Provenance of Chinese Loess : Evidence from Stable Lead Isotope

Twenty-seven samples of typical loess and paleosol strata collected in nine different regions of the Chinese Loess Plateau (CLP) were fractionated into PM1.0, PM2.5, PM10 and Total Suspended Particulates (TSP) (particulate matter with aerodynamic diameters less than 1.0, 2.5, 10 and ~30 μm, respectively) by a resuspension chamber at the Desert Research Institute (DRI; Reno, NV, United States). The amounts and isotope ratios of lead (Pb) were quantified in the loess samples. Our size-segregated analysis demonstrated that the Pb isotopic composition in the loess-paleosol deposits was preserved after grain-size sorting and that therefore the isotope ratio can serve as a proxy for source tracing. A similar pattern of Pb isotope ratios was observed for sediment collected from potential source regions and the loess samples suggested that the Gobi and deserts in southern Mongolia and northern China are major sources for the deposits in the CLP. No significant deviation of Pb isotope amount was found between the nine samples of loess and paleosol strata, implying the stability of loess sources during the glacial and interglacial regime.


InTRoduCTIon
Loess is recognized as an important paleoclimate archive, presenting a terrestrial equivalent to the deep-sea oxygen isotope record of Quaternary glacial-interglacial cycles (Porter 2001).The loess-paleosol sequence in the Chinese Loess Plateau (CLP) has long been documented as a continuous history of dust activity over 2.4 Ma (Liu 1985;An et al. 1991).Due to its eolian origin, the loess in the CLP is one of a few direct records of paleowinds that not only aid greatly in reconstructing synoptic-scale paleoclimatology (Porter and An 1995), but also provide climate information on the dust-generated area (Sun and An 2002).
The upwind regions, such as the Gobi (stony desert) and sand deserts in the northern and northwestern China are commonly regarded as potential loess sources to the CLP (Liu 1985).However, large divergences regarding the specific origins of the loess have recently been reported.
For instance, previous studies suggested that the loess is a highly mixed material from the Gobi and other deserts near the CLP and inland basins in northern and northwestern China (Gallet et al. 1996;Jahn 2001); however, Zhang et al. (1997) claimed that it originated from several source regions, e.g., western deserts (mainly the Taklamakan Desert) and northern low-and high-dust deserts (such as the Badain Juran and Tengger deserts).Comparisons between isotopes, elements and mineral compositions of fine fractions of the loess in the CLP showed that the Gobi in south Mongolia and the adjoining Gobi and sand deserts (e.g., the Badain Jaran Desert, Tengger Desert, Ulan Buh Desert, Hobq Desert and Mu Us Desert), rather than the three inland basins (including Tarim, Tsaidam and Jungar), are the major loess source regions (Sun 2002).In addition, strontium (Sr) and neodymium (Nd) isotopes in fine fractions of the Chinese loess also indicated that the north edge of Tibet (e.g., the Tsaidam Basin, the Badain Juran and Tengger deserts) is the source region which mainly contributed to the loess in the Terr.Atmos. Ocean. Sci., Vol. 22, No. 3, 305-314, June 2011 CLP (Chen et al. 2007).Based on the results of electron spin resonance signal intensity analysis and crystallinity index (CI) measured from fine-grained quartz, Sun et al. (2008) suggested that both the Gobi desert in southern Mongolia and the sandy deserts in northern China (primarily the Badain Juran and Tengger deserts) are the major sources of eolian loess in the CLP during the last glacial-interglacial cycle.
Even though there were no significant variation of the sources reported during the deposition of loess and paleosol in the CLP (Gallet 1996;Jahn 2001), Zhang (1997) pointed out that the sources of Chinese loess varied between the north and the west source regions during the glacial and interglacial regimes.Sun (2008) further revealed a transition of the loess sources from the Gobi desert in southern Mongolia during the glacial period to the sandy desert in northern China during the interglacial regime.Such discrepancies have instigated a series of intensive researches to obtain a full understanding of the provenance of loess in the CLP as well as their climatic and environmental implications.
Lead (Pb) isotopic composition is a powerful tool for source identification because both minerals and rocks have distinct Pb isotope ratios, subject to their geological derivations.In addition, the Pb isotope ratios would not be affected, to a measurable extent, by physical or chemical fractionation processes (Bollhöfer andRosman 2000, 2001;Veysseyre et al. 2001).Application of Pb isotopes as tracers has been successfully differentiated silt sources for the late Wisconsin loess in the central Great Plains of eastern Colorado (Aleinikoff et al. 1999).Comparisons of Pb isotopic composition of Chinese loess with the ice core and marine sediments also assisted establishment of linkage between the provenance of those sediments and the Asian inland regions (Biscaye et al. 1997;Jones et al. 2000;Pettke et al. 2000;Godfrey 2002;Stancin et al. 2006;Klemm et al. 2007).In addition, the Pb isotopic ratios have been recently demonstrated the relationship of the provenance changes (Sun and Zhu 2010).
However, loess is derived from protoliths, of which the geochemical characteristics are size-dependent.Since loess was transported by winds of different strength or underwent different transport distances in geological times, its geochemical characteristics may be changed by grain-size sorting.Previous studies on the provenance of loess using Pb isotopes focused mainly on bulk samples, and therefore no data about the distribution of Pb isotopic compositions among different size fractions of Chinese loess could be referenced.No one knows whether there is any systematic change in the source area which would be indicated by variations in wind strength or distance to the source or wind direction.
In this study, samples were collected in nine loess and paleosol strata at the Xi Feng Loess Profile.The Pb isotopes in different size fractions of loess were precisely quantified.
The objectives of this paper are to: 1) evaluate impacts of grain-size sorting on Pb isotopic compositions; and 2) track potential source regions of loess over the CLP with the Pb isotopic ratios.

