High-temperature thermomagnetic properties of vivianite nodules , Lake El ’ gygytgyn , Northeast Russia

Vivianite, a hydrated iron phosphate, is abundant in sediments of Lake El’gygytgyn, located in the Anadyr Mountains of central Chukotka, northeastern Russia (6730 N, 17205 E). Magnetic measurements, including mass-specific low-field AC magnetic susceptibility, fielddependent magnetic susceptibility, hysteresis parameters, temperature dependence of the induced magnetization, as well as susceptibility in different heating media, provide ample information on vivianite nodules. Electron microprobe analyses, electron microscopy and energy dispersive spectroscopy were used to identify diagnostic minerals. Vivianite nodules are abundant in both sediments of cold (anoxic) and warm (oxic) stages. Magnetic susceptibility of the nodules varies from 0.78× 10−6 m3 kg−1 to 1.72× 10−6 m3 kg−1 (average=1.05× 10−6 m3 kg−1) and is higher than the susceptibility of sediments from the cold intervals. Magnetic properties of vivianite are due to the respective product of oxidation as well as sediment and mineral inclusions. Three types of curves for high-temperature dependent susceptibility of vivianite indicate different degrees of oxidation and inclusions in the nodules. Vivianite acts as a reductant and reduces hematite to magnetite and masks the goethite–hematite transition during heating. Heating vivianite and sulfur mixtures stimulates the formation of monoclinic pyrrhotite. An additive of arsenic inhibits the formation of magnetite prior to its Curie temperature. Heating selective vivianite and pyrite mixtures leads to formation of several different minerals – magnetite, monoclinic pyrrhotite, and hexagonal pyrrhotite, and makes it difficult to interpret the thermomagnetic curves.


Introduction
Vivianite, Fe 3 (PO 4 ) 2 8H 2 O, is a hydrated iron phosphate that has long been identified in a variety of natural environments.This authigenic mineral forms when anoxic environments provide readily available ferrous iron -usually dissolution products of iron oxides, desorption of iron-bearing silicates, and inorganic phosphorus.It has been found as a secondary mineral in a wide range of situations including weathering products of hydrothermal deposits, Fe-bearing ore veins, reducing soils and aquifers, and lacustrine and marine reduced sediments.The presence of vivianite in lake sediments is well documented, and it has been the subject of detailed studies in many lacustrine environments from large lakes such as Lake Baikal (Fagel et al., 2005;Sapota et al., 2006), to small mesotrophic lakes in the Canadian prairie (Manning et al., 1991(Manning et al., , 1999) ) to Holocene muds in Norway (Rosenquist, 1970) to Lago Maggiore, Italy (Nembrin et al., 1983).The formation of authigenic vivianite in lake sediments is not fully understood, but appears to be influenced by redox conditions, pH values, dissolved elements -primarily Fe and P, but often including impurities of Ca, Mn, and Mg -organic matter content, and sedimentation rate (Sapota et al., 2006).Vivianite is usually a stable mineral in environments over the pH range of 6 to 9 and low Eh values of less than 0.0 (Nriagu and Dell, 1974).Vivianite is found as small disaggregate crystals to large nodules, often several centimeters in diameter.In air vivianite readily oxidizes, turning from opaque or white to vivid blue in a short time, eventually altering to metavivianite (kerchenite) (Rodgers, 1986) or santabarbaraite (Pratesi et al., 2003), as the original Fe +2 becomes completely replaced by Fe +3 .The magnetic properties of vivianite are well known (Meijer et al., 1967), and although it is paramagnetic in natural environments and surface temperature, it becomes antiferromagnetic at exceedingly low temperature with a Néel temperature of ∼ 12 K (Meijer et al., 1967;Frederichs et al., 2003).Maximum magnetic susceptibility (MS) of vivianite (6.62 × 10 −6 m 3 kg −1 ) is observed at the Néel temperature (Frederichs et al., 2003).In this paper we investigate the magnetic properties, especially those at high temperature (up to 700 • C), of vivianite nodules found in the sediments of Lake El'gygytgyn in northeastern Siberia.Magnetic behavior of vivianite particles and nodules is important to the overall interpretation of magnetic properties of lacustrine sediments that are used as environmental proxies.At room temperature the susceptibility of El'gygytgyn Lake nodules is close to 1 × 10 −6 m 3 kg −1 ; this is higher than MS of cold stage sediment, but lower than MS of warm stage sediment.Sediment of high and low magnetic parameters shows different curves of temperature dependence magnetic susceptibility.Cooling curves plot above heating curves for samples from cold climate intervals, indicating formation of new higher susceptibility phases possibly as the result of vivianite alteration.
Although vivianite is paramagnetic at room temperature, studies of nodules of the material investigate the presence or absence of magnetic grains within the vivianite and oxidation, and shed light on the magnetic properties of lake sediments where vivianite is common.

