LA-ICP-MS analysis of trace and rare-earth element distribution in calcite fracture fillings from Forsmark, Simpevarp and Laxemar (Sweden)

Concentrations and spatial distribution of trace elements in secondary minerals provide valuable information about mobility controlling processes in natural fractures. Important examples include rare-earth element contents that act as analogues for the retention of trivalent actinides such as Am/Cm or Pu(III). The secondary phases (carbonates) investigated in this study originate from exploration drilling bore cores of the Swedish Nuclear Fuel and Waste Management Company SKB (Forsmark, Simpevarp and Laxemar, Sweden). Here, high-resolution element analysis (Micro-X-ray Fluorescence-Spectrometry (µXRF) and Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS)) is applied to scan for Na and the trace elements Mn, Fe, Sr, Pb, Th and U as well as the rare-earth elements Y, La, Ce and Yb associated with carbonate fracture fillings. High resolution element maps highlight growth zones and microstructures within the samples, which are not detected by the usual point and line measurements. Evidence of phase-dependent partitioning is observed. The partition coefficients, D, determined from formation water and carbonate data were compared to experimentally generated coefficients and values derived from a 17-year precipitation experiment carried out at the Äspö Hard Rock Laboratory (HRL). Distribution coefficients of the light rare-earth elements La and Ce have been found to be relatively high in the studied samples, whereas the coefficients of distribution of Sr and U are remarkably low. Overall, the results of this work show that the secondary calcite formed in deep granitic fractures coprecipitated periodically with significant amounts of radionuclide analogues (i.e., rare-earth elements).


Introduction
Ever since nuclear power plants have been part of the civil energy mix, the responsible long-term disposal of the radioactive waste and isolation from the biosphere has been a major challenge. Even though radiotoxicity gradually declines over geological time (Joyce 2017), the long persistence of nuclear waste and its decay products require a carefully selected concept for the long-term storage (Ojovan et al. 2019). Crystalline rocks are widely discussed as one suitable host rock formation for the deep geological disposal of spent nuclear fuel (SNF). To this day, major investigation campaigns on crystalline host rock are carried out for example at the Fennoscandian Shield of Sweden and Finland (Andersson et al. 2013;Milnes et al. 2008), at Archean granite-gneisses in Siberia (Laverov et al. 2016), or within underground research laboratories such as the Grimsel Test and Henderson 2013;Füger et al. 2019;Smrzka et al. 2019). Hence, the study of the spatial trace element distribution within the fracture filling minerals can help to reconstruct the paleo-hydrogeological evolution of the site Milodowski et al. 2018;Mizuno et al. 2022). In the context of SNF storage, these findings allow us to estimate the potential TE retention under prospective hydrogeologic and climatic changes.
At Forsmark, Simpevarp and Laxemar the fracture systemof the granitoid bedrock is filled with assemblages of hydrothermal, sedimentary and low temperature fracture fillings formed at intermittent fluid flow events. Low temperature carbonates in Forsmark, Simpevarp and Laxemar show enrichments in REE and other elements, e.g., Mg, Mn, Fe, Sr, Th and U (Landström and Tullborg 1995;Drake et al. 2012Drake et al. , 2018Maskenskaya et al. 2015). TE analyses of these fracture coatings have been carried out mainly on ground and dispersed bulk samples or as point measurements on intact rocks.
In contrast to previous works, this study aims to gain a more detailed understanding of the structural incorporation of TEs in calcite fracture-fill. High-resolution spatial distributions allow for a precise evaluation of the formation conditions and their influence on element retention. In this study, µXRF and LA-ICP-MS were used to examine the distribution patterns of the trivalent REEY, especially Y, La, Ce and Yb in association with other common elements (Na, Fe, Mn, Sr, Pb, Th, U). Comparing calcite compositions and groundwater compositions from the same (or adjacent) borehole sections enable calculation of apparent TE/Ca partitioning coefficients, which signify the role of solid solutions in the mobility of RN under natural conditions. Based on these results, precipitation rate and plausibility of the interpretations were estimated.

