Introduction

Due to the rapid urbanization and industrialization occurring along river basins in recent times, rivers are subjected to increasing amounts of stress from water pollution and environmental deterioration. Many pollutants discharged into rivers are readily accumulated in sediments, and in any part of the hydrological cycle less than 1 % of pollutants are actually dissolved in the water leaving more than 99 % to be stored in the sediments (Hakanson 1992; Salomons and Stigliani 1995). For this, sediments have become increasingly recognized as a carrier and a possible source of contamination in aquatic systems.

Environmental magnetism is a convenient, effective and economical monitoring technique, and has received attention in the field of urban environmental pollution, particularly in identifying the behaviour of anthropogenically derived particulate matter (Matzka and Maher 1999; Muxworthy et al. 2002; Kapička et al. 2003). Numerous studies have been conducted to investigate the distribution of heavy metals and organic pollutants in sediments (Dauvalter and Rognerud 2001; Farkas et al. 2007; Katahira et al. 2009), and the results of these studies indicate that magnetic materials in river sediments originate from various sources. However, despite their potential applications in source tracing, the magnetic properties of river sediments have attracted less attention (Sinha et al. 2007; Yang et al. 2007).

Lanzhou, the largest city along the Yellow River, has become one of the most polluted cities in the world due to rapid industrial development and topographic condition. The Yellow River is the only water supply for the industries and residents of Lanzhou. Many tributaries surrounding Lanzhou empty into the Yellow River and are polluted due to the agricultural, industrial and life waste water flowing through the urban area. These waters gravely influence the water quality in the upstream area of the Yellow River. However, little information regarding the magnetic characteristics and sources of pollutants in the river sediments in Lanzhou is available. In this paper, the spatial variations of the magnetic properties and their relationship with organic matter and heavy metals (As, Co, Cr, Cu, Ni, Pb, Zn and Fe) in the surface sediments in Lanzhou were studied with the aims of delineating the sources of pollutants and examining the applicability of magnetic measurements for the mapping of heavy metals in urban river sediments within a small area.

Materials and methods

Sample collection

Sediment cores were obtained from eight tributaries flowing into the Yellow River in the urban area of Lanzhou (Fig. 1). Sediment cores are located in the upstream (outer city sediment), midstream and downstream (where the tributaries debouch into the Yellow River). The sediment samples were obtained from river sections with very low flow velocity, where silts and suspended particulates form easily. In July 2008, a total of 23 samples were collected in the Sier River (represented as SE-1, SE-2 and SE-3), Qili River (Q-1, Q-2 and Q-3), Shuimo River (S-1, S-2 and S-3), Luoguo River (LG-1, LG-2 and LG-3), Dasha River (DS-2 and DS-3), Paihong River (D-1, D-2 and D-3), Lanni River (L-1, L-2 and L-3) and Yuer River (Y-1, Y-2 and Y-3). Furthermore, the river path at the segment of each core is relatively straight, wide and distant from industrial outlets and populated sewage discharge sites.

Fig. 1
figure 1

Sketch map of the sampling sites of river sediment in the urban area of Lanzhou

Full size image

Laboratory methods

Magnetic measurements

Freeze-dried and homogenised samples were passed through a 0.25-mm mesh sieve to remove refuse and small stones. An aliquot of ~5.5 g was packed into plastic pots for the following series of magnetic measurements. Low- and high-field magnetic susceptibility (χ lf and χ hf) derived frequency-dependent susceptibility (χ fd), and χ fd percentage (χ fd %) measurements were obtained using a dual-frequency (470 and 4,700 Hz) Bartington Instruments MS2-B Susceptibility Meter. Anhysteretic Remanent Magnetization (ARM) was imparted using a DTECH AF demagnetizer with a peak AF field of 100 mT and DC bias field of 0.05 mT. The measurements are expressed as the susceptibility of ARM (χ ARM) by the steady field. Stepwise backfield remagnetization of saturation isothermal remanent magnetization (SIRM) at 1T was carried out using three reverse fields (i.e. −20, −100 and −300 mT). IRMs were imparted using an MMPM10 pulse magnetizer. All remanence measurements were made using a Minispin magnetometer. SIRM, HIRM (HIRM = [(SIRM + IRM−300)/2]/mass), and SOFT (SOFT = [(SIRM + IRM−20)/2]/mass) are expressed on a mass-specific basis. HARD % and SOFT % were used to determine the respective percentages of HIRM and SOFT concentration. Magnetic hysteresis loops and Curie temperature (T C) were measured using a Variable Field Translational Balance (VFTB) with a maximum field of ±1 T. Saturation magnetization (Ms), saturation remanence magnetization (Mrs) and coercivity (Bc) were determined after subtraction for a paramagnetic contribution.