Geologic setting
Vivianite was studied from sediments of Lake El'gygytgyn, located in the Anadyr Mountains of central Chukotka, northeastern Russia (67 • 30 N, 172 • 05 E) (Fig. 1).The lake is situated in a 3.6 Ma impact crater located in a series of volcanic rocks (rhyodacite ignimbrites, rhyolite to andesite tuffs, and basalt flows) of the late Cretaceous age (Bely and Belaya, 1998;Bely and Raikevich, 1994;Gurov et al., 2007;Layer, 2000).Sediment input to the lake is controlled by a series of some 50 small inlet streams draining the crater; output from the lake is only by one stream, the Enmyvaam River, flowing southeast from the crater to the Bering Sea (Nolan and Brigham-Grette, 2007).Preliminary studies and short cores collected from the lake (Brigham-Grette et al., 2007) attested to the suitability of lake El'gygytgyn sediments to provide a detailed climate record for this high arctic site.Additional deep drilling was done in 2008-2009, when over 318 m of sediment core and 200 m of impact breccia was extracted by ICDP at site 5011 in the lake (Melles et al., 2011), providing a vast amount of material to be studied for paleoclimate information (Melles et al., 2012, and papers within this issue).Geomorphological data indicate that the crater has never been glaciated (Glushkova, 2001) and no hiatus in sedimentation is observed (Melles et al., 2011).Chronology of the entire package has been established using a paleomagnetic timescale with detailed tuning to the marine oxygen isotope record and insolation variations (Melles et al., 2012;Nowaczyk et al., 2013).Biological, physical and chemical proxies reflect glacial/interglacial climatic conditions (Melles et al., 2012).During cold climatic stages conditions included perennial lake ice, oxygen-depleted bottom waters, absence of bioturbation and enhanced preservation of organic matter (Melles et al., 2007), resulting in the dissolution of magnetic minerals (Nowaczyk et al., 2002).A perennial ice cover on the lake obviously restricted the transport of coarse-grained sediments but enabled finer particles to be transported to central part of the lake basin through cracks or moats around the shore during summer (Asikainen et al., 2007).
Sediments from cold climate stages are enriched in Al 2 O 3 , MgO, TiO 2 , Fe 2 O 3 , Ni, Cr, and have high MS (Minyuk et al., 2007(Minyuk et al., , 2013)).During warm climate periods conditions include seasonal ice cover, decomposition of organic matter in bottom waters, oxic conditions, high level of bioturbation, low dissolution of magnetic minerals, and input of less chemical altered material.The resulting sediment is characterized by high content of CaO, Na 2 O, SiO 2 , K 2 O, Sr, and low values of organic carbon, sulfur, and nitrogen.
Vivianite has been identified extensively in the earlier, shorter cores (Asikainen et al., 2007;Minyuk et al., 2007) using both observational and laboratory techniques, and interpreted as forming under anoxic conditions when phosphorus and iron were both plentiful.Fine-grained dispersed vivianite has been recognized using low-temperature magnetic measurements down to 10 K (Murdock et al., 2012).Due to the ubiquitous occurrence of vivianite in cold (anoxic) and warm (oxic) stage sediments, Minyuk et al. (2007) conclude that its formation is controlled by diagenetic microenvironments and not influenced by large-scale climate conditions.

