Qualitative phase characterization of the "reaction products-calcite matrix" system
First, the capability of combined µXRPD and µXRF imaging to characterize the different reaction products (CaOxs and CaPs), within the newly formed system in all treatment methodologies (single treatment and sequential treatments), was studied.
Analysis of the maps’ average µXRPD patterns for all Noto Limestone samples analyzed, untreated, and treated were first conducted. The average diffraction patterns of the untreated limestone samples show predominantly the peaks associated with calcite, with occasional quartz. This trend remains consistent across treated samples' diffraction patterns, as illustrated in Fig. 3. In these patterns, the main peaks of the reaction products were weak. This result aligns with prior literature and reflects how the reaction products of both solutions are in a low wt% fraction if compared to the abundant calcite matrix [47].
Focusing specifically on the reaction products from the AmOx solution (Fig. 4a and b), the CaOxs, low-intensity peaks appear in some surface-proximate diffraction patterns (as shown in Fig. 4c), where the highest crystallization activity for these phases usually occurs, indicating their presence without definitive identification. The challenge in their identification arises due to the nano-crystalline nature of the main CaOx phases— whewellite (CaOx monohydrate) and weddellite (CaOx dihydrate)—together with their low fractions if compared to the calcite matrix [44]. Even with our high-resolution XRPD patterns, discerning these CaOxs in all patterns remains difficult.
The sole examination of µXRF imaging does not facilitate their identification either, as CaOxs share a similar elemental composition to the calcite matrix (Ca, C, O). However, in the case of the AmOx-treated samples, the high-resolution µXRF Ca maps have shown, as seen in Fig. 4d, low-intensity (~ 70–100 in the scale bar) and high-intensity areas (corresponding to ~ 170–250 in the scale bar). An unusually big gap in the intensity of Ca through the elemental map was here registered. These intensity variations throughout the Ca maps could, therefore, be correlated with different calcium contents in crystalline phases within our "reaction products-calcite matrix" system. Given that CaOxs contain less calcium (around 38% CaO) than calcite (approximately 55% CaO) [48], these intensity fluctuations in the elemental map could suggest a different distribution of oxalate and calcite phases within the stone system, corresponding respectively to the low and high-intensity area.
Combining the µXRF maps with µXRPD data, we could generate area-specific average diffraction patterns by selecting pixels from areas exhibiting these calcium intensity variations. These patterns reveal peaks attributed to CaOxs with good resolution, as shown in Fig. 4e. The area-specific diffraction pattern exhibited a high signal-to-noise ratio and a consistently flat baseline, which helps us confirm not only the crystallization but also facilitates the unambiguous identification of whewellite and weddellite.
The situation becomes even more complex in the case of DAP-treatment samples (Fig. 4f and g). Like CaOxs, CaPs are nanometric-sized products and low in concentration [43]. Additionally, the complex reaction of DAP solutions with calcite does not yield only hydroxyapatite (HAP) but results in a mixture of by-products with varying Ca/P ratios in concentration distributed within the matrix based on reaction conditions [45]. These by-product phases share similarities in elemental composition and crystalline structure [46], making their phase identification even more complex. Consequently, map average diffraction patterns of DAP-treated samples lack identifiable peaks even in our diffraction patterns, as shown in Fig. 4h, which hinders the confirmation of the presence of CaPs.
Moving on to µXRF imaging, a situation like the AmOx-treated sample was observed in the elemental maps of calcium distribution in samples treated with DAP. This is coupled with localized high phosphorus intensities (Fig. 4i). Given the known high Ca/P ratio in HAP (around 1.76), correlating calcium intensity variations with areas of high phosphorus concentration implies the potential CaP presence. Confirming this involves associating the µXRF imaging with the µXRPD data, generating area-specific diffraction patterns for pixels showing Ca/P elemental correlation. By generating specific diffraction patterns for these correlated areas, the identification of CaPs was achievable. The resulting diffraction patterns showed well-resolved HAP peaks that supported its visualization and identification. These patterns exhibit also good resolution at the low angular range where by-products of the reaction reside. The result is that this combined technique aids in identifying traditionally difficult-to-characterize [21] by-products such as octacalcium phosphate (OCP) and other CaP phases associated with HAP crystallization, such as Ammonium Dihydrogen Phosphate ADP, and carbonate-substituted HAP (C-HAP), as shown in Fig. 4l, providing crucial information about the composition of the DAP "reaction products-calcite matrix" system.
