Calcium Carbonate Hexahydrate (Ikaite): History of Mineral Formation as Recorded by Stable Isotopes

Abstract

Calcium carbonate hexahydrate (ikaite) is a rare mineral that forms as metastable species in the organic-carbon-rich sediments of the King George Basin, Bransfield Strait, Antarctica, as a consequence of early diagenetic decomposition of organic matter under cold water (−1.4 °C) and high pressure (200 bar) conditions. Large crystals grow in the sediment immediately below the diagenetic transition between microbial sulfate reduction and methanogenesis at ~320 cm below sea floor (bsf). This process is reflected in the dissolved sulfate, total carbon dioxide, and methane concentrations, as well as in the carbon, hydrogen, and oxygen isotope chemistries of the interstitial fluids and dissolved gases of the host sediment. The ikaite crystal faithfully records in its zonal structure the changing carbon isotope ratio of the total dissolved carbon dioxide pool as it gradually diminishes during methanogenesis (δ¹³C_ikaite = −17.5 to −21.4‰). These changes in the crystal’s host environment follow general Rayleigh carbon isotope fractionation. The oxygen isotopes of the ikaite carbonate (δ¹⁸O_ikaite = 1.46 to 4.45‰) also show a strong zonal distribution, unrelated to temperature of formation, but perhaps controlled by the degree of recrystallization of ikaite to calcite. The crystal water of the ikaite is depleted 11‰ in ²H/¹H (VSMOW) relative to the coexisting interstitial water, which is in excellent agreement with the isotope fractionation of other hydrated minerals. In addition to the in situ temperature and pressure, nucleation of the ikaite crystals in the Bransfield Basin sediments may be induced by the high alkalinity, high phosphate concentrations, and dissolved organic compounds. Intense microbial metabolism generates such compounds; of these, aspartic acid and glutamic acid may play an important role, as they do in biological and extracellular carbonate mineral precipitation. All indications are that low temperatures (such as of polar environments), high calcium carbonate supersaturation caused by interstitial methanogenesis, and a sufficiently large supply of dissolved phosphate and amino acids favor metastable ikaite formation. These conditions, modified by recrystallization, may be preserved in calcite glendonites, thinolites, and other calcitic pseudomorphs derived from ikaite and found throughout the ancient sedimentary record.

1. Introduction

Calcium carbonate hexahydrate is a monoclinic, hydrated carbonate mineral (CaCO₃·6H₂O; Dana Class: 15.01.04.01 Hydrated Carbonates; Nickel–Strunz classification: 05.CB.25) that is composed of Ca–CO₃ surrounded by six water molecules [1,2]. Calcium carbonate hexahydrate has been known from synthesis studies since the 1800s [3,4,5]. Naturally occurring calcium carbonate hexahydrate is less commonly reported and was originally discovered by Hans Pauly in Ikka Fjord, Ivigtut, Greenland [6], after which the mineral name ikaite was applied (granted by International Mineralogical Association in 1963). Natural occurrences of ikaite are severely limited under Earth’s surface conditions because its stability is favored by high pressures and low temperatures. Yet, in many environments, ranging from deep-sea trenches and fans [7,8], glacial fjords and shelves [6,9], to polar basins [10,11,12,13,14], ikaite forms as a metastable species, sometimes in preference over calcite and aragonite. The formation of ikaite may be favored over calcite and aragonite by a combination of calcium-rich, high-alkaline fluids with elevated calcite-inhibiting inorganic compounds, e.g., calcium, magnesium, phosphate, sulfate [15,16], and/or organic additives [2,17,18,19,20,21,22,23,24].

Ikaite also occurs in a wide range of other environments, including sea ice [25,26,27,28,29], speleothems [30], alkaline springs or lakes [21,31,32], in riverbeds (due to anthropogenic pollution [33]), and in foods (e.g., frozen shrimp [34,35] and cheese [36]).

The ancient sedimentary record contains ample evidence, such as in glendonites, that ikaite was present in low-latitude marine sediments from the Neoproterozoic Era [37,38] Ediacaran Period (635–538.8 Mya) through to the present. It is now established that calcite hexahydrate is the precursor of thinolites, glendonites, gennoishi, jarrowites, and other calcitic pseudomorphs, nodules, and concretions seen worldwide. Some Ediacaran glendonites are silicified, replacing calcitic pseudomorphs that had resulted from earlier replacement of ikaites; new data suggest that they formed during an early Ediacaran cold period [39,40,41,42,43,44,45,46].

We examine here the detailed conditions of ikaite formation and its use as a paleoenvironmental indicator based on interstitial fluid chemistry, isotopic signatures, and diagnostic amino acid compositions retained in the crystal structure. Earlier, we reported calcite hexahydrate forming in recent organic-carbon-rich sediments of the King George Basin, Bransfield Strait, Antarctica [10], (Figure 1). In this back-arc basin, large, euhedral, monoclinic crystals of ikaite were repeatedly found in sediments from water depths of >1900 m. The crystals were chemically similar, as well as optically and structurally identical to the mineral ikaite from the Ikka Fjord, Greenland [6,9,47], hence our usage of this mineral name.

In the King George Basin, we postulate that ikaite forms as a consequence of diagenetic decomposition of sedimentary organic matter. These conditions are reflected in the interstitial fluid chemistry, including the carbon, oxygen, and hydrogen isotope ratios and the dissolved methane in the host sediment. As we discuss, the ikaite crystals grow in the sediment just beneath the diagenetic transition zone between microbial sulfate reduction and hydrogenotrophic methanogenesis at ~320 cm below seafloor (bsf). We also postulate that the ikaite growth is limited within the methanogenic zone to a ~65 cm depth interval from ~320 to 385 cm bsf. With our limited radiometric dating, this 65 cm interval corresponds to an ikaite growth period of <1000 years. The preferential formation of ikaite over other carbonate polymorphs, e.g., calcite, aragonite, and vaterite, is predicated on the in situ pressure and temperature (>200 bar and −1.4 °C, respectively). Nucleation may be induced by organic compounds, rich in acidic amino acids, which result from intense methanogenic bacterial metabolism, but also from a range of other potential inhibitors presented in the paper.

This study was undertaken with the RV Polarstern in the area of the initial discovery in the King George Basin, Bransfield Strait, Antarctic [10,48]. Interstitial fluids and gases were obtained and analyzed from (1) a giant box core, ANT 1138-4 [49,50], that penetrated to a sediment depth of 1225 cm bsf; (2) a 760 cm gravity core, ANT 1327-1; (3) a 50 cm short box core, ANT 1324-2; and (4) a single sample from a 2050 cm giant piston core, ANT 1344-1. The water depth at ANT 1138-4 was 1985 m, and ikaite crystals were encountered repeatedly between 714 and 727 cm bsf, suggesting that ikaites may occur as an extensive subsurface pavement of crystals. The Bransfield Strait is an active back-arc basin; its morphotectonic setting is described by [51,52,53]. For comparison, two samples of glendonite samples from Spitzbergen and the Canadian Arctic were also collected and analyzed as a C–O isotope profile. Furthermore, we report on additional ikaite, glendonite, thinolites, and other pseudomorph samples from the U.S. (Hoh Fm, Astoria Fm, Arctic, DSDP 583 and 584) and compare their C–O isotope compositions with the existing worldwide data on glendonites.

2. Methods

2.1. Shipboard Analyses

Interstitial waters from the mineral’s host sediment of the three stations were collected by pressure filtration at the temperature of the bottom water close to 0 °C with a technique similar to that described by [54]. The total dissolved carbon dioxide (DIC) was determined by acidification, He-stripping of the interstitial waters, and online CO₂ gas chromatography, while dissolved sulfate was determined gravimetrically as BaSO₄, and dissolved ammonium was measured photometrically. The results are reported in units of millimoles (mM) of interstitial water (

Table 1. Chemical composition of interstitial fluids and gases in Bransfield Strait cores ANT 1138-4 (1), 1324-2 (2), and 1327-1 (3).