Materials and method
Vivianite nodules and pieces were collected down continuous core (Core 1A, 1B of ICDP site 5011) from 5.67 to 28 m in composite depth with ages 125.1-682.5 kyr (Melles et al., 2012).
Both sediment and nodules were retained for study, with the nodules separated from lake sediment by sieving, using a 250 µm sieve.The nodules were analyzed using a range of thermomagnetic and mineralogical methods at the North-East Interdisciplinary Scientific Research Institute of the Far East Branch of the Russian Academy of Sciences.Mass-specific low-field AC MS, as well as field-dependent and frequency-dependent MS were measured on a MFK1-FA Kappabridge (AGICO Ltd., Brno, Czech Republic) with sensitivity 3 × 10 −8 SI.Hysteresis parameters, including saturation magnetization (J s ), induced magnetization (J i ), saturation remanence (J rs ), coercive force (B c ), and remanence coercivity (B cr ) were measured by an automatic coercive spectrometer (Burov et al., 1986).Sensitivity of magnetic moment J rs is 1 × 10 −8 Am 2 , sensitivity of magnetic moment J i is 1 × 10 −6 Am 2 , and maximum specimen volume is 1.92 cm 3 .
The temperature dependence of the induced magnetization (J i -T ) was measured on a Curie express balance (Burov et al., 1986) in field of 500 mT with heating rate of 100 • min −1 and sensitivity of magnetic moment J i 3 × 10 −8 Am 2 , and maximum specimen volume 0.07 cm 3 .Two heating runs were used.Temperature-dependent susceptibility (K-T) of crushed nodules was measured continuously from room temperature up to 700 • C and back to room temperature using a MFK1-FA Kappabridge equipped with a CS-3 hightemperature furnace (AGICO Ltd., Brno, Czech Republic) with sensitivity 1 × 10 −7 SI and maximum specimen volume 0.25 cm 3 .The heating and cooling rates were 10-12 • C min −1 .Software CUREVAL ver.8.0.1.(http://www.agico.com/)was used for resolution of susceptibility into ferromagnetic and paramagnetic components based on Curie-Weiss law (Hrouda, 1994) and for Curie temperatures estimation using the two-tangent method (Petrovský and Kapička, 2006).
Crushed vivianite nodules were heated in air.Powders from several vivianite nodules were mixed with sucrose (organic carbon), carbamide (nitrogen compound (NH 2 ) 2 CO), metallic powders of arsenic, and elemental sulfur, with the additives never being more then 5 % of the total material.Samples were then heated continuously from room temperature to 700 • C and cooled back to room temperature.Additives were used to simulate some chemical conditions in bottom sediments.Melles et al. (2007) show that El'gygytgyn Lake sediment, accumulated during cold climates, is enriched in total organic carbon, total sulfur, and total nitrogen.We found framboidal pyrite and fine-grained greigite in the sediment.Some sulfide grains included in vivianite nodules and as heating products of decomposition of pyrite and greigite can affect the high-temperature property of vivianite.Chalcopyrite (FeAsS) and impurity of arsenic in some pyrite framboids were determined by energy dispersive spectroscopy.Although arsenic was not found in El'gygytgyn vivianite nodules it could potentially be adsorbed onto vivianite (Thinnappan et al., 2008).To study the effect of vivianite on other minerals during heating, selected vivianite samples were heated with hematite and goethite: both of which were also detected in the sediments.
Electron microprobe (EMP) analyses of polished nodules mounted in epoxy resin were performed using "Camebax" microprobe (manufactured by Cameca Instruments in France) under an accelerating voltage of 25 kV and electronbeam spot size of 4 µm.Crystal LIF for FeK α and MnK α lines, and crystal PET for PK α lines were used.The standards used were hematite (Fe), apatite (P) and MnTiO 3 (Mn).Specimens for scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) analyses were mounted on aluminum stubs and carbon coated.The SEM-EDS analyses have been carried out using QEMSCAN system including scanning electron microscopy (EVO-50) with energy dispersive X-ray spectroscopy Quantax Esprit (Bruker, Germany).
The chemical composition and major elements in the lake sediments minus the large (> 250 µm) nodules were analyzed using a multichannel X-ray fluorescence (XRF) spectrometer SRM-25 (former USSR) and S4 Pioneer X-ray fluorescence spectrometer (Bruker, Germany).Elemental compositions were determined using the fundamental parameters method (Borkhodoev, 2002).

Results
Vivianite nodules range in size up to 3 cm.They are composed of crystal aggregates often having spherical morphology, sometimes crystal clusters or flat crusts.Surfaces of nodules are darker than inner parts, with a striking bluish to greenish color appearing on the surface with oxidation.Many spherical nodules are empty in the central parts or consist of visible crystals clusters; morphologies of inner and outer parts of typical nodules are illustrated in Fig. 2.

Microprobe analyses -nodules and sediments
Electron microprobe quantitative analyses of 10 selected nodules reveal the presence of Fe 2 O 3 , P 2 O 5 and MnO.The average content of Fe 2 O 3 is 35.18 %, with a range from 30.18 to 39.4 %.P 2 O 5 displays values between 21.23-29.28% (average 25.02 %) with MnO concentration varying from 0.67 to 6.34 % (average 2.06 %) (Table 1).The ratio of phosphorus to manganese in the vivianite samples range from 4.61 to 37.45.
Major elements determination was performed on sediment samples from which large (> 250 µm) vivianite nodules were removed (Minyuk et al., 2007(Minyuk et al., , 2013)).Figure 3 shows the distribution of Fe 2 O 3 , P 2 O 5 , and MnO along the depth profile.Distribution of both P 2 O 5 and MnO are synchronous in this section, whereas Fe 2 O 3 appears to vary independent of P 2 O 5 and MnO, with some intervals showing strong correlations, but other regions where low Fe 2 O 3 content corresponds to the phosphorus and manganese highs.For all samples the correlation coefficient (Pearson) between P 2 O 5 and MnO is 0.55; between Fe 2 O 3 and P 2 O 5 it is 0.55 (Fig. 4).
Distribution of phosphorus and manganese shows few levels with higher concentrations of these elements -depth intervals 8. 71-9.21, 11.51-12.19, 13.19-13.51, 18.39-18.85, 20.95-21.17, 24.31-24.73m (Fig. 3, Table 2).Geochemical data suggest that fine-grained vivianite minerals are present at these levels.Smear slides prepared from sediments in the indicated regions reveal blue, greenish vivianite aggregates and broken pieces of fine concretions.A maximum amount of the oxides in the studied interval are Fe 2 O 3 is 13.54 %, P 2 O 5 is 4.02 %, MnO is 0.91 %.The phosphate mineral apatite is not responsible for phosphorus spikes because of the absence of correlation between phosphorus and calcium, which are the main elements of apatite.

