Probing the intrinsically oil-wet surfaces of pores in North Sea chalk at subpore resolution

Abstract

Pore surface properties control oil recovery. This is especially true for chalk reservoirs, where pores are particularly small. Wettability, the tendency for a surface to cover itself with fluid, is traditionally defined by the angle a droplet makes with a surface, but this macroscopic definition is meaningless when the particles are smaller than even the smallest droplet. Understanding surface wetting, at the pore scale, will provide clues for more effective oil recovery. We used a special mode of atomic force microscopy and a hydrophobic tip to collect matrices of 10,000 force curves over 5- × 5-μm² areas on internal pore surfaces and constructed maps of topography, adhesion, and elasticity. We investigated chalk samples from a water-bearing formation in the Danish North Sea oil fields that had never seen oil. Wettability and elasticity were inhomogeneous over scales of 10s of nanometers, smaller than individual chalk particles. Some areas were soft and hydrophobic, whereas others showed no correlation between hardness and adhesion. We conclude that the macroscopic parameter, “wetting,” averages the nanoscopic behavior along fluid pathways, and “mixed-wet” samples have patches with vastly different properties. Development of reservoir hydrophobicity has been attributed to infiltrating oil, but these new results prove that wettability and elasticity are inherent properties of chalk. Their variability, even on single particles, must result from material originally present during sedimentation or material sorbed from the pore fluid some time later.

The global economy is based on hydrocarbons, and even with the best of intentions it would take at least a decade to develop viable, sustainable alternatives. Continuing high demand for oil, depletion of easy-to-get oil reserves, and concerns about even higher CO₂ production from coal burning drive the hunt for methods to increase the amount of recoverable oil. Modern methods produce significantly more oil than the ≈5% that comes simply from overburden compression (lithostatic pressure), but between 25 and 75% is often left behind, in the chalk formations of the North Sea and elsewhere around the world (1). Better understanding about the fundamental controls on surface wetting and adhesion is one key to improving oil recovery.

Chalk is composed predominantly of calcite, CaCO₃, as remnants of coccoliths, the micrometer-scale platelets produced by some species of marine algae (Fig. 1). The biogenic coccolith elements (marked C) can be rhombohedral, or they can have some other form (2). These are clearly distinguishable from the larger, rhombohedral crystals (marked R in Fig. 1), whose mineralogical form and texture indicate that they have formed by recrystallization after deposition in the original calcareous mud. The euhedral crystals are often 10–1,000 times larger than the biogenic fragments. Chalk particles range in size from several millimeters to a few tens of nanometers, and the pore spaces between them vary over a similar range.

Fig. 1.

SEM image of chalk from a North Sea core drilled through a water-bearing zone. We see coccoliths remnants (marked C), many loose coccolith elements, and larger rhombohedral crystals of inorganically grown calcite (R). The gray square represents an area that could be covered by the maps presented here. The square is for reference only. It is not possible to make AFM maps from an SEM sample.

Fluid flow through the pores is controlled by the properties of the individual particle surfaces (3). Wettability, the tendency of a surface to cover itself with a fluid, plays a key role (3–5). Surfaces can be hydrophilic or hydrophobic, i.e., water- or oil-wet, thus favoring flow of either water or oil. Some researchers have proposed that chalk pore surfaces are “mixed-wet,” meaning that they behave midway in the water- to oil-wet spectrum (4, 6). Oil typically migrates from a source rock, often organic-rich mud or shale, into the chalk, so mixed-wettability has been explained as a response by the pore interfaces to invading oil. Studies have assumed that original, internal pore surfaces are bound by pure mineral crystal faces (7–9). Most previous work on chalk has investigated how various crude oil components interact with pristine calcite (10, 11). Several researchers have attempted to define the controls on wettability and explain its role in oil recovery (4, 8, 12). Studies have tried to link micrometer-scale contact angle measurements from pure calcite, silicon, or mica substrates with atomic force microscopy (AFM) (8, 10, 11, 13, 14). To simulate reservoir conditions, mica surfaces have been aged by exposure to various crude oil components before contact angle experiments. One study examined the modifying effect of drilling muds and surfactants (15). Reservoir wettability is generally described from a macroscopic perspective, but information about the forces at play among chalk, oil, and water, at resolution that matches the size of the pores, i.e., at the nanometer scale, will offer new perspectives on the processes that control oil release.