Geology
Simpevarp and Laxemar site is located ca. 320 km southof Stockholm at the Swedish east coast (Fig. 1). The nuclear power plant of Oskarshamn and the Äspö Hard Rock Laboratory (HRL) are in close proximity to the site. The Forsmark site is located ca. 350 km to the north of Oskarshamn, adjacent to the Forsmark nuclear power plant (Fig. 1). The sites are situated in the Svecokarelian Orogen of the Fennoscandian Shield. The predominantly igneous bedrock (ca. 1.91-1.75 Ga) underwent numerous cycles of mostly Proterozoic tectonic deformation and magma intrusions and was uplifted later in time (Guenthner et al. 2017). The sites belong to different tectonic units and were affected by early ductile and later brittle deformation, i.e., Sveconorwegian faulting, late-and post-Caledonian faulting, and Permian extension (Hermansson et al. 2007;Viola 2008;Viola et al. 2009;Saintot et al. 2011;Tillberg et al. 2020Tillberg et al. , 2021. Simpevarp and Laxemar belong to the Transscandinavian Igneous Belt (TIB) which was formed ca. 1.85-1.66 Ga ago (Högdahl et al. 2004) in the context of the volcanic activity related to the subduction alongside the Svecokarelian orogeny, to which Forsmark belongs (Wahlgren 2010). The TIB consists of granite to quartz monzodiorite to diorite gabbros. These areas experienced further major periods of igneous activity viz. the granitic magmatism ca. 1.45 Ga ago linked to the Hallandian orogeny and the intrusion of dolerite dykes (ca. 0.9 Ga) in the TIB due to the Sveconorwegian orogeny (Söderlund et al. 2005;Wahlgren et al. 2007). The long-term uplift of the region enabled the manifestation of a sub-Cambrian unconformity followed by several cycles of loading and unloading. Present regoliths are assumed to be deposited by the end of the last glaciation and after deglaciation (Lundin et al. 2004;Tröjbom and Söderbäck 2006;SKB 2008;Nyman et al. 2008;Brydsten and Strömgren 2010).
During the Quaternary, periods of glaciation, glacial reboundand shoreline displacements shaped the area. Along with the glacial waters, a range of meteoric, lacustrine, brackish, marine, and fresh water, so called end members, infiltrated into deeper aquifers. The paleo-hydrogeological setting has been influenced by the mixing processes of these end members with deep saline waters (Laaksoharju 2004;Laaksoharju et al. 2008aLaaksoharju et al. , 2008cGómez et al. 2014).
As a result of the complex groundwater evolution in combination with upwelling hydrothermal fluids multiple generations of fracture fillings developed. At both sites, a set of hydrothermal fracture fillings is followed by two younger generations of fracture fillings of low-temperature groundwater precipitates (Generation 3/4 in Forsmark, generation 5/6 in Simpevarp and Laxemar). In this study, low temperature carbonates were investigated (see e.g., Drake et al. 2006Drake et al. , 2009bSandström et al. 2008 for detailedoverviews of the fracture filling generations).

Methods
Calcite crystal samples were hand-picked from drill cores taken at the SKB test sites Simpevarp, Laxemar and Forsmark and named according to borehole number and depth in meter (Table 1). Formationwater samples were taken from the nearest possible sampling location in the same borehole (Table 1). The hydrogeochemical data of the formation water samples were provided by the SKB (Table 2) (Laaksoharju et al. 2008b(Laaksoharju et al. , 2009. Calcite samples of < 3 cm length were impregnated in Araldite 2020 epoxy resin (Huntsman Corporation, Salt Lake City, Utah, USA) and polished to create a planar surface. For characterization via Polarized Transmitted Light Microscopy (PTLM), thin sections were polished to a thickness of 30 μm. Images were generated using the VHX-6000 Keyence digital microscope (Keyence Corporation, Osaka, Japan) and the research microscope Zeiss Axioplan 2 (Carl Zeiss AG, Oberkochen, Germany).
Micro-X-ray Fluorescence-Spectrometry (µXRF) (M4 TORNADO PLUS µXRF, Bruker Corporation, Billerica, Massachusetts, USA) was used to generate spatially resolved element map scans of the planar sample surfaces. Map scans were measured in one cycle with overlapping spots (spot size 20 μm, spot distance 10 μm) under vacuum condition (20 mbar). The acceleration voltage was 40 keV at 130,000 cps, the time per pixel 15 ms and the energy deviation between 0.4-0.8 meV. Samples with well-distinguished zonation were selected for further analysis.
Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS) analysis were carried out with a LSX-213 C2 + Laser Ablation System (Teledyne The samples were taken in close proximity to the fracture fillings listed in Table 2. * Additional groundwater data from (Laaksoharju 2004) were added for drilling KSH01A Section 150-200 m n.a. no data available, < LOD data below limit of detection pH (field) 6.74 n.a n.a n.a n.a n.a pH ( Speed was 10 μm/s, energy fluence 10 J/cm 2 . Due to the destructive character of the method, direct measurement repetition on the same sample spot is not possible. However, the measurement of the MAC-S 3 standard prior to and after each calcite scan ensured homogeneity and comparability of the measurements. Element specific detection limit (LOD) ranges within one map scan are given exemplarily in the Appendix, Table 9. The LA-ICP-MS data were processed with the data analysis software Iolite v3 (School of Earth Science, University of Melbourne, Melbourne, Australia) running in IGOR Pro 6 (WaveMetrics Inc., Oregon, USA). Single line scans were hand processed based on a similar approach with Microsoft Excel 2016 (Microsoft Corporation, Washington, USA). Calcium contents were determined by calculating the difference between the stoichiometric values of Ca minus the cumulative trace elements content (Eq. 1): Element maps and single line scans allowed the identification of generations of overgrowth. To clearly assign each measurement point of the mappings to a growth zone, multivariate analyses were performed using the software R (version 3.6.1) and RStudio® (version 1.2.5019) with basic packages as well as ggplot2 and psych. Only elements for which a zonation was visible in the element map (Mn, Fe, Sr, Y, La, Ce, Yb, U) were included and the data were logtransformed and auto scaled prior to any multivariate calculations. First, a principal component analysis was performed for each sample. In a next step, only principal components whose scores showed a (spatial) zonation were included in a k-means cluster analysis of the corresponding principal component scores. Different numbers of k were chosen and for each of them the cluster analysis was calculated ten times. The solution to work with was chosen based on a combination of a reasonable number for k (zones not spatially connected should be separate clusters) and the most frequent solution of the ten repetitions.
For single line scans, the contents of the outermost overgrowth were extracted from the hand processed data as the mean of the ten outermost laser spots (100 µm). For map scans, the module profile points of the software Iolite was (1) c Ca calcite = 400, 400 ppm − c(ΣTE calcite ) used to extract data points alongside the youngest overgrowth with an estimated thickness of several tens µm.
The REE fractionation was calculated based on the Chondrite normalized La and Yb content of the calcite and the Chondrite normalization factors taken from McDonough and Sun 1995. For sample KSH01A_212, the REE fractionation was evaluated based on Chondrite normalized La/Y ratios as for this sample, no Yb contents were measured.
The partition ratios of trace elements between the formation water and the youngest overgrowth were calculated based on the Henderson-Kracek division coefficient D TE (Eq. 2), (Henderson and Kracek 1927): whereas η is the molar amount of the subscripted element in the superscripted phase.
The precipitation rate R P was calculated based on Eq. 3 (DePaolo 2011). Experimentally driven TE/Ca partitioning ratios of this work, the equilibrium partition coefficient K eq and the forward partition coefficient K f taken from DePaolo 2011 and the calcite dissolution rate R b adapted from (Chou et al. 1989) were applied to Eq. 3:

Morphology, overgrowth, zonation of the carbonate fracture fillings
Single calcite crystals from Forsmark and Simpevarp show euhedral crystal habitus. The habitus of the single growth zones at the samples KSH01A_205_A/B, that consist of several concentric, accordant growth zones, changes between elongated c-axis (c ≈/ > a) and elongated a-axis (c < a).
