The δ30Si peak value discovered in middle Proterozoic chert and its implication for environmental variations in the ancient ocean

The silicon isotope composition of chert has recently been used to study the historic evolution of the global ocean. It has been suggested that Precambrian cherts have much higher δ30Si values than Phanerozoic cherts do and that the former show an increasing trend from 3.5 to 0.85 Ga, reflecting a decrease in ocean temperatures. However, cherts have various origins, and their isotopic compositions might be reset by metamorphic fluid circulation; thus, different types of cherts should be distinguished. Here, we present a new set of δ30Si data for cherts from early and middle Proterozoic carbonate rocks from Northern China. We found that cherts of 1.355–1.325 Ga show a peak range of 2.2–3.9‰. Based on these results, we propose that from the Archean to the middle Proterozoic, there was a drastic decrease in silicon content and an increase in the δ30Si value in ocean water due to a temperature decrease and biological activity increase. After that period, the silicon content of the ocean was limited to a low level by a high degree of biological absorption, and their δ30Si values varied in a small range around a significantly lower value.

The silicon isotope composition of chert has recently been used to study the historic evolution of the global ocean. It has been suggested that Precambrian cherts have much higher δ 30 Si values than Phanerozoic cherts do and that the former show an increasing trend from 3.5 to 0.85 Ga, reflecting a decrease in ocean temperatures. However, cherts have various origins, and their isotopic compositions might be reset by metamorphic fluid circulation; thus, different types of cherts should be distinguished. Here, we present a new set of δ 30 Si data for cherts from early and middle Proterozoic carbonate rocks from Northern China. We found that cherts of 1.355-1.325 Ga show a peak range of 2.2-3.9‰. Based on these results, we propose that from the Archean to the middle Proterozoic, there was a drastic decrease in silicon content and an increase in the δ 30 Si value in ocean water due to a temperature decrease and biological activity increase. After that period, the silicon content of the ocean was limited to a low level by a high degree of biological absorption, and their δ 30 Si values varied in a small range around a significantly lower value.
Chert is a sedimentary rock composed mainly of microcrystalline quartz. It occurs in sedimentary strata from the early Archean to the present. Its chemical and isotopic characteristics have been widely investigated to study the conditions of its formation. Recently, Si isotopes in chert have been used as an important tracer of environmental conditions in the global ocean [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18] . Song and Ding (1990) suggested distinguishing sedimentary facies of chert with silicon isotope compositions 1 . Ding et al. (1996) indicated that different genetic types of cherts have different silicon isotope characters 2 . They discussed variations in silicon content and silicon isotope composition in marine water according to the data from chert formed in shallow marine environments. According to the silicon isotope variation of chert, Robert and Chaussidon (2006) developed a temperature-evolution curve for the Precambrian ocean 3 . It has been suggested that Precambrian cherts have much higher δ 30 Si values than Phanerozoic cherts and that the former show a generally increasing trend from 3.5 to 0.85 Ga, thus reflecting a decrease in seawater temperature 3 . However, these statements have been challenged because cherts can have various origins and their isotopic compositions might have been reset by metamorphic fluid circulation [4][5][6][7][8][9] . Thus, different types of cherts need to be considered separately 4,[9][10][11][12] because their isotope composition might record various distinct processes [13][14][15][16][17][18] . In addition, the factors that affect silicon isotope composition of chert have been investigated in detail [13][14][15][16][17][18] . It was found that peritidal cherts are enriched in 30 Si, but that basinal cherts, which are associated with banded iron formations (BIF), are depleted in 30 Si; this difference is attributed to Si having been derived from hydrothermal sources in the BIFs 10 . Now, it is widely accepted that the formation of chert is a complicated process and that the Si isotope composition of chert is dependent on the relative contributions of various Si sources and the effects of different forming processes [9][10][11][12][13][14][15][16][17][18] . However, the relative contributions of the various Si sources to the global ocean vary throughout geological history. Thus, specific types of chert might be representative for reconstructing the Si isotope composition of the ocean for different geological periods. For example, during the Archean period, sea-floor weathering and submarine hydrothermal fluids dominated the Si input to the ocean, and continent inputs were negligible; thus, the Si isotope compositions of quartz bands in BIF are likely the best approach for tracing the Si isotope compositions of the ocean. In contrast, in the Proterozoic and Phanerozoic periods, the Si

Results
The Si and O isotope compositions of the cherts and the C and O isotope compositions of dolomites obtained in this study are listed in Table 1 and shown in Fig. 4. All isotope compositions are reported in delta-notation relative to a standard material, i.e., with an average of 23.0 ± 3.5 (1 SD) ‰. These results also show that they were formed in a shallow marine environment but might have been slightly affected by diagenesis.