Depth	Core	DIC	SO4	NH4	PO4	Ca	Mg	Mg:Ca	CH4	δ13CH4	δ13CO2	δ2H-CH4	δ2H-H2O	δ18O-H2O
cm	no.	mM	mM	mM	mM	mM	mM	(mol)	ppb wt.	‰, VPDB	‰, VPDB	‰, VSMOW	‰, VSMOW	‰, VSMOW
seawater		2.2	28.00	0.00	0.003				1		−0.1		0.0	0.0
0.5	2	4.2	27.33	0.14
3.5	2	5.4	26.61	0.18
6.5	2	6.6	26.11	0.20
10.0	3	6.8	25.66	0.34	0.075	10.40	52.62	5.1	39	−69.5
12.0	2	7.6	25.58	0.23
15.5	2	9.5	24.82	0.23
20.5	2	10.6	24.36	0.27
25.5	2	11.8	22.67	0.61	0.082				32	−51.4	−18.2
26.0	3	12.4	23.95	0.30		10.40	52.83	5.1
37.5	2	14.5	22.51	0.34
51.0	3	18.0	20.46	0.85	0.207	10.53	52.13	5.0	39	−58.9	−19.8
78.0	1	17.8	19.94	0.61	0.131						−18.4		−4.9	0.0
112.0	3	29.1	14.45	1.02	0.360	10.70	51.39	4.8	50	−60.3	−22.8
133.0	1	26.4	15.12	1.02	0.191	9.64
203.0	3	39.5	5.92	1.76	0.262	8.83	50.20	5.7			−20.6			0.0
235.0	1	42.9	5.54	1.82	0.542	10.35			82	−59.3
322.0	3	53.3	0.20	2.23	0.509	9.60	49.62	5.2	7340	−99.7	−23.7
366.0	1	55.1	0.10	2.32	0.387	9.21
370.0	3	54.6	0.05	2.38	0.522	9.98	50.36	5.0	14890	−100.7
412.0	3	56.6	0.05	2.57	0.426	9.78	50.73	5.2	17465	−101.9
431.0	1	57.9	0.05	2.70	0.275	8.38					−17.5		−7.0	−0.4
468.0	3	55.8	0.05	2.74	0.364	9.38	50.61	5.4	22540	−100.7
512.0	3	55.7	0.05	2.74	0.253	8.53	50.57	5.9	46380	−100.2
520.0	1	59.2	0.05	2.96	0.284	8.33
552.0	3	56.6	0.05	2.92	0.331	8.50	51.96	6.1	28740	−98.4
633.0	3	55.7	0.05	3.07	0.224	7.30	52.17	7.1	16950	−94.9
658.0	1	59.7	0.05	3.09	0.235				10720	−95.8		−200.0		−0.1
692.0	1	59.7	0.05	3.20	0.215	6.82					−13.7
728.0	3	58.7	0.05	3.12					8564	−90.3
752.0	1	59.1	0.05	3.20	0.185						−11.9		−5.2	−0.1
753.0	3	57.9	0.05	3.22	0.157	7.50	53.82	7.2	15480	−88.5
790.0	1	60.7	0.05	3.20	0.158	5.93
812.0	1	60.0	0.05	3.20	0.158	5.71			7395	−93.2	−11.7	−205.0		−0.1
908.0	1	59.8	0.05	3.20	0.134	5.83
1033.0	1	59.8	0.05	3.48	0.101	6.48					−7.9		−5.6	−0.1
1111.0	1	62.5	0.05	3.67	0.106	7.08			4338	−89.2		−198.0
1162.0	1	64.2	0.05	3.87	0.101				5247	−86.5	−5.7	−194.0		−0.1

). They represent a composite downcore dataset with increased frequency of samples near the sediment–water interface, the sulfate reduction zone (SRZ)–methanogenic zone (MZ) transition, and the ikaite crystal horizon. Note, for simplicity, the species charges are understood and not added, e.g., NH₄⁺ is reported as NH₄.

Hydrocarbon gases were extracted from fresh sediment core sections using a vacuum/acid degassing procedure [55]. Gas concentrations were quantified by FID gas chromatography and are reported in parts per billion (ppb wt.), i.e., as weight of the hydrocarbon gas per weight of wet sediment (Table 1). Interstitial water samples for stable isotope measurements were drawn into vacutainers and treated with mercuric chloride. Mineral fragments of ikaite, generally from the crystal surface, were placed in glass ampoules, flushed with nitrogen, and sealed rapidly.

2.2. Shore-Based Analyses

Isotope measurements were made of dissolved inorganic carbon, DIC (¹³C/¹²C), interstitial water (¹⁸O/¹⁶O and ²H/¹H), and methane (¹³C/¹²C and ²H/¹H) on variably spaced samples throughout the composite depth of coring. The crystal-hydration water of “in vitro” decomposed ikaite mineral fragments, collected and kept in glass ampoules, were analyzed for ²H/¹H (¹⁸O/¹⁶O of the hydration water was not measured). The calcium carbonate residue of bulk ikaite after decomposition was analyzed for ¹³C/¹²C and ¹⁸O/¹⁶O ratios. In addition, a microsampling traverse of 7 points was made with a fine drill across the lateral growth axis of a 43 mm long × 17 mm wide ikaite single crystal (sample ID K735) and, similarly, at 3 points along the surface of the crystal in the direction of longitudinal growth (c*-axis) [47,56]. For comparison, similar microsampling traverses were made on two glendonite samples (calcite pseudomorphs) from Spitzbergen and Canada. These microsamples were analyzed for ¹³C/¹²C and ¹⁸O/¹⁶O of the calcium carbonate.

2.3. Isotope Methods

The interstitial H₂O was reduced to hydrogen for ²H/¹H measurement by freezing the H₂O over onto zinc-filled 6 mm glass tubes, which were evacuated, sealed, and heated to 460 °C [57]. In a similar manner, for the H₂O released from the ikaite fragments, decomposition was accelerated by slight heating (<40 °C), was frozen over (under vacuum) onto the zinc tubes, and then reduced to hydrogen gas for ²H/¹H measurement. The DIC in the interstitial waters was released by acidification and collected at 25 °C for C- and O-isotope determinations. The bulk and traverse-sampled calcium carbonates of the ikaite were prepared for C- and O-isotope ratios by the conventional H₃PO₄ acidification method of [58].

Methane was partitioned by gas chromatography (GC) and then combusted with CuO at 880 °C [59]. The ¹³C/¹²C of the resultant CO₂ was measured directly by isotope ratio mass spectrometry (IRMS). The H₂O from the methane combustion was subsequently reduced on zinc, and the ²H/¹H was measured on the evolved hydrogen gas by GC-IRMS. The stable carbon and hydrogen isotope results (Table 1) are reported in the usual δ-notation of δ²H, δ¹³C, and δ¹⁸O, whereby

$$\mathsf{\delta }{}_{ }{}^{\mathrm{n}}\mathrm{X}=\left(\frac{{}_{ }{}^{\mathrm{n}}\mathrm{R}{}_{\mathrm{sa}}-{}_{ }{}^{\mathrm{n}}\mathrm{R}{}_{\mathrm{st}}}{{}_{ }{}^{\mathrm{n}}\mathrm{R}{}_{\mathrm{st}}}\right)\times {10}^3$$

(1)

and R_sa are the ²H/¹H, and ²H/¹H and ¹⁸O/¹⁶O ratios in the sample relative to R_st, the corresponding ratios in the standards Vienna Peedee belemnite (VPDB) and Vienna standard mean ocean water (VSMOW). The isotope fractionation factors (α_A−Β) are defined as

$${\mathsf{\alpha }\mathrm{X}}_{\mathrm{A}-\mathrm{B}}=\left({\mathrm{R}}_{\mathrm{A}}/{\mathrm{R}}_{\mathrm{B}}\right)=\frac{({\mathsf{\delta }}_{\mathrm{A}}+{10}^3)}{({\mathsf{\delta }}_{\mathrm{B}}+{10}^3)},$$

(2)

whereby the isotope having the larger mass is more concentrated in chemical species A than in B. This relationship generally implies isotopic equilibrium, which may not necessarily be the case in all the reactions discussed here.