The sample KFM04A_233 and KSH01A_205_B show inclusions of Fe-oxides/-hydroxides. The sample aliquots of KFM04A_233 and KSH01A_205_A/B consist of single euhedral crystals, whereas the others are healed fracture fillings. The samples KLX04_669 and KLX19A_414/428 display calcite fracture-fill attached to wall rock and display aggregates of anhedral carbonate crystals. Sample KLX04_669 shows traces of wall rock alteration. The fracture wall is rimmed by the oldest generation of euhedral quartz followed by calcite ( Fig. 2A andB). Samples from Laxemar (KLX19A_414/428, Fig. 2C and D) show a greenish contact zone (GCZ) within the fracture filling calcite. At a depth of 414 m, the fracture filling shows anhedral growth with a sudden change of crystal habitus approximately in the center of the fracture (at the GCZ). At 428 m below the surface, the greenish contact zone is expanded to about 3-5 mm with a more symmetrical structure of the growth zones on both sides.

Simpevarp
The two samples of the borehole KSH01A, Sect. 205 m, show analogue overgrowths with similar element assemblage but different absolute contents.
At least seven distinct periods of precipitation, with growth zone 1 being the oldest overgrowth, were identified by morphology and trace element assemblage. Based on the qualitative element content, the strictly statistically generated cluster analyses revealed seven and eight clusters shown in Fig. 3. These are, however, not to be confused with the growth zones which were interpreted based on element content, morphology and mineralogy of the fracture fillings. Overall, sample B shows lower TE contents. In both samples, enrichment patterns of light rare-earth elements (LREE) follow the opposite trend as Fe and Mn. At intervals, distribution patterns of Sr are congruent with Mn and Fe. Uranium is strongly enriched in the two youngest growth zones. Within sample A, similar distribution patterns for Pb, Th and U are apparent. Growth zone 4 of sample A shows signs of face dependent partitioning. The growth zone 5 shows a gradual decline in Na, Mn, Fe, Sr and heavy rareearth elements (HREE), but increase in LREE. The degree of fractionation of REE differs throughout the samples and reaches highest ratios in zone 2 (median Chondrite normalized ratio: 58.27 for sample A and 5.45 for sample B) and the youngest part of zone 5 (median Chondrite normalized ratio: 236.68 for sample A and 68.79 for sample B). Further similarities are evident for zone 2 and the youngest part of zone 5, e.g., both show strong enrichment in LREE.
The third analyzed sample of borehole KSH01A originates from a depth of 212 m and contains two growth zones. As Yb contents are often below detection limit in this sample, the La/Y ratio is used as indicator of REE fractionation. The older, Fe, Sr and REE enriched zone 1 is subdivided into zone 1a with high TE contents (see Appendix, Figure 9) and zone 1b with decreasing TE contents and REE fractionation in favor of LREE (La/Y ca.1-12.5). Within the adjoined zone 2, TE contents are generally low. The La/Y ratio decline to < 1. The crystal is surrounded by a rim enriched in TEs and REE. La/Y ratios reach values up to > 100.
In general, the observed calcite TE contents are higher compared to sample measurements performed by Drake and Tullborg (2009a), Drake et al. (2012) and Maskenskaya et al. (2015) from comparable depths. Only Th and U contents are lower than expected values based on the former study. The fact that the number of growth zones varies with the depth might be related to a later opening of the deeper fracture or fewer events of fracture reactivation. Model calculations identified a change of groundwater types at approximately 200 m depth at the site (Laaksoharju 2004). It is therefore proposed, that the samples precipitated from different formation waters.
The combination of high Mn content, relatively low Sr content and Chondrite normalized La/Yb, respectively La/Y ratios of the samples from Simpevarp (Table 3) indicates the affiliation of the fracture fillings from drill core KSH01A with generation 5a or 5b/6 calcites, as described by Drake and Tullborg 2009a. This generation has been dated to max. age of ~ 160 Ma Ivarsson et al. 2020). REE adsorption on highly reactive surfaces such as colloids (Ozaki et al. 2006;Pourret et al. 2009) and clay   (Kretzschmar and Schäfer 2005;Bouby et al. 2008;Norrfors et al. 2016) is widely observed in ground waters. The complexation of REE with organic ligands in particular enables the transport of REE with infiltrating waters and leads to an overall increase in the formation water (Torres and Choppin 1984;Xiong 2011). Natural calcites with a Chondrite normalized REE curve depleted in heavy rareearth elements (HREE) have been linked to precipitation from organic-rich waters under low-temperature conditions in contrast to flat REE profiles which trace back to complexation by inorganic HCO 3 − . Mathurin et al. 2014 applied speciation modeling to show, that the REEs in meteoric groundwaters are strongly bound to humic substances.