ExPERImEnT mEThodS
The CLP is one of the most extensive areas of loess deposition in the world.It spans an area of approximately 440000 km 2 , predominantly in the provinces of Shanxi, Shaanxi and Gasnu, between 33 -40°N and 98 -115°E (Liu 1985) (Fig. 1).Xi Feng (35°45'N, 107°49'E), a city of Shaanxi province, is located in the central CLP.Its profile contains more than 30 loess-paleosol layers and represents an exceptionally long record of the Asian dust deposits for 2.5 Ma.Loess samples were taken from nine representative loess and paleosol horizons, which are respectively named as S0, L1LL1, L1LL2, S1, L2, S2, S5, L9, and L15.Three parallel samples (~2 kg) were collected in each horizon.The loess samples were sieved through Tyler 30, 50, 100, 200, and 400 mesh sieves to obtain ca. 5 g of materials for resuspension.
The sieved loess samples were re-suspended in a chamber facility at the Desert Research Institute (DRI) at Reno, NV, United States.The experimental procedures are described in detail elsewhere (Chow et al. 1994).In the resuspension chamber, the loess were fractioned into four sizes, including PM 1.0 , PM 2.5 , PM 10 and Total Suspended Particulates (TSP) (particulate matter with aerodynamic diameters less than 1.0, 2.5, 10 and ~30 μm, respectively), and collected onto 47 mm Teflon-membrane filters (2 mm pore size, Pall Sciences, Ann Arbor, MI, USA).
Amounts of Pb isotopes in the re-suspended samples were measured.The analytical procedures were described as follows.A whole filter sample was cut into pieces and transferred to a clean digestion tube.Fifty microliters of ethanol plus 4 ml of a 1 : 1 nitrite acid (HNO 3 ) and water (in volume ratio) were added into the tube.The sample was gradually digested in an oven at 95°C for 65 minutes.After cooling down to room temperature, the digestion solution was brought up to 30 ml with water.Reagent blanks and control samples were prepared in the same way.The Pb isotopes were quantified using an inductively coupled plasmamass spectrometer (ICP-MS, X Series, Thermo Elemental).Two thallium isotopes ( 203 Tl and 205 Tl) were used as internal standards (IS) to correct for any instrumental drift.A digested NIST SRM 981 (Common Lead Isotopic Standard) was run per 10 samples in order to demonstrate the stability of the instrument.The Pb isotopic ratio (Pb Ratio corr ) was corrected by the following equation: where Pb Ratio meas is the measured ratio in the filter sample, NIST cert is the certificated ratio of the NIST SRM 981 sample, and NIST meas is the measured ratio in the NIST SRM 981 sample.Analytical uncertainty (σR) of the Pb ratio was calculated by adding in quadrature the uncertainties of the measured Pb ratio for the filter sample and the NIST SRM 981 sample as specified in Bevington (1969) with the equation of: where σS m and σN m are the standard deviation of the measured Pb ratio for a sample and the NIST SRM sample, respectively; N c is the certified Pb ratio for the NIST SRM sample; S m is the measured Pb ratio for a sample; and N m is the measured Pb ratio for the NIST SRM sample.