Conventionally, wettability is determined from the angle that the edge of a droplet makes with a surface. On submicrometer particles, the contact angle definition becomes meaningless because the particles are smaller than it is possible to make a droplet. We took a different approach. As a first step toward understanding the innate wetting behavior of chalk, we examined samples that had never been exposed to oil but that had the same geological history as oil-bearing zones in the formation. We assumed that, although chalk particles are predominantly calcite in the bulk, the surfaces could very well have a completely different composition. Thus, wettability of the chalk particles might be determined by the material that was on them when they were deposited as calcareous mud, 60 million years ago, or by the material that eventually covered them from the pore water. Mixed-wettability might result from the inherent surface composition of the chalk particles themselves rather than from invading oil. The composition and properties of the chalk interface with the fluid are largely unknown. Our hypothesis is that pore–solid interface character is a critical parameter for oil extractability.

The purpose of this work was to define the physical and chemical properties of chalk at the subpore scale. Our aims were (i) to determine local (nanoscale) adhesion forces and surface elasticity and to produce maps of these parameters over areas of a few micrometers, and (ii) to define the relationship between surface forces and macroscopic wettability.

AFM Force Mapping

Force mapping, developed and implemented by Asylum Research, allows imaging on rough surfaces where height varies over many micrometers, because each point is an independent measurement. The surface is probed vertically with a constant force, and the tip lifts completely away before moving to the next location, so rapid tracking by piezoelectric scanner feedback is not necessary. Although it takes ≈50 min to generate a dataset with acceptable resolution, advantages are that loose material is not dragged over the surface, very rough surfaces are accessible, and physical properties can be mapped. We used gold-coated tips functionalized with a layer of hexadecanethiol. Because of the strong bonding between the thiol group and gold, the hydrocarbon end of the molecule (hexadecyl) is exposed, producing a hydrophobic tip (16). The force curves record tip-sample separation, so each curve is corrected for cantilever bending. A force map is constructed from data from 10,000 curves, each representing 1 pixel in a 50- × 50-nm² area that is a little larger than typical AFM imaging (17). However, many chalk particles are >50 nm in diameter, so this pixel size is small enough to reveal coccolith elements or single, euhedral crystals. The gray square on Fig. 1 shows an example scanning area and its scale relationship.

We investigated samples from a water-filled zone in Maastrichtian chalk from the Danish North Sea sedimentary basin. This stratigraphic layer contains several oil reservoirs, but geological evidence indicates that the area sampled has never been exposed to gas or oil. The rock is very pure calcite (97.5%, rest mostly quartz), and we used stringent sample-handling procedures to avoid contamination by fingers and extended air exposure. We also tested the behavior of freshly cleaved, inorganically precipitated Iceland spar and chalk from the same Maastrichtian layer exposed onshore near the Danish city Aalborg, to verify that drilling mud contamination was not responsible for what we saw. Our intention was to examine pore surfaces. Scanning electron microscopy (SEM) offered no evidence for cleavage across crystals. Fractures propagate along grain boundaries and through pore spaces. Although it would have been very nice to be able to image the same site by both AFM and SEM, this was not possible. AFM resolution is much higher than SEM, and the chalk is extremely friable and very rough. It is not possible to mark an AFM imaging site. If only one particle dislodges during transfer from one microscope to the other, the site appearance changes.

Force Curves, Adhesion, and Indentation.

From a matrix of force curves, maps can be produced that show a range of properties. Fig. 2 is a curve obtained from a single location on a chalk sample as a tip approaches (red curve moving left). The distance 0 μm is set where it first feels a repulsive force, but it pushes further into the surface until indentation pressure reaches a predefined trigger called the maximum force (set at 500 pN on this plot). The trigger is chosen so the cantilever bends slightly, to be sure the tip makes contact. What constitutes the actual surface is a matter of definition at the nanometer scale. Once maximum force is reached, the tip retracts again (blue curve moving right) until there is no pressure on it (force at 0 pN). The curve passes through 0-μm distance and into the negative-force regime, indicating that it adheres to the surface. At some point during retraction, the cantilever spring force overcomes the adhesion force and the tip snaps free, either by a single or by a series of release events. In Fig. 2, partial release is followed by full release. Once the tip breaks free, the curve returns to its flat, equilibrium position, where no forces are felt.

Fig. 2.

A single AFM force–distance curve from the data matrix used to create Fig. 3A-C, showing tip approach (red) and withdrawal (blue). It has been corrected for cantilever bending. The surface is defined as the point where the tip presses with a force of 500 pN (at maximum force). Adhesion, the force needed to disengage the tip, is defined as the minimum force (−210 pN). Indentation has two parts, elastic and plastic deformation. Adhesion work is derived from the integrated area (gray) between the approach and retract curves.