It is therefore proposed, that infiltrating surface waters loaded with organic complexes contributed remarkably to the formation waters of overgrowth 2 and 5a at sample KSH01A_205A/B. (see Appendix, Figure 10 and Table 4).

Laxemar
The area scan of sample KLX04_669 from borehole KLX04 reveals a series of four precipitation events. Along the fracture wall, secondary euhedral quartz of ca. 30-50 µm formed followed by three generations of calcite overgrowth. The oldest calcite (zone 1) contains especially high REE contents and relatively high contents of Pb, Th and U. It is disrupted by growth zones 2 and 3 which indicates the disruption of the fracture. The calcite growth zone 2 contains particularly high median Mn and Fe contents. The zone is crossed by numerous micro-veins (calcite growth zone 3), with local accumulations and lamination perpendicular to the growth direction. Parallel and perpendicular fractures in relation to the precipitation front have been linked to crystallization pressure (Noiriel et al. 2010). In comparison to the generation 2 calcite, the micro-veins are enriched in REE with median Chondrite normalized La/Yb ratios of 43.48. Chondrite normalized REE contents reveal an LREE enrichment within the zones 1 and the vein filling carbonate and a flat profile for zone 2. Based on the overall low REE concentration and flat Chondrite normalized profile (see Appendix, Figure 11, Table 6), it is concluded, that the calcites of zones 2 precipitated from formation water with minor share of typically organic-rich surface waters (Torres and Choppin 1984;Xiong 2011).
Samples from borehole KLX19A consist of two generations of calcite. Within the older growth zone 1, TEs are homogeneously distributed. At 414 m, TE contents are elevated along the wall rock/fracture filling contact (most prominent for Na, Th). Within this sample, interspaces between the calcite crystals of growth zone 1 are filled with secondary calcites that are enriched in U but depleted in REE, Sr and Th (see Appendix, Figure 12, Table 7).
Generation 1 calcites are overgrown by large, euhedral calcite crystals, which indicate prolonged and slow precipitation. The µXRF element mapping shows TE incorporation is irregular throughout the growth zone (Fig. 4B). Two crystal-types with clear REE fractionation were identified at 414 m. Type 1 carbonates are enriched in LREE (median Chondrite normalized La/Yb ratio: 26.33) associated with Pb and Th, whereas HREE associated with Mn and Fe are preferentially incorporated in type 2 carbonates (median Chondrite normalized La/Yb ratio: 0.66). At 428 m (Fig. 4A), cluster analysis of the LA-ICP-MS scanned area show three cluster within growth zone 2 with highly differing Fe, Sr, REE, Pb, Th and U contents (Fig. 4D).
The fracture filling generations are divided by a ca. 1-mm-thin greenish contact zone (GCZ) with strongly trace elements enriched rims (Fig. 4C). Low Ca contents indicate other mineral phases than calcite. Maximal difference between LREE and HREE is reached within the rims of the GCZ with median Chondrite normalized La/Yb ratios of 5.5 at 414 m and up to 300 at 428 m.
Deposition of clay minerals and adsorbed TEs seems to be a likely scenario to explain the extremely high TE contents. LREE and TEs such as Th and U predominantly bind on organic ligands and clay minerals. However, in present day fracture waters from bore hole KLX19A, DOC concentrations of 1.5-2.2 mg/l were measured in the corresponding depth. Analogue to present day concentrations, organic ligands have been presumably low in this depth in the past (Laaksoharju et al. 2008b(Laaksoharju et al. , 2009. The fractionation in association with organic ligands in the formation water seems therefore less probable. A more likely scenario is the former presence of biofilms. At the Laxemar and Simpevarp sites, paleo-biofilms of estimated Late Pleistocene age were described by Heim et al. 2012. They too linked the accumulations of LREE and TEs (Mg, Al and Fe) to the complexation with organic matter, but also associated them with the transport of Fe-rich colloids (Heim et al. 2012 and references therein). Authigenic clay-minerals, as they are accumulated along the rims of the GCZ, have been proposed to be part of fossilized microbial mats (Sallstedt et al. 2019).
