media: A critical review. Water Resources Research, 51(9), 6804- 6845. Monte, M. (1993). The influence of environmental conditions on the reproduction and distribution of epilithic lichens. Aerobiologia, 9(2), 169-179. Montero, F. (2015). Respiration. In M. Gargaud, W. M. Irvine, R. Amils, H. J. Cleaves II, D. L. Pinti, J. C. Quintanilla, … M. Viso (Eds.), Encyclopedia of Astrobiology (pp. 2172–2175). Berlin, Heidelberg: Springer. Neidhardt F.C. (1996) Escherichia coli and Salmonella. Cellular and Molecular Biology, 1, 14. Nemati, M. and Voordouw, G. (2003). Modification of porous media permeability, using calcium carbonate produced enzymatically in situ. Enzyme and Microbial Technology, 33(5), 635–642. Nivens, D. E., Chambers, J. Q. anderson, T. R., Tunlid, A., Smit, J. and White, D. C. (1993). Monitoring microbiol adhesion and biofilm formation by attenuated total reflection/Fourier transform infrared spectroscopy. Journal of microbiological methods, 17(3), 199-213. Noël, P. (1970). Les carrières françaises de pierre de taille. Société de diffusion des techniques du bâtiment et des travaux publics. Nokhal, T. H. and Mayer, F. (1979). Structural analysis of four strains of Paracoccus denitrificans. Antonie van Leeuwenhoek, 45(2), 185–197. Nokhal, T. H. and Schlegel, H. G. (1983). Taxonomic study of Paracoccus denitrificans. International Journal of Systematic and Evolutionary Microbiology, 33(1), 26-37. Northeast Laboratory Services (2016). Technical product information, Tryptic Soy Broth (USP Formulation). 2 p.p. O'Donnell, A. G., Young, I. M., Rushton, S. P., Shirley, M. D. and Crawford, J. W. (2007). Visualization, modelling and prediction in soil microbiology. Nature Reviews Microbiology, 5(9), 689-699. Ortega-Calvo, J. J., Ariño, X., Hernandez-Marine, M., and Saiz-Jimenez, C. (1995). Factors affecting the weathering and colonization of monuments by phototrophic microorganisms. Science of the Total Environment, 167(1-3), 329-341. Palmer, J., Flint, S. and Brooks, J. (2007). Bacterial cell attachment, the beginning of a biofilm. Journal of industrial microbiology and biotechnology, 34(9), 577-588. Pedersen, K. (2000). Exploration of deep intraterrestrial microbial life: current perspectives. FEMS microbiology letters, 185(1), 9-16. Phillips, A. J., Lauchnor, E., Eldring, J., Esposito, R., Mitchell, A. C., Gerlach, R., ... and Spangler, L. H. (2013). Potential CO2 leakage reduction through biofilm-induced calcium carbonate precipitation. Environmental science & technology, 47(1), 142-149. Phillips, J. D. (2016). Biogeomorphology and contingent ecosystem engineering in karst landscapes. Progress in Physical Geography, 40(4), 503-526. Poortinga, A. T., Bos, R., Norde, W. and Busscher, H. J. (2002). Electric double layer interactions in bacterial adhesion to surfaces. Surface science reports, 47(1), 1-32.