Scanning electron microscopy and energy dispersive spectroscopy
In most cases, the vivianite concretions studied are not homogeneous, but contain irregular oxidized patches as well as discrete mineral inclusions.Ten polished nodules were studied by electron microscopy (SEM) and energy dispersive spectroscopy (EDS).Backscatter images with accompanying EDS elemental scans are shown in Fig. 5.As can be observed in Fig. 5a and b, the lighter colored parts of the backscattered electron images have higher contents of Fe and P. Some of the studied vivianite nodules were attracted to a hand magnet.A few of these "magnetic" nodules were crushed and the resulting magnetic extract was studied using SEM and EDS.In a few grains sulfides of Fe, presumably greigite, were found (Fig. 5c), identified by the presence of iron and sulfur, and the absence of phosphorus.A semiquantitative composition of sulfide is Fe = 46.20-50.23 norm.wt %, S = 35.62-33.26norm.wt %, Ni = 2.31-2.77norm.wt %, and a few percent of Mg, Al, K, and Si.In some nodules ilmenite was identified (Fig. 5d) by the combined presence of iron and titanium, with a composition of , Al and Si < 1 %.Although rare, in a few samples Fe-rich phases were found, with iron (Fe = 82.87-92.90norm.wt %) the main constituent and only a few percent P, Mn, and Si.All the nodules investigated were found to include minor minerals with high contents of Si, Al, and Na.On the electron images these minerals look like dark patches and show up as several different forms and sizes.
Selected nodule samples, all from zones of low susceptibility, were measured for susceptibility in a range of magnetic fields from 5 to 700 A m −1 .There is considerable scatter in the measurements at low fields (< 100 A m −1 ), probably due to relatively high measuring error at very low fields (Hrouda et al., 2006) when dealing with paramagnetic material.Above 100 A m −1 samples show very little change in susceptibility -less than a few percent -whereas one sample shows a consistent susceptibility but some 10 % higher than the original value.

High-temperature dependence of magnetic susceptibility
Selected nodule samples were continuously measured for susceptibility as the temperature was raised from room temperature to 700 • C and cycled back to 50 • C in air, with repeated cycles on the same samples (Fig. 7).The behavior of the nodules can be divided into three general classes: those with heating and cooling curves that are not reversible and producing phases other than magnetite, those with nonreversible curves, but producing only magnetite, and curves that are nearly reversible.

Nonreversible curves with other phases than magnetite
In these samples heating and cooling curves are not reversible during cycling to 700 • C. On initial heating low susceptibility is measured until 500 • C when there is an increase of susceptibility with a marked peak occurring at about 560-570 • C, followed by a sharp drop at the Curie point of magnetite (Fig. 7a).The cooling curve displays a strong increase in susceptibility from the Curie temperature of magnetite to 500 • C followed by a continuous decrease.
After the first heating run the susceptibility approximately doubled, suggesting the formation of magnetite within the  vivianite sample.The heating curve of the second run displays an increase in susceptibility at 330-350 • C and a sharp drop at 630 • C (Fig. 7a).The cooling curves are much higher than the heating curves, with a sharp increase in susceptibility from 630 • C reaching a maximum at temperatures of 280-330 • C, before a rapid decrease to room temperature.Heating and cooling curves of the third run are almost reversible and have the same shape as the cooling curve of the second run, but the sharp drop of susceptibility on heating is not complete until 650 • C. Hysteresis data collected after each cycle are shown in Fig. 7b and tabulated in Table 4. Coercive force decreases from 50.3 to 20 mT and coercivity of remanence decreases from 91 to 56 mT after heating runs, suggesting the formation of a lower coercivity magnetic phase.

Nonreversible curves producing magnetite
In these samples heating and cooling curves are not reversible (Fig. 7c).K-T warming curves show a gradual decrease in susceptibility and a sharp drop close to 580 • C, marking the presence of magnetite.The cooling curves display a strong increase in susceptibility at the Curie temperature of magnetite, with a small but steady increase below 525 • C.After the first heating cycle, susceptibility shows an increase of 1.4 times the original.The heating curve of the second run is similar to the heating curve of the second run of the first curve class (Fig. 7a) and displays an increase in susceptibility at 330-350 • C and a sharp drop at 615 • C. In the cooling path there is a marked increase in susceptibility starting at 615 • C and reaching a maximum at 320 • C (Fig. 7c).On continued cooling susceptibility makes a sharp drop before leveling off at a value nearly double the initial susceptibility.Hysteresis behavior after the first heating is shown in Fig. 7d, along with hysteresis parameters listed in Table 4. Ratios of J rs to J s and B cr to B c plot within the pseudo-single-domain field of Day et al. (1977).