Furthermore, cogenetic calcite and pyrite have been found related to microbial activity (Tullborg et al. 1999;Drake et al. 2015;Drake et al. 2018b). Pyrite precipitation consumes free Fe 2+ cations in formation waters and could be an explanation for low Fe contents in proximity of the Fe enriched rim of the GCZ at sample KLX19A_428. Cogenetic pyrite has been described as characteristic for generation 5/6 calcites as defined by Drake et al. 2009b. This fact supports the assumption, that calcites investigated in this work precipitated under low-temperature conditions at max. 160 Ma ago (Drake et al. , 2018Ivarsson et al. 2020).
Thus, the results show that • REE fractionation is highly variable. Chondrite normalized La/Yb ratios reach outliers of < 1 and > 100. However, preferential LREE incorporation is predominating. • The calcites investigated from Simpevarp (205-212 m depth) presumably belong to generation 5/6 fracture fillings as defined by Drake et al. (2009b). • The GCZ is a feature of fracture fillings below 400 m at the Simpevarp and Laxemar sites. The zone stands out by extremely high TE contents. The origin of those might result from a combination of coprecipitating minerals in association with biofilms, and the sorption of TEs on inorganic ligands. • Th contents often lie below the detection limit. In areas where both elements are sufficiently enriched, Th and U follow the same distribution patterns. As a result, uranium which is commonly above detection (mean

Forsmark
One central line scan and five perpendicularly arranged line scans were carried out at the sample KFM04A_233. The element maps indicate at least one major change in the hydrochemical properties of the formation water. An early generation of highly TE-loaded calcite is followed by a generation with fluctuating TE content. The spatially resolved results show, that the younger growth zone is divided into at least two subzones. Mn and Yb are of higher abundances in closer proximity of growth zone 1. At the center of this overgrowth, light rare-earth elements (LREE) are enriched. Towards the outer rim of zone 2, LREE decrease whereas other TEs such as Fe, Yb, Pb, Th and U reach higher contents (Appendix, Figure 8). However, this study emphasizes the necessity of detailed element mappings to describe the element incorporation precisely and draw differentiated conclusions. Further studies on sample materials from Forsmark are therefore foreseen.

Face-dependent partitioning
The concentric growth zone 4 of sample KSH01_205_A shows areas of distinct TE enrichment (Fig. 5A). These distinctions can be found exclusively within this sample but not at the equivalent growth zone of the same geochronologic episodes within sample KSH01_205_B. This finding indicates sector related zoning as described by Reeder and Paquette (1989) rather than a change in precipitation environment. Within growth zone 4 at least five sectors are visible (Fig. 5A). The statistical analysis confirms the heterogeneity within the growth zone. A cluster analysis of TE contents in the sample assigned the sectors a and b as well as sector B belong to the same cluster, whereas sector A has a uniform signature as growth zone 5a and sector C the same as growth zone 5b (Fig. 5B). Sector b contains enrichment of the TEs Mn, Fe and Sr, whereas the other TEs compared in this study (Na, Y, La, Ce, Yb, Pb, Th, U) are relatively enriched in sector a. Analogue to growth zone 4, sector B of growth zone 5 stands out by enrichment of divalent TEs. However, sectors A and C do not follow the same pattern. While HREE and U are enriched in sector A, Na, LREE, Pb and Th are enriched in sector C. The observations are congruent with the studies of Reeder and Paquette (1989) on synthetic and natural calcite crystals, who found Mn, Fe and Sr to be enriched in the same sector within one growth zone. In addition to these findings, our data suggest, that all other TEs investigated in this study are relatively enriched in the remaining sectors. This observation leads us to the conclusion, that differences in partitioning regarding nonequivalent crystal faces (face dependent partitioning) applies to those elements as well. The structural separation between Mn, Fe, Sr and the trivalent REE (Y, La, Ce and Yb) suggest, that Mn and Fe must be in divalent form. Studies of Elzinga et al. (2002) and Withers et al. (2003) on synthetically grown single-crystal REE-doped calcite show that HREE occur in a six-fold coordination, whereas the incorporation of LREE in the Ca site in calcite requires a sevenfold coordination. The seven-fold coordination, however, leads to disruption of local structures and charge disequilibrium. As one possible mechanism of charge compensation, Elzinga et al. (2002) andMarques Fernandes et al. (2008) refer to co-substitution of Ca 2+ with monovalent cations, e.g., Na + + REE 3+ as it is most distinct in sector C. From the similar behavior of Pb with LREE and U with HREE it can be reasoned that site-specific partitioning occurs in dependence of the ionic radius.
