As it was mentioned in the introduction, occurrence of vivianite in Hungary is common in terrestrial sediments characterized by reductive environments of high iron(II) and orthophosphate and low S2− concentration of pore water, where pH and Eh conditions are also sufficient (Szakáll et al. 2005). The point is whether the geochemical features in the study area are adequate for the formation of vivianite.
Chemical and physico-chemical conditions of groundwater
In the area of the chemical plant, the “basal clay” forming the floor of the first aquifer was exposed by several drills; however, vivianite has not been found in it. The chemical composition of the pore water in the “basal clay” has not been known; however, chemical composition and physical–chemical features of the first and second aquifers are well studied. Chemical constituent and physico-chemical parameter data listed in Table 3 suggest extremely high orthophosphate concentration in the groundwater within a distance of 150 m NE of the well cluster No. ICE-113 in the first aquifer (2–6 m beneath the surface) with a concentration range between 10 and 40 mg/dm3 and with a maximum value of 80 mg/dm3, while the average orthophosphate concentration for the whole area of the plan is as low as 0.5 mg/dm3 (Table 3). Because of the limitation of available data (n < 10), it is difficult to determine obvious trends by time series analysis (e.g., Mann–Kendall test); it can be stated, however, that in the studied time span orthophosphate concentration in the first aquifer increased in the area of the sewage pond, while no trend could be experienced in the monitoring well cluster No. ICE-113. Moreover, in the vicinity of the sewage pond increased orthophosphate concentration (5–20 mg/dm3) could be detected even in the second aquifer, which possibly shows an increasing trend. Simultaneously to the orthophosphate concentration, sulfate concentration also increased in the first and the second aquifers at the sewage pond (150–400 and 100–270 mg/dm3, respectively); time series of the concentrations suggest increasing–possibly increasing trend, excepting the well cluster No. ICE-113, where a decreasing trend is noticed. This decrease in sulfate can be in relationship with diluting processes at plume margin, and it can be a consequence of increasing sulfate reduction following the iron reduction in the Terminal Electron Accepting Process sequence; as a consequence, HS− is formed from the dissolved orthophosphate (Miao et al., 2012). The sulfate reaction can take place under circumneutral pH conditions with considerable H+ consumption; or due to microbiological activity by requisition of organic carbon or H2, which finally results in pH decrease (Nriagu 1972; Postma 1981; Heiberg et al. 2012; Egger et al. 2015; Rothe et al. 2014, 2016).
Accordingly, values of Eh and dissolved oxygen have decreased in the sewage pond area (particularly, in the first aquifer), while contents of dissolved nitrate and total iron have multiplied. In the area of the former sewage pond ORP values measured in the wells installed into the first aquifer after purging have varied between −365.8 (calculated Eh: −151 mV) and −65.5 mV (calculated Eh: 149 mV) (with an average of −205 mV ORP and of 8 mV Eh), for the last 5 years (2014 and 2018); while in a monitoring well near the site of well cluster No. ICE-113 screened in the first aquifer the ORP values have ranged from −82.5 (calculated Eh: 133 ± 5 mV) to 142.6 mV (calculated Eh: 359 ± 5 mV) (with an average of 22 mV ORP and of 238 mV Eh). The average Eh distribution map calculated from the ORP results of the first aquifer clearly shows that redox potential on the site of the former pond is definitely low even nowadays, which indicates reductive conditions, as it was mentioned previously. This reductive condition tends to change just in the environment of the well cluster No. ICE-113 as it is indicated by ORP, as well as Eh values ranging from 0 to 50 mV and from 200 to 300 mV. The ORP values vary from −50 to −150 mV (Eh of 80–190 mV) in the second aquifer and are also lower than the background in the surroundings of the sewage pond.
A USGD Excel sheet (Jurgens et al. 2009) was used for identifying redox processes in the groundwater. Unfortunately, however, the redox state of iron and manganese was not analyzed, and dissolved sulfur was analyzed as sulfate, therefore, certain forms and the total amount of sulfides is not known. It is obvious, however, that under the sewage pond there are mixed (anoxic) redox conditions in the first aquifer, while mixed (oxic-anoxic) conditions are characteristic for the well cluster No. ICE-113 and the background; then again, everywhere in the second aquifer mixed (anoxic) redox conditions can be detected. Therefore, groundwater data clearly show that in the area of and some meters surrounding the pond physical and chemical conditions dramatically changed compared to the wider environment, and it may have an effect on geochemical and biochemical processes in the area. Distribution maps (Fig. 9) show that the concentration of dissolved oxygen is depleted not only at the pond but at the northern part of the study area, too. Possibly, this depletion has not been the result of the decomposition of the deposed organic material. It is supported by the fact that, parallel to a decrease in DO, an increase in dissolved total iron can be detected in the pond area, which suggests increasing microbial Fe reduction. This phenomenon can also be observed in the second aquifer; although it is not so well-marked there.
Groundwater in the first aquifer is neutral (pH 7.0–7.4); however, slightly alkaline (7.5–7.8) in the site of the former pond and at the well cluster No. ICE-113. In the second aquifer, pH is 7.2–7.3 in the site of the former pond, and it is 7.3 at the well cluster No. ICE-113.
The pH values do not show a noticeable anomaly, which suggests that the matter released from the pond did not induce an essential modification in the acid–base system. It is important since the formation of vivianite is highly pH-regulated (Liu et al. 2018). Accordingly, XRPD detected vivianite with higher degree of crystallization, which is characteristic for vivianite formed at about pH 7 (Liu et al. 2018). However, pH values of water in the vicinity of the well cluster No. ICE-113 and the pond are slightly increased, which suggests intensification of sulfate reduction. Nitrate concentrations are elevated in both aquifers, but regarding the first aquifer dissolved nitrate input comes from other sources, as well.