Geochemical element analysis

After grinding over a 200 mesh screen, the materials were pressed into pellets and the metals were analyzed using a PW2403/00 X-ray fluorescence spectrometer. The test environment temperature was 23 °C, and relative humidity was 56 %. Eight common elements were selected from the 41 determined geochemical elements for study, i.e., arsenic (As), cobalt (Co), chromium (Cr), copper (Cu), nickel (Ni), lead (Pb), zinc (Zn), and ferrum (Fe). For the details of the XRF measurement procedure and the signal deconvolution method, the reader is referred to the work of Boyle (2000).

Total organic carbon measurements

The high-temperature K2Cr2O7 method was used for TOC. With excessive H2SO4, K2Cr2O7 oxidizes organic carbon, and the Cr6+ is titrated with FeSO4. The TOC is determined by the titration. The reactions are as follows:

$$2{{\text{K}}_2}{\text{C}}{{\text{r}}_2}{{\text{O}}_7} + {\text{3C}} + { 8}{{\text{H}}_2}{\text{S}}{{\text{O}}_4} \to {2}{{\text{K}}_2}{\text{S}}{{\text{O}}_4} + {\text{2C}}{{\text{r}}_2}{\left( {{\text{S}}{{\text{O}}_4}} \right)_3} + {\text{3C}}{{\text{O}}_2} \uparrow + \,{8}{{\text{H}}_2}{\text{O}}$$

The remaining Cr8+ titrate with FeSO4, according to the number of organic C variation before oxidation, may be used to calculate the content of organic C, and FeSO4 solution titration reactions as follows:

$$ {{\text{K}}_ 2}{\text{C}}{{\text{r}}_ 2}{{\text{O}}_ 7} + {\text{ 6FeS}}{{\text{O}}_ 4} + { 7}{{\text{H}}_ 2}{\text{S}}{{\text{O}}_ 4} \to {{\text{K}}_ 2}{\text{S}}{{\text{O}}_ 4} + {\text{ C}}{{\text{r}}_ 2}{\left( {{\text{S}}{{\text{O}}_ 4}} \right)_ 3} + {\text{ 3F}}{{\text{e}}_ 2}{\left( {{\text{S}}{{\text{O}}_ 4}} \right)_ 3} + { 7}{{\text{H}}_ 2}{\text{O}} $$

All parameters were measured in the Key Laboratory of Western China’s Environmental System (Ministry of Education), Lanzhou, China.

Results

Magnetic properties

The measurement results are depicted in Table 1. The low-frequency magnetic susceptibility χ lf, one of the most commonly used magnetic parameters, represents the total contribution of ferrimagnetic minerals (Thompson and Oldfield 1986; Dearing 1999; Wang et al. 2008). The samples have relatively high mean χ lf values of 106.43 × 10−8 m3 kg−1, ranging from 35.46 × 10−8 to 371.73 × 10−8 m3 kg−1. SIRM is related to the concentration of magnetic minerals which carry remanent magnetizations, and is largely controlled by ferrimagnetic and canted-antiferromagnetic minerals without the influence of paramagnetic and diamagnetic material (Thompson and Oldfield 1986). The mean SIRM value is 1,129.15 × 10−5 Am2 kg−1, ranging from 432.63 × 10−5 to 2,759.55 × 10−5 Am2 kg−1. The soft parameters are approximately proportional to the concentration of the ferrimagnetic minerals and maghemite within a sample, while HIRM estimates the antiferromagnetic component in a sample (Oldfield and Richardson 1990). SOFT, HIRM and Ms have similar variations with χ lf and SIRM. The magnetic concentration-dependent parameters indicate that the river sediment samples contain a moderately high concentration of magnetic minerals especially in downstreams of the Paihong, Shuimo, Sier, Yuer Rivers, midstream of the Lanni River, and upstream of the Qili River. The percentage of frequency-dependent susceptibility χ fd % is related to the superparamagnetic (SP) ferrimagnetic component (Dearing et al. 1996), particularly grains ranging between 0.01 and 0.025 mm in size (Maher 1988; Dearing et al. 1997). A common feature among all magnetic particles contained in river sediment is their low values of χ fd % (2.55 %), which indicates that magnetic grains in the SP state do not contribute significantly to the general magnetization (Maher 1986). The SOFT and HARD percentages may serve as approximate indicators of the relative dominance of ferrimagnetic and canted-antiferromagnetic sample components (Oldfield and Richardson 1990). Compared with the HARD percentage, the considerably higher SOFT percentage indicates that ferrimagnetic minerals dominate the samples’ magnetic characteristics.