2.4. Amino Acids

For the analysis of amino acids, 135 mg of the homogenized, dry, ikaite material was weighed into heavy-walled glass ampoules, flushed with purified nitrogen for 3 min after adding 2 mL of 6N HCI, sealed, and then hydrolyzed at 110 °C for 22 h. Appropriate amounts of norleucine (1.25 mmole) were added as internal standard prior to hydrolysis [60]. The hydrolysate was evaporated to dryness at reduced pressure and the residue taken up in 2 mL of sodium citrate buffer (pH 2.0, 0.18 N Na⁺). Aliquots of 250 µL were injected into an automatic amino acid analyzer (Biotronik) for separation and detection. Amino acids were separated by ion exchange chromatography using a sodium citrate 4-buffer system for elution, the q-phthalaldehyde method for detection [61,62], and a computing integrator (Spectra Physics 4100) for peak integration and quantification [63]. The concentrations of amino acids were calculated from norleucine recovery, which ranged from 78 to 90 % in similar analytical runs. Amino sugars were not quantified here.

3. Diagenetic Environment

3.1. Nutrient Stoichiometries

The unconsolidated sediments from the King George Basin, in which the ikaites are found, are organic-rich, rapidly deposited, and undergo diagenesis in strongly anoxic conditions. Dissolved sulfate decreases rapidly below the water–sediment interface downcore, from a normal seawater concentration of 28 mM to exhaustion at ~320 cm bsf (Figure 2, Table 1) as it is utilized by sulfate-reducing bacteria (SRB) in the oxidation of organic matter. In this initial depth interval of sulfate consumption, methane concentrations are low, at low background levels of <100 ppb wt. (Figure 2, Table 1). Following the depletion of dissolved sulfate at ca. 320 cm bsf, methane concentrations quickly increase up to >46 ppm wt. The depth separation of the sulfate reduction and methane formation zones (SRZ and MFZ) is typical for anaerobic marine sediments that are rich in organic carbon [64], δ¹³C_CH4, and δ²H_CH4 measurements, i.e., from −95.8 to −83.8‰ and from −200 to −194‰, respectively (Figure 2, Table 1). This indicates that the methane is of archaebacterial origin and a product of the hydrogenotrophic methanogenic pathway [64]. A further consequence of organic matter decomposition is the steady regeneration of dissolved DIC and ammonium (Figure 2, Table 1). Ammonium rapidly accumulates up to 2 mM in the sulfate-reducing zone and continues to rise, albeit more slowly, in the methanogenic zone to over 3.88 mM at the greatest depth cored (2055 cm bsf). DIC also increases rapidly in the SRZ from 2.2 mM to 53 mM at 320 cm bsf. However, in contrast to the ammonium, its concentration below the SRZ increases only slightly to a maximum of 64 mM at 1162 cm bsf. In the sulfate-reduction zone, dissolved phosphate increases from seawater values (~2 µM) to over 0.5 mM (Figure 2) but then drops steadily below the SRZ to 0.1 mM at 1162 cm bsf.

The depth distribution of dissolved DIC is controlled by various and sometimes competing diagenetic processes that can be followed using stable isotope signatures of the gas and fluid reservoirs. To understand ikaite formation in the King George Basin, it is essential to first discuss the dissolved DIC habitat from which the mineral is derived. The initial rise in DIC to 53 mM (Figure 2) is largely due to organic matter remineralization related to bacterial sulfate reduction in the SRZ. This is evident in the highly significant negative stoichiometric correlation between DIC and SO₄, and between NH₄ and SO₄ (core ANT 1138-2, Equations (3) and (4), Figure 3):

DIC (mM) = −1.79 SO₄ (mM) + 53.5; r = 0.997

(3)

and between ∆NH₄ and −∆SO₄ (Figure 3):

NH₄ (mM) = −0.079 SO₄ (mM) + 2.26; r = 0.993

(4)

The slight deviation from the ideal regeneration ratio of ∆DIC:−∆SO₄ = 2:1 suggests that the anaerobic oxidation of methane by sulfate-reducing bacteria consumes sulfate at the base of this zone [64,65]. Below 320 cm bsf and within the zone of methanogenesis, the input to the DIC reservoir by remineralization of the organic matter slightly exceeds the depletion of the DIC pool by methanogenesis. This is reflected by the continued slight increase in DIC observed at depth, whereas ammonium continues to increase with depth (Figure 2).

The depth distribution of phosphate in core ANT 1138-2 is very different from both DIC and NH₄. In the SRZ, phosphate linearly tracks the increased ammonium due to organic matter remineralization, up to a maximum of 0.39 mM (Figure 4), with the stoichiometry of

NH₄ (mM) = 0.155 PO₄ (mM) + 0.018; r = 0.989.

(5)

However, beneath the SRZ, phosphate decreases continually in concentration down to 0.1 mM. This bifurcated relationship is illustrated in Figure 4, with the clear departure of phosphate from both NH₄ and DIC beneath the SRZ. On the other hand, phosphate tends to track the loss of calcium below the SRZ (Figure 5), which may be due to the uptake by calcium orthophosphates. The molar ratio of phosphate tracking Ca loss below the SRZ is almost 1:1, suggesting such a process.

Some Neoproterozoic strata, potentially associated with the “Snowball Earth” [66], also contain glendonites, e.g., Dalradian slates and metalimestones, Scotland [37,38], and some lithologic units in the early Doushantuo, China, which are dominated by phosphorites [45,46].

3.2. Carbon Isotope Variation in DIC

The carbon isotope ratio of DIC (δ¹³C_DIC) measured downcore shows an interesting isotopic reversal (Figure 2). Firstly, there is an excursion in δ¹³C_DIC from 0‰ to −23.7‰ between the sediment–water interface and the base of the SRZ (Table 1). Beneath the SRZ, δ¹³C_DIC is continually enriched in ¹³C, from −23.7‰ to −5.7‰, with increasing sediment depth. This characteristic isotope depth distribution can be related primarily to two separate diagenetic processes: (1) isotopic mixing of different bicarbonate sources during remineralization, and (2) isotopically fractionated removal during methanogenesis. These are described below.

3.2.1. Linear Isotope Mixing

Between the sediment surface and the base of the SRZ at 320 cm bsf, the δ¹³C_DICtotal represents a mixture of dissolved inorganic carbon from incorporated bottom seawater (DIC_sw, 2.2 mM, δ¹³C_DICsw ~ 0‰) and DIC_org (δ¹³C_DICorg ~ −24‰) that has been derived from the decomposition of organic matter (δ¹³C_org ~ −25‰) in the sediments through diagenesis. The resultant carbon isotope composition of this mixture is approximated by the linear mixing in Equation (6):

f_total·δ¹³C_DICtotal = f_sw·δ¹³C_DICsw + f_org·δ¹³C_DICorg,

(6)

where f_a is the fraction of the concentration (mM) and f_total is f_sw + f_org.

The isotope mixing line, drawn in the SRZ (Figure 6, right panel), is generated using the actual measured DIC concentration (f_total) increasing with depth. Although there is more variability in the actual measured values, these mixing-model values adequately describe the δ¹³C_DICtotal distribution down to near the base of the SRZ.

3.2.2. Kinetic Isotope Effects Associated with Methanogenesis

Beneath the SRZ, deeper than 360 cm bsf, the observed δ¹³C_DIC profile is modulated by carbon isotopic fractionation due to methanogenesis. The higher conversion rate by methanogens of ¹²C_DIC over ¹³C_DIC into methane leads to a substantial shift to ¹³C-enriched values in the carbon isotope ratio of residual DIC in the MFZ (Figure 6) [67,68]. This intense, archaebacterial-mediated kinetic isotope effect is easily observed in the accumulation of methane in the sediments that is depleted in ¹³C, i.e., δ¹³C_CH4 ~ −100‰ (Figure 2 and Figure 6, Table 1).

