INTERPRETATION OF THE ANOMALOUS GROUNDWATER CHEMISTRY AND ²³⁴U/²³⁸U ACTIVITY RATIO DISEQUILIBRIUM IN THE NORTHERN PART OF THE BALTIC REGION

4.1. A new reality for talik system existence by ²³⁴U/²³⁸U activity ratio analysis

As mentioned previously, one of long-standing positions suggests that isotope-geochemistry anomaly formation on EH groundwater developed in paleo permafrost conditions. It was confirmed by the non-equilibrium fractionation model for stable environmental isotopes and chemistry [17, 18]. Besides, the high uranium isotope ²³⁴U/²³⁸U activity ratio, in principle, confirms permafrost existence [19]. The third proof of this hypothesis may be due to the display of high values of helium-4 concentration in groundwater [55, 56]. Because alpha particles are ⁴He isotope nuclei, the phenomenon is related to U and Th geochemical anomalies. High helium content sites in Estonian groundwater directly connect with the radioactive decay in the Cm-V aquifer systems rocks. The α-recoil of ²³⁸U balances the uranium decay series by daughter products: ²³⁴U and ²²⁶Ra to ²³⁴Th and ²³⁴U to ²³⁰Th. The uranium decay is limited by secular equilibrium. Decay products and their parent isotopes represent a steady-state equilibrium, where the decay constants define the maximum possible intermediate daughter activity [57]. ²³⁴U occurs due to the decay series of ²³⁸U via ²³⁴Th, therefore, the activity ratio of uranium isotopes (AR = a₂₃₄U/a²³⁸U) forms a secular equilibrium in mineral and groundwater close to one. However, the studies revealed that uranium isotopes AR in groundwater also occur in a disequilibrium state because the mineral surface is in reality damaged or influenced by weathering processes of rock [52, 58–60]. Surface water mostly has AR ranging 1–2 [52, 61] and oceanic water 1.15 [52, 62, 63]. Deviations of the activity ratio from the secular equilibrium in groundwater are usually above 1 (often up to 2–3 and in exclusive cases >10). In general, there are two types of isotopic AR abundance variations: 1) it arises due to the α-recoil of the daughter ²³⁴U isotope from the parent ²³⁸U and 2) by preferential leaching from weathered rocks under oxidizing conditions in recharge sites by meteoric water. In the oxidizing environment, a hexavalent uranium form has higher solubility, and groundwater can form enhanced AR if the down-gradient flow has difficulties discharging. Therefore, such phenomena significantly depend on the changes in thermodynamic conditions on redox boundary in aquifers. At greater depths, quadrivalent uranium species dominate, ²³⁴U leaching is depleted and uranium may precipitate. Mainly ²³⁴Th isotope recoil can produce ²³⁴U atoms [2]. The ²³⁴U isotopes are more mobile during the weathering of host rocks and can be leached into the water phase more efficiently than their parent nuclide ²³⁸U. Change of oxidizing conditions to reduced during the down-gradient flow of groundwater from the redox front results in the secondary precipitation of uranium into the aquifer matrix and groundwater depletion of ²³⁸U. Thus, AR in the groundwater increases [58, 64–66].

The ‘α-recoil – weathering’ model requires an additional consideration of mixing processes (mainly on the local sites for down-flow by leakage through aquitards) and the development of the Rayleigh progressive partitioning model for the permafrost condition. It is possible that in permafrost conditions the ²³⁸U (as a heavier by atomic weight uranium isotope) could preferentially precipitate with carbonates during the reduction of U(VI) to U(IV) into an aquifer. In the partitioned residual groundwater, the accumulation of ²³⁴U isotope occurs. The experimental studies of uranium isotopic fractionation in the formation of ice crystals at the partial freezing of trapped fluid were carried out in the Federal Center for Integrated Arctic Research, Russian Academy of Sciences [67].