Discussion
The increasing trend of δ 30 Si from the early Proterozoic to middle Proterozoic is consistent with the trends reported by previous studies 2,3,5-10 . However, together with the new data provided in this study, the middle Proterozoic δ 30 Si peak becomes a prominent feature in the Si isotope record of Precambrian cherts, which implies major environmental variations in the ancient ocean (Fig. 4).
Combining data obtained using SiF 4 and MC-ICP-MS methods in this study and previous studies 10, [24][25][26][27][28][29][30] , the δ 30 Si variation trend from the late Archean to present is plotted for chert that formed in shallow marine environments (Fig. 5). The upper limit of δ 30 Si values of chert increases gradually from 1.8‰ at 2.53 Ga to 3.9‰ at 1.335 Ga and then decreases drastically to 2.0‰ at 1.104 Ga. After 1.104 Ga, the upper limit of δ 30 Si values for chert fluctuates between 1.5‰ and 2.5‰.
Chert is normally formed by the recrystallization of a precipitated amorphous silica precursor in the diagenetic process. To test the possibility of using O and Si isotope compositions of chert to trace those of contemporary   marine water, the relationship between the δ 18 O and δ 30 Si values of chert and those of the amorphous silica precursor must be evaluated first. It is known that the O isotope composition of silica would be reduced during diagenesis due to two factors. First, the chert nodular and band in the limestone are considered to form in the groundwater of mixed meteoric-marine coastal systems during diagenesis 31 . Due to the involvement of meteoric water, the O isotope composition of groundwater would be lighter than that of contemporary marine water, causing a reduction in the δ 18 O value of chert to some extent. Second, as the diagenetic temperature is normally higher than that of marine water, the O isotope fractionation between silica and water at the diagenetic stage would be smaller than that in the precipitation stage, which would cause a δ 18 O reduction in the chert. As shown by the microscopic and SEM examinations (Fig. 3), the slight reduction in the δ 18 O value of chert indicates that the studied cherts were formed in the diagenetic process and their δ 18 O values cannot represent those of the amorphous silica precursor. In contrast to the O isotope composition, the silicon isotope composition would not change during early diagenesis for following reasons: (1) Chert bands and nodules are commonly confined in layers of sedimentary strata, which indicate that silica does not move over long distances during diagenesis.   relative Si isotope enrichment of chert to seawater. According to the theory of isotope fractionation, Δ 30 Si Ch-SW is temperature dependent. Thus, to determine a proper Δ 30 Si Ch-SW value, precipitation temperatures of the amorphous silica precursor of chert should be evaluated beforehand.
Based on the O isotope composition of chert, Robert and Chaussidon (2006) 3 suggested the seawater temperatures were 70 °C at 3.5 Ga and 35 °C at 0.85 Ga. However, the inferred high Paleoarchean temperatures are controversial [34][35][36][37][38] , partly because δ 18 O determinations of the Paleoarchean temperature rely on the assumption that δ 18 O of the Archean ocean was similar to that of an ice-free modern ocean. However, it has been suggested by a number of researchers that the δ 18 O value of the global ocean could have varied significantly over time [35][36][37][38] . Recently Hren et al. 34 Northrop and Clayton (1966) 40 , we obtained a diagenetic temperature range of 25~56 °C (averaging 36 ± 7 °C) for dolomite in the early Proterozoic and a diagenetic temperature range of 12~50 °C (averaging 26 ± 9 °C) for dolomite in the middle Proterozoic. Assuming the diagenetic water still has a δ 18 O value of − 10‰, and using the O isotope fractionation equation (1000lnα Chert-H2O = 3.09 × 10 6 T −2 − 3.29) of Knauth & Epstein (1975) 41 , we obtained a diagenetic temperature range of 19~43 °C (averaging 34 ± 7 °C) for chert in the early Proterozoic and 1~63 °C (averaging 17 ± 15 °C) for chert in the middle Proterozoic. The calculated average diagenetic temperature (34 °C) for early Proterozoic chert is slightly lower than that (36 °C) for early Proterozoic dolomite, and the calculated average diagenetic temperature (17 °C) for middle Proterozoic chert is significantly lower than that (26 °C) for middle Proterozoic dolomite. These observations may be caused by a difference between chert and dolomite during O isotope exchange process with diagenetic solution. The cherts are more resistant to O isotope exchange than dolomite in the diagenetic process. Thus, the calculated diagenetic temperature for dolomite may be more representative. Based on these considerations and assuming the diagenetic temperature is a little higher than the sedimentary temperature on the sea floor, we estimate the ocean temperature as 30 °C in the early Proterozoic and 20 °C in the middle Proterozoic.