Carbon isotope fractionation during methanogenesis by carbonate reduction in a closed system leading to DIC enriched in ¹²C can be described by the general Rayleigh-type equation [69]:

R_t = R_o·f(α_c − 1),

(7)

where R is the ¹³C/¹²C ratio of the initial DIC (R_o) and the residual DIC (R_t) of the fraction DIC remaining reservoir (f) as f goes from 1 to 0. The Rayleigh equation, approximated by [70], has been converted to δ-notation as

$$\mathsf{\delta }{}_{ }{}^{13}\mathrm{C}{}_{{\mathrm{DIC}}_{\mathrm{t}}}=\left(\mathsf{\delta }{}_{ }{}^{13}\mathrm{C}{}_{{\mathrm{DIC}}_{\mathrm{i}}}+{10}^3\right)f^{\left(\mathsf{\alpha }-1\right)}-{10}^3,$$

(8)

where δ¹³C_DIC(i) is the initial DIC carbon isotope value (−24‰), δ¹³C_DIC(t) is the carbon isotope value of the DIC reservoir at time t, and f is the fraction of the DIC precursor pool remaining after uptake by hydrogenotrophic methanogenesis and αC_methane-DIC. The difficulty in solving Equation (8) lies in the proper estimation of the DIC fraction that has been removed from the precursor pool during bacterial carbonate reduction, i.e., 1 − f. As a first approximation, this DIC fraction can be estimated by comparing the DIC with the corresponding ammonium concentration that accumulates in the interstitial water, as described below.

The accumulation of ammonium may be assumed to represent the degree of organic matter remineralization at each depth interval. This is generally true because ammonium remains largely unaffected in these sediments by secondary reactions, such as precipitation or further metabolism. Admittedly, absorption of minor amounts ammonium will introduce some error into this estimation. However, comparison with the phosphate concentrations reveals that this error is, in fact, small. In the SRZ, where the DIC is unaffected by methanogenesis, the DIC is stoichiometrically related to NH₄ as follows:

DIC (mM) = 22.2 NH4 (mM) + 2.81; r = 0.989,

(9)

which corresponds to a molar DIC:NH4 ratio of 1:0.045 (closed symbols in Figure 4), and a Redfield-based C:N of 105:4.7. Clearly, the amount of ammonium regenerated compared to DIC is less than that predicted by the normal diagenetic model. In the zone of methanogenesis (open symbols in Figure 4), the molar DIC:NH₄ ratio shifts markedly to a value of 1:0.21, which corresponds to a C:N of 22.1:4.7. This decrease in C:N is a direct consequence of the combined effects of: (1) the continued addition of ammonium and bicarbonate to the interstitial fluid by remineralization and (2) the preferential removal of DIC by methanogens during methanogenesis via the hydrogenotrophic pathway.

The predicted DIC concentration, i.e., that which would have been present in the deeper sediments (>320 cm bsf) without any removal by methanogenesis via carbonate reduction, can be calculated directly from the ammonium concentration using Equation (9). To do this, we assume that the release of DIC from the metabolizable organic matter in the methanogenic zone (prior to methanogenesis) follows the same C:N regeneration ratio as that observed in the SRZ, i.e., DIC:NH₄ = 1:0.045 (Figure 4 and Figure 6, left panel). We also assume that the ammonium concentration is only affected by addition through remineralization. The difference between the predicted and the measured DIC concentrations in the methanogenic zone is therefore the fraction of bicarbonate consumed as a result of methanogenesis or other DIC removal. If there is more DIC than predicted by the corresponding NH₄, then the DIC is in excess. The opposite situation is a DIC deficit. As shown in Figure 6 (left panel), there is a clear and strong DIC deficit within the zone of methanogenesis that increases from 0 at the upper boundary (~350 cm bsf) to 24.5 mM at the base of the core ANT 1138-4 (1162 cm bsf). This corresponds to a change in the DIC reservoir with depth from f = 1.0 to 0.72, wherein

f = DIC_measured/DIC_predicted.

(10)

Using Equation (8) and taking δ¹³C_DIC(i) = −24‰ and αC_methane-DIC = 1.06 for hydrogenotrophic methanogenesis [64], the change with depth for the predicted δ¹³C_DICt can be calculated. Figure 6 (right panel) shows that this predicted change in δ¹³C_DIC of the residual DIC reservoir due to methanogenesis adequately models the actual measured δ¹³C_DIC profile in the methanogenic zone. Calcium carbonate hexahydrate precipitation is apparently not a major influence on the δ¹³C_DIC, nor on the magnitude of the DIC reservoir.

An independent confirmation for this DIC:NH₄ stoichiometric model is provided by the corresponding carbon isotopic shift in the accumulating methane (δ¹³C_CH4). The preferential ¹²C depletion of the DIC substrate by methanogens enriches the remaining DIC in ¹³C. As a consequence, the methane formed subsequently (deeper) becomes isotopically enriched in ¹³C (δ¹³C_CH4 shifts from −101.9‰ to −86.5‰, Table 1). Figure 2 clearly demonstrates this parallel isotope tracking of δ¹³C_CH4 to δ¹³C_DIC with increasing depth.

4. Calcium Carbonate Hexahydrate Precipitation

4.1. Ikaite Formation

It is our contention that the precipitation of the calcium carbonate hexahydrate was initiated at the base of the SRZ in response to increased carbonate supersaturation and inhibition of calcite and/or aragonite formation. Ikaite crystals formed in this zone were subsequently buried to greater depth, with some shorter-term additional growth. At the prevailing conditions of the Bransfield Strait surface sediments, −1.4 °C and ca. 0.2 kbar, the metastable CaCO₃·6H₂O forms rather than calcite or aragonite. Based on the works of [20,21], ikaite at −1.4 °C would be the stable carbonate mineral only at pressures above 3.1 kbar. This metastable state accounts for the observed rapid decomposition at shipboard conditions of ikaite to vaterite, calcite or amorphous calcium carbonate, and H₂O [71,72]. It may also be responsible for the possible in situ recrystallization of the center portions of ikaites to sparry calcite and thus incipient pseudomorphism. Methanogenesis possibly initiates ikaite crystallization, analogous to that reported for calcite precipitation by other investigators [67,73,74,75]. The removal of bicarbonate by hydrogenotrophic methanogens, starting around 320 cm bsf, tends to raise the pH and thereby vastly increase the degree of supersaturation (ICP/K_sp >> 1). The resulting carbonate precipitation proceeds according to the following reaction:

Ca²⁺ + 2HCO₃⁻ + 3H₂O + 8H⁺ ⟶ CH₄ + CaCO₃·6H₂O_(s).

(11)

This possible association of ikaite and methane formation is supported indirectly by the close spatial relationship between the onset of methanogenesis and the suspected initial point of formation for the ikaite crystals.

An alternative explanation for the initiation of ikaite precipitation at the base of the sulfate zone is the accumulation of excess DIC, resulting from the anaerobic oxidation of methane. Methane, diffusing upward to the base of the sulfate zone, is consumed by methanotrophs, and returned in part to the interstitial water as DIC [64,65,76]. This anaerobic methane oxidation maintains the methane concentration at trace levels (<0.04 ppm) in the uppermost 300 cm bsf. The excess DIC in the methane oxidation zone is only weakly evident from the C:N stoichiometric model (Figure 4). The DIC concentration measured around 350 cm bsf, i.e., at the depth where significant methane consumption is occurring, is higher by only about 1 mM relative to that predicted from the ammonium concentration by Equation (9). This metabolic DIC derived from ¹²C-enriched methane (δ¹³C_CH4 ~ −100‰) should also be relatively enriched in ¹²C. To some extent, such a ¹²C-DIC isotope shift is observed around 300 cm bsf and fits the δ¹³C_DIC model (Figure 6, right panel).

4.2. Radiocarbon Dating

Radiocarbon dates were obtained from two organic matter samples on an adjacent sediment core (Station 14882-278). Samples at 610 and 810 cm bsf yielded ages of 3560 yBP and 6530 yBP, respectively, giving an average sedimentation rate over this interval of ~0.67 mm y⁻¹ (Erlenkeuser, 1984, unpublished data). Assuming that the ikaite crystal grew in the sediment depth interval of 320 to 385 cm bsf (65 cm), this average sedimentation rate corresponds to a crystal growth period of ~965 years. If the crystal growth was constant over this period, which certainly may not be true, then the estimated average growth rate for the ikaite crystal c*-axis (43 mm long) is ~0.045 mm/year. A single radiocarbon date on a carbonate sample of the ikaite outer rim (δ¹³C_ikaite = −18.6‰), collected at 720 cm bsf, gave a conventional, uncorrected age of 4600 yBP (Geyh, 1984, unpublished data). This, indirectly, provides the most recent age of the dissolved bicarbonate pool carbon incorporated into the ikaite. The corresponding age of the sediment at 720 cm bsf interpolated from the 610 and 810 cm bsf samples (3560 and 6530 yBP) was 5193 yBP. This rough agreement between the age dates of the sedimentary organic matter and interstitial fluid DIC precludes the possibility that substantial vertical transport and/or exchange in the interstitial water had occurred.