70

Potter, K., Kleinberg, R. L., Brockman, F. J. and McFarland, E. W. (1996). Assay for bacteria in porous media by diffusion-weighted NMR. Journal of Magnetic Resonance, Series B, 113(1), 9-15. Powelson, D. K. and Mills, A. L. (1998). Water saturation and surfactant effects on bacterial transport in sand columns. Soil Science, 163(9), 694-704. Pro Roc (1998). Roches de France: pierres marbres granits grés et autres roches ornamentales et de construction. Pro Roc, Ternay, France. Pujalte, M. J., Lucena, T., Ruvira, M. A., Arahal, D. R. and Macian, M. C. (2014). The Family Rhodobacteraceae. In E. Rosenberg, E. F. DeLong, S. Lory, E. Stackebrandt and F. Thompson (Eds.), The Prokaryotes: Alphaproteobacteria and Betaproteobacteria (4th ed., pp. 440–512). Berlin, Heidelberg: Springer. Qajar, J., Francois, N., and Arns, C. H. (2013). Microtomographic characterization of dissolution- induced local porosity changes including fines migration in carbonate rock. SPE Journal, 18(03), 545- 562. Qian, C., Wang, R., Cheng, L., and Wang, J. (2010). Theory of Microbial Carbonate Precipitation and Its Application in Restoration of Cement‐based Materials Defects. Chinese Journal of Chemistry, 28(5), 847-857. Rajab, M., Yazdian, F., Rasekh, B. and Rashedi, H. (2016). Effect of Metal Nanoparticles on Biological Denitrification Process : A Review. Journal of Applied Biotechnology Reports, 3(1), 353–358. Randall, D. and Tsui, T. K. (2002). Ammonia toxicity in fish. Marine Pollution Bulletin, 45(1–12), 17–23. Rebata‐Landa, V. and Santamarina, J. C. (2006). Mechanical limits to microbial activity in deep sediments. Geochemistry, Geophysics, Geosystems, 7(11). Rebata-Landa, V. and Santamarina, J. C. (2012). Mechanical effects of biogenic nitrogen gas bubbles in soils. Journal of Geotechnical and Geoenvironmental Engineering, 138(2), 128-137. Rijnaarts, H. H., Norde, W., Lyklema, J. and Zehnder, A. J. (1995). The isoelectric point of bacteria as an indicator for the presence of cell surface polymers that inhibit adhesion. Colloids and Surfaces B: Biointerfaces, 4(4), 191-197. Rijnaarts, H. H., Norde, W., Lyklema, J. and Zehnder, A. J. (1999). DLVO and steric contributions to bacterial deposition in media of different ionic strengths. Colloids and Surfaces B: Biointerfaces, 14(1), 179-195. Roberts, K. (2009). Pore-Scale Analysis of DNAPL Dissolution and Biomass Distribution. PhD Thesis, Louisiana State University, Baton Rouge, Louisiana, USA, 62 p.p. Robertson, L. a and Kuenen, J. G. (1984). Aerobic denitrification: a controversy revived. Archives of Microbiology, 139(4), 351–354. Rockhold, M. L., Yarwood, R. R., Niemet, M. R., Bottomley, P. J. and Selker, J. S. (2002). Considerations for modeling bacterial-induced changes in hydraulic properties of variably saturated porous media. Advances in water resources, 25(5), 477-495. Rodríguez, K., and Araujo, M. (2006). Temperature and pressure effects on zeta potential values of reservoir minerals. Journal of colloid and interface science, 300(2), 788-794. Rodriguez-Navarro, C., Rodriguez-Gallego, M., Chekroun, K. B. and Gonzalez-Muñoz, M. T. (2003). Conservation of ornamental stone by Myxococcus xanthus-induced carbonate biomineralization. Applied and Environmental Microbiology, 69(4), 2182-2193.