Concerning Δ 30 Si Ch-SW , there has been a number of investigations on Si isotope fractionation during abiotic silica precipitation [15][16][17][42][43][44][45] . Early experimental studies on abiotic solid-fluid silicon isotope fractionation yielded Δ 30 Si solid-fluid values ranging from − 2.0‰ to − 1. Assuming that the temperature of the ocean is 40 °C, 30 °C and 20 °C in the Archean, early Proterozoic and since the middle Proterozoic, respectively, and Δ 30 Si Ch-SW are − 0.5‰, − 1.0‰ and − 1.2‰ in the Archean, early Proterozoic and since the middle Proterozoic, respectively, δ 30 Si values of ocean water are calculated from δ 30 Si values of chert (Fig. 5). Figure 5 shows that the upper limit of inferred δ 30 Si NBS-28 values in ocean water increases gradually from 2.8‰ at 2.53 Ga to 5.1‰ at 1.335 Ga and then decreases drastically to 3.2‰ at 1.104 Ga. After 1.104 Ga, the upper limit of inferred δ 30 Si NBS-28 values in ocean water fluctuates between 2.7‰ and 3.7‰.
As shown in Fig. 6, during the Archean period, the input sources of dissolved Si to the ocean are submarine hydrothermal fluid and sea-floor weathering, and the output paths are chemical precipitation (to form C cherts) and silicification of the precursor sediments or rocks (to form S cherts) 4-6 . The Si concentration in ocean water remains in its saturated concentration at a given temperature, but the δ 30 Si SW increases gradually due to Si isotope fractionation between dissolved Si and precipitated SiO 2 . When a steady state is reached, δ 30 Si Out (the average δ 30 Si value of all output Si) will be equivalent to δ 30 Si In (the average δ 30 Si value of input Si), and δ 30 Si SW will be equal to (δ 30 Si Out -Δ 30 Si Out -SW ), where Δ 30 Si Out -SW is the relative silicon isotope enrichment of the output Si to the ocean water (δ 30 Si Out -δ 30 Si SW ). Because the average δ 30 Si value of Si in the submarine hydrothermal fluid and Si from sea-floor weathering is ~− 0.3‰ and Δ 30 Si Out-SW is ~− 0.5‰ at a temperature of 40 °C 15 , the δ 30 Si SW value of that period is approximately 0.2‰.
The Si cycle in the modern ocean (Fig. 6) is quite different 46 . First, dissolved Si from the continents (in rivers 47 and groundwater) has become a dominant input source (6.4 Tmol Si/a) to ocean Si, and the Si input from submarine hydrothermal fluid (0.6 Tmol Si/a) and sea-floor weathering (1.9 Tmol Si/a) has become less significant 46 . δ 30 Si in is calculated as ~0.78‰ using the equation In the equation, f represents the relative fraction of each Si source and the subscripts Cont, SFW and SHF indicate continent, sea-floor weathering and submarine hydrothermal fluid, respectively. Second, the biological absorption of Si has become a dominant path for Si output from the ocean and Si contents in modern ocean water (0.05 mg/L~0.2 mg/L for shallow seawater and 0.3 mg/L~3.5 mg/L for deep seawater) are 2 orders of magnitude lower than those of the saturated concentration in seawater 2 . At steady state, the Scientific RepoRts | 7:44000 | DOI: 10.1038/srep44000 amount and δ 30 Si value of output Si from ocean water would be equal to those of input Si. Thus, δ 30 Si Out would also be ~0.78‰ at present.
The silicon isotope fractionations of diatoms-seawater and sponges-seawater have been experimentally studied. The determined Δ 30 Si Diatom-SW is commonly − 1.0∼ − 1.1‰ [48][49][50] , but Δ 30 Si Sponge-SW varies from − 1.1‰ to − 3.7‰ 51,52 . Because diatoms are much more abundant in the ocean than sponges, we assume Δ 30 Si Out-SW in the modern ocean is ~− 1.2‰. From the above estimation, the δ 30 Si SW value of the modern ocean can be calculated as ~1.98‰, which is very close to the value (1.9‰) of surface ocean water inferred previously 53 .