4.3. Carbon Isotope Zonation of Ikaite

An exceptionally large, 43 × 15 mm calcium carbonate hexahydrate crystal, collected from Antarctic sediment core ANT 1138-4, allowed us to verify the history of the ikaite mineral formation envisioned above. Carbon isotope profiles of subsamples, taken along a traverse of lateral crystal growth (Figure 7,

Table 2. Carbon and oxygen profiles across ikaite crystal (sample K735) and glendonites.

Sample No.	Distance across Crystal (mm)	δ13Cikaite Carbonate (VPDB)	δ18Oikaite Carbonate (VPDB)	Temperature (°C) (Grossman 2012)
Bransfield Strait ikaite (sample K735)
1	0.0	−17.49	1.58	8.71
2	0.0	−17.60	1.46	9.19
3	0.0	−18.10	1.60	8.63
4	1.6	−18.33	2.32	5.85
5	3.0	−19.95	3.04	3.19
6	5.6	−20.85	4.07	−0.39
7	9.1	−21.35	4.45	−1.65
8	11.6	−20.25	3.91	0.15
9	14.9	−20.46	2.75	4.25
10	15.9	−18.10	1.67	8.36
Spitzbergen glendonite
1		−11.03	−4.51
2		−13.41	−2.95
3		−16.07	−1.85
Canada glendonite
1		−17.23	−6.86
2		−17.61	−5.83
3		−18.42	−5.04

), demonstrate the response of the calcium carbonate hexahydrate to the changing carbon isotope composition in the interstitial waters from which it was derived. The carbon isotope ratio of the innermost (oldest) section of the crystal is that which is most depleted in ¹³C (δ¹³C_ikaite = −21.4‰, sample 7, Figure 7). Proceeding outwards to the younger growth, the carbon isotope ratio shifts progressively by +3.3‰ to ¹³C-enriched δ¹³C_ikaite values of ~−18.1‰ (samples 4 and 10, Figure 7). Analyses of three samples, drilled from the current outside crystal surface along the longitudinal growth axis, yielded δ¹³C_ikaite values from −17.5‰ to −18.1‰ (samples 1, 2, and 3). These values agree with the outermost (youngest) samples from the transect. Assuming that the ikaite formed with constant isotope effects and in equilibrium conditions during its growth, then the δ¹³C_ikaite shift of +3.9‰ (i.e., sample 1 vs. 7), from the innermost ikaite to its surface, indicates that the δ¹³C_ikaite mirrors changes in δ¹³C_DIC as the crystal grew, as indicated in Figure 7.

The depth below seafloor at which the ikaite precipitation was initiated and proceeded is difficult to establish. This is partly because the carbon isotope effect between δ¹³C_DIC and δ¹³C_ikaite associated with ikaite formation is poorly known. Some authors [77,78,79] have suggested that the carbon isotope fractionation for ikaite-DIC (αC_ikaite-DIC) is small in the range of αC_ikaite-DIC = 1.0006 to 1.003. The αC_ikaite-DIC could be calculated directly from our isotope measurements, provided that the crystal is in equilibrium with the current interstitial fluid bicarbonate pool. At the present depth of ikaite occurrence (714–727 cm bsf), the δ¹³C_DIC is ca. −12.8‰. Using the δ¹³C_ikaite ~ −18‰ for the outer crystal surface, then, according to Equation (2), αC_ikaite-DIC would be approximately 0.995, i.e., the ikaite is ¹²C-depleted relative to the interstitial fluid DIC. This carbon isotope partitioning would indicate a reverse isotope effect, i.e., the solute phase is enriched in the heavier isotope relative to the mineral phase—a highly unlikely situation.

The formation of other hydrated carbonate minerals, for example, gaylussite (Na₂CO₃·CaCO₃·5H₂O) or trona (Na₂CO₂·NaHCO₃·5H₂O), reported by [80], have carbon isotope fractionation factors estimated at −1.4 °C to be αC_{gaylussite-DIC} = 1.0035 and αC_trona-DIC = 1.002, respectively. For comparison, based on the work of [81], the fractionation αC_calcite-DIC would be 1.0011 at −1.4 °C. The revision by [82] would set αC_calcite-DIC = 0.9995 at −1.4 °C. Unfortunately, there is less information available on the isotope fractionation for other hydrated carbonate minerals, such as nahcolite (NaHCO₃) or eitelite (Na₂Mg(CO₃) found in alkaline lakes [83], for comparative purposes.

These estimates of the ikaite-DIC carbon fractionation do not take into account rates of crystallization, which, as pointed out by [84], can lead to kinetic isotope effects at high precipitation rates, potentially decreasing the overall carbon isotope fractionation.

Similar to ikaite, the transformation process of vaterite to calcite or aragonite and possibly amorphous calcium carbonate, is potentially inhibited in the presence of phosphate [85] and stabilized by sulfate [86]. Furthermore, the vaterite can be more enriched in ¹³C by freezing and rapid growth rates [87,88].

Accordingly, we propose that the ikaite surface is not in equilibrium with the current interstitial DIC at 714–727 cm bsf where it is recovered; therefore, the calculated αC_ikaite-DIC of 0.995 is incorrect. Rather, the ikaite formed at an earlier stage, when it was situated shallower in the sediment column at lower δ¹³C_DIC, likely in the methanogenic zone, beneath the sulfate-reduction/anaerobic methane oxidation zone.

Applying a more reasonable range for αC_ikaite-DIC to be between 1.001 and 1.003, and assuming equilibrium, then the current ikaite surface with δ¹³C_ikaite = −17.5‰ was formed in the sediment where the δ¹³C_DIC of the interstitial fluid bicarbonate was between −18.5 and −20.4‰. Correspondingly, the interior of the ikaite crystal (δ¹³C_ikaite = −21.4‰) would have formed from the interstitial fluid bicarbonate with a δ¹³C_DIC between −22.4 and −24.3‰, depending on the chosen αC_ikaite-DIC. This indicates that the δ¹³C_DIC of the bicarbonate from which the ikaite was formed was about 10‰ enriched in ¹²C compared with the bicarbonate δ¹³C_DIC at the present depth of ikaite occurrence (727 cm bsf). Choosing a median αC_ikaite-DIC value of 1.002 and using the surface and interior δ¹³C_ikaite values of −17.5‰ and −21.4‰, respectively, gives a range in δ¹³C_DIC from −19.5‰ to −23.4‰. This implies that the ikaite single crystal grew while it was situated between 330 and 400 cm bsf (Figure 6, right panel). It is notable that the depth of 330 cm bsf corresponds to just beneath the base of the SRZ and methane oxidation zones. This assumes, of course, that the present interstitial fluid profiles have remained constant, i.e., in steady state, over the last few thousand years. Ikaite formation at the base of the SRZ has also been reported by others [89]. As the ikaite crystal grew between 320 and 385 cm bsf, it remained in contact with the interstitial fluid bicarbonate pool. The δ¹³C_DIC of this pool was being progressively enriched in ¹³C due to the faster removal of light bicarbonate (¹²C) by the methanogens. This led to the pronounced carbon isotope zonation profile measured across the crystal growth structure (Figure 7). Once the crystal growth ceased near the base of the sulfate zone (ca. 385 cm bsf), the interstitial δ¹³C_DIC continued to be depleted in ¹²C due to carbon isotope fractionation during methanogenesis until the carbon isotope separation between the crystal surface and the DIC became the 5.6‰ that is measured now. As mentioned above, the postulate that the current crystal surface is in equilibrium with the DIC at 720 cm bsf could only be tenable if the fractionation factor αC_ikaite-DIC would be 0.994, which is the opposite of the isotope effect expected for carbonate mineral precipitation in hydrated and anhydrous phases.