Nearly reversible curves
For the sample shown in Fig. 7e, the initial heating and cooling curves of the first run are almost reversible.There is a decrease in susceptibility up to 700 • C, with a cooling curve that is slightly lower but not markedly different.Susceptibility here includes up to 85 % due to a paramagnetic component, producing a noisy curve.Curie points of neither magnetite nor hematite are clearly visible on the graph.The heating curve for the second run is very similar to the first cooling curve.But, a marked change is observed in the second cooling run, with a strong increase in susceptibility observed at 620 • C, reaching its maximum at 270 • C, followed by a sharp decrease (Fig. 7e).Hysteresis behaviors after the first and second runs are plotted in Fig. 7f.There are moderate increases in B c and B cr from the first heating run to the second run (Table 4).The B cr /B c and J rs /J s ratios after second heat- ing suggest formation of a superparamagnetic domain magnetic phase.

Induced magnetization versus temperature
Additional material from three samples discussed above was also studied for the behavior of magnetization (J i ) with heating to 700 • C (Fig. 8).The initial heating curve for each sample shows a small but distinctive "hump" marked by change in slope or a slight increase in J i at 180-200 • C and a decrease at 320-340 • C.There is, additionally, a slight hint of a minor increase in J i at 580 • C (Fig. 8).The second set of heating curves shows no humps, but continuous decay of the magnetization with increased temperature to 580 • C after which the curves appear to flatten out.
In comparing the high-temperatures curves of MS of the first heating run (Fig. 7a, c, e) to the induced magnetization of first heating (Fig. 8a, b, c), several differences are noted.The visible increase in susceptibility at 500 • C for sample EV294 (Fig. 7a) is not observed on the J i heating curve (Fig. 8a temperature (Fig. 8) are nowhere evident on initial heating curves of susceptibility (Fig. 7).

High-temperature behavior of vivianite with additive material
To further investigate the behavior of the vivianite nodules with other common materials in lake sediments, we studied high-temperature susceptibility of vivianite mixed with sucrose (organic carbon), carbamide (nitrogen), metallic powder of arsenic, and elemental sulfur.
No difference is seen in the heating curves for crushed vivianite samples without additive or those with sucrose or carbamide added (Fig. 9a, b, c).All the heating curves show small but distinct drops near 580 • C. Cooling curves all show a considerable increase in susceptibility starting at 580 • C, with a continued increase to sharp peaks around 500 • C, and then decay as temperature decreases for the plain vivianite and the vivianite plus carbamide samples.The vivianite and sucrose sample also increases below 580 • C, but continues to increase up to room temperature.All samples finish with susceptibility 1.5 to 2 times higher than the initial value.
A vivianite sample representative of first type K-T nonreversible curves with increased MS during heating was heated with arsenic.The initial vivianite run (Fig. 9d) has a large increase in susceptibility between 500 • -600 • C, indicating the formation of magnetite.When heated with arsenic this increase is not seen, although a small increase occurred at lower temperatures around 400 • C (Fig. 9e).It appears that arsenic suppresses the formation of magnetite prior to the Curie temperature.The cooling curve for the arsenic-added sample indicates the formation of magnetite between 580 • C and 700 • C and shows a steep increase to 450 • -500 • C as seen in the previous examples.A similar result of arsenic suppressing magnetite formation can be observed in the heating curves of plain chalcopyrite and chalcopyrite with arsenic (Fig. 9f).
There is a noticeable increase in susceptibility at 150-170 • C on the heating curves of vivianite with added sulfur (Fig. 9g).On cooling, increases in susceptibility are seen at Curie temperature of magnetite and monoclinic pyrrhotite.Heating the vivianite and sulfur mixture to temperatures of 200, 400 and 600 • C and cooling each time, shows that all heating and cooling curves are irreversible (Fig. 9h).The formation of pyrrhotite is seen only after the final 600 • C heating and cooling run.On all these runs the final susceptibility increases, but always less than a factor of 2.
There are very different heating results when sulfur is added to goethite or hematite (Fig. 10b, e).After heating hematite and goethite with sulfur, susceptibility increases between 400 and 610 times.The newly formed mineral is magnetite, indicated by the large increase in susceptibility at 580 • C (Minyuk et al., 2011).
Mixtures of vivianite and either hematite or goethite show vivianite to play the reducing role, but less than sulfur does.Heating a hematite and vivianite mixture (1 : 1) shows that during heating magnetite is formed (Fig. 10c).On cooling the susceptibility curve sharply increases at both 580 • C and 685 • C, producing marked Hopkinson peaks for magnetite and hematite.The final MS increases four-fold after the heating and cooling cycle.Goethite-vivianite mixture shows the same cooling curves as for pure goethite, but without visible goethite-hematite transition at temperature 360 • C on heating curves (Fig. 10f).There is a small increase in susceptibility at 470 • C and the expected large drop at 580 • C. On cooling, both the 580 • C and 360 • C increases are noted, with final susceptibility again being four-fold larger than the initial.
Mixtures of pyrite and vivianite were also tested to further investigate changes in oxide mineralogy.The K-T curve of pyrite (Fig. 11a) is very similar to the heating and cooling curves of a 1 : 1 mixture of vivianite and pyrite (Fig. 11b).Very similar curves are also obtained when the mixture is 1 : 5 (Fig. 11c).In this case the Hopkinson peaks of monoclinic pyrrhotite are higher than magnetite ones.Cooling curves of mixture 1 : 10 are different, with no sharp increase in susceptibility at monoclinic pyrrhotite Curie temperature obvious (Fig. 11d).Susceptibility gradually increases from 600 to 280 • C on cooling, but on a second run heating curve there is a decrease in susceptibility at Curie temperature of monoclinic pyrrhotite.Distinct Hopkinson peaks of magnetite are observed on all curves of second runs.
K-T curves of mixture 1 : 50 yield another set of very different results (Fig. 11e).On cooling of the first run, susceptibility increases at 280 • C; the same increase is seen on the cooling curve of second run.Such increase suggests a formation of hexagonal pyrrhotite during cooling.During the second heating run, the hexagonal pyrrhotite transforms into monoclinic pyrrhotite, showing increased susceptibility at 220 • C and a drop at 300-320 • C (Fig. 11e).This may reflect the so-called λ-transition from antiferro-to ferromagnetic behavior as described by Dekkers (1989b), and Kontny et al. (2000).On the cooling curves of both runs, no increase in susceptibility at the Curie temperature of monoclinic pyrrhotite is visible.Sharp magnetite Hopkinson peaks are evident on the heating and cooling curves of the second runs.