Table 1 Magnetic parameter values of river sediment samples from urban area of Lanzhou
Full size table

Thermomagnetic curves, showing the Curie point temperature of magnetic minerals, are sensitive to determine the composition of mineral phases in soils and sediments (Thompson and Oldfield 1986). Figure 2 shows that the thermomagnetic (M–T) curves for heating and cooling processes of the sediment samples from the Paihong River (D-1, D-2 and D-3) and Qili River (Q-1, Q-2 and Q-3) are non-reversible, as the cooling curve is above the heating curve, with a Curie temperature of 580–600 °C indicating the original prevalent contribution of magnetite and production of new magnetite grains during the heating process. The heating curves of the Luoguo River (LG-1, LG-2 and LG-3) sediment samples are non-reversible, with the cooling curve being below the heating curve, and a Curie temperature of 580–600 °C indicating the dominance of original magnetite. In some samples (e.g. D-1, D-2, D-3, Q-3, LG-1, LG-2 and LG-3), the heating curves reduce steadily before 120 °C suggesting the existence of goethite. Furthermore, magnetization shows a clear decrease in the range of 300–450 °C in sample D-1 implying that the maghemite converted to hematite. The peaks of heating curves indicate the oxidation in the range of 450–550 °C such as D-1, D-2, D-3 and Q-3. Some heating cycles (e.g. D-1, Q-3 and LG-3) decrease to almost zero at 680 °C indicating the existence of hematite.

Fig. 2
figure 2

Thermomagnetic curves of river sediments in the urban area of Lanzhou

Full size image

Hysteresis is a unique physical phenomenon of ferromagnetic material used to describe the relationship between magnetic remanence obtained in a pulsed direct current magnetic field and the density of the external magnetic field. The hysteresis loop may be interpreted as a magnetic mineralogical signature based on its shape (Thompson and Oldfield 1986; Evans and Heller 2003). The height of a loop is a function of concentrations and types of magnetic minerals in a natural sample, and the width of a loop is entirely controlled by the ‘hardness’ of minerals (Xia et al. 2011). The ‘SOFT’ minerals, such as magnetite and pyrrhotite, produce steep and narrow loops while haematite has a broader loop. A reversible linear relationship between magnetization and the applied field without any loop produced is characteristic of paramagnetic or diamagnetic minerals (Thompson and Oldfield 1986). The hysteresis loops of typical samples and their corresponding hysteresis parameters are depicted in Fig. 3. Most hysteresis loops are thin, closed and the magnetization almost saturates until the applied magnetic field reaches 300 mT, and Bc ranges from 4.91 to 11.27 mT indicating that the magnetic minerals are dominated by low-coercivity ferromagnets (mainly magnetite). The value of Ms is high in the downstream of the Paihong River and the upstream of the Qili River and is low in the Luoguo River implying that these samples have relatively higher magnetite contents than others.

Fig. 3
figure 3

Magnetic hysteresis loops for river sediments in the urban area of Lanzhou

Full size image

The thermomagnetic analyses indicate that the main contribution originates from magnetite-like phases. The day plot is useful in determining the domain state of the particles (Day et al. 1977). The SD–PSD–MD boundaries being defined for magnetite by Dunlop (2002) are used to classify the domain state. As depicted in Fig. 4a, the bulk magnetic grain size of the samples is in the pseudo single domain (PSD). However, a mixture of multi domain (MD) and single domain (SD) grained particles cannot be excluded. Bc is in the range of 5–12 mT for the majority of the samples used in this study. SD grain size magnetite has higher Bc values than MD minerals (Evans and Heller 2003), which suggests that larger grains (PSD/MD) are dominant in the sediment samples.