4.4. Oxygen Isotope Zonation

The stable O-isotopes of the interstitial water in core ANT 1138-4 remain constant with sediment depth of δ¹⁸O_H2O around −0.1 ± 0.1‰ (Figure 2, Table 1). This is consistent with values for Antarctic bottom waters (δ¹⁸O_H2O of −0.29‰, VSMOW) [90]. The uniformity of δ¹⁸O_H2O with sediment depth indicates either that the bottom water incorporated into the sediments has not changed significantly over the time spanned by the core, as supported by the ¹⁴C radiometric dates, or that vertical fluid movement has exchanged and homogenized the interstitial water. The latter is unlikely, in consideration of the dissolved nutrient and δ¹³C_DIC profiles, which most certainly would also have uniform distributions resulting from diffusive mixing.

The cross-sectional profile of the carbonate oxygen isotope ratios in ikaite crystal K735 (δ¹⁸O_ikaite) systematically change along the 16 mm cross section sample traverse (Figure 7), analogous to that observed for δ¹³C_ikaite. The three surface samples along the length of the crystal (numbers 1, 2, and 3 in Figure 7) have a consistent δ¹⁸O_ikaite values of +1.58, +1.46, and +1.60‰, respectively. The outermost (younger) samples in the cross section (numbers 4 and 10), with δ¹⁸O_ikaite values of +2.32 and +1.67‰, respectively, are similar to the surface δ¹⁸O_ikaite values. However, these samples are significantly depleted in ¹⁸O, in comparison with the innermost (older) section (sample 7), which has a δ¹⁸O_ikaite of +4.5‰. Relative to the interstitial water, the oxygen isotope fractionation between carbonate ikaite and water ranges from αO_ikaite-H2O of 1.0046 (innermost, sample 7) to 1.0016 (outermost, sample 4).

The oxygen isotope separation between the ikaite carbonate and the formation water can potentially provide information on the processes of precipitation, recrystallization, and formation temperature. However, the determination of formation temperature is complicated by several factors, including lack of equilibrium and the possible partitioning of oxygen between the carbonate oxygen (δ¹⁸O_ikaite) and the contamination (mixing and/or exchange) by the hydration water oxygen. In addition, the magnitude of the isotope fractionation between ikaite and the interstitial water (αO_ikaite-H2O) may also be controlled by factors which include the degree of Ca, DIC saturation, and rate of formation [91]. Furthermore, it is clear that the simple application of Equations (2) and (12) to estimate temperatures of hydrated or cryogenic carbonates, such as ikaite, monohydrocalcite, or calcium carbonate hemihydrate, may not be valid. Recognizing the limitations, the standard ¹⁸O-paleotemperature expression for the equilibrium precipitation of calcite from seawater (e.g., as in [92], based on [93,94]) can be very tentatively used to compare the calculated vs. actual temperature of the ikaite formation, i.e.,

$$T \left(°\mathrm{C}\right)=15.7-4.36\left({\mathsf{\delta }}^{18}{\mathrm{O}}_{\mathrm{calcite}}-{\mathsf{\delta }}^{18}{\mathrm{O}}_{\mathrm{water}}\right)+0.12{\left({\mathsf{\delta }}^{18}{\mathrm{O}}_{\mathrm{calcite}}-{\mathsf{\delta }}^{18}{\mathrm{O}}_{\mathrm{water}}\right)}^2,$$

(12)

where δ¹⁸O_calcite is relative to VPDB and δ¹⁸O_water is relative to VSMOW. Recognizing these cautions and constraints, the corresponding ikaite formation temperature calculation using Equation (12) gives a range in the formation temperature, from −1.7 °C for the innermost (oldest) part to 9.2 °C for the surface (youngest) of the crystal (Figure 7, Table 2). The warmest values are clearly unreasonable, considering the known sediment temperatures in the Bransfield Strait. Furthermore, it is unlikely that the shift is due to a restricted interstitial water reservoir that became enriched in ¹⁸O with time, nor the interstitial water of the sediment column exchanging completely over time and with a varying δ¹⁸O_H2O. The change in δ¹⁸O_ikaite across the crystal also cannot be explained by partial recrystallization of ikaite to calcite because microscopic, X-ray diffraction (XRD) and crystallographic measurements have confirmed that the recovered crystals are entirely intact ikaite [47].

The precipitation of ikaite is associated with an oxygen isotope fractionation of the carbonate oxygen that is lower than that of calcite, as represented by the outer, most recent section of the crystal (αO_ikaite-H2O = 1.0016). Rickaby et al. (2006) [95] determined experimentally over a δ¹⁸O_water range from −3.6 to 17.9‰ SMOW, that the oxygen isotope fractionation between the ikaite hydration water δ¹⁸O_hydration and δ¹⁸O_water was constant at ~−2.9‰ (2 °C), i.e., an αO_{hydration-water} of 1.0029. In contrast, and curiously, the carbonate δ¹⁸O_ikaite in the Rickaby et al. (2006) [95] experiments did not vary with δ¹⁸O_water and was constant at ~11‰. This calculates as an αO_ikaite-H2O of 0.9619 to 0.9812 compared with the αO_ikaite-H2O of 1.0046 to 1.0016 for our Bransfield Strait ikaite. Studies of vaterite (μ-CaCO₃), a metastable polymorph of calcium carbonate formed via dehydration [96], exhibit oxygen isotope fractionation of an αO_vaterite-H2O similar to calcite, i.e., ~1.028 at 25 °C [97,98,99]. Grasby (2003) [88] found an αO_vaterite-H2O of ~ 1.035, similar to αO_calcite-H2O at 2 °C. Some support for this is suggested from the isotope fractionation of mirabilite (Na₂SO₄·10H₂O), which has an αO_{mirabilite-H2O} of 1.0020 at 0 °C [100] and for gypsum (CaSO₄·2H₂O) an αO_gypsum-H2O of 1.003–1.0041 [101,102].

Although there is evidence that the ikaite crystal was intact prior to sampling, we cannot fully eliminate the possibility that the metastable ikaite crystal had slowly decomposed in situ to vaterite, aragonite, calcite, or amorphous calcium carbonate and H₂O. In this case, the ikaite crystal initially may have had a uniform δ¹⁸O_ikaite distribution of around +1.5‰, but potential recrystallization to vaterite/calcite proceeds gradually outward from the older, center of the ikaite crystal. Re-establishment of the oxygen isotope equilibrium for the newly formed calcite near the center of the crystal, with respect to the sediment temperature of −1.4 °C, would predict an oxygen isotope fractionation between the ikaite calcite and interstitial water of 5.2‰, which is comparable to the measured 4.5‰. Recrystallization would not affect the distribution of carbon isotopes in the ikaite zonal structure owing to slow carbon exchange rates, carbon reservoir size, and conservative transfer.

4.5. Hydrogen Isotopes in Ikaite Hydration Water

Similar to the oxygen isotopes, the stable hydrogen isotopes of the interstitial water in core ANT 1138-4 remained constant, with sediment depth with a δ²H_H2O around −5.7 ± 0.9‰ (Figure 2, Table 1) and were consistent with the values for Antarctic bottom waters (δ²H_H2O of −3.3‰, VSMOW) [90]. Hydrogen isotope measurements of the ikaite crystal water were made only on bulk samples of the mineral. From a total of five measurements on two samples, the bulk hydrogen isotope ratio for the crystal water (δ²H_ikaite) was around −14.4 ± 2.9‰ (VSMOW). The technique employed here could not isolate the possible influences of inclusion water, disequilibrium growth, or postformation exchange effects on the ²H/¹H of the crystal water. The corresponding measurements of δ¹³C_DIC = −17.5‰ and δ¹⁸O_ikaite = +1.4‰ (VPDB) indicate that the bulk sample was from the outer portion of the hydrate, as noted earlier (Section 4.3). Relative to the present isotope ratios of the interstitial water in the core, the bulk ikaite sample was depleted in deuterium by 11‰, i.e., the hydrogen fractionation factor between the crystal-hydrate H₂O (ikaite) and the interstitial water (H₂O) was αH_ikaite-H2O = 0.989. Although the hydrogen isotope effects associated with the formation of ikaite are unknown, the depletion of deuterium in the hydrate crystal water over that of the interstitial water is in excellent agreement with those typically observed for several other hydrated minerals. A deuterium depletion in the crystal H₂O of various salts relative to the solution H₂O, e.g., αH_CuSO4-H2O = 0.978 (25 °C), was reported by [103]. Hydrogen fractionation for gypsum was found by [104] to be constant with the temperature: αH_gypsum-H2O = 0.985 between 17 and 57 °C. Also comparable with ikaite is gaylussite (Na₂CO₃·CaCO₃·5H₂O), a similar hydrated carbonate discussed above with respect to its carbon isotope fractionation, having αH_{gaylussite-H2O} = 0.987 at 18 °C [80]. However, the latter authors reported larger fractionations for nahcolite (estimated 20 to 90‰ depletion) and trona (calculated at −1.4 °C as a 110‰ depletion), while borax formed essentially without fractionation. In contrast to the trend observed for ikaite and other hydrates, deuterium was found to be enriched in mirabilite (Na₂SO₄·10H₂O) relative to the formation solution, αH_{mirabilite-H2O} = 1.019 [100].