Environmental significance of vivianite
In Lake El'gygytgyn sediment vivianite is the more abundant authigenic mineral among other iron-bearing minerals such as pyrite, greigite, siderite, and Fe-Mn aggregates.The nodules and concretions of vivianite amount to a few grams per sample along the core profile (weight of sample is 6 grams in average).Morphological features of nodules are variable, suggesting different growth conditions.Some nodules are empty or filled with crystals in central parts, indi-cating growth of the concretions from the surface towards the center.Other nodules consist of crystal clusters with randomly oriented fragile crystals, suggesting that nodules formed before sediment compaction.
Large nodules were found in sediments of cold (anoxic) and warm (oxic) climate stages, indicating that formation of nodules is not controlled by large-scale climate oscillations (Fig. 3).Vivianite nodules are distributed in different lithological facies A, B, C representing glacial (dark gray to black finely laminated silt and clay), interglacial (olive gray to brownish massive to faintly banded silt) and "super" interglacial (finely laminated reddish-brown silt) periods (Melles et al., 2012).Nodule occurrence appears to be controlled by the diagenetic microenvironment (Stoops, 1983;Stamatakis and Koukouzas, 2001); anoxic microenvironments are often found around fossil fish bones and organic detritus.
Other distributions seen as spikes of P 2 O 5 , MnO and Fe 2 O 3 occur along the depth profile (Fig. 3).In general, enrichment in these elements is in cold (anoxic) stage sediments as 6.2. 6.4, 6.6, 7.4, 8.4, partly 10 and 12, and in anoxic layers of 11.3.Often spikes of Fe 2 O 3 coincide with maximum MnO/Fe 2 O 3 ratios, indicating reducing conditions in the lake (Mackereth, 1966).
Magnetic behavior of vivianite nodules is important to overall interpretation of magnetic archives of lacustrine sediments that are used as environmental proxies.In the study of Lake El'gygytgyn, MS is an important parameter, and is used for core description, correlation, and dating (Melles et al., 2012;Nowaczyk et al., 2007Nowaczyk et al., , 2013)).Vivianite is paramagnetic at room temperature and becomes antiferromagnetic (weakly magnetic) only at low temperatures (< 20 K).But, samples studied here show that the nodules have magnetic susceptibilities similar to many of the bulk sediments, and often higher than the susceptibility of sediments from cold intervals.For example, MS in general shows low values in stage 6, but at depths 7.05-7.26m and 9.13-9.33m MS has slightly increased (Fig. 3).Simultaneously, this depth is enriched in P 2 O 5 and MnO that may enhance the formation of vivianite.Correlation coefficients between P 2 O 5 and MS in intervals 7.05-7.26m and 9.13-9.33m are 0.47 and 0.68, respectively.In stage 7.4 at depth 11.59-12.19m, the high P 2 O 5 content positively correlates with increased MS values (r = 0.58), indicating fine-grained vivianite is also present.Large vivianite nodules are absent in Fe-P-Mn intervals, suggesting possible precipitation of vivianite directly from lake water (e.g., Dean, 2002) that can be related to events of climate and environmental change.
In most cases vivianite concretions are not homogeneous, but contain irregular oxidized patches.In backscattered electron images the oxidized parts have lighter color and record higher Fe and P content (Fig. 5a, b).The oxidized parts are often located along cracks and in outer parts of nodule grains, and possibly consist of metavivianite or santabarbaraite.Some oxidation spots are depleted in Mn, as was reported by Fagel et al. (2005) for Baikal vivianite.On the other hand, Pratt (1997) reports that the product of vivianite auto-oxidation on cleaved surfaces is ferric hydroxide.Sapota et al. (2006) mention that some Baikal vivianite microconcretions are covered with yellow-brown iron oxides, likely goethite.In a calcareous medium, vivianite is oxidized to poorly crystalline lepidocrocite (Roldán et al., 2002).However, the products of oxidation appear to have little influence on the magnetic properties of vivianite due to the weak MS of any of the oxidation products.
Lepidocrocite is paramagnetic at room temperature with a low-field susceptibility of 57.8 × 10 −8 m 3 kg −1 (e.g., Hirt et al., 2002), goethite is antiferromagnetic with susceptibility 0.5-1.5 × 10 −6 m 3 kg −1 (e.g., Dekkers, 1989a).MS of nodules (n = 54) exhibit values between 0.78 × 10 −6 m 3 kg −1 and 1.72 × 10 −6 m 3 kg −1 (average = 1.05 × 10 −6 m 3 kg −1 ).Numerous nodules include grains of sulfides, ilmenite, iron, titanomagnetite, and/or clay minerals; magnetic susceptibility could be increased and magnetic inclusions could even increase the attraction of nodules to a magnet.Positive correlation between MS of nodules and MS of sediment suggests that during growth the nodules capture grains from sediments.But this included material is a secondary source of MS because the susceptibility of the nodules is higher than that of the sediment from low-magnetic intervals.