The face dependent partitioning observed in this study gives strong evidence, that.
• REE (Y, La, Ce, Yb) and divalent cations (Mn, Fe, Sr) do not coprecipitate in the same sector. • HREE/LREE partitioning might occur. Pb and Th are thereby associated with LREE, U with HREE.

Plausibility assessment of trace element partitioning
Fracture fluids exhibit a unique, depth-dependent compositional fingerprint of the different infiltrating and hydrothermal mixing waters (Gimeno et al. 2008;Drake et al. 2012;Gómez et al. 2014). To allow for the estimation of D TE ratios in natural systems, bulk formation water data are commonly used under the assumption, that the analyzed physicochemical parameters are identically with those of the precipitation solution of the calcites (outermost growth zones of calcites fracture fillings at Laxemar  have been dated to an age of Late Miocene by Ivarsson et al. (2020). This assumption simplifies not only the temporal fluctuations since the precipitation of the outermost calcite growth zones, but also the spatial disparities between the formation waters in single fractures. However, this approach is still a meaningful addition to laboratory and field derived experiments.
Here, partition coefficients D TE between TE concentration of the formation water and TE content of the outermost overgrowth of single calcite crystals are discussed regarding their plausibility in comparison to laboratory-derived partition coefficients (Curti 1999 and references therein) and previous field studies conducted at the Äspö HRL (Drake et al. 2018 Fig. 6). The latter field studies were carried out on recent precipitated calcites (i.e., 17 years growth period) and recent formation water in bore holes at more than 400 m depth (pH 7.4-7.7, 14 °C) (Drake et al. 2018).
At Simpevarp and Laxemar, two recent groundwater types are dominant in borehole KSH01A (Laaksoharju 2004). The transition between the freshwater type and the underlying brackish groundwater type is located at ca. 200 m. Therefore, data from the formation water of the same borehole section and freshwater type groundwater data taken between 150 and 200 m depth were used to calculate D TE ratios for the samples KSH01A _205_A/B ( Table 2, Laaksoharju 2004).
Overall, D TE ratios based on formation water data from < 200 m depth fit best with previously found ratios calculated by Drake et al. (2018) based on 17-year precipitation experiment carried out at the Äspö HRL. Partitioning coefficients of Mn and Fe are comparably higher and lower for Sr when calculated based on formation water data from > 200 m depth. D Na coefficients are in good agreement with field-derived data and at the lower end of laboratoryderived data. REE and U concentrations are only available for formation water > 200 m. The discrepancies of Y uptake between samples from the same site in Simpevarp is striking, as the ratios in 205 m depth match with laboratory-derived ratios while those from 212 m depth correspond with field observations from the area. Partition coefficients for La, Ce and U are in good agreement with laboratory experiments and high in comparison with previous field studies at the Äspö HRL.
For sample KSH01A_212, mean D TE ratios of Mn, Fe and Sr are below or at the low end of the range of laboratory driven values. The low Sr uptake is contrary to previously found ratios at the Äspö area, which even exceed laboratoryderived ratios. Partition coefficients for La and Ce are in good agreement with laboratory experiments and high in comparison with previous field studies. D Na and D U are in good agreement with field-derived coefficient range.
For sample KFM4A_233 (Forsmark), D TE ratios of Mn, Fe, Sr and U diverge significantly from laboratory and field driven ratios (Fig. 6). Only La and Ce partitioning is in the expected range.
Based on these results, the comparison with partition ratios generated under laboratory conditions should be treated with caution as they are conducted under wellknown, abiotic conditions (i.e.; p = 1atm , T = 25 • C , defined aqueous speciation, supersaturation conditions, competing cation and others) and therefore represent only a rough approximation of natural systems. Under laboratory conditions, most elements exist in free cationic state, complexation on the other hand is common in the formation waters.