The studies show that the forming ice is depleted by ²³⁴U, and the residual part of water is enriched by radiogenic uranium atoms. A parent nuclide ²³⁸U and a daughter product ²³⁴U, connected by a single decay chain, in the solid phase formation behave differently, which confirms the existence of uranium in water in two forms: dissolved uranium in a form of individual compounds and uranium, which is in a mineral particle separated by the recoil nucleus ²³⁴Th from the rock at entering the water. Thus, ²³⁴U/²³⁸U AR values in forming ice and residual water differ. The volume of unfrozen subpermafrost residual water becomes enriched with the isotope ²³⁴U by saturation on the dispersed solid particles and in total uranium relative to the values of the original sample. The forming ice, in turn, is depleted in radiogenic ²³⁴U atoms and total uranium. As shown by the freezing tests for borehole groundwater AR, the initial value of 5.89±0.02 was depleted for ice to 4.29±0.08 and enhanced for residual water to 5.99±0.02 [67]. In a regional conceptual model, most of the melted groundwater could be discharged during the permafrost thawing compared to a deep groundwater stagnant flow regime. A similar fractionation mechanism in paleo permafrost groundwater developed in the Estonian Cm-V and O-Cm aquifer systems for groundwater stable oxygen and deuterium isotopes.

In principle, an analogical chemical and isotopic distillation model for primordial uranium isotopes reduction proved in sandstone-hosted U ores, and the laboratory experiments suggest that ²³⁸U is removed from the solution preferentially compared to ²³⁵U during the partial reduction of U(VI) to U(IV) into minerals such as uraninite [68].

The uranium content in groundwater may vary from 10–6 to 1 ppm [52]. Usually, a higher uranium content corresponds to lower AR values. Due to the valence of uranium species in the oxidizing environments, U content is 1 ppb, and in the reducing ones it is 0.06 ppb [52, 69, 70]. In Sweden, at Stripa site natural uranium in the crystalline basement shallow groundwater is strongly oxidized, and waters have U contents of up to 90 ppb. In a deeper zone, the U content decreases below 1 ppb because of changing redox conditions into reducing. The intermediate-deep groundwater type is Na-Ca-Cl with saturated calcium content, depleted bicarbonates and high pH. Stable δ¹⁸O and δ²H values are depleted compared to modern shallow groundwater. The uranium isotopes AR value in the deep groundwater increases up to 7–10, related to long residence times (25–30 ka) estimated by radiocarbon dating 71]. However, AR change is strongly influenced by the U distribution within fractures, the extent of the rock–water interface, and the amount of ²³⁸U in the solution. Dissolved radiogenic ⁴He in the groundwaters increases with their depth of origin and depends on the U content of the granite and its fracture porosity [71].

The groundwater in Kopli and Viimsi Peninsulas during the periglacial infiltration of paleo-meteoric water, in oxidizing conditions, may have produced AR up to 25–40. As a result, the uranium content formed in the range 0.05–0.25 ppb. Such an amount of ²³⁸U is closer to the values found in the reducing groundwater environment. It indicates that high AR values are related to a long groundwater residence time due to the ²³⁴Th-recoil release of ²³⁴U. In permafrost, a significant part of α-recoils ²³⁴U was sealed beneath underground ice and after the thawing of the permafrost layer caused an excessive ²³⁴U content in the aquifer system. After the Ice Sheet retreat, the groundwater in the studied aquifer system was predominant in the reducing conditions. Modernmeteoric water recharge was absent, except in places near the slopes of the buried valley on the northern coast of the Gulf of Finland. Due to periodic climate changes during the Pleistocene, the secondary quadrivalent uranium precipitation could occur multiple times are changed the redox barriers at glacialinterglacial boundaries. In the northwestern part of Estonia, the same oxidizing process could occur through a deep-set talik system, which continues to the south. Since the end of cryogenic processes, the formation of uranium isotope AR anomaly in the entirely reducing condition began.

The isotopic δ¹⁸O content of groundwater shows that continuous permafrost existed on the BAB territory with a local open talik net during the Late Pleistocene. The conceptual talik location scheme was reconstructed in a northern part of the BAB, which does not disturb a general opinion about the isotope-geochemistry features (Fig. 2). The talik system framework was grounded on the seabed and onshore lowlands, where significant erosion occurred. The bathymetry revealed the main lineament zones and eroded sites: Gulf of Finland, North-Central Baltic Proper, Gulf of Riga, Liivi Lowland and Alutaguse-Peipus Lake Lowland. Southwards of Peipus Lake, to Russian territory, talik is connected with the Pskov buried valley lowland. The general branches of the talik system could induce lateral flow under sub-permafrost aquifers and vertical discharge zones. The groundwater flow from the sub-permafrosted zone upwards is limited by regional Kotlin and Lontova aquitards and permafrost terrains slabs, which may locally occur in discontinuous permafrost sites.