The concentration of dissolved HCO3− is particularly important since it fundamentally influences vivianite formation. As Table 3 shows at the well cluster No. ICE-113 HCO3− concentrations of the first and second aquifer are equal to background values; however, they are multiplied under the sewage pond.
In our opinion, the high orthophosphate content of the groundwater is the result of the decomposition of the biomass deposed in the late 1950s and the early 1960s, and this assumption is supported by the map of the dissolved orthophosphate distribution in the groundwater (Fig. 9). In this period, the amount of the processed and then deposited plant litter might be several thousand tons per year. The depth of the sewage pond is unknown; however, the increased (5–20 mg/dm3) orthophosphate concentration in the second aquifer suggests that there could be a direct communication between the aquifers.
Transportation of orthophosphate
The next point is whether the dissolved orthophosphate could filtrate from the pond to the site if the well cluster No. ICE-113 is under natural hydrogeological conditions characteristic for the study area. The upper part of the vivianite-bearing “basal clay” (which can be found from 4.0 to 9.0 m below the surface) can be regarded as clayey fine- and medium-grained silt, while its lower part is silt. The K filtration coefficient of the “basal clay” was calculated using the grain-size distribution curve; this method is widely accepted in hydrogeological-environmental geological practice; however, it should be carefully applied. Several mathematical formulae have been published for estimating filtration coefficient using grain-size distribution curves; however, these can be applied for sediments possessing a given distribution and sorting. We used Devlin’s Excel with a VBA code file (Devlin, 2015). According to the calculations, characteristic filtration coefficient values are 8.2 × 10–9–1.4 × 10–8 m/s and 1.0 × 10–8–1.8 × 10–8 m/s for the more clayey upper, and for the sandy and vivianite-bearing lower part of the “basal clay”, respectively. Therefore, the “basal clay” can be regarded as a leaky confining bed.
Unfortunately, site-specific data for the specific yield of “basal clay” is not known; however, according to Singhal and Gupta (2010) effective porosity (or specific yield which can be regarded as the same) for silt may range from 5 to 20%. Due to high uncertainties, a conservative approach (advection) was applied to predict the fate and transport processes of the dissolved orthophosphate. Considering the calculated horizontal hydraulic gradient of the first aquifer (0.004), the above-mentioned filtration coefficient, and the 5–20% effective porosity values characteristic for silts, we can use the formula
$$ v\, = \,\frac{{K{\text{d}}h}}{{n{\text{d}}l}}, $$
(1)
where v is the seepage velocity of the groundwater in the “basal clay” as a leaky confined bed; K is the hydraulic conductivity; n is the specific storage; and dh/dl is the hydraulic gradient. According to the calculation, during 50 years propagation of the groundwater could range from 3 to 22 m in the lower, silty part of the “basal clay”; considering that the well cluster No. ICE-113 traversing the vivianite-bearing layer is located 15 m of the margin of the former pond, it can be stated the dissolved orthophosphate could be advectively transported by groundwater from the former pond to the site of the drilling.
Origin of iron
Formation of vivianite requires proper quantity of iron in the pore water. In the first and the second aquifer, the concentration of the total dissolved iron is about 1 mg/dm3 in the background, while definitely increased beneath the sewage pond in the first aquifer (1.5–3.5 mg/dm3) and especially in the second aquifer (4–11 mg/dm3). Temporal change in dissolved iron in the underground water suggests intensification of iron reduction, which is not a simple facilitating but indispensable factor of vivianite formation. Considerable amount of reduced sulfur in the system may inhibit this process because it favors the formation of iron sulfides. Precipitation of vivianite is possible under a relatively narrow range of pH-Eh conditions (5–9 pH and from + 200 to − 200 mV Eh) (Lemos et al. 2007; Dill and Techmer 2009); moreover, low S:Fe ratio also favors this process (Rothe et al. 2016). Depending HCO3− activity (logaHCO3 ≥3) siderite may also form at the expense of vivianite, quasi competing with it for the dissolved Fe2+ (Dill and Techmer 2009); however, XRPD analysis did not detect siderite in our samples. This experience can be explained by low kinetics of formation and growth of siderite (which is also characteristic for vivianite) (Postma 1981), and particularly by the fact that HCO3− concentration in the well cluster No. ICE-113 is almost equal to the background concentration. A considerable increase in HCO3− activity may favor formation of siderite at the expense of that of vivianite (Dill and Techmer 2009). In the surroundings of the well cluster No. ICE-113 the concentrations of HCO3− are favorable for the formation of vivianite, while toward the pond, because of multiplied HCO3− concentrations, formation of siderite can be expected.
As a source of reduced Fe, chlorite and micas should also be considered. On the altered crystal edges and surfaces exposing the octahedral sites occupied by Fe2+, the action of dissolved phosphate in basic pH may induce dissolution of cations and subsequent precipitation through crystallization of a stable product. In such a situation, the action of orthophosphate in basic solution is initiating the reaction on chlorite crystallites, supported by the BSE observations, that vivianite is mostly formed on the surface of chlorite rich aggregates.
Another geological factor may also contribute to vivianite formation. It is highly possible that the water level in the pond was considerably higher than the static water level in the first aquifer; therefore, mobile inorganic components, such as orthophosphate, radially spread from the pond. As Fig. 10 shows the upper surface of the "basal clay" can be dominantly found at 90.0–90.5 m a.s.l. and slopes to the south, however, here and there it is 1.0–1.5 m higher. This kind of elevation of the "basal clay" occurs at the well cluster ICE-113. East of the pond, however, the first aquifer is much thicker, and physical, chemical, and lithological conditions characteristic for this bed do not allow the formation of vivianite.