71

Rodriguez-Navarro, C., Jroundi, F., Schiro, M., Ruiz-Agudo, E., and González-Muñoz, M. T. (2012). Influence of substrate mineralogy on bacterial mineralization of calcium carbonate: implications for stone conservation. Applied and environmental microbiology, 78(11), 4017-4029. Roels S. (2000) Modelling unsaturated moisture transport in heterogeneous limestone, PhD thesis, Katholieke Universiteit Leuven, Leuven, Belgium. 211 p.p. Roels, S., Carmeliet, J., Hens, H. and Elsen, J. (2000). Microscopic analysis of imbibition processes in oolitic limestone. Geophysical research letters, 27(21), 3533-3536. Roels, S., Elsen, J., Carmeliet, J. and Hens, H. (2001). Characterisation of pore structure by combining mercury porosimetry and micrography. Materials and structures, 34(2), 76-82. Roels, S., Carmeliet, J. and Hens, H. (2003). Modelling unsaturated moisture transport in heterogeneous limestone (Part 1. A mesoscopic approach). Transport in porous media, 52(3), 333-350. Rong, H. and Qian, C. (2013). Microstructure evolution of sandstone cemented by microbe cement using X-ray computed tomography. Journal of Wuhan University of Technology-Mater. Sci. Ed., 28(6), 1134-1139. Rust, C. M., Aelion, C. M., and Flora, J. R. (2000). Control of pH during denitrification in subsurface sediment microcosms using encapsulated phosphate buffer. Water Research, 34(5), 1447-1454.7 Saranraj, P. (2013). Bacterial biodegradation and decolourization of toxic textile azo dyes. African Journal of Microbiology Research, 7(30), 3885-3890. Satik, C., and Yortsos, Y. C. (1996). A pore-network study of bubble growth in porous media driven by heat transfer. Transactions-American society of mechanical engineers journal of heat transfer, 118, 455-462. Schmid, T., Helmbrecht, C., Panne, U., Haisch, C. and Niessner, R. (2003). Process analysis of biofilms by photoacoustic spectroscopy. Analytical and bioanalytical chemistry, 375(8), 1124-1129. Schmidt, H., Vetterlein, D., Koehne, J. M., and Eickhorst, T. (2015). Negligible effect of X-ray μ-CT scanning on archaea and bacteria in an agricultural soil. Soil Biology and Biochemistry, 84, 21-27. Schmidt, S. I., Cuthbert, M. O. and Schwientek, M. (2017). Towards an integrated understanding of how micro scale processes shape groundwater ecosystem functions. Science of The Total Environment, 592, 215-227. Seki, K., Miyazaki, T. and Nakano, M. (1998). Effects of microorganisms on hydraulic conductivity decrease in infiltration. European Journal of Soil Science, 49(2), 231-236. Seymour, J. D., Codd, S. L., Gjersing, E. L. and Stewart, P. S. (2004a). Magnetic resonance microscopy of biofilm structure and impact on transport in a capillary bioreactor. Journal of Magnetic Resonance, 167(2), 322-327. Seymour, J. D., Gage, J. P., Codd, S. L. and Gerlach, R. (2004b). Anomalous fluid transport in porous media induced by biofilm growth. Physical review letters, 93(19), 198103. Seymour, J. D., Gage, J. P., Codd, S. L. and Gerlach, R. (2007). Magnetic resonance microscopy of biofouling induced scale dependent transport in porous media. Advances in water resources, 30(6), 1408-1420. Shapleigh (2006). The denitrifying Prokaryotes. In M. Dworkin, S. Falkow, E. Rosenberg, K.-H. Schleifer and E. Stackebrandt (Eds.), The Prokaryotes (3rd ed., pp. 232–249). New York, New York: Springer. Shapleigh, J. P. (2013). Denitrifying prokaryotes. In E., Rosenberg, E. F., DeLong, S., Lory, E., Stackebrandt and F., Thompson (eds.). The prokaryotes (4th ed. pp. 405-425). Berlin Heidelberg: Springer.

