Introduction

Sediment is a particulate matter that transports by the flow of air, water or glaciers and eventually settles down as deposits. Beach sands and river deposits are examples of sediments. In rivers, lakes and oceans, sediments are most often transported by water. It creates river bed, which preserves its own history that is the direction and magnitude of water flow, parent rock lithology and other factors such as environmental pollution. Natural processes responsible for the formation of sediments can be altered by anthropogenic activities. Many man-made materials enter into water bodies through atmospheric deposition, runoff from land, or direct discharge into the water. Energy and mineral consumption by humans is the main cause of contamination of the biosphere. Under certain conditions the contaminants in the bottom sediments may be released back into water and enter into the food chain. Elements are transferred from soil/sediments to surface water to aquatic biota to human. These contaminants may pose a risk to environment on a large scale and hence need to be monitored at regular intervals. Anthropogenic sources of trace elements in sediments are mostly associated