Fig. 4
figure 4

Scatter plot of river sediments in the urban area of Lanzhou: a Day plot of Mrs/Ms vs Bcr/Bc, b scatter diagram of Bcr vs SIRM/χ

Full size image

A bivariate plot combining SIRM/χ and Bcr illustrates the dominance of magnetite in the SD range (Fig. 4b) (Peters and Thompson 1998). Points are grouped with respect to SIRM/χ indicating contributions of the MD to SD range.

Spatial variations of heavy metal concentration and organic matter contents

The concentrations of the heavy metals in the 23 river sediment samples as well as the minimum, maximum and mean values, and enrichment factor (r) are shown in Table 2. In the samples used in this study the average concentrations for each element (i.e. As 22.57 mg kg−1, Co 11.7 mg kg−1, Cr 190.29 mg kg−1, Cu 102.45 mg kg−1, Ni 41.82 mg kg−1, Pb 58.16 mg kg−1, Zn 315.28 mg kg−1 and Fe 31.15 g kg−1) are higher than the background levels in the soils of sierozem (i.e., As 11.5 mg kg−1, Co 10.0 mg kg−1, Cr 59.3 mg kg−1, Cu 20.3 mg kg−1, Ni 33.0 mg kg−1, Pb 18.2 mg kg−1, Zn 61.3 mg kg−1 and Fe 28.3 g kg−1).

Table 2 Statistics of heavy metal contents of river sediments in the urban area in Lanzhou
Full size table

Based on the work of Adami (Adami et al. 2000) and Adamo (Adamo et al. 2005), the analytical data corresponding to each sampling site may be used in the computation of the enrichment factor (r) defined as the following ratio:

$$ r \, = \, \left( {{C_{\text{sed}}} - \, {C_{\text{back}}}} \right)/{C_{\text{back}}} $$

where C sed is the content of a particular metal in each sample obtained from the Lanhzou River, and C back is the sierozem content. The r values for each sampled site are reported in Table 2. Due to the fact that the heavy metals As, Co, Ni, and Fe present mean values below unity, the concentrations of these metals are below the contamination level. In contrast, the metals which are most highly linked to anthropogenic activities, Cr, Cu, Pb, and Zn, have mean values above unity, i.e., their concentrations are at least three times higher than the background sediment. On the basis of the mean value of r, the sediments are enriched in metals in the following order: Zn > Cu > Cr > Pb.

The metals with r > 1 may be considered as a guide to anthropogenic metal pollution, and are used for evaluating the degree of pollution of surface river sediments through computation of a “total enrichment factor”, R for each site, by means of averaging the r values of the all (n) indicator-metals (in this case: Cu, Zn, Pb, Cr, n = 4):

$$ R=\left( \sum r \right)/n $$

The resulting R values are shown in Fig. 5 providing an integrated index of local metal pollution and from the figure a synthetic view of the spatial distribution of this integrated indicator may also be obtained. The classification of the examined sites as severely polluted (R > 3.0), moderately polluted (2.0 < R < 3.0) and slightly polluted (R < 2.0) makes it evident that the midstream and downstream of the Yuer River are severely polluted, the upstream of the Qili River and downstream of the Lanni and Shuimo Rivers are moderately polluted, and all others are slightly polluted (Fig. 5).

Fig. 5
figure 5

Total enrichment factor and TOC % of river sediment in the urban area of Lanzhou

Full size image

The sedimentary organic matter is also subjected to considerable influence from human activities, and is often a pollutant itself (Glikson et al. 1995). Organic matter, quartz, feldspar and calcium carbonate are all diamagnetic and thus have the potential to dilute the magnetic signal. Figure 5 shows that the organic matter content is not very high in any of the sediment samples with the exception of the downstream of the Shuimo River; therefore, it may be seen that the diluting effect of organic matter content is not significant.

Discussion

Sedimentary environment

Previous studies (Wang et al. 2011a, b) have suggested that no apparent positive correlations exist between magnetic parameters (χ lf, SIRM, χ ARM, χ ARM/SIRM, χARM/χ) and grain size, especially finer fractions, implying that grain size is not the key factor influencing the magnetic properties at least not in this study.