The proposed talik system theory may explain various isotopic-geochemistry and chemical types distributed in groundwater of the O-Cm and Cm-V aquifer systems. Aquifer recharge with precipitation and surface water can occur from lakes and rivers near the talik location. Flow paths may as well connect in uplands through supra- and intra- permafrost units (Figs 2, 3).

A strongly depleted δ¹⁸O content is observed in the north Tallinn zone, where its values are down to –22.5‰ (Tables 1, 2). It is noteworthy that a depleted δ¹⁸O isotopic composition in the Tallinn intake (Kopli and Viimsi Peninsulas) sites is inherent with high uranium isotope ratios and formed by permafrost conditions [18, 19].

In the Cm-V aquifer system hydraulically connected with the underlaying basement, processes in a frozen underground state took place for a long time. The prolonged existence of permafrost in a frozen underground state probably caused the accumulation of ²³⁴U isotope in a solution like a daughter product of the ²³⁸U from the disturbed crystal lattices of minerals [19]. As a result of permafrost degradation, a quick filling of the aquifer with surface waters and the entrance of accumulated ²³⁴U from the crystal lattices of minerals into an aqueous phase occur. It is possible that in northwest Estonia, an additional superposed inflow with the groundwater having extremely high AR values leaked from an alum-shale layer (O-Cm) of the sub-permafrosted part. Obviously, the intrusion of the Last Glacial Maximum (LGM) ice sheet meltwater into aquifers cannot result in a high ²³⁴U/²³⁸U isotopic activity ratio. Some authors suggested that the Pleistocene ice sheet meltwater intruded into an aquifer can exchange reduced environment to oxidizing [74].

Äspo sites in Sweden present an excellent illustration of a similar situation [75]. Authors discuss that in reducing conditions, the quadrivalent ²³⁴U is leached in preference to ²³⁸U, resulting in the rock matrix obtaining a negative shift of uranium activities ratios [75]. This shift is the precondition to the high solubility of ²³⁴U, situated in mineral crystal lattices (produced by uranium and ²³⁴Th from fracture surfaces into solution), resulting in groundwater with elevated AR. For the Laurentide Ice Sheet retreat, the ice meltwater inflow rate promotes the (AR) results in a secular equilibrium state near 1 [76].

In the eastern part of the Gulf of Finland within the Baltic-Ladoga lowland segment (area of the Luga Bay, the Koporye Bay and Saint Petersburg), the substrate is mainly composed of the Upper Vendian–Lower Cambrian impermeable clays, siltstones and sandstones. Valleys are V-shaped, widths of 800–3000 m, depths down to 90–110 m b.s.l. cut in the pre-Quaternary bedrock 50–80 m. Near the Koporye Bay coast site for Sosnovyi Bor (Leningrad District, Russia) wellfield, the ²³⁴U/²³⁸U ratio of Lontova clays is 0.85, in the weathered core of basement rocks it is 0.75. The uranium isotope ratio of the Cm-V aquifer paleogroundwater varies from 0.48 to 0.65 (Table 3) [54]. The modern meteoric recharge increases uranium ratio up to 1.1 due to oxidation conditions (buried valleys and outcrops sites). Most likely, in the vicinity of Koporye Bay (Cm-V outcrop), near the former periglacial talik, an active infiltration recharge existed. Uranium-234 was intensively leached during the Pleistocene time and removed from the aquifer system by discharge flow to the surface water body. Leaching occurred from low U-Th content host rocks by recharge–discharge flow interaction either sub-, supra-permafrost aquifer or surface waters. Koporye Bay talik could have been connected as a branch to the main talik of the Gulf of Finland (Fig. 2). On the contrary, the Cm-V aquifer system is isolated only by the Kotlin aquitard in northern Estonia because the Lontova Blue clay offshore is absent. So, we may conclude that low ²³⁴U/²³⁸U ratios in the Sosnovyi Bor are caused by long-term active dissolving of uranium from the host rocks where the uranium content is low.

Table 3. Isotopes and chemistry of the Tallinn and Sosnovyi Bor intake for the Cm-V aquifer system. Modified after [18, 19, 49, 50, 54].

