Hydrological Behaviour of Tritium on the Former Semipalatinsk Nuclear Test Site ( Kazakhstan ) Determined using Stable Isotope Measurements

Tritium and stable isotope (deuterium 2H and 18O) concentrations have been determined in natural waters collected from shallow lakes, wells, streams and rivers inside and in the vicinity of the former Semipalatinsk Nuclear Test Site (NE Kazakhstan). The Semipalatinsk Test Site (STS) was one of the main proving grounds for the testing of nuclear weapons by the former Soviet Union. Tritium activity concentrations have been determined by liquid scintillation counting, while hydrogen isotopic composition have been determined using a GV-Isoprime mass spectrometer coupled to an elemental analyzer. Tritium activity concentrations recorded in lake waters (in most cases >10 Bq L-1) were significantly higher than those in well, stream and the Irtysh River waters. In lake waters, enrichments in deuterium and 18O (δD and δ18O varying between –5 and –64 ‰ V-SMOW and –8.4 and +5.5 ‰ V-SMOW, respectively), and high salt concentrations, strongly suggest that significant evaporation has occurred. In contrast, deuterium and tritium signatures of ‘common’ surface and underground waters at the STS were mostly typical of present-day isotope backgrounds of natural waters in NE Kazakhstan. In STS, come salt lakes like Bajansor and Tumatsor with elevated tritium activity from 12 to 15 Bq L-1 lie close to the Global Meteoric Water Line. The potential tritium source for these lakes is residual concentration of tritium after former nuclear test in STS. The study provides evidence to show that export of tritium from underground nuclear test areas and tritium enrichment produced by evaporation are both important determinants of tritium concentrations in standing waters on the Semipalatinsk test site.


Introduction
The Semipalatinsk Test Site (STS), located in the north-east of Kazakhstan, was one of the main proving grounds for the testing of nuclear weapons by the former Soviet Union.Over the period 1949-89, a total of 456 nuclear tests were conducted at the STS [1].In the former STS, elevated levels of tritium ( 3 H) are often recorded in natural waters (streams, wells and lakes) close to sites where nuclear tests were conducted, especially in the Degelen Mountains (site of approximately 239 underground tests), in the Balapan tests fields (107 explosions) and close to "atomic lakes" formed by sub-surface detonations (e.g., the Balapan, Telkem 1 and Telkem 2 lakes) [2][3][4].Previous studies have shown that tritium contaminated aquifers act as source terms for underground and surface waters.In addition, some export of tritium has been observed to more distant wells and rivers [4].
The analysis of chemical or isotopic tracers, such as deuterium ( 2 H) and 18 O, can greatly benefit the interpretation of tritium migration in surface and/or underground waters.Indeed, the stable isotopes of hydrogen and oxygen within the water molecule are natural candidates for an almost ideal water tracer.In the present study, we report tritium activity concentrations, δD and δ 18 O values in natural water samples from STS.In addition, we demonstrate the use of fractionation of δD and δ 18 O to identify the contribution of different water sources (meteoric, lake, underground or river waters) and hydrological processes (mixing, recharge or evaporation) to tritium concentrations.
Analyses of tritium in waters from a northern region of the former STS were supported by measurements of hydrogen and oxygen isotopic ratios ( 2 H/ 1 H and 18 O/ 16 O, respectively) in the same waters, with a view to identifying possible tritium exchanges at the surface water/atmosphere interface.Particular attention was paid to establishing whether or not evaporation processes in salt lakes lead to a significant increase in the content of heavy hydrogen isotopes ( 2 H and 3 H) in the brines.