5. Implications of Ikaite Occurrences

5.1. Formation Conditions and Cold Temperature Recorders

Calcium carbonate pseudomorphs, commonly referred to as thinolites, glendonites, or gennoishi, etc., have been described worldwide from environments comparable to the Bransfield Strait basin, but also from Pleistocene glacial lakes, deep-sea trenches, and pre-Cambrian continental shelves and shelf basins (for reviews see: [44,45,46,105,106,107,108]). Their pseudomorph crystal form and frequent association with organic-rich and phosphorite-containing sediments in colder or polar environments provide evidence that these carbonates are relics of recrystallized ikaites [106]. Although often speculated to be an indicator of colder paleoenvironments, ikaite pseudomorphs are not necessarily unique to such conditions. As noted earlier, [20,21] demonstrated the metastable character of ikaites and these minerals have been reported from various sedimentary environments regardless of temperature, notably the Nankai Trough [7].

The chemical literature of the l9th and early 20th century abounds with reports on hydrated CaCO₃ phases, the calcite hexahydrate being the one most commonly synthesized by artificial nucleation [3,4,5,109]. The precipitation of ikaite in preference over calcite or aragonite is not exclusively temperature-controlled, rather it was suspected that the presence of inorganic and/or organic crystallization nuclei or inhibitors may determine which carbonate phase is formed. Additives such as sucrose, protein, mollusk and arthropod blood, and polyphosphates, but also Mg, Co, and Ni, were all observed to favor hexahydrate precipitation over that of anhydrous calcium carbonate at low temperatures [2,110,111].

Early studies indicated that CaCO₃·6H₂O precipitation could also be induced without additives at subzero temperatures, under carefully chosen conditions of maintaining supersaturation [18]. Works by [20,21] showed that CaCO₃·6H₂O is a phase more stable than aragonite and calcite at high water pressures of up to 6 kbars at 25 °C. Experimental studies thus indicate that the formation of calcium carbonate hexahydrate is strongly favored by low temperatures, and that increased water pressure or certain additives, which do not necessarily enter into the crystal lattice, or both, could lead to metastable precipitation at appreciably higher temperatures. We think that herein lies an explanation for their formation in disparate temperature regimes.

Natural, sedimentary ikaite is known to occur as metastable species within the approximate temperature range from −2 to +7 °C [106]. Experimentally, ikaite has been formed at higher temperatures up to 12 °C [112] and even 35 °C [22,113,114], with and without high pH, salinity, alkalinity, and inorganic precipitation additives, such as phosphate or magnesium [9,16,115,116,117,118]. Generally, however, these crystals quickly decompose to vaterite, aragonite, and calcite [16,19,22,119,120,121].

It is well understood that orthophosphate, and in particular CaPO₄ and CaHPO₄^o, can inhibit calcite precipitation and thereby promote ikaite formation [19,122,123,124,125,126]. Phosphate concentrations in the Bransfield Strait interstitial fluids are up to 0.54 mM (Table 1), which is 100–500 times greater than the typical 1–4 µM in seawater. The threshold for phosphate inhibition of calcite formation was reported by [125,126,127] to be ~5 µM, i.e., ~100 times lower than the interstitial fluids around the SRZ (Figure 4). In contrast, experiments demonstrated that phosphate was not important for the precipitation of ikaite in Ikka Fjørd [114] or sea ice [116,120]; rather, it is the high degree of carbonate supersaturation, pH, and potentially being coupled with Mg inhibition. Interestingly, the authors of [114] also noted that hydrated Mg carbonates, i.e., lansfordite (MgCO₃·5(H₂O)) and nesquehonite (MgCO₃·3(H₂O)), may coprecipitate with ikaite.

Calcium and magnesium in the Bransfield Strait interstitial fluids range from 5.7 to 10.5 mM and from 49.6 to 53.8 mM, respectively (Table 1), which is similar to the present-day mean ocean water values (Ca 10.3 mM and Mg 53 mM [128]). Experiments by [22] at 5 °C, using a pH range of 9–10.5, and sodium ratios have shown that ikaite is formed at Mg concentrations above ~5 mM, i.e., 10 times lower than present in the interstitial fluids. Similarly, experiments by [23] showed a significant suppression of calcite formation at Mg > 5 mM. They also demonstrated that MgSO₄ and SO₄ (also in combination with Mg) substantially inhibited calcite at concentrations of ~3 mM and 10 mM, respectively. In contrast, [129] reported that the highest Mg:Ca is associated with the least-stable ikaite and the Mg may, in fact, destabilize ikaite. The molar Mg:Ca may also be an important factor for calcite inhibition and therefore promotion of ikaite formation [112]. The Bransfield Strait interstitial fluid Mg:Ca (molar) varies from 4.8 to 7.2 in the interstitial fluids (Table 1) and is similar to modern seawater with a mean Mg:Ca of 5.1 [130] and range of 4.5 to 6.5 [128].

The inhibition of calcite formation by sulfate has been proposed by several studies [22,23]. Considering that the ikaite in the Bransfield Strait forms at the base of the SRZ, which was also observed by [89], it may perhaps suggest that all carbonate formation, not just calcite, is suppressed in the SRZ. However, although it seems to contradict the support from several studies which have concluded that ikaite forms in the presence of sulfate [131], there is evidence that sulfate can enhance the inhibition effects of Mg [15,23,125,127].

Although we cannot fully resolve the parameters critical for ikaite formation in this paper, it is noteworthy that the ikaite formed in the Bransfield Strait sediments is at the depth where the phosphate concentration is at a maximum, Mg is at a minimum, and is directly beneath the transition zone of sulfate-reduction/anaerobic methane oxidation and methanogenesis (Figure 2).

5.2. Amino Acids

In addition to inorganic inhibitors, we suggest that naturally occurring organic compounds, characteristically enriched in acidic amino acids and other nonprotein amino acids, could favor the formation of ikaites in environments such as the Bransfield Strait basin.

This effect is clearly evident in the amino acid spectrum of the hydrolysable organic residue isolated from ikaite crystals (

Table 3. Amino acids of hydrolysable residue from ikaite crystals, plankton, and sediments from the Bransfield Strait, Antarctica. Concentrations in µg/g.

Amino	Plankton	Sediment	Ikaite
acid	residues/1000
cyst	3	3	94 **
tau *	1	1	28 **
meto	3	3	40 **
asp	108	142	209 ****
thr”	57	62	42
ser”	75	87	72
glu	101	79	61 ****
alpha-AAA	<1	< 1	8
gly	87	212	242 ****
ala	106	93	44
val	71	52	29
met	13	9	6
alpha-APA	<1	< 1	3 ***
allo-ileu	<l	1	1
ileu	56	33	11
leu	105	47	10
tyr	28	15	3
phe	53	27	8
beta-ala	3	10	77 *,**
gamma-ABA	<1	4	7 **
orn	<l	5	8 **
lys	70	60	64
his	15	13	9
arg	45	44	4

Concentrations in µg/g; * complex mixture with taurine; ** degradation products: beta-ala from aspartic acid taurine, cysteic acid from cysteine and methionine, gamma-ABA from glutamic acid, ornithine from arginine, and/or bacterial biomass: alpha-DPA and alpha-AAA; *** beta-ala: here very likely from degradation of aspartic acid; **** amino acids in Ca-carbonate-secreting organic templates; amino acids of silica-secreting tissue concentrated in Bransfield Strait plankton.