Note: borehole 1A is located on the Aegna Island (Tallinn Bay offsore); Sosnovyi Bor site is located on the eastern coast of Koporye Bay (Leningrad District); ٭ uranium (²³⁴U–²³⁸U) isotopic dating of age; ٭٭ the average value of δ¹⁸O and δ²H stable isotopes.

A very high ²³⁴U/²³⁸U ratio (<20) is fixed in the groundwater near the Arkhangelskaya kimberlite pipe space at the M.V. Lomonosov deposit on the White Sea coast. The Vendian rocks are characterized by elevated or anomalous U, Th and K contents concerning the background values. This halo of non-equilibrium uranium is localized in the nearpipe space and revealed in Vendian aquifer system rocks consisting of siltstone and sandstone and rocks of the epiclastic unit in the pipe at the contact with host rocks [77]. The authors explained the deviation of the isotopic equilibrium of uranium by the dynamics and long-term circulation of underground water along the tectonic fractures bounding the pipes.

The distribution of radioactive isotopes in the Vendian aquifer system is well studied in the Northern Dvina River delta (NDB) at the Arkhangelsk–Severodvinsk–Novodvinsk site [49]. The study area is presented by non-continuous permafrost, which is confined by peat soils up to 15 m. The average annual temperature in this territory is –0.5 – –1.0°C. Here the Vendian aquifer system and intermediate deep old brackish groundwater of Vendian Mezen aquifer inversely discharge to the above layered fresh Vendian Padun aquifer, which is recharging by modern meteoric precipitation. The maximum (AR) values of Vendian aquifers groundwater in deeper, reducing zone are 4.8–7.2 and the adjusted ¹⁴C age range is 17–33 ka. Brackish groundwater is more depleted with δ¹³C and δ¹⁸O contents –15.5 to –16.6‰ and –14.4‰, respectively, TDS values between 4.3 to 9 g/L. In the oxidizing recharge zone freshwater, δ¹³C and δ¹⁸O contents are –11.7‰ and –13.8‰, respectively. The fresh groundwater residence time is estimated from 0.3 to 16.4 ka. The medium AR (1.3–5.9, average 3.0) is characteristic of this water. Near the redox barrier, old saltwater has ages ranging from 17 to 33 ka with TDS from 4 to 13 g/L and high AR (4.8–7.2, average 5.9). Below the redox barrier, in reduced lenses at the buried paleo-valley bottom, the elevated concentrations of uranium are preserved. The precipitated U(IV) uranium concentration in rocks reaches 20 ppm, and AR in rocks decreases to 0.5–0.9. But, complete precipitation does not occur because recoil continues, and both uranium isotopes enter the water. In these areas, the most dangerous aspect is the flow of groundwater from the underlying horizons, since during the operation of water supply wells it can lead to the creation of local zones of oxidizing conditions in the perforated boreholes screen zone and the transition of uranium into a solution [49].

Cyclical glaciations and marine transgressions strongly influenced the geochemical evolution of groundwater systems of the Arkhangelsk region. In the lower part of the Vendian Mezen aquifer, brines are found at >1 km depth. Near the modern sea coast, shallow groundwater is enriched by heavy stable isotopes, average values δ¹⁸O = –7.7‰ and δ²H = –98.9‰ [50].

We made the correction of the ¹⁴C ages of the groundwater at the Tallinn intake using Eq. (1), based on the background value of δ¹³C = –15.0‰ PDB used in [78], where δ¹³C and ¹⁴C are in per mil and pmC in the groundwater sample, respectively:

At the Tallinn intake, the Cm-V aquifer system groundwater ²³⁴U/²³⁸U AR varies from 3.6 up to 26 at the 50–140 m depth range (Table 3). The increase of uranium AR slightly corresponds to TDS elevation from 0.35 up to 0.95 g/L. In Holocene, strongly reducing conditions prevail in the Cm-V aquifer, with oxidation–reduction potential Eh varying from –150 to –250 mV. Redox environment transition to oxidizing mainly appears near buried valleys and Eh values vary from –50 to +100 mV [40].