Maps made from maximum-force data resemble conventional AFM height images. A map of adhesion force is constructed by plotting the maximum deflection during retraction (such as −210 pN from Fig. 2) for each x,y point over the imaging area. Maps of surface hardness can also be produced. The distance the tip travels after it feels first repulsion is called indentation. The part that recovers during retraction is called elastic indentation (Fig. 2 Inset), and the part that does not recover is the plastic indentation or deformation. Force curves convolute the properties of the sample surface, the tip apex, and the fluid medium in which the curves are recorded. To minimize complication, we kept the parameters for tip and fluid as constant as possible. Samples were submerged in calcite-saturated solution at equilibrium with atmospheric CO₂, so the only nonconstant in the tip/surface/fluid system was the sample surface. A sudden change in topography, such as a steep step, increases attraction, adding forces from all surfaces adjacent to the tip, those beside, and those underneath.

Freshly cleaved calcite is hydrophilic; a droplet of pure water spreads immediately, so we used inorganically grown Iceland spar as a model for a water-wetting surface. Force mapping showed very flat, smooth terraces with step height varying by less than a few nanometers (Fig. S1). Mean adhesion was ≈37 pN, with variability (determined as standard deviation) ± 35 pN. A typical curve is shown in Fig. 3G. The indentation depth on fresh calcite was <0.5 nm, which is the detection limit for this type of cantilever, meaning that the surface is hard, neither elastically nor plastically deformed. To convert the fresh sample to a hydrophobic surface, we exposed it to octane and collected a series of curves with the same tip. Fig. 3H shows a typical curve. Strong adhesion (>400 pN) pulls the hydrophobic tip toward the oil-wet surface.

Fig. 3.

AFM force maps and curves from two samples of chalk. Maps are generated from 10,000 curves taken in a 100 × 100 grid over a 5- × 5-μm² area. The scale for all maps is the same. (A) Map, derived from height (maximum force) data, that was passed through a Photoshop emboss filter to enhance edges. (B) Adhesion force map of the same area. Two domains, chosen to represent a hydrophobic and a hydrophilic area, have been outlined in A. For patch I, adhesion is high; for patch II, it is close to 0 nN. (C) Elastic deformation on the same surface, constructed from the indentation data. Some soft sites are adhesive, others are not. (D) Emboss-filtered height map from a different chalk sample showing a coccolith fragment (C) and a euhedral rhombic calcite crystal (outlined; R). (E) Adhesion map. The coccolith elements are more hydrophilic than the sutures. (F) Indentation map. The coccolith crystal elements are hard with soft domains between. (G) Single-force curve from freshly cleaved, inorganically grown calcite. (H) The same sample after exposure to octane. (I–L) Curve from the nonadhesive, hard patch II (I) and from the adhesive, relatively soft patch I (J); from a nonadhesive, hard spot, on the face of the calcite crystal (K) (R in D); and from a relatively adhesive, very soft spot on a particle nearby (L). (M) Adhesion force histogram for the entire data matrix from a fresh surface of inorganically grown calcite. The peak represents tip–water drag. (N) Adhesion force histogram for the whole data matrix for the first chalk sample (A–C). The subpeak near 20 pN represents tip–water drag.

Using the same setup, we mapped a fresh chalk surface. Fig. 3A shows a representative 100- × 100-point maximum force map from a randomly selected 5- × 5-μm² area. The original map (Fig. S2), which resembled a typical AFM height image, was passed through a Photoshop emboss filter to enhance feature edges at the expense of the height information. The features cover a range of 3.7 μm from highest (Lower Left quadrant) to lowest points. One can recognize particles, but the edges are blurred by the width of the tip. Fig. 3B is the adhesion map of the same area. Particle edges cause sporadically high values (pink spots). However, the most important aspect is the patchy nature of adhesion. Some domains are very adhesive (red and black) whereas others are not (light blue). From the height-derived image (Fig. 3A), we see that these domains are not distinguished by any particular feature, nor are any of the other adhesion-similar sites. Patches I and II have been outlined for reference. Patch II (compare Fig. 3 A with B) has nearly no adhesion. Its force curves resemble those of calcite (compare Fig. 3 I with G) whereas patch I force curves resemble those of the octane-exposed surface (compare Fig. 3 H and J).