, Figure 8). Its outstanding feature is the dominance of aspartic acid, glutamic acid, and glycine. Together, they constitute more than 50 mol %. If we include beta-alanine, which is considered a degradation product of aspartic acid, then they account for more than 60% of all amino acids. Such dominance is diagnostic for the composition of biogenic carbonate-secreting tissues as well as for selectively sorbed organics thought to constitute organic templates, which induce inorganic calcite precipitation from seawater [132,133]. The potential influence of cold-active/alkaline-active enzymes, such as phosphatase, α-galactosidase, protease, α-amylase, β-galactosidase, and β-glucanase, on ikaite formation is also uncertain, but they have been found together [134,135,136]. In addition, certain rare amino acids from bacterial metabolism (alpha-AAA, alpha-APA, and gamma-ABA) were also present in significant quantities in the ikaite residue. It is likely that they originate from the methanogenic microbial population present in the Bransfield Strait sediments. The presence of these amino acids may eventually constitute a tracer for this particular diagenetic environment. Peculiar to the Bransfield Strait are the relatively high amounts of serine and threonine, two amino acids known to be concentrated in biogenic silica-secreting tissues. These are not unexpected in the diatom-rich plankton and sediments of the high latitudes. We are uncertain at this stage if any diagnostic amino acids are preserved in ancient pseudomorphs or glendonites. It certainly would be a worthwhile project to search for such organic tracers in ancient ikaites/glendonites.

5.3. Glendonites

In an initial attempt to trace to ancient environments, the isotope signal established during the formation of ikaite, and its subsequent recrystallization, we profiled two samples of calcite pseudomorphs (glendonites from Spitzbergen and Canada) for δ¹³C_glendonite and δ¹⁸O_glendonite. The carbon isotope profiles across the two pseudomorphs show that the δ¹³C_glendonite became progressively ¹³C-depleted, moving from the inside to the outside of the sample (Figure 9, Table 2). The δ¹⁸O_glendonite of the two samples became enriched in ¹⁸O from inside to outside. Although the magnitude of the C- and O-isotope shift in the glendonites is similar to that of the Bransfield Strait ikaite, the values of δ¹³C and δ¹³⁸O differ. In particular, the ikaite became ¹³C-enriched, moving from inside to outside. Analogously, Rogov et al. (2017) [137] showed profiles across glendonites with δ¹³C_glendonite from ~−21 to −17‰ and δ¹⁸O_glendonite ranging from ~−3 to −12‰ in response to changing diagenetic fluids. Wang et al. (2020) [46] reported spot analyses of four individual glendonites. Samples from the interior to exterior had δ¹³C_glendonite changing from −27.9‰ to −1.7‰ and δ¹³⁸O_glendonite from −9.4‰ to −12.5‰.

The δ¹³C_carbonate and δ¹⁸O_carbonate for ikaites and glendonites worldwide of the Phanerozoic age have been compiled by [46,138]. The isotopes measured in these papers are measured on bulk, spatially undifferentiated samples, i.e., not profiles of crystals or pseudomorphs. Thus, they may not be directly comparable to the more systematic inter-glendonite transects we report here. Figure 10 shows the very wide spread in isotope values for ikaites and glendonites with δ¹³C_carbonate ranging from −45 to +5.8‰ and δ¹⁸O_carbonate ranging from −17 to +0.3‰ (

Table 4. Maximum and minimum carbon and oxygen isotope ratio values of glendonites in databases from Rogov et al. [138] and Wang et al. [46].

	δ13Ccarbonate (‰, VPDB)		δ18Ocarbonate (‰, VPDB)
	Minimum	Maximum	Minimum	Maximum
Rogov et al., 2021 [138]
Paleozoic	−27.7	0.4	−23.3	1.6
Mesozoic	−44.6	−0.1	−15.1	0.0
Cenozoic	−31.9	8.5	−17.0	4.9
Wang et al., 2021 [46]
Paleozoic	−27.7	0.6	−16.9	0.3
Mesozoic	−45.2	0.0	−14.7	0.0
Cenozoic	−25.2	5.8	−11.6	−0.8

). The most ¹²C-depleted glendonites (−30 to −45‰) are clearly strongly influenced by DIC derived from methane oxidation and have invariant δ¹⁸O_carbonate of around 0 to −5‰. Otherwise, in Figure 10, to a first approximation, as the ikaite and glendonite δ¹³C_carbonate values become more ¹³C enriched, they are accompanied by δ¹⁸O_carbonate depleted in ¹⁸O. Although both the δ¹³C_carbonate and δ¹⁸O_carbonate are responding to changing diagenetic fluid isotope compositions and environmental conditions, e.g., temperature, the explanation for this trend is not immediately apparent. It is unclear whether this glendonite δ¹³C_carbonate–δ¹⁸O_carbonate behavior is coincidental or whether there is a causal, sympathetic mechanism that systematically enriches/depletes the carbonate ¹⁸O and ¹²C. Similar trends have been reported for glendonites in Siberia [139]. We suspect that these trends are controlled more by the environmental conditions under which the ikaite structure changes to calcite. Specifically, the trend in oxygen isotopes would be reversed if such recrystallization occurred under higher or lower temperatures than that at which the initial ikaite formed. The carbon and oxygen isotopes would respond if the change was accompanied by a drastic change in fluid chemistry, such as from meteoric waters or fluids charged with different partial pressures of DIC.

As in the case of ikaite formation and decomposition, kinetics can substantially influence the isotope composition [91,102,140,141,142]. The comprehensive Phanerozoic glendonite databases from [46,138] clearly show the wide range in δ¹³C_glendonite (~−45 to +8.5‰) and δ¹⁸O_glendonite (~23 to +5‰) (Table 4) that also encompass our ikaite and glendonite results, but not our more recent ones reported elsewhere [46] (Figure 10). In addition, [107] generated a very similar δ¹⁸O_carbonate vs. δ¹³C_carbonate plot, largely using the same ikaite and glendonite data as in Figure 10. At this point, it is reasonable to make the interpretation that the most strongly ¹³C-depleted δ¹³C_glendonite values reflect the ¹²C-rich DIC derived from the aerobic or anaerobic oxidation of ¹²C-enriched microbial methane [11,14,139,143,144]. Subsequently, the resultant glendonites are influenced by DIC generated by diagenetic sources. The δ¹⁸O_glendonite reflects diagenetic and other alteration influences, but the explanation appears more complicated and calls for further investigations.

There are many indications that low temperatures, such as of a polar environment, climate-related cold spells, high calcium carbonate supersaturation caused by interstitial methanogenesis, and a sufficiently large supply of dissolved organics containing acidic amino acids, favor metastable ikaite formation. Under these conditions we speculate that there might be a partial trade-off, such that an insufficient supply of dissolved organics is compensated for by extremely low temperatures in order for ikaite to form or that higher temperatures may permit ikaite formation only under total pressures of several hundred bars and abundant dissolved organics. The large crystal size and macroscopic dimensions of the pseudomorphs allow a detailed reconstruction of the events of ikaite formation and recrystallization.

There is continued evidence and support that glendonites can serve as cold environmental indicators, such as glaciomarine settings [45,46,95,145,146,147,148,149,150,151,152,153]. However, there are some concerns that the δ¹⁸O_glendonite of glendonites are not robust recorders of low-temperature environments, due to various processes such as dissolution and re-precipitation [129,137,154]. There is also uncertainty with the formation of ikaite and its subsequent conversion to calcite, including possible intermediates and mixtures with monohydrocalcite, vaterite, and aragonite [7,118,155].

It is still a puzzle as to why we only encounter ikaite crystals at ~720 cm bsf, possibly as a “pavement” in the Bransfield Strait. Outstanding questions remain in finding a convincing explanation for why ikaite stops growing in the sediment below ~385 cm bsf and why, in the case of the Bransfield Strait, new ikaite crystals are not continuously and currently forming in shallower portions of the sediment package at the base of the SRZ.