Many papers have reported the δ 30 Si values of cherts formed during the Proterozoic [1][2][3][8][9][10][24][25][26]32 , but no model of the Si cycle in the Proterozoic ocean has been presented. Here, we present a conceptual model for the Si cycle of the Proterozoic ocean based on known and inferred boundary conditions (Fig. 6). Similar to the conditions in the Archean ocean, submarine hydrothermal fluid and sea-floor weathering are still important input sources of dissolved Si to the Proterozoic ocean, but the amounts of these inputs decreased as the hydrothermal activity and ocean temperature decreased from their Archean to Proterozoic values. Further, the input of dissolved Si from the continents became significant as supercontinents appeared in the early Proterozoic. For the output of dissolved Si from the ocean in the Proterozoic Eon 54 , chemical precipitation was still a major pathway, but biological absorption may have also played a significant role.
The Si concentration in ocean water remains in its saturated concentration at a given temperature (30 °C for the early Proterozoic and 20 °C for the middle and late Proterozoic). When a steady state is reached, δ 30 Si Out will be equivalent to δ 30 Si In , and δ 30 Si SW will be equal to (δ 30 Si Out − Δ 30 Si Out -SW ), where Δ 30 Si Out -SW is − 1.0‰ for the early Proterozoic and − 1.2‰ for the middle and late Proterozoic. The estimated δ 30 Si SW value of that period would be approximately 1.47‰ (Fig. 6).
From the discussion above, extreme values of δ 30 Si for chert and δ 30 Si for seawater cannot be explained for an ocean at steady state conditions. Thus, this peak in δ 30 Si indicates an extraordinary period at non-steady state suggesting a scenario at a transition stage.
From the Archean to present, there should be a transition period of the Si cycle in the ocean. In that period, the D Si in ocean water is reduced by approximately 2 orders of magnitude below that of the saturated concentration in seawater 2 , and the δ 30 Si SW value first rises from ~0.2‰ to ~5.1‰ and then decreases to ~1.98‰. The rise of δ 30 Si SW is caused by Rayleigh fractionation when SiO 2 precipitates from ocean water (Fig. 7). One mechanism that causes the D Si reduction in ocean water should be a decrease in ocean temperature. As temperature decreases, the saturated Si concentration in ocean water would be reduced, causing additional SiO 2 precipitation. The saturated SiO 2 concentration in ocean water is ~363.5 mg/L, ~221.2 mg/L and ~178.9 mg/L at 40 °C, 30 °C and 20 °C, respectively 55 . When the temperature of seawater decreases from 40 °C to 20 °C, the fraction of dissolved Si remaining in the seawater (f) will be reduced to ~0.687. In the Rayleigh fractionation process, it will cause an increase of ~0.2‰ in δ 30 Si SW . It seems that the decrease in seawater temperature alone cannot explain the significant increase in the δ 30 Si SW value, and other mechanisms should be considered. Another mechanism causing a D Si decrease in ocean water is the increase of Si absorption activities by biological species, which can reduce the D Si in ocean water 2 orders of magnitude lower than that of the saturated concentration. In the Rayleigh fractionation process, the combined effect of these two types of mechanisms can cause δ 30 Si SW to increase ~4.0‰ when f is reduced to 0.01. It is known that diatoms and radiolarians are Si-fixing organisms that were active in the Phanerozoic 46 ; sponges were active in the Phanerozoic 46 and the later Proterozoic 54 . Assuming their appearance is the start of a drastic decrease of D Si in ocean water, we should observe a δ 30 Si SW peak value in the later Proterozoic or early Phanerozoic. However, according to the data here, the δ 30 Si SW peak appeared in the middle Proterozoic (1.325~1.355 Ga) instead, which indicates the drastic decrease in ocean water D Si happened prior to 1.355 Ga.
It is known that microbes were the dominant biological species in the Precambrian. Stromatolites are found in early Archean strata from 3.5 Ga 56 and are very well developed in Proterozoic strata [19][20][21][22][23]57 . REE and C, O, Nd isotope compositions have been used to study the formation conditions of stromatolite-bearing sediments, particularly the effect of biological activities 22,23,[56][57][58] . The early and middle Proterozoic chert-bearing dolomites investigated in this study are all rich in stromatolites, showing a close correlation between silica precipitation and biological activities. Moreover, macroscopic eukaryotic fossils were recently discovered in the 1.56 Ga Gaoyuzhuang Formation in the Yanshan area of Northern China 59 . If some Proterozoic species are capable of absorbing or precipitating Si from ocean water, the Si content in the Proterozoic ocean water would be drastically reduced causing δ 30 Si SW to rise significantly. Thus, the high peak in δ 30 Si SW values in the middle Proterozoic ocean water may reflect a drastic reduction in Si content caused by a rapid increase in biological activity in the ocean. After that peak period, the D Si reduction rate in ocean water decreased gradually, and the δ 30 Si SW value decreases to a significantly lower value at steady state.