The combination of techniques proves effective for qualitative phase analysis of both CaOxs (whewellite and weddellite) and CaP (HAP and by-products), even in the more complex scenario of sequential treatment samples (DAP + AmOx). Despite the additional challenges posed here because of the simultaneous presence of CaOxs and CaPs, this approach facilitated the visualization and characterization of the newly crystallized phases (oxalate and phosphate) within the complex systems generated by these treatments, as shown in Fig. 5.
In the end, focusing solely on the average µXRPD pattern, the identification of reaction products was limited mainly because of their low concentration compared to the dominant calcite matrix. In contrast, examining µXRF data alone resulted challenging in samples lacking elemental markers for identifying CaOxs and could only hint at the crystallization of CaPs when the markers were present. Integrating both techniques enabled the production of area-specific diffraction patterns associated with those zones exhibiting calcium intensity variations or high Ca/P correlation, allowing for the identification of specific crystalline phases characteristic of the newly formed “reaction products-calcite matrix" system.
Phases’ penetration depth and spatial distribution of the reaction products
The combination of techniques used in this study yields an invaluable advantage in generating comprehensive µXRPD maps. Phase-specific distribution maps were generated by isolating distinct peaks associated with the identified crystalline phases in the diffraction patterns, providing detailed insight into the spatial distribution of each phase. SR-µXRPD mapping allows micrometric individual mapping of each reaction product distribution within the "reaction products-calcite matrix" systems, offering exceptional spatial resolution. Such data not only visualizes the penetration depth of the crystallization but also localizes where the chemical reactions take place.
Our analysis consistently reveals phase-specific color maps accurately depicting the areal distribution of CaOxs and CaPs crystalline phases across all examined samples, without artifacts.
In the context of AmOx-treated samples, the areal distribution data through RGB correlation maps reveals a distinct distribution pattern of CaOx phases: a gradient from the surface, with weddellite (WED) overlaying whewellite (WHE). This observation underscores the spatial relationship between these phases shown in Fig. 6a. The data underscored a crystalline network following the solution diffusion path within the stone's microstructure, originating from the surface interface. These findings highlight the CaOx areal arrangement given by the treatment. Such data offer, for the first time, an opportunity to establish correlations between the phases of CaOx formed and their spatial distribution, allowing interpretations of the consolidation effect. Currently, literature only suggests that the distribution and penetration of distinct CaOxs significantly affect resultant system properties [49].
The application of µXRPD mapping of DAP reaction products has revealed unprecedented details regarding the areal distribution of CaP phases within the stone, particularly highlighting hydroxyapatite (HAP) and its associated by-products (ADP, OCP, and C-HAP). As shown in Fig. 6b, HAP distribution follows a gradient from the surface, less homogeneous than the CaOxs network. The by-products, less identified in prior studies, exhibit a distribution pattern linked to HAP, extending to similar depths within the stone and localized in spots within its porous network. It is important to acknowledge that the representation of these CaP phases in the color maps might be subject to potential inaccuracies (either underestimated or overestimated) because of the mixture of phases within our systems. CaP peaks are collocated in the diffraction pattern close to the peaks of other phases complicating their isolation and mapping. Despite this, the ability to pinpoint the spatial distribution of both HAP and its by-products remains crucial. The literature already suggests correlations between the presence of these by-products, such as ADP indicating acidification, and changes in the stone's condition [10]. Therefore, having data on CaP distribution allows for a deeper exploration of these correlations, facilitating a better understanding of the localized effects of CaP phases on the stone's chemical changes and conditions.
In the case of sequential treatment, the mapping of CaOxs and CaP phases, and the examination of phase coexistence through RGB correlation maps, highlight distinct variations in both penetration depth and areal distribution compared to single treatments. The data (Fig. 6c) showed that, in sequential treatments, HAP and by-products are dispersed throughout the stone bulk rather than follow a gradient from the stone surface. CaOxs seemly exhibit different penetration depths compared to single treatments, although they are still primarily located on the surface. Additionally, the maps illustrate sparse areas of coexistence between CaOxs and CaPs. These observations strongly indicate interactions between the solutions’ reactions. The significance of obtaining these data lies in its capacity to shed light on these interactions, thereby enhancing our understanding of combining treatment effects on stone microstructures absent in the literature.