Enhanced accumulation of a more soluble uranium-234 isotope and high ²³⁴U/²³⁸U AR formation could begin in the periglacial period before LGM. The drop in the sea level opened a highly elevated cliff in the Estonian western and northern parts. Steep (150 m amplitude elevation) landscape could be formed toward the eroded seabed of the North-Central Baltic Proper and the Gulf of Finland valleys. Connected lake and river systems established a new recharge/discharge boundary for periglacial groundwater flow. A redox front separated above (near-surface) layered oxidized zone in the northwest to north-east part from the deeper aquifer part with reducing conditions in the south-eastern direction. In deeper parts (300–500 m), oxidation environment might be possible only by mixing supra- and sub-permafrost aquifers groundwater in open Liivi Bay (Gulf of Riga) and Alutaguse-Peipus talik systems. It probably caused a prolonged (10–20 ka) enhanced accumulation of the ²³⁴U atoms in the sub-permafrosted aquifer due to α-recoil from parent ²³⁸U aquifer rocks in leaching or damaged crystal lattices of U-Th rich minerals. Later on, groundwater with high (AR) values was moved toward the discharge sites with a reduced decay regime. A natural process on the periglacial reduced Eh conditions occurred for a long time leading to the dis-equilibrium of ²³⁴U/²³⁸U (AR ≫ 1) in the groundwater of the O-Cm and Cm-V aquifers (Figs 4, 5).

For example, the maximal AR values 40–56 are observed on the Saaremaa Island and Viimsi Peninsula [19]. The latest publications indicate that in southern Finland, the Middle Weichselian glaciation and the interstadial did not occur [79]. The Late Weichselian glaciation Marine isotope stages (MIS2, 29–11 ka ago) were preceded by a nearly 90 ka long, poorly known non-glacial period, featuring tundra climate with permafrost. A new Middle Weichselian paleoenvironmental scenario had much longer periglacial conditions before LGM in Finland, indicating that permafrost could have easily penetrated >450 metres depth b.s.l. [79]. Therefore, the single LGM glaciation evidence during the Weichselian time in the EH is substantial [80].

Helium ages reveal that south Estonia groundwater may originate in the Middle Weichselian periglacial time. The evaluation of the southern part of Alutaguse-Peipus Talik, Varska site, for O-Cm (depth 463 m) and Cm-V (545 m) aquifer system groundwater radiogenic ⁴He ages are 81 and 86 ka (MIS5 a-b substages), respectively. Groundwater at the Häädemeeste site close to the Liivi Bay Northern talik (depth 610 m) ⁴He age is 150 ka corresponding to the Middle Pleistocene Saalian Glaciation by MIS6 [81]. In eastern Estonia, the middle part of Alutaguse-Peipus Talik (Alatskivi borehole) ¹⁴C adjusted age is younger (26.8 ka) compared to the Tapa well on the Pandivere Highland northern (33.1 ka) [18]. In the depth of 280 m Alatskivi groundwater, TDS is 1.5 g/L, while Tapa is 0.5 g/L in 242 m depth. Such difference could be caused by talik conditions, which promote the inversion. These ages support prolonged periglacial permafrost conditions on the EH (Last Glacial and Saalian times).

A good relationship between the ²³⁴U and ⁴He contents was explicitly demonstrated for Canada’s periglacial conditions [76]. Freshwater is devoid of radiogenic ⁴He, shows AR close to equilibrium (1), saline older groundwater contains higher ⁴He content and increased AR up to 6 [76]. The authors suggested that opening new fractures in periglacial sites provided additional surfaces for host rocks, increasing ²³⁴U and ⁴He migration and escaping the water [76]. A similar process may occur on EH – high AR and ⁴He contents anomalies are associated on the West Estonian Lowland, North Estonia coastal plain and Alutaguse-Peipus Lake Lowland. These areas are frames to the biggest taliks. The reality of open talik systems, which functioned as a redox barrier for radioactive elements transformation during the LGM, is evident. In a periglacial environment, the sub-permafrost O-Cm and Cm-V aquifer systems, and basement fractures zone were connected to a single hydraulically joint unit, where the groundwater was altered by series of isotope-geochemistry fractionation into a rare hydrogeologic body. The relationship of the ²³⁴U and ⁴He excess in the aquifer explains the existing redox boundary from the external sources of talik system between supra- and sub-permafrost units.