Methods
Si and O isotope analysis of chert samples. For Si and O isotope analysis, ~100 mg of a chert band or nodule was selected from a polished section of each specimen. The sample was crushed and ground to a powder of − 200 mesh. Then, the sample powder was reacted with 6 N HCl in Teflon beakers to dissolve small amounts of carbonate. The remainder was washed at least in triplicate with Milli-Q water. Then, the remainder was transferred to a Pt crucible, dried at 105 °C in an oven and then calcined at 1000°C in a muffle furnace to remove organic C impurities. The δ 30 Si of the starting ocean water is assumed to be 0‰ and the Si isotope fractionation factor between SiO 2 precipitate and ocean water (α Pre-SW ) is 0.999.
Oxygen isotope analyses were carried out using the BrF 5 method (Clayton and Mayeda, 1963) 60 , and silicon isotope analyses were carried out using the SiF 4 method (Ding, 2004) 61 . Approximately 10 mg of pretreated chert was placed in a Ni reactor in a metal vacuum line and reacted with BrF 5 at a temperature of approximately 500°C to produce gaseous O 2 and SiF 4 .
The O 2 gas was separated from SiF 4 , BrF 5 and BrF 3 by evaporating at liquid nitrogen temperature. Then, O 2 gas was converted to CO 2 by reacting with a carbon rod at 700 °C. Finally, the CO 2 gas was collected for O 2 isotope measurement.
SiF 4 was separated from the BrF 5 and BrF 3 by evaporating at dry ice-acetone temperature. The separated SiF 4 was purified further by passing it through a Cu tube containing pure Zn particles at a temperature of 60°C. This procedure removed trace amounts of the remaining active F-bearing compounds (BrF 5 and BrF 3 ). Then, the purified SiF 4 was collected for silicon isotope measurement.
The isotopic measurements were carried out with a MAT-253 mass spectrometer.
For O 2 isotope measurement, the NIST Standard Reference Material for O isotopes (NBS-28) was used directly as the working standard in this study. The precision of the O isotope measurement is better than ± 0.2‰ (2σ ). The O isotope compositions of all samples are reported as δ 18 O values relative to the V-SMOW standard.
For Si isotope measurements, international reference material for Si isotopes (NBS-28) and two Chinese national standards for Si isotopes (GBW04421 and 04422) were used as working standards in this study. The long-term reproducibility of the silicon isotope measurements is better than ± 0.1‰ (2σ ). The silicon isotope compositions of all samples are reported as δ 30 Si values relative to the NBS-28 standard.
C and O isotope analysis of dolomite samples. The continuous-flow isotope-ratio mass spectrometric method was used for C and O isotope analysis of dolomite 62 . The system consists of a Thermo-Finnigan GasBench II equipped with a CTC Combi-Pal auto-sampler and linked to a Finnigan MAT 253 mass spectrometer.
Approximately 100 mg of dolomite was taken from the specimen and ground to a powder of − 200 mesh. Approximately 100 μ g of dolomite powder was loaded manually into a 12 ml round-bottomed borosilicate exetainer and sealed using butyl rubber septa. Four national reference materials (GBW04405, GBW04406, GBW04416 and GBW04417) were routinely loaded. The exetainers were automatically flushed with grade 5 He by penetrating the septa using a double-hole needle at a flow rate of 100 mL/min. Then, 4-6 drops of phosphoric acid were deposited in each exetainer. The exetainers were placed onto an aluminium tray and kept at 72 °C for 24 h. Subsequently, the sample gas was introduced into the mass spectrometer through the standard 100 μ L sample loop, CO 2 was separated from other components using a gas chromatographic column heated to 70 °C, and the peak corresponding to this CO 2 was passed via an open split to the mass spectrometer.
The calculated external precision is typically ± 0.2‰ (2σ ) for δ 13 C and δ 18 O. The C isotopic compositions of the dolomite samples are reported as δ 13 C values relative to the V-PDB standard. The O isotopic compositions of the dolomite samples are reported as δ 18 O values relative to V-PDB and V-SMOW standards.