Spatial variation in magnetic properties of river sediment

The rivers in Lanzhou may be divided into three types according to the variation of their magnetic properties. The first type is characterized by a significant increase of the ferrimagnetic mineral concentration from the upstream to downstream, and is known as the downstream polluted type. Thermomagnetic curves for this type showed that the magnetization reduces steadily during heating from 250 to 450 °C (Thompson and Oldfield 1986) implying the existence of maghemite. In the upstream the magnetization reduces at 120 °C and between 600 and 680 °C, which indicates that the sediments contain goethite and hematite. In the downstream, the signal of hematite disappears. The hysteresis loops also suggest that a magnetite concentration ratchet from the upstream to downstream. Samples from the Paihong, Sier, Yuer and Shuimo Rivers belong to this first type. The second type of river is characterized by the gradual decrease of the ferrimagnetic mineral concentration from upstream to downstream. Samples from the Qili and Lanni Rivers belong to this type. The magnetic mineral concentration reduces from upstream to downstream, and the thermomagnetic curves also reveal the magnetite characters significantly in the upstream and midstream; in addition, the signal of hematite is presented strongly in the downstream rather than the upper reaches. The hysteresis loops suggest that the magnetite concentration gradually reduces from the upstream to downstream. The third type has relatively low magnetic mineral content and nearly identical magnetic phases (magnetite-dominated), as evidenced by general magnetization, thermomagnetic and high-field activities, such as in the Luoguo and Dasha Rivers; therefore this third type of river is known as the less affected type.

Relationship between magnetic parameters and organic matter content

Table 3 shows that the association between organic matter content and χ fd, χ ARM, χ ARM/SIRM are significant at 0.01 level. χ fd is a reasonably effective indicator of superparamagnetic (SP) material concentration in mixed-domain assemblages. The presence of stable single domain (SSD) grains can be estimated by χ ARM (King et al. 1982; Maher 1988). The ratio χ ARM/SIRM values are sensitive to grain size in ferrimagnetic minerals (Maher 1988; Oldfield and Yu 1994; Dearing et al. 1997). A strong linear correlation exists between the organic matter content and χ fd, χ ARM or χ ARM/SIRM of bulk samples, which suggests that the organic matter is related to the SP ferrimagnetic grain proportions and/or fine-grained ferrimagnetic materials in urban rivers, and implies that they share the same source. A similar relation was documented in the work of Xie (1999, 2001) and Shilton (2005). Magnetic assemblage may be portioned between that associated with the inorganic mineral fraction (perhaps from primary earth materials, such as soil and concrete residues derived from water erosion, among other sources) and a magnetic fraction associated in some ways with the organic fraction (including discharged industrial wastewater, solid waste and airborne dust derived from combustion and processed organic materials). A close link exists between the χ fd percentage value and the contribution of soil-derived materials (Maher 1986; Thompson and Oldfield 1986). Typically, primary earth materials present the variable χ fd percentage value and close associations with organic matter content. In contrast, anthropogenic materials present low χ fd percentage values and weak associations with organic matter content (Xie et al. 1999). The low χ fd and χ ARM/SIRM values and high TOC concentrations of river sediments indicate that anthropogenic materials are the major source. The results of this study also suggest that simple, rapid, and non-destructive magnetic measurements may provide useful information concerning the river sediment organic matter content.

Table 3 Pearson correlations between different parameters of river sediments in the urban area of Lanzhou
Full size table

Relationship between magnetic parameters and heavy metal contamination

As shown in the results and discussions above, river sediments are mixtures of natural and anthropogenic components and both contain magnetic mineral fractions with specific magnetic properties. The high values of χ lf and SIRM, low χ fd %, and the presence of coarse-grained magnetite as the primary magnetic carrier reveal the dominance of ferrimagnetic minerals with anthropogenic origins. In order to appraise the relationship among the variables of heavy metals, the Pearson correlation coefficient (Table 3) and tree diagram (Fig. 6) are constructed (in which the variables are standardized, then processed using SPSS software). The correlation between magnetic parameters (χ lf, SIRM, Ms, SOFT) and heavy metals (Cr, Ni, Pb, Zn, Fe, Cu) reveals a positive relationship. However, the above-mentioned heavy metals are relatively weakly correlated with mineral composition ratios (χ ARM/SIRM, χ ARM/χ). Many previous studies (Beckwith et al. 1986; Charlesworth and Lees 1997; Geogeaud et al. 1997; Bityukova et al. 1999; Desentfant et al. 2004; Gautam et al. 2005a, b) have reported possible linkages between magnetic properties and heavy metals, and described the fact that both were derived from the same source as being the main reason for their significant correlation. The results of this study reflect the fact that in Lanzhou River sediments the magnetic minerals and elements originate from the same pollution source.