Impact of Grain-Size Sorting on Pb Isotopic Composition
Table 1 shows the Pb isotope ratios for PM 1.0 , PM 2.5 , PM 10 and TSP in the loess samples.Good consistence was found for the ratios among three radiogenic isotopes 206 Pb, 207 Pb and 208 Pb in the same strata, while large variations were seen for 206 Pb/ 204 Pb, 207 Pb/ 204 Pb and 208 Pb/ 204 Pb.This could be explained by the limits on analytical precision of 204 Pb owing to its low natural abundance (Walder and Furuta 1993).Herein, ratios among three radiogenic isotopes 206 Pb, 207 Pb and 208 Pb will be discussed.
Mean 207 Pb/ 206 Pb and 208 Pb/ 206 Pb ratios of the four particle-size fractions in the nine strata are shown in Fig. 2. The experimental uncertainties were calculated on the basis of values from the three parallel samples collected in the same strata.The uncertainties for 207 Pb/ 206 Pb and 208 Pb/ 206 Pb were 0.05 and 0.03, respectively.Small variations were seen for the Pb isotope ratios among the four particle-size fractions.With a difference less than < 0.0005, average 207 Pb/ 206 Pb ratios for PM 1.0 , PM 2.5 , PM 10 and TSP samples were 0.8373, 0.8374, 0.8372 and 0.8377, respectively.The largest variation of 207 Pb/ 206 Pb (0.02) and 208 Pb/ 206 Pb (0.013) occurred in the samples collected at L2 and L9, respectively.But of noted, all these variations were within the range of experimental uncertainties and thus are eligible.
Loess is a product of rock weathering; its coarse and fine fractions represent the minerals of different origins.Among them, the coarse fractions are composed mainly of primary minerals, while fine fractions are the mixtures of secondary minerals formed during chemical weathering.Since the Pb isotope ratios would not be affected, to a measurable extent, by physical or chemical fractionation processes (Bollhöfer andRosman 2000, 2001;Veysseyre et al. 2001), the primary minerals and secondary minerals of loess should have similar lead isotope composition, which cause the similarities of lead isotopes among different size fractions.Therefore, the lead isotope composition of loess reflects the characteristics of materials in its source regions, and they can be used as potential provenance tracers of loess.