Chalk adhesion values (Fig. 3N) range widely, from 0 ± 5 pN (pale blue) to 3.3 nN (pale pink), with a broad peak at ≈200 pN. The overall mean is ≈240 pN with standard deviation of 160 pN. Because of the very inhomogeneous nature of chalk surfaces, standard deviation should be understood as variability over the map, not as error or uncertainty. For example, median adhesion in patch I is 570, with variability ± 60 pN, whereas in patch II, it is 37, with variability ± 17 pN, very similar to freshly cleaved, inorganic calcite. Patch I is >15 times more adhesive than patch II. Some domains have sharply defined edges where adhesion properties change within 1 pixel (50 nm), whereas for others, properties change gradually. On the force curve from patch I (Fig. 3J), adhesion is ≈600 pN when the tip retracts, but it does not spring free in a single event. At ≈280 nm, a second event releases the tip. The exact cause is not known, but it is known (18, 19) that adhesion can extend for several hundred nanometers away from the surface when organic molecules stick to each other, stretch, and snap free. We see this on the curve for octane (Fig. 3H). Curves for patch II and clean calcite (Fig. 3 I and G) show no detectable adhesion.

A map of the elastic properties is presented in Fig. 3C. Mean indentation over the whole surface is 16 nm, with variability of 40 nm, but some domains are hard, and others are very soft. For some patches, indentation and adhesion correlate. Adhesive patch I is soft, with indentation of 37 nm and variability ± 7 nm (green–blue on Fig. 3C) whereas nonadhesive patch II is hard, indentation 4 ± 2 nm (dark blue). For other domains, indentation is not correlated with adhesion. Some high-elasticity sites (green to white) are among the least adhesive. Because indentation area is on the same scale as radius of curvature for the tip (30 nm), we can assume that the tip apex is round, so interaction resembles a hard ball indenting a soft surface. Using a simple Hertz model, we can estimate the relative difference in Young's modulus (20), a measure of stiffness:

and

where F_Hertz represents the force exerted by the tip, E_surface is Young's modulus for the surface, ν_surface represents the Poisson ratio of the sample, R_tip is the radius of curvature of the tip apex, and δ_elastic is the elastic indentation. By using measured indentation values, the stiffness for patch II is ≈30 times higher than for the softer patch I.

Fig. 3 D–F is from a different chalk sample. The height-derived image (Fig. 3D) shows a segmented band (marked C) with the correct size and shape to be a coccolith fragment, such as those on Fig. 1. Above and to the right of Center, is a rhomb (R), typical morphology for recrystallized calcite, such as the crystals in Fig. 1. Maps of adhesion and indentation (Fig. 3 E, F, and K) show that one face of the rhomb is hard with no adhesion, whereas the other faces are somewhat adhesive. Another particle close by is somewhat adhesive (small black area, Fig. 3 E and L) and very soft (green, Fig. 3 F and L). Its plastic indentation is 70 nm, and elastic indentation is 105 nm. On the coccolith fragment at the Bottom of the image, the ridges of the individual elements are hard and nonadhesive. The sutures between are adhesive and soft.

Chalk particles have heterogeneous surface properties, even at submicrometer scale. Although the coccoliths in North Sea chalk are undoubtedly calcite, the surface properties of this biogenic material are often very different from those for the euhedral, inorganically formed crystals. Some areas, such as patch II, behave the same as inorganic calcite whereas others, such as patch I, are many times more deformable and adhesive. In many cases, the force curves (such as Fig. 3J) prove that the material responsible for tip attraction has complex structure.

Correlation of Adhesion, Hydrophobicity, and Wettability.

AFM tips, functionalized to be hydrophobic, identified sites of high adhesion, so these were directly interpreted as hydrophobic and patches with low adhesion, as hydrophilic. Measurements at many sites were between hydrophobic and hydrophilic, even from one pixel to the next. It is not possible to define precise borders for the patches because resolution is limited by the size of the tip. With some assumptions, we can estimate maximum contact area. If the sample is flat, if the radius of curvature is <30 nm, if the tip's soft coating is at most 2 nm thick and if the coating is squashed flat at the tip apex, the contacting surface makes a circle with a diameter of ≈22 nm, yielding a contact area ≈365 nm². In such an area, there are many molecules responsible for the tip–sample interaction. Thus, each pixel in the adhesion map is determined by the adsorption density and the hydrophobic/hydrophilic nature of many individual molecules.