Quantitative phase analysis and Orientation of the reaction products
The high resolution of µXRPD maps also enables us to obtain information on quantitative phase concentration and orientation at the scale of the studied area (ROI = 400 x 400 µm²). Acquiring such data at this scale, with a clear view of the phases’ distribution, provides essential insights for assessing localized effects and evaluating treatment performance.
Table 1
Quantitative Assessment of the Reaction Products Areal Amounts (Mean % ± Standard Deviation) on Surface Maps for Various Treatment Methodologies.
| Reaction Products |
| WHE | WED | HAP | C-HAP | OCP | ADP |
AmOx- treated | 25.4 ∓ 2.71 | 1.54 ∓ 0.47 | - | - | - | - |
DAP-treated | - | - | 22.2 ∓ 3.91 | 1.48 ∓ 0.61 | 7.62 ∓ 1.39 | 0.26 ∓ 0.13 |
DAP + AmOx treated | 21.5 ∓ 3.47 | 2.76 ∓ 0.59 | 16.2 ∓ 2.17 | 1.16 ∓ 0.45 | 5.38 ∓ 1.72 | 0.51 ∓ 0.18 |
For quantitative assessment of newly crystallized reaction products, image analysis of phase-specific µXRPD maps was used to quantify the areal fraction of reaction product observed in the 2D map (shown in Table 1). In the case of AmOx-treated samples, this quantification confirmed existing literature by highlighting predominant whewellite crystallization, while uncovering unexpectedly high content of weddellite in some surface areas. Similarly, in samples treated with DAP solution, through quantitative evaluation, we categorized the main reaction products (HAP and OCP, 22.2% and 7.6%, respectively) as well as reaction by-products into minor (between 1% and 0.1%) and trace (less than 0.1%) products, based on their concentrations. The localized quantitative data derived from µXRPD map image analysis, of the sequentially treated samples, revealed that the % of crystallization of the reaction compounds is, generally, in line with the one in the single treatments.
Additionally, µXRPD distribution maps are generated by selecting Bragg peaks of the crystalline phase. As each peak on the diffraction pattern corresponds to a specific set of crystallographic planes within the crystalline material, mapping a peak shows the areal distribution of a specific crystal plane. Correlation through RGB correlation of maps associated with different peaks of a single phase is here used to reveal the crystal orientation of the reaction products. Indeed, the lack of merging colors in the RGB correlation maps in specific areas was used to detect the crystal orientation of a selected crystalline plane. For instance, in the case of CaOxs (both whewellite and weddellite), they exhibit randomly distributed crystallites within the stone matrix, showing up as a single cohesive yellow color on the maps (Fig. 7a). On the other hand, CaPs, primarily HAP, in DAP-treated samples show preferential orientations in some areas, as distinct green (HAP d310) and red (HAP d002) color distributions show up on the correlation maps (Fig. 7b). This observation becomes particularly intriguing when studying sequential treatments where no significant orientation of CaPs can be detected (Fig. 7c). Here the data show how CaPs crystallized differently. This proves an interaction between the solutions (the DAP-stone system is disturbed by the crystallization of CaOxs,) and the reactivity of the CaPs phases to be disturbed by the interaction.
Overall, our experimental setup prioritized high spatial resolution and simultaneous µXRF and µXRPD acquisition, but this compromised the features of the collected X-ray diffraction patterns hindering, for example, a proper Rietveld analysis. The preclusion of Rietveld’s full-profile fit hinders any structural analysis of the crystalline phases, along with a phase quantification (in a multi-phase system) based on the refined scale factor of each crystalline component. However, this limitation is here efficiently overcome, as 1) the crystal structure of the co-present crystalline phases, in the system under investigation, is already well known and 2) the relative (areal) fraction of the crystalline components can be obtained through the µXRPD map analysis, along with their average crystallites orientation. In this light, the lack of the Rietveld full-profile fit does not diminish the description of the complex polycrystalline system under investigation and opens a new route for other scenarios in which the application of full-profile fit analysis is precluded.