A reverse direction AR trendline appears at the sites with fully or partially cut aquitards near the coast (paleo valleys). Intense depletion of AR occurs due to meteoric precipitation at the onshore parts and the saline water intrusion from the seaside. These processes are confirmed by δ¹⁸O and TDS in groundwater (Fig. 5 and Table 3). The TDS versus ²³⁴U/²³⁸U activity ratios analysis shows that a high uranium isotope activity ratio belongs to the groundwater of the Ca-Na-Cl-HCO₃ and Na-Ca-Cl-HCO₃ types. Between AR values 10–26, the groundwater salinity increases up to 0.43–0.84 and more g/L. The Na/Cl ions ratio for these chemistry types is very low (0.45–0.6) and shows a high groundwater metamorphism degree which evidences long residence times.

The influence of meteoric groundwater on groundwater near buried valleys using a stable isotope δ¹⁸O ratio was proven in multiple studies [17, 19, 29, 32, 33, 82]. Modern groundwater tends to enrich the oxygen ratio isotopic values towards the average annual precipitation (–11.4 per mil). Pleistocene time groundwater is usually depleted (down to –23 per mil). The relation between uranium AR and oxygen-18 shows that modern groundwater is closer to the AR equilibrium, while Pleistocene age groundwater AR is significantly higher. In cases of modern and paleogroundwater mixing, the depletion of uranium AR values occurs. It could be described using end-members of this study from Viimsi and Kopli Peninsula. The borehole No. 26 AR 2.7 and δ¹⁸O 12‰ represent groundwater with a higher portion of modern-meteoric water. In borehole No. 20, AR = 26 and δ¹⁸O 21.3 ‰ indicate paleogroundwater. In places where strongly by stable isotopes depleted groundwater is observed, almost none of the meteoric water influence occur.

In the sites where the groundwater from the reducing zone is mixing from perched modern waters via buried valleys, the AR values are intensively lowered (Fig. 6). Evolution of the initial periglacial groundwater on northern Estonia developed by influence of LGM and Baltic Ice Lake meltwater, modern meteoric recharge and Baltic Sea intrusion. That is proved according to the depletion of AR values and also by ¹⁴C data.

LGM groundwater background parameters of the Tallinn intake are δ¹⁸O –22.5‰, δ¹³C –16‰ and TDS 0.35–04 g/L. The highest observed AR (26) in boreholes are distant from paleo valleys. Groundwater extraction caused background values to change, toward freshening, a drop of AR (down to 3.0), enrichment of δ¹⁸O‰ and δ¹³C‰ content to –12.0‰ and –12.4‰, respectively. Increased ¹⁴C content is observed in these samples (up to 54.7 pmC) (Fig. 6). The sea intrusion results in elevated TDS up to 0.95 g/L groundwater, δ¹⁸O may increase up to –19.4‰, δ¹³C to –15‰ and ¹⁴C to 16.8–33.6 pmC. Sea intrusion influence is observed in boreholes No. 14 and 22 at Kopli Peninsula and No. 27 at Viimsi Peninsula (Fig. 6, Table 3). Change in the chemical composition of this groundwater indicates a significant increase in TDS and SO₄/Cl ratio during 1977–1988 [17].

4.2. The cryogenic groundwater identification based on stable isotope-geochemistry

In a periglacial environment, groundwater flow and isotope-geochemistry zonation was strongly affected by cryogenesis in the O-Cm, Cm-V aquifers and basement fractures. Therefore, a hydraulically connected Lower Ordovician-Cambrian-Vendian-Basement groundwater (O-Cm-V-B) should be analyzed as a single unit. Generalized data provide the evidence that permafrost may strongly influence the groundwater composition (Table 1). Most samples show elevated pH values up to 8.6 (average 7.6–8.1). An increase in calcium, magnesium and bicarbonate content makes the water oversaturated with carbonate minerals (calcite and dolomite), resulting in their precipitation. The decrease of these ions follows the precipitation: north Estonia alkalinity content is down to 42 mg/L averaging between 154 and 190 mg/L (Table 1), similar to permafrost terrains in Yakutia, Russia [9, 10, 12]. The chloride content in the Cm-V aquifer groundwater average values is 209–295 mg/L, which may result from brine rejection in the relic residual water fraction. As mentioned before, seawater intrusion may also be the case during the interglacial time maritime conditions and present when seawater was mixed with in situ O-Cm-V-B unit groundwater.