Environmental Context, Sampling and Analysis
The climate in the Semipalatinsk area (NE of Kazakhstan) is continental: summers are hot, dry and windy and winters are cold with snow.Especially low mean annual precipitations are recorded in the city of Semipalatinsk (located some 100 km from the former STS): 280 mm y -1 between 1977 and 2007 (see: http://climexp.knmi.nl/).As a consequence of this aridity in apparently "closed" (no surface inflow or outflow) and shallow lakes (that have formed within 1-2 km large, flat basins), the water is progressively evaporated under the combined action of heat and wind.Evidence of brine ingress from these lakes can also be found in some local streams and wells.
Lake, stream and well water sampling was carried out during field campaigns in the northern portion of the STS, close to the village of Maiysk, as part of the SEMIRAD 2 (a NATO Science for Peace project) investigation of this area.Campaigns were undertaken during the summers of 2005 and 2006 (Fig. 1).In this part of the STS and especially in Technical Area 4a, testing of radiation dispersal devices resulted in substantial contamination by longlived radionuclides -mainly 90 Sr.However, no underground or subsurface tests with thermonuclear devices were conducted in this portion of the STS and any tritium present would have either ingressed from adjacent areas or have been deposited by local and global fallout.Water samples were taken twice in two lakes within the study area (Jamantus and Baïansor Lakes) before (samples 2006-1 and 2006-4, respectively) and after rain episodes (samples 2006-6 and 2006-5, respectively) to check the effect of rain on the isotopic signature of the lake waters.
Further samples were taken from the Chagan River (southeast of the STS), a tributary of the Irtysh River that drains the Balapan underground test area (samples 2005-3 and 2006-10).This was undertaken because exportation of tritium from Balapan to the Chagan River was invoked by Mitchell et al. [4] to explain the relatively high tritium activity measured in this river (36 ± 3 Bq L -1 ).In fact, other measurements in this river closer to Balapan indicated tri-tium levels as high as 281.000Bq L -1 [5] and levels of this magnitude could only have resulted from tritium migration from the underground tests at Balapan.Similarly, streams draining the Degelen Mountains test area contain tritium levels of ~140.000Bq L -1 [4,6].Finally, sampling was undertaken in the Irtysh River: upstream (samples 2005-1 and 2005-2) and downstream of the Chagan River outfall (sample 2005-4); further downstream, at Kurchatov City (samples 2005-12 and 2006-9).In all cases, approximately one litre of water was collected at each sampling site and these were sealed as sub-samples in separate, clean, airtight bottles for tritium and deuterium analyses.Tritium activities were measured at University College Dublin, Ireland, following filtration through 0.45 μm cellulose nitrate membranes.A 5 ml aliquot of each sample was added to high quality 30 mL polyethylene counting vials pre-loaded with 10 mL aliquots of water-miscible liquid scintillant (Ultima Gold™ AB supplied by Packard) and shaken vigorously for a few minutes.Counting was carried out using an LKB-supplied, Wallac Quantulus low background liquid scintillation counter, the settings of which had been optimised for low-level tritium measurement [4].The minimum detectable 3 H concentration was ~2 Bq L -1 .
Separate 50 mL aliquots were prepared for deuterium analysis at IRSN, Cadarache, France.Samples were filtered through <0.2 μm nylon membranes and stored in polyethylene bottles free of air.Isotopic analyses were carried out at the Centre de Recherches Pétrographiques et Géochimiques (CRPG) in Nancy (France).The hydrogen isotopic composition of the water samples was recorded using a GV-Isoprime mass spectrometer coupled to an elemental analyzer.For this analysis 3 μL of each sample was injected over a Cr reactor at 1100 °C under He flux and analysed in continuous flow mode. 2 H/ 1 H and 18 O/ 16 O ratios in the water samples are expressed as δD and δ 18 O relative to Vienna Standard Mean Ocean Water (V-SMOW).The overall analytical reproducibility attained was better than ± 2 ‰.Concentrations of main cations (e.g., Na + , K + , Ca 2+ , and Mg 2+ ) and chloride were also measured at CRPG using Atomic Absorption Spectrometry (AAS), following the methodology described by Yeghicheyan et al. [7].

Tritium in Water
The tritium levels found in all water samples were low (<16 Bq L -1 ).The levels in the six well water samples, in the stream traversing the SEMIR-AD 2 area, in one Chagan River sample upstream of Balapan, and in the Irtysh River at Kurchatov, were below the detection limit of the method employed (<2 Bq L -1 ).The levels measured in one SEMIRAD 2 area stream sample, in two samples collected from the Irtysh River, in one salt lake sample, and in one sample collected from the Chagan River, were a little higher, but still <6.5 Bq L -1 .The highest tritium levels (up to 15 Bq L -1 ) within the SEMIRAD 2 area were found in lake waters.Tritium concentrations in these lakes were lower by 1 or 2 Bq L -1 after rain, but remained higher than those expected to result from global tritium fallout in the northern hemisphere.In contrast, the very low levels of tritium in the well waters sampled were comparable to those expected in surface waters as a result of global fallout, though lower than those found in other wells located in the village of Sarzhal, southeast of the STS (4.1 to 4.7 Bq L -1 ) [4].The latter were measured in 2002 and at that time somewhat higher levels of tritium were also found in the Chagan and Irtysh Rivers (36 ± 5 and 9 ± 3 Bq L -1 , respectively).