Interpretation of thermomagnetic data
Investigations of the behavior of vivianite upon cyclic heating and cooling reveal the mineralogical changes that can occur in this system.The formation of new mineral phases -predominantly magnetite and/or hematite -occurs when vivianite samples are repeatedly heated and cooled between room temperature and 700 • C. Such conditions occur in nature during peat deposit fires, where vivianite often occurs (Matukhina et al., 1986).
There are three types of K-T curves (Fig. 7).The characteristic features of all vivianite nodules are observed in the cooling curves of second and third runs.These curves show an increase in susceptibility from 620-650 • C to 250-300 • C followed by a sharp drop.
J i -T curves show increased J i during heating at temperatures from 180-200 • C (Fig. 8).The hump on the curves possibly reflects the dehydration of vivianite and the ensuing oxidation of Fe 2+ .
For interpretation of TMA results we use data from the literature of differential thermal analyses.Natural vivianite shows weight loss steps at 105, 138, 203, 272 and 437 • C that are attributed to dehydration (Frost et al., 2003).Marincea et al. (1997) point out that endothermic peaks recorded on the differential thermal analysis DTA curve at temperatures of 183 • C and 205 • C mark a major loss of structurally bound H 2 O and the beginning of oxidation of F 2+ to Fe 3+ .The exothermic peak recorded on the DTA curve at 270 • C corresponds to another phase of oxidation.Rodgers and Henderson (1986) report the loss of structural water combined with the oxidation of Fe 2+ spanning 65 to 315 • C, and the formation of alpha-FePO 4 , Fe(PO 3 ) 3 and, occasionally, Fe 2 O 3 , as marked on DTA curve at 660 • C. On K-T curves there are no visible signs of dehydration of vivianite at these temperatures, indicating that this process does not influence MS.
K-T curves of the first type show increased susceptibility on heating at temperature of 500 • C (Fig. 7a).The hump on heating curves is possibly attributed to the transformation of sulfides from pyrite to magnetite during heating.But pyrite transformation during heating usually begins at temperatures of 420-450 • C and forms magnetite and monoclinic pyrrhotite (e.g., Wang et al., 2008;Tanikawa et al., 2008).Thermomagnetic study of vivianite and pyrite mixtures shows that, depending on the mixture composition, monoclinic or hexagonal pyrrhotite is produced (Fig. 11).In our case, neither monoclinic pyrrhotite (high pyrite content) nor hexagonal pyrrhotite (low pyrite content) form.
A plausible explanation can be obtained with goethite as an oxidation product of vivianite.Goethite shows no such hump on heating curves (Fig. 10d), but on heating a mixture of vivianite and goethite a similar increase in susceptibility is seen after 500 • C (Fig. 10f).
Continued cycling of samples from the first type shows in the second run drop of MS at ∼ 620 • C. On the third run the temperature has increased to 650 • C. If this is indicative of a Curie temperature (630-650 • C), it is lower than that of hematite and closer to that of maghemite ( Özdemir, 1990;Gehring et al., 2009).According to J rs /J s and B cr /B c ratios pseudo-single-domain particles were formed during the first heating (Day et al., 1977).The decrease of B c and B cr after few cycles of heating and cooling, a decrease in J rs /J s , and an increase in B cr /B c indicate the formation of lower coercivity minerals and larger domain-state particles (Table 4, sample EV294).
Thermomagnetic data show that quantities of magnetic minerals are not generated during heating and cooling.MS does not increase by more than two-fold after heating.