Two deviations from the GMVL are observed on the Craig diagram (Fig. 7) permafrost and snow meltwater. The northern part of Estonia Cm-V groundwater chemical facies varies mainly between Ca-Na-Cl, Ca-Na-Cl-HCO₃, and Na-Ca-HCO₃-Cl. Facies differences could be explained by hydrochemical stratification, which occurs during the cryogenic concentration. Assuming that the permafrost formation is a gradual and slow process, brine rejection and mineral precipitation should result in various groundwater types [9, 10, 17]. Groundwater in the upper part of the aquifer is fresh, Na-HCO₃-Cl type; deeper, the residual portion is mainly sodium chloride facies. Cryogenic groundwater from the Tallinn intake is Ca-Mg-Cl, Ca-Na-Cl, and Ca-Cl type. Exclusive Ca-Na-Cl type groundwater is found in the deepest BAB aquifers (>1.5 km depth) under stagnant conditions (Lithuania, Kaliningrad and Poland); therefore, it is highly unlikely that it could be the source of dissolved solids for the Cm-V aquifer system in northern Estonia. The northward migration of these deep brines is stopped by the uplifted horst barrier as the low permeability lineament zone of the Liepaja–Riga–Pskov. The northwestern part unit is subdivided into two bodies, the Cm-V and O-Cm aquifer systems, where Na-Ca-Cl-HCO₃ and Na-HCO₃-Cl facies are observed, respectively (Fig. 8). Different chemical types may evolve due to the cryogenic hydrochemical stratification mentioned before. In the (O-Cm-V-B) body, the overlaying O-Cm water was likely exposed to more cryogenesis cycles, resulting in multiple carbonate precipitation occasions and depletion of calcium and bicarbonate. The lower part (Cm-V) is also more depleted δ¹⁸O (down to –22.5‰), compared to O-Cm (~–19‰), but more concentrated in TDS. The Rayleigh distillation trendline could easily explain this distribution towards depletion in the cases of residual water fraction F > 0.5 [2]. The same isotopic depletion scenario is applicable in other northern Estonia (O-Cm-V-B) groundwater as well. Northward from 1 g/L TDS boundary in the central part of Estonia, chemical and isotopic inversion occurs: δ¹⁸O is depleted by 3‰, while chloride content rises to 3 and more g/L. Southwards from the freshwater boundary (1 g/L), δ¹⁸O values are enriched by 6‰, Cl_– content increases to 10–12 g/L (Fig. 1).

The evolution of the cryogenic groundwater could be well established on Na/Cl and Br/Cl ratios diagram (Fig. 9) [15, 83, 84]. Groundwater of the O-Cm aquifer in the Br/Cl range 0.002–0.004 is a Na-HCO₃-Cl type, and the samples with the ratio >0.004 are Na-HCO₃. This groundwater δ¹⁸O content varies between –19.0 and –18.5‰. Chemical type transformation is due to carbonate precipitation, calcium, magnesium and partly bicarbonate being removed from the solute. Ca-dominated initial water (before freezing) could have resulted in Na-HCO₃ diverse water facies and manifest in the O-Cm freezing trend. The evolution of the O-Cm groundwater composition coincides with the data from Scandinavian Shield, Aspo and Palmottu sites [83].

The Cm-V aquifer groundwater seems to be altered by increasing Br/Cl and a substantial decrease of Na/Cl ratios, following seawater freezing and evaporation trends. Groundwater with Br/Cl values over 0.004 is Ca-Na-Cl type δ¹⁸O values varying from –21.5 to –20.7‰. The evaporation process, which occurred in the deep BAB groundwater formation [37], is unlikely to be the case in the Cm-V aquifer groundwater alteration because it evolved during the Pleistocene when the climate was cold, causing periglacial permafrost formation. The evaporation should have enriched δ¹⁸O values, while data shows the opposite. The predicted freezing trend of the Cm-V groundwater could result from the wasted bicarbonate content in calcite precipitation. The initial water, in this case, should have been less Ca-dominated and more saline, compared to the O-Cm. The rejection of dissolved solids increases calcium, sodium and chloride content in residual fluid, but due to lack of bicarbonate, further mineral precipitation does not occur; the solute concentration may develop without disruptions. Such a process could lead to the formation of Cl-HCO₃-Ca-Na chemical type groundwater and explain the palaeo-freezing trend of the Cm-V, O-Cm aquifer systems and in the crystalline basement water through permafrost (Fig. 9).