An important question is how to correlate hydrophobicity with the macroscopic concept, wettability, which is typically determined from the contact angle between a fluid droplet and a solid or from imbibition, like in the Amott test. A purely oil-wet surface produces a water droplet with a contact angle of nearly 180°, whereas the contact angle for a completely water-wet surface is said to be 0. For angles between, the surface is called mixed-wet (21). Using a model based on forces between two cylinders, Drummond and Israelachvili (21) described the relationship between work and contact angle by using the Young-Dupré equation:

where γ_ow represents the interfacial tension between oil and water, θ is the contact angle, and W represents the adhesion work per unit surface area. The relationship is universal because it depends only on W and γ_ow (22). Although an AFM tip interacting with a chalk surface is not the same as two cylinders interacting, it is a reasonable approximation, especially for sites where only the tip apex interacts.

Adhesion work is determined by integrating the area between the approach and retract curves. Only the part of the curve representing adhesion should be included (gray area, Fig. 2). For force curves collected in fluid, drag on the cantilever causes an artifact. As it moves toward the surface, the cantilever bends away, and as the tip retracts, the cantilever bends toward the surface. We see the effect in the force curve as a slight offset between the approach curve, which is used to set the force scale at zero, and the retract curve. The offset gives a measure of the water drag influence. It is particularly clear in Fig. 3G, and it is visible on the others. The offset gives rise to the small, narrow peaks at 5–20 pN observed in the force histograms (Fig. 3 M and N). Cantilever bending from water drag contributes significantly to calculated work so we derived a correction factor from the clean, fresh calcite, which is completely hydrophilic, so adhesion should be 0. Cantilever–water drag caused ≈10 pN difference (Fig. 3M), which adds ≈10 aJ (1 aJ = 10⁻¹⁸ J) to the calculated adhesion work (Table 1, first column). Subtracting tip–water drag from the calculated adhesion work, which for freshly cleaved calcite is 10 aJ, gives a contact angle of 0°, as it should be for a hydrophilic surface. Tip–water drag is constant within curves, but it varies between curves. It was estimated and subtracted before contact angles were determined.

Table 1.

Adhesion and contact angle derived from force curve parameters for several areas

Parameters	Calcite	Chalk (Fig. 3A– C)	Chalk (Fig. 3D–F)	Chalk patch I	Chalk patch II
Adhesion force, overall mean, pN	37	240	180	570	37
Variability in adhesion force, pN; standard deviation	35	160	240	60	17
Indentation ± variability, nm	<0.5	16 ± 40	23 ± 45	37 ± 7	4 ± 2
Uncorrected adhesion work, aJ	10	15	12	48	1.3
Work of tip–water drag*, aJ	10	6	4	6	6
Adhesion work per unit area†, mJ/m2	0	25	22	115	3
Derived contact angle, °	0	64	60	180	0

Uncertainty arises from the difficulty in determining the contact area between the tip and sample.

*From average separation between approach and retract curves (Fig. 2, gray area).

^†Assuming tip contact area of 365 nm² and correcting for adhesion contribution of tip–water drag.

Chalk surface properties vary at a local scale. Hydrophobic and hydrophilic patches alternate over the space of less than 1μm. Patch I, representative of many hydrophobic areas (dark red and black), has a mean work of adhesion corrected for tip–water drag of 42 aJ. By using the interfacial tension between hexadecyl and water [46 mJ/m² (23)], a contact area of 365 nm² and Eq. 3 gives a contact angle of 180° or W/γ_ow >2. Patch I is therefore completely hydrophobic, or oil-wet, according to the convention used by Drummond and Israelachvili (21). Adhesion for patch II, representing hydrophilic regions (pale blue), gives a contact angle of 0°, as for inorganically grown calcite. Typically, chalk is described macroscopically as mixed-wet, implying that surface properties are averaged for the whole sample. Thus, for comparing one of our local scale analyses with another, we can consider the differences from the mean for each surface as a measure of their inhomogeneity. Table 1 presents a summary. For the two chalk samples, mean work of adhesion, 9 and 8 aJ, gives contact angles, 64° and 60°, which are in the range for mixed-wet surfaces.

Salathiel (6) originally proposed that mixed-wettability means a continuous pathway of oil-wet surfaces interconnected through pores. Skague et al. (24) categorized wettability for sandstones into three subclasses defined by size and location of the oil-wet patches relative to the size of the pores and the channels between them. A “fractionally wet” sample was defined to have variable size water- and oil-wet domains. In mixed-wet large sandstones, only the large pores are oil-wet, and in mixed-wet small, the small pores are also oil-wet. The pores in our chalk samples are ≈100 times smaller than sandstone (24), but their randomly distributed, variable size hydrophobic and hydrophilic patches put them decidedly into the fractionally wet class. For macroscopic determination by imbibition, the sample would appear water-wet, which is consistent with macroscopic data from our own samples and those of others (25); water imbibes easily into freshly fractured chalk surfaces.