Deuterium and 18 O in Waters
The results for 2 H and 18 O in water are presented in Table 1.No sampling/measurement of meteoric water in the studied zone was undertaken.However, δ 2 H and δ 18 O measured in rainwater samples have been reported for Omsk, Novosibirsk and Barabinsk (55° latitude in Central Russia, i.e. some 700 km away from studied area).These range between -120 and -100 ‰ and -16 and -12 ‰ V-SMOW, respectively [8] and, there being no indications to the contrary, were considered likely to be indicative of the isotopic composition of rain falling on north-eastern Kazakhstan.The δ 2 H and δ 18 O signatures (mean ± SD) for water collected from the Chagan and Irtysh rivers and from wells in the north east of the STS are similar to those expected for rainfall: rivers δ 2 H -102.4 ± 4.3 ‰ and δ 18 O -12.4 ± 4.5 ‰ V-SMOW; wells δ 2 H -100.8 ± 7.2‰ and δ 18 O -13.5 ± 0.9‰ V-SMOW (Fig. 2).It follows that the river and well waters can be considered as parts of the same water reservoir with only one sample (2006-9) collected from the Irtysh River showing a substantially atypical δ 18 O value of -2.8 ‰ V-SMOW.Nevertheless, δ 2 H and δ 18 O values tended to be a little higher than those expected on the basis of rainwater from Central Russia and this may be due to a 'latitude effect' with precipitations more enriched in heavy isotopes over Central Russia because of the lower mean temperature of the air mass and resulting higher fractionation factors than in the NE of Kazakhstan [9].However, the fact that well, river, streams (samples 2005-6 and 2006-3) and some lake samples (Bajansor and Tumatsor) lie very close to the Global Meteoric Water Line (GMWL: δD = 8*δ 18 O+10 [10] confirms the assumption that those water bodies derive from meteoric water. Waters collected from lakes (samples 2005-10, 2005-13, 2005-9, 2006-1, 2006-6, 2006-8) were more variable in composition than the river and well waters, but generally were much less depleted, or even enriched, in 2 H and 18 O -the δ 2 H and δ 18 O varying between -5 and -64 ‰ V-SMOW and -8.4 and +5.5 ‰ V-SMOW, respectively.Streams on the site drain some lakes, as well as the surrounding terrain, and produced water samples with levels of 2 H and 18 O depletion intermediate between those seen in the larger rivers and the lake waters.The mean δ 2 H for stream water was -76.3 ± 18.4 ‰ V-SMOW and the mean δ 18 O was -7.9 ± 2.9 ‰ V-SMOW.The lakes and stream waters (2005-15 and 2006-2) are heavily 18 O and deuterium enriched with respect to GMWL (Fig. 2).The trend of increasing δD and δ 18 O observed in these samples defines an evaporation line.During the course of evaporation, the remaining water of a desiccating water body will gradually become enriched in the heavy isotopes [11][12].It is noteworthy that the stable isotopic signatures of lakes were distinct from those of wells in the same area.This suggests that either negligible or very slow exchange occurs between lake and shallow underground waters in the studied zone.Also, no significant changes of the isotopic signature of the Jamantus waters were observed after rain episodes (Table 1), indicating that the input of summer meteoric water has little influence on the heavy isotope signature of these evaporated water reservoirs.