Sr for example is mostly available as Sr 2+ , Fe mainly as Fe 2+ and with increasing depth as FeHS + (Laaksoharju et al. 2009 As mentioned above, it is unlikely that physiochemicalproperties remained comparable since the point of precipitation. However, even with constant trace element concentration, other parameters such as temperature T, pH value and precipitation rate R P can limit or benefit the TE incorporation. DePaolo (2011) described the influence of these parameters mathematically in Eq. 3. The equation was used in this work, to back-calculate TE/Ca ratios of precipitation solutions and compare them to recentgroundwater data. The example of Sr/Ca and Mn/Ca ratios of the precipitation Fig. 6 Partition ratio of the trace elements Na, Mn, Fe, Sr, Y, La, Ce and U between fracture water and outer most growth zone of calcite crystals from bore hole KFM04A (233 m) and KSH01A (205 m and 212 m). For comparison, partition ratios from laboratory experiments carried out at 25 °C and 1 atm (Curti 1999 and references therein) and field-driven experiments carried out at the Äspö HRL (Drake et al. 2018) were added. Missing data are marked in the diagram solution as a function of log R P under defined T and pH is given in Fig. 7. The pH values of the recent formation waters range from 7.19 in Forsmark to 7.24 in Simpevarp. We therefore used equilibrium TE/Ca partitioning coefficients of K eq = 0.035 for strontium and K eq = 60 for manganese and forward kinetic fractionation factor for TE/Ca in the precipitation reaction is K f = 0.24, R b = 7 × 10 -7 mol/m 2 /sec (values taken from DePaolo (2011) used for experimental data of Lorens (1981) obtained at T = 25 °C, pH ≈ 7.4, P = 1 atm).
Under these assumptions, the best fit between recalculated TE/Ca ratios of the precipitation solution and recent ground waters could be found for sample KSH01A_212 and waters from above 200 m under medium to fast precipitation rates (logRp ≈-5). Even though, Sr/Ca ratios areslightly lower than expected from groundwater data (Fig. 7). For the samples KSH01A_205_A/B and groundwaters above 200 m, Sr/Ca ratios fit with a logR P ≈ − 7, whereas Mn/Ca ratios would be rather high in comparison to recent groundwaters (power of ten above). However, Mn rich waters are not uncommon at the site (Tullborg et al. 1999). Especially through microbial activity, local Mn enrichments can occur. For groundwater data below 200 m, logRp < − 8 is reasonable. The results of KFM04A show, under the given T and pH values, precipitation solution and recent groundwater TE chemistry differ largely.
Overall, these results show, that the partitioning ratios as generated in this and many otherstudies should be handled with caution. The interrelation between D TE , respectively, TE/Ca and R P given in Fig. 7 shows, that even slight changes of the precipitation environment can influence significantly the trace element incorporation into the calcite crystal lattice during the secondary phase formation and retention.

Conclusions
The high-resolution maps created in the course of this work allow for the detection of zonation and microstructures within the samples. Local enrichment and variability of elementcontents measured by LA-ICP-MS are in many cases not detectable with µXRF. Thus, it becomes clear, that single point measurements or line scans reveal a limited view of the sample which could lead to misinterpretations. Especially face-dependent partitioning, as it was found at our samples, cannot be covered by point measurements.
From the change of paleo-hydrogeological conditions determined by trace elements, it can be concluded that the retention of rare-earth elements in calcite is high even with fluctuating precipitation conditions. Special attention should be given to the greenish contact zone found in samples from Laxemar at approximately 410-430 m, where clay mineral rims are incorporated into a calcite-dominated aggregate growth. The particularly high trace elements contents in this area are promising in terms of understanding the increased uptake of trace elements.
Taken that lanthanides and trivalent actinides show a similar chemical behavior, and that the observed concentrations and apparent partitioning of the lanthanides in the studied calcites are high, the retention potential of radionuclides in calcite facture fillings in crystalline rock environments for deep geological disposal of radioactive waste should be considered to be promising.