72

Shemesh, H., Goertz, D. E., Van der Sluis, L. W. M., de Jong, N., Wu, M. K. and Wesselink, P. R. (2007). High frequency ultrasound imaging of a single-species biofilm. Journal of dentistry, 35(8), 673-678. Siewerdsen, J. H. and Jaffray, D. A., (1999). Cone-beam computed tomography with a flat-panel imager: effects of image lag. Medical Physics 26 (12), 2635–2647. Sijbers, J. and Postnov, A., (2004). Reduction of ring artefacts in high resolution micro-CT reconstructions. Physics in Medicine and Biology 49, N247–N253. Sills, G. C. and Gonzalez, R. (2001). Consolidation of naturally gassy soft soil. Geotechnique, 51(7), 629- 639. Song, F., Koo, H. and Ren, D. (2015). Effects of material properties on bacterial adhesion and biofilm formation. Journal of dental research, 94(8), 1027-1034. Taylor, S. W. and Jaffé, P. R. (1990). Biofilm growth and the related changes in the physical properties of a porous medium: 3. Dispersivity and model verification. Water resources research, 26(9), 2171- 2180. Terada, A., Yuasa, A., Tsuneda, S., Hirata, A., Katakai, A., and Tamada, M. (2005). Elucidation of dominant effect on initial bacterial adhesion onto polymer surfaces prepared by radiation-induced graft polymerization. Colloids and Surfaces B: Biointerfaces, 43(2), 99-107. Thieme, J., Schneider, G. and Knöchel, C. (2003). X-ray tomography of a microhabitat of bacteria and other soil colloids with sub-100 nm resolution. Micron, 34(6), 339-344. Thullner, M. (2010). Comparison of bioclogging effects in saturated porous media within one-and two- dimensional flow systems. Ecological Engineering, 36(2), 176-196. Tobler, D. J., Maclachlan, E., and Phoenix, V. R. (2012). Microbially mediated plugging of porous media and the impact of differing injection strategies. Ecological Engineering, 42, 270-278. Torkzaban, S., Tazehkand, S. S., Walker, S. L., and Bradford, S. A. (2008). Transport and fate of bacteria in porous media: Coupled effects of chemical conditions and pore space geometry. Water Resources Research, 44(4). Torrentó, C., Cama, J., Urmeneta, J., Otero, N. and Soler, A. (2010). Denitrification of groundwater with pyrite and Thiobacillus denitrificans. Chemical Geology, 278, 80–91. Tufenkji, N., and Elimelech, M. (2004). Deviation from the classical colloid filtration theory in the presence of repulsive DLVO interactions. Langmuir, 20(25), 10818-10828. Tufenkji, N. and Elimelech, M. (2005). Breakdown of colloid filtration theory: Role of the secondary energy minimum and surface charge heterogeneities. Langmuir, 21(3), 841-852. Umar, M., Kassim, K. A. and Chiet, K. T. P. (2016). Biological process of soil improvement in civil engineering: A review. Journal of Rock Mechanics and Geotechnical Engineering, 8(5), 767-774. Updegraff, D. M. (1982), Plugging and penetration of reservoir rock by microorganisms, paper presented at International Conference in Microbial Enhancement of Oil Recovery, U.S. Dept. of Energy, Bartlesville, Oklahoma, USA. Van Der Star, W. R. L., Taher, E., Harkes, M. P., Blauw, M., Van Loosdrecht, M. C. M. and van Paassen, L. A. (2009). Use of Waste Streams and Microbes for in situ Transformation of Sand Into Sandstone. In: C. F. Leung and R. F. Shen (eds). Ground Improvement Technologies and Case Histories. (pp. 177–182). Singapore: Research Publishing Services. van der Wal, A., Norde, W., Zehnder, A. J. and Lyklema, J. (1997). Determination of the total charge in the cell walls of Gram-positive bacteria. Colloids and surfaces B: Biointerfaces, 9(1-2), 81-100.

73

Vandevivere, P. and Baveye, P. (1992a). Effect of bacterial extracellular polymers on the saturated hydraulic conductivity of sand columns. Applied and Environmental Microbiology, 58(5), 1690-1698. Vandevivere, P. and Baveye, P. (1992b). Relationship between transport of bacteria and their clogging efficiency in sand columns. Applied and environmental microbiology, 58(8), 2523-2530. Vandevivere, P. and Baveye, P. (1992c). Saturated hydraulic conductivity reduction caused by aerobic bacteria in sand columns. Soil Science Society of America Journal, 56(1), 1-13. van Loosdrecht, M. C., Lyklema, J., Norde, W. and Zehnder, A. J. (1989). Bacterial adhesion: a physicochemical approach. Microbial Ecology, 17(1), 1-15. van Loosdrecht, M. C., Lyklema, J., Norde, W. and Zehnder, A. J. (1990). Influence of interfaces on microbial activity. Microbiological reviews, 54(1), 75-87. Van Oss, C. J., Giese, R. F. and Costanzo, P. M. (1990). DLVO and non-DLVO interactions in hectorite. Clays Clay Miner, 38(2), 151-159. van Paassen, L. A., Daza, C. M., Staal, M., Sorokin, D. Y., van der Zon, W. and van Loosdrecht, M. C. M. (2010). Potential soil reinforcement by biological denitrification. Ecological Engineering, 36(2), 168– 175. Vlassenbroeck, J. (2010). Advances in laboratory-based X-ray microtomography, PhD thesis, Ghent University, Ghent, Belgium, 278 p.p. Walsh, J. H. (2001). Ecological considerations of biodeterioration. International Biodeterioration & Biodegradation, 48(1-4), 16-25. Wang, J., Dewanckele, J., Cnudde, V., Van Vlierberghe, S., Verstraete, W., and De Belie, N. (2014). X- ray computed tomography proof of bacterial-based self-healing in concrete. Cement and Concrete Composites, 53, 289-304. Wang, J., Ersan, Y. C., Boon, N., and De Belie, N. (2016). Application of microorganisms in concrete: a promising sustainable strategy to improve concrete durability. Applied microbiology and biotechnology, 100(7), 2993-3007. Warren, L. A., Maurice P. A., Parmer and Ferris F. G. (2001). Microbially Mediated Calcium Carbonate Precipitation: Implications for Interpreting Calcite Precipitation and for Solid-Phase Capture of Inorganic Contaminants. Geomicrobiology Journal, 18(1), 93–115. Warscheid, T., and Braams, J. (2000). Biodeterioration of stone: a review. International Biodeterioration & Biodegradation, 46(4), 343-368. Washburn, E. W. (1921). The dynamics of capillary flow. Physical review, 17(3), 273. Watson, J. V. (1999). The early fluidic and optical physics of cytometry. Cytometry Part A, 38(1), 2-14. Weaver, L., Webber, J. B., Hickson, A. C., Abraham, P. M. and Close, M. E. (2015). Biofilm resilience to desiccation in groundwater aquifers: A laboratory and field study. Science of the Total Environment, 514, 281-289. Webster, A., and May, E. (2006). Bioremediation of weathered-building stone surfaces. Trends in Biotechnology, 24(6), 255-260. Wheeler, S. J. (1988). A conceptual model for soils containing large gas bubbles. Geotechnique, 38(3), 389-397. Whiffin, V. S., van Paassen, L. A. and Harkes, M. P. (2007). Microbial Carbonate Precipitation as a Soil Improvement Technique. Geomicrobiology Journal, 24(5), 417–423.

74

Whitehead, K. A. and Verran, J. (2006). The effect of surface topography on the retention of microorganisms. Food and bioproducts processing, 84(4), 253-259. Whitman, W. B., Coleman, D. C. and Wiebe, W. J. (1998). Prokaryotes: the unseen majority. Proceedings of the National Academy of Sciences, 95(12), 6578-6583. Wiggli, M., Smallcombe, A. and Bachofen, R. (1999). Reflectance spectroscopy and laser confocal microscopy as tools in an ecophysiological study of microbial mats in an alpine bog pond. Journal of microbiological methods, 34(3), 173-182. Willey, J. M., Sherwood, L. M., Woolverton, C. J. (2009). Prescott’s Principles of microbiology. New York: McGraw-Hill, 817 p.p. Williams, S. C., Hong, Y., Danavall, D. C. A., Howard-Jones, M. H., Gibson, D., Frischer, M. E., and Verity, P. G. (1998). Distinguishing between living and nonliving bacteria: evaluation of the vital stain propidium iodide and its combined use with molecular probes in aquatic samples. Journal of Microbiological Methods, 32(3), 225-236. Wilkins, S. W., Gureyev, T. E., Gao, D., Pogany, A., and Stevenson, A. W. (1996). Phase-contrast imaging using polychromatic hard X-rays. Nature, 384(6607), 335-338. Wilson, P. S., Reed, A. H., Wood, W. T., and Roy, R. A. (2007). Low frequency sound speed measurements paired with computed x-ray tomography imaging in gas-bearing reconstituted natural sediments. In Proceedings of the 2nd International Conference and Exhibition on Underwater Acoustics Measurements: Technologies and Results (pp. 21-29). Heraklion, Greece. WTCB, (1997). Technische voorlichting 205: Natuursteen WTCB, Brussel, Belgium. Xi, C., Marks, D., Schlachter, S., Luo, W. and Boppart, S. A. (2006). High-resolution three-dimensional imaging of biofilm development using optical coherence tomography. Journal of biomedical optics, 11(3), 034001. Xia, L., Zheng, X., Shao, H., Xin, J., Sun, Z. and Wang, L. (2016). Effects of bacterial cells and two types of extracellular polymers on bioclogging of sand columns. Journal of Hydrology, 535, 293-300. Yarwood, R. R., Rockhold, M. L., Niemet, M. R., Selker, J. S. and Bottomley, P. J. (2006). Impact of microbial growth on water flow and solute transport in unsaturated porous media. Water resources research, 42(10). Yee, N., Fein, J. B. and Daughney, C. J. (2000). Experimental study of the pH, ionic strength and reversibility behavior of bacteria–mineral adsorption. Geochimica et Cosmochimica Acta, 64(4), 609- 617. Zamarreňo, D. V., May, E. and Inkpen, R. (2009). Influence of environmental temperature on biocalcification by non-sporing freshwater bacteria. Geomicrobiology Journal, 26(4), 298-309. Zappala, S., Helliwell, J. R., Tracy, S. R., Mairhofer, S., Sturrock, C. J., Pridmore, T., Bennett, M., and Mooney, S. J. (2013). Effects of X-ray dose on rhizosphere studies using X-ray computed tomography. PloS one, 8(6), e67250. Zavilgelsky, G. B., Abilev, S. K., Sukhodolets, V. V., and Ahmad, S. I. (1998). Isolation and analysis of UV and radio-resistant bacteria from Chernobyl. Journal of Photochemistry and Photobiology B: Biology, 43(2), 152-157. Zumft, W. G. (1997). Cell biology and molecular basis of denitrification. Microbiology and Molecular Biology Reviews : MMBR, 61(4), 533–616.