Table 2
Concentrations of main cations and chloride (in mg L -1 ) in wells, lakes, streams, Chagan and Irtysh River waters sampled in July 2005 and 2006

Tracing Water Evaporation using Deuterium and Cations Concentration
Salt lake waters were characterized by lower levels of deuterium depletion (-5.0 to -63.8 ‰ V-SMOW) and high cations concentrations, whereas stream waters were characterized by deuterium depletion (-59.9 to -98.5 ‰ V-SMOW) and lower cations concentrations.The other hydrological compartments (rivers and wells) are found to be also deuterium depleted (-89.6 to 110.3 ‰ V-SMOW), and exhibit even lower cation concentrations.Thus, our data suggest a trend between stable hydrogen isotope ratios and cation content (Fig. 3), the end-members of which are: 1) The Irtysh River and the well waters, with low cations concentration and low stable hydrogen isotope ratios; 2) Salt lake waters, with high cation content and deuterium enrichment.The salt lake data indicate that evaporation processes produce not only an enhancement in Ca 2+ and Na + concentrations, but also a marked enrichment of deuterium in the remaining water.Conversely, the Irtysh River (constant δ 2 H and cation concentrations) shows no evidence of evaporation -probably due to the great size of this water body.
The enrichment in heavy isotopes in residual waters as a result of the isotopic fractionation that accompanies evaporation was recognized long ago [10] and since then it has been widely utilized to establish the water balance of lakes, as reviewed by several authors [11,[13][14].Although both isotopic data from the present study and high cation and chlorine concentrations clearly point to the importance of evaporation as the mechanism producing heavy isotope enrichment in water, the lack of hydrological data did not allow us establish the water budget of lakes (inflow and outflow) using the isotopic signatures.

The Effect of Evaporation on Tritium Water Activity
The combined chemical, isotopic and radiological data suggest that the analysed waters are characterised by at least three distinct tritium and deuterium 'signatures' depending on the water sources and/or hydrogen isotopes fractionation processes, as follows (Fig. 4). 1) Most samples exhibit low tritium activity (with tritium concentrations below the detection limit of 2 Bq L -1 plotted arbitrarily as 1 Bq L -1 ) and low deuterium content.The signatures of these water bodies, comprising mainly streams, well waters and the Irtysh River, are assumed to reflect the hydrological and present-day radiological backgrounds of surface and underground waters in NE Kazakhstan, the main source being precipitation.
2) Lake water samples with high tritium activity concentrations characterized by enrichment in the heavier hydrogen and oxygen isotopes, mainly due to evaporation.In the past, when there was a large contrast in the environmental 3 H concentration between rainfall and other components of the water cycle, tritium was mainly used as a hydrological tracer to investigate lake water balance.Thus, the effect of evaporation on tritium lake waters could not be distinguished from atmospheric input.However, although no data on the molecular impact of evaporation on lake tritium activity has been published, at least two experimental studies have been undertaken which show the fractionation of tritium caused by isotope effects during evaporation [15] and the production of tritiated water using evaporation and condensation [16].
3) Salt lakes (Bajansor and Tumatsor) with elevated tritium activity from 12 to 15 Bq L -1 , which lie close to GMWL line.One of potential tritium sources for these lakes is residual concentration of tritium after nuclear test in STS.
4) The higher tritium concentrations recorded in the Chagan River (6.5 Bq L -1 in the present study, but 36 Bq L -1 [4] and up to 281 kBq L -1 [5][6] are not accompanied by any enrichment of deuterium -indeed, deuterium appears to be depleted in these waters.For this river, the main source of tritium remains the Balapan test field area that is located upstream.

Conclusions
Our data, involving both tritium, deuterium and 18 O isotope measurements, suggest that while some of the waters in the STS (e.g., in the Chagan River) have been contaminated by tritium from local underground nuclear tests, increases in tritium concentrations observed in other water bodies (e.g., in the salt lakes) can be explained by additional contribution from other sources.Specifically, the present study confirms that evaporation of standing water can lead to additional increasing of tritium activity in some salt lake waters, and that the consequent isotopic enrichment is quantifiable using deuterium measurements.

Fig. 1 .
Fig. 1.Map of the STS with the site boundaries indicated (dash line) and the location of the various wells, lakes, streams and rivers sampled in the course of the present study.

Table 1
Tritium concentration (in Bq L -1 ) and stable isotopic ratios of oxygen and hydrogen (δ 18 O and δ 2 H, in ‰ V-SMOW) in wells, lakes, streams, Chagan and Irtysh Rivers waters sampled in July 2005 and 2006