Influence heating media and Fe-bearing minerals on thermomagnetic properties of vivianite nodules
The behavior of the susceptibility of vivianite upon heating with organic carbon, nitrogen, sulfur and arsenic was monitored.The same as vivianite, these additives occur in Lake El'gygytgyn sediments of cold (anoxic) stages, and may influence on high-temperature properties of vivianite.Vivianite mixed with sucrose (organic carbon) or carbamide (nitrogen) shows similar thermomagnetic curves to vivianite without the additive.Arsenic suppresses the generation of magnetite at temperatures lower than the Curie temperature of this mineral (578 • C).But, magnetite that formed at higher temperature is of the same amount as that which had formed after heating vivianite with sucrose and carbamide.Specific thermomagnetic curves are produced for vivianite mixed with elemental sulfur.All first heating curves P. S. Minyuk et al.: Vivianite nodules, Lake El'gygytgyn, Northeast Russia show a small hump in susceptibility at temperatures of 150-170 • C. Interpretation of this hump is highly ambiguous.Such increase in susceptibility may reflect λ-transition from antiferro-to ferromagnetic behavior of pyrrhotite.Kontny at al. (2000) report temperatures of 160 • -210 • C for this transition.According to Li and Franzen (1996), thermomagnetic heating-cooling curves for a peak-type pyrrhotite in the temperature range between 20 and 400 • C are reversed from this.Our data indicate that the mineral phase formed here is unstable.K-T curves after heating to 200 • C and to 400 • C exhibit irreversible thermomagnetic behaviors (Fig. 9h).A plausible explanation is that the hump reflects the dehydration of vivianite that was invisible without sulfur.
After heating the sulfur-vivianite mixture to 600 • C, monoclinic pyrrhotite, marked by Hopkinson peak on cooling curve at a temperature 320 • C, is formed.
Such behavior is different from thermomagnetic curves of hematite and goethite mixtures with sulfur showing formation of magnetite in large amounts (Fig. 10b, e).
Thermomagnetic data are widely used to diagnose magnetic minerals, but heating media and the presence of other minerals in the samples can complicate interpretation of the results.The influence of vivianite on the thermomagnetic behavior of hematite, goethite and pyrite was investigated.Hematite-vivianite mixture (1 : 1) shows that vivianite acts as a reductant and stimulates formation of magnetite (Fig. 10c).During heating of the goethite-vivianite mixture (1 : 1), the vivianite masks the goethite-hematite transition but stimulates the formation of magnetite (Fig. 10f).Behavior of a pyrite and vivianite mixture during heating depends on the specific pyrite-vivianite content.Thermomagnetic curves for mixtures with vivianite content up to 80 % are nearly similar to those for pyrite (Fig. 11a, b, c).Pyrite-vivianite mixing ratio 1 : 10 show formation of magnetite and monoclinic pyrrhotite during heating but lacking the sharp Hopkinson peak of pyrrhotite on cooling.A second heating run produces more magnetite.Heating and cooling of a pyrite-vivianite mixture (1 : 50) exhibits the formation of hexagonal pyrrhotite.This pyrrhotite is unstable during a second heating and transforms into unstable monoclinic pyrrhotite and magnetite, as seen on the cooling curve of the second run.

Conclusions
Vivianite nodules are abundant in both sediments of cold (anoxic) and warm (oxic) stages.MS of vivianite nodules varies from 0.78 × 10 −6 m 3 kg −1 to 1.72 × 10 −6 m 3 kg −1 (average = 1.05 × 10 −6 m 3 kg −1 ) and is higher than the susceptibility of sediments from cold intervals.This susceptibility of vivianite can then obscure the environmental signal determined from rock magnetic properties.
Magnetic properties of vivianite are due to respective products of oxidation as well as sediment and mineral inclusions.
There are three types of curves of high-temperature versus susceptibility for vivianite indicating different degrees of oxidation and inclusions in nodules.
Vivianite acts as a reductant and reduces hematite to magnetite, while also masking the goethite-hematite transition during heating.
Adding sulfur to vivianite stimulates formation of monoclinic pyrrhotite, whereas the addition of arsenic suppresses the formation of magnetite below its Curie temperature.
Heating selective vivianite and pyrite mixtures shows formation of different minerals -magnetite, monoclinic pyrrhotite, and hexagonal pyrrhotite -that make it difficult to interpret the thermomagnetic curves.

Fig. 7 .
Figure 7 4 Fig. 7. Susceptibility versus temperature (in air) curves (a, c, e) of representative vivianite samples of groups (the arrows indicate the heating and cooling curves; cursive numbers show the heating and cooling runs) and typical examples of hysteresis loops after heating run (uncorrected for paramagnetism) of vivianite (b, d, f) where dashed line indicates.

Fig. 10 .
Figure 10 2 Fig. 10.Susceptibility versus temperature curves of hematite and goethite mixed with sulfur and with vivianite (1 : 1).The arrows and cursive numbers indicate the heating and cooling runs.

Table 1 .
Electron microprobe analyses of selected vivianite nodules from Lake El'gygytgyn.

Table 4 .
Hysteresis parameters of vivianite nodules.SampleDepth, m B c ,mT B cr , mT J s , Am 2 kg −1 J rs , Am 2 kg −1 J rs /J s B cr /B c Italic font -parameters uncorrected for paramagnetism.