Fig. 6
figure 6

Cluster analysis

Full size image

The tree diagram is shown in Fig. 6, in which the average distance between heavy metals (Pb, Ni, Cu, Cr and Zn) and magnetic concentration parameters (χ lf, SIRM, SOFT and Ms, χ ARM) is smaller than others, which suggests a strong correlation between them. In addition, the Pb, Cu, Cr and Zn levels are higher than the background levels in sierozem soils. The study results confirm the fact that these four heavy metals are dominated by non-natural sources, which is possibly attributable to different industrial sewage and other anthropogenic origins. At the same time, the distances among heavy metals (As, Co, Fe) and other magnetic parameters are small, and heavy metal contents of As, Co and Fe below the background levels, implying that these elements are mainly soil-derived. Therefore, the close correlations between magnetic concentration-related parameters (e.g. χ, SIRM, SOFT and Ms) and heavy metals Cr, Ni, Pb, Zn and Fe may suggest mixed contributions of magnetic material from urban and natural sources, probably urban sources mainly.

The anthropogenic input is reflected in the Pb, Zn, Cr and Cu concentrations, as well as the magnetic particulate matter derived from automobile exhaust, coal combustion, alloy, mining, electrofacing, metallurgy, paint, petroleum and other industrial sources (Alloway 1990; Tiller 1992; Stroganova et al. 1998; Siegel 2002). However, the difference is not significant between contaminated and non-contaminated samples with low Fe concentrations in sediment. In addition, the Fe concentration in sediment samples is dominated by the primary material. Nevertheless, the magnetic signal increases are due to the input of magnetic particles which exposed to anthropogenic influences, thus, despite the low Fe concentrations in the samples, the significant correlation with magnetic susceptibility may be attributed to the influence of anthropogenic magnetic particulate matter. On the other hand, the high correlation between magnetic susceptibility and Cr, Ni, Pb and Zn (Table 3) may be attributed to the incorporation of these elements in the crystalline structure of Fe mineral molecules (Kukier et al. 2003). For As and Co, the correlation with magnetic susceptibility is poor, which may be attributed to the fact that these elements do not show an anthropogenic increase, nor does a paragenetic material produced with magnetic particulate matter exist. In addition, the metals analyzed all display significant correlations with Fe, with the exception of the element As. The poor correlation in this case may be attributed to the different sources of Fe and other heavy metals. Thus, the same source for heavy metals and magnetic minerals is the condition of using magnetism for the monitoring of river sediment pollution.

The total enrichment factor, however, which is determined by the enrichment of the studied metals above the unity (R > 2), has a significant correlation with the susceptibility values. It is well known that the Lanzhou River sediments have been exposed to severe pollution. The results of this study highlight the fact that magnetism may serve as an effective proxy for using heavy metal pollution to monitor river sediment pollution in urban areas.

Conclusions

The magnetic properties of river sediments in the urban area of Lanzhou are dominated by coarse-grained and low-coercivity magnetite, which is associated with maghemite and haematite. The rivers in Lanzhou may be divided into three types according to the contribution of anthropogenic activities (i.e., downstream polluted, upstream polluted and less affected).

Significant correlations were observed for organic matter content and the parameters χ fd %, χ ARM and χ ARM/SIRM, showing that organic matter content is influenced by magnetic grain size. Some magnetic parameters show the potential to be used as proxies for distinguishing source apportionment in river sediments.

Positive correlations between magnetic parameters χ lf, SIRM, SOFT, Ms and heavy metals Cr, Cu, Ni, Pb, Zn and Fe indicate that environmental magnetism is an effective measurement in monitoring heavy metal pollution and mapping in polluted areas.