dominant Sources of Loess and Their Glacial-Interglacial Fluctuations
The Pb isotopic compositions in the loess collected in the nine strata of the CLP are compared with eolian sand samples from the source regions in order to identify the dominant sources of loess (Fig. 3).Since the Pb isotopic compositions for eolian sand samples in the source regions   are generally for their silicate fraction, we discuss first the difference of lead isotope ratios between bulk samples and silicate fractions of loess.Table 2 compares the Pb isotopic compositions for bulk samples and silicate fractions of eight loess samples, which were analyzed at the same time by Jones et al. (2000).It can be seen that the Pb isotopic compositions of loess were systematically lower in the silicate fractions than in the bulk samples.All these data were translated into those of the silicate fraction by subtracting a factor of 0.010 and 0.012 for 207 Pb/ 206 Pb and 208 Pb/ 206 Pb, respectively.
As a successful source tracer of loess, the Pb isotopic ratios in sediments from different source regions can be distinguished.Analysis of Pb isotopes in the eolian sand samples of the potential source regions, including the Takalamagan Desert, Mu Us Desert, Ulan Buh Desert, Hobq Desert, Otindag sand-land, Horqin sand-land, Tengger Deserts, and the Gobi in southern Mongolia, have shown that the Pb isotopic ratios were consequently varied (Bicaye 1997;Li 2007).Based on the Pb isotopic compositions, four major source regions were identified: 1) the west source regions as represented by the Taklamakan desert; 2) the Mu Us  Pb ratios of the eolian sand in the Mu Us Desert were 50% and 60% higher than those of the loess in the CLP.They are two completely different end-members and located at opposite ends of the Pb isotope growth line.Our results obviously indicate that the Mu Us Desert is unlikely to be the major source contributing region to the loess deposited in the CLP.In addition, the 207 Pb/ 206 Pb and 208 Pb/ 206 Pb ratios of the Hobq Desert, the Ulan Buh Desert and the sand-land in northeastern China had a 30% and 40% higher than those of the loess in the CLP.In fact, the dusts from the Hobq and Ulan Buh Deserts are very limited, and the Otindag and Horqin sand-lands are not located in an upwind position of the CLP.Both of them are thus unexpected as the potential sources of Chinese loess (Sun et al. 2008).The two Pb isotopic compositions of the loess were much close to those of eolian sand in the Taklamagan Desert (with an increase of 10%).Unfortunately, we cannot determine lower Pb isotope ratios in any samples collected in all potential source regions.As a result, no solid conclusion can be drawn regarding whether these regional sources were fully-or partially-dominated in the contributions of the CLP.
Conversely, the Pb isotopic compositions of the loess in the CLP, in fact, fell in the range of the Gobi and deserts in southern Mongolia and northern China, thus possibly representing the main source regions of the Chinese loess.Our conclusion is consistent with the findings from Sun (2002) that the Gobi in southern Mongolia and the adjoining Gobi and sand deserts (including the Badain Jaran Desert, Tengger Desert, Ulan Buh Desert, Hobq Desert and Mu Us Desert), rather than the inland basins, are the dominant source areas of the CLP.Further, Sun et al. (2008) support our finding that the fine-grained dust deposits on the CLP originate mainly from the Gobi desert in southern Mongolia and the sandy deserts in northern China (primarily the Badain Juran and Tengger deserts), rather than from the Taklimakan desert in western China, at least during the last climatic cycle.Chen et al. (2007) reported the Nd-Sr isotopic compositions of the loess.Their results also suggest that the main source regions of the last glacial loess in the CLP are the Badain Jaran Desert, Tengger Desert, and Qaidam Desert, even though the Qaidam Desert region is not covered in our study.
Vertical variations of Pb isotopic ratios in different horizons can also be seen in Fig. 2.Only small deviations between the loess and paleosol samples were found.In the TSP fraction, the highest amounts of 207 Pb/ 206 Pb (0.8434) and 208 Pb/ 206 Pb (2.0823) were seen in the samples of L1L1 and S1, respectively, while the lowest were in S5 (0.8338) and L2 (2.0602), respectively.The differences of maximum to minimum values were again below the experimental uncertainties.This implies that the dominant source regions of the loess in the CLP were relatively stable and had no significant variations during the glacial and interglacial regime.

SummARy
The grain-size impact on the Pb isotopic composition of loess in the central CLP is insignificant, and therefore our result suggest that a Pb isotopic composition can be used as a proxy for source tracing.The comparisons of Pb isotopic compositions between loess and eolian sand in the arid and semi-arid regions in the northern and northwestern China evidenced that the Gobi and deserts in southern Mongolia and northern China are the main source regions of loess in Note: All data are from Jones et al. (2001); Bulk = bulk loess; Sil = extracted silicate fraction.
the CLP.No obvious deviations of Pb isotopes were obtained between the nine horizontal layers in the loess and paleosol samples, implying the loess source regions were stable during the glacial and interglacial regime.

Fig. 3 .
Fig. 3. Pb isotopic ratios plot ( 207 Pb/ 206 Pb vs. 208 Pb/ 206 Pb) in Chinese loess and desert dust from different source areas.The data for eolian sand samples of Takalamagan Desert, Mu Us Deserts, Ulan Buh Desert, Hobq Desert, Otindag Daka and Horqin sand-land are from Li (2007); and the data of Tengger Deserts and the Gobi Desert in southern Mongolia (SM) and north China (NC) are from Bicaye et al. (1997).The growth curve of Pb ore in north China are plotted based on the data of Zhu (1995).

Table 1 .
Pb isotope ratios for PM 1.0 , PM 2.5 , PM 10 and TSP in the loess samples.

Table 2 .
Differences of 207 Pb/ 206 Pb and 208 Pb/ 206 Pb between bulk loess and the extracted silicate fraction.) the Gobi and deserts in southern Mongolia and northern China; and 4) the Hobq, and Ulan Buh deserts and the sand-lands in northeastern China.The 207 Pb/ 206 Pb and 208 Pb/ 206