Until now, it has been assumed that mixed-wettability develops when oil migrates from the source rock into the reservoir. Our results show that hydrophobic patches are an innate property of chalk. Published studies proposing mechanisms for development of mixed-wettability at the single-pore scale are based on measurements from pure crystalline surfaces that are converted from water-wet to oil-wet by treating with oil or other materials (10, 26–29). Some research has used samples from chalk reservoirs or from outcrops and modified wetting behavior by introducing or reintroducing crude oil (30–33). There are no reports in the literature of natural samples examined at pore-level resolution.

One returning question in wettability discussions is how crude oil turns an initially hydrophilic surface to a hydrophobic one. Calcite, even exposed only to air, is covered with a layer of water that is ≈1.5 nm thick. The first water is tightly bonded to, and structurally ordered by, the underlying surface (34). The oil would have to replace the associated water, and the structurally attached water, before it could interact with the ionic CaCO₃ surface, where it would establish anchoring points for other oil-wet compounds (26). Various theories have been proposed for how this could occur. One suggests that mineral edges penetrate the water film, allowing direct contact with asphaltenes (11). Another proposes that small, amphiphilic molecules from the crude oil migrate to the hydrated surface (11). However, the force maps prove that some areas on chalk particles are already hydrophobic before oil enters the reservoir. This is true for domains such as patch I (≈500 nm across) and for single points (≈50 nm) where adhesion is in the mixed-wet range. Intrinsic hydrophobic patches mean the scenario for how oil gets into chalk is simplified.

Implications

Our results show that the surface properties of chalk are heterogeneous. The internal pore surfaces vary in wettability, and there is soft material that is not necessarily correlated with hydrophobicity. The nature and origin of this material are not yet known. Pevear (35), Strand et al. (30), and others have reported that small amounts of clay and silicates affect wettability, but these minerals are also hydrophilic when they are fresh and clean. The force curves show that the adhesive material can be pulled several hundreds of nanometers out from the surface. The indentation data show vast differences also in elasticity and plastic deformation, even over distances of 50 nm. If all chalk particle surfaces were as clean as freshly cleaved calcite (such as Fig. 3G), elasticity differences would be negligible. Chalk is predominantly biogenic calcite. Coccoliths growing in sea water today are protected from dissolution by an organic coating that can be removed by an oxidizing agent, and coccolith growth is controlled by a complex polysaccharide (2, 36, 37). Therefore, we propose that this organic material, and other organic compounds entrained in the original sediments, remain attached to the coccolith fragments, controlling their wettability throughout geological time and aiding oil emplacement.

Methods

We used an Asylum Research MFP-3D AFM with software from Atomicforce DE to generate and analyze the hundreds of thousands of force curves. To be able to compare maps from different sites, we developed a procedure from preliminary experiments, optimizing parameters such as tip properties, scan size, data collection rate, noise level, time and memory use. We used 1-μm pull length, 4-Hz scan rate, 500-pN trigger force, and 100 × 100 data points over areas 5 × 5 μm². Tips were Olympus biolevers with a nominal spring constant of 30 pN/nm. For each experiment, deflection sensitivity was established on a hard surface, and the spring constant (20–30 pN/nm) was estimated from a thermal spectrum fit. Gold-coated biolevers were functionalized in an ethanol solution of 1 mM CH₃(CH₂)₁₅SH for >24 h to ensure complete monolayer coverage (38–40). The method was tested on a gold-coated silicon wafer by observing the contact angle of a water droplet. We verified tip stability and measurement reproducibility by repeating many scans.

Chalk composition was determined by using X-ray diffraction and Rietveld refinement. Experiments began with a piece (≈5-mm diameter) broken from a sample taken from inside a drill core to avoid drilling mud contamination. Freshly cleaved (34), pure, inorganic Iceland spar calcite (Chihuahua, Mexico) served as control. Samples were glued to clean glass substrates with epoxy (DANA LIM 334) and cured for >6 h. Data were collected at 23 °C during immersion in solutions saturated with crushed Iceland spar in MilliQ water (18 MΩ/cm) at equilibrium with atmospheric CO₂. We investigated at least 5 sites on each sample. To prove that epoxy did not affect surface composition, we examined a sample mounted under a gold spring. Results from the mechanically held sample were indistinguishable from those of the glued samples.