75

9.APPENDIX

9.1. Bioclogging experiments

6 4

2 3 1 5

Figure 9.1: Dual setup for bioclogging experiments with red and blue arrows illustrating the flow paths. (1) syringe pump, (2) pressure release valves, (3) manometer, (4) laptop connected to manometer, (5) Stent pump and (6) flow cell containing the samples.

To induce bioclogging after the confirmation that the bacteria do adhere to the rocks, a new more elaborated flow setup has been developed. It can do double experiments at one moment (figure 10.1). Unfortunately it was not possible to solve all the problems by the end of the project. The idea behind the setup is comparable with the initial setup used during the bacterial adhesion experiments. The samples would be impregnated with water under vacuum and would reside within the flow cell, in a sleeve. On this sleeve a confining pressure would be performed to prohibit the flow along the samples. It has however been upgraded by a syringe pump, pressure release valves, manometers, new stent pumps and new flow cells. The syringe pump will pump at a constant flow rate. The manometer will measure the pressure needed for the flow to penetrate the porous rocks in the flow cells. The pressure release valve acts as a safety mechanisms that will leak water to decrease the pressure in the system when it surpasses 100 PSI.

This setup needed the purchase of new equipment like the syringe pump, stent pumps and the manometers. This asked a lot of time, but the flow cells were the largest problem as they leaked and could not bear the high pressure. Two out of three available flow cells leaked at the top. The other leaked internally by bringing water used for the confining pressure within the sleeve. For this reason no confining pressure could be maintained around the samples, making any reliable flow test impossible. The “older” flow cell used during the bacterial adhesion experiments was not operational anymore, because it was not compatible with the new stent pumps. The old stent pumps were fine for the previous experiments, but broke during by the creation of this setup.

The idea was that the bacteria would clog the porous limestones within the flow cells, when a solution with a very slow flow rate and a high concentration of bacteria was flushed through it for several hours. Bioclogging would reduce the permeability, and increase the pressure needed to flow the solution with a particular rate through the rock samples. The manometer would measure the pressure and would display this on the pc. This pressure is a proxy for the level of clogging and are both proportionally related. When the pressure would reach 100 PSI, the experiment would end, as the safety pressure release valves would stop it. At this stage most pores assumed to be clogged. µCT would thereafter scan the flow cells to reveal the cause of the clogging. No visualization of clogging would conclude that the bacteria clogged the pores physically, while clogging by gas or MICP would be visible on a scan. The experiment would be executed on both rocks to find a potential link with the pore structure. The

76

results of this experiment would act as a framework to start monitoring the bioclogging process in 4D using the EMCT scanner of UGent. It could reveal any potential flow path diversions and detect the locations of the initial clogging and its development over time.

It's key in such experiments to inhibit any external air bubbles within the system. Two Hamilton valves that can changes the flow path to water or other liquid reservoirs will be added and this could provide a refill of the syringes without the introduction of air within the system. The preparation of the sample and flow cell would be done under water to exclude any trapped air bubble. This setup would also act to redo the bacterial adhesion experiments, with a higher precision. The syringe pump could maintain a constant flow rate and change the “droplets” unit to volume units. One sample would be taken during the flow of a certain volume, after which the syringe pump would stop for a few minutes. The manometer would show then how the pressure would decrease, and when it would reach a certain low level. The syringe pump would restart again and a new sample would be collected.

Everything at this moment is provided to successfully execute these experiments. The last problems with the leaking flow cells will be solved within the upcoming months after which everything will be conducted according this plan.

9.2. Poster with preliminary results, presented at the Interpore Conference (8-11 May 2017, Rotterdam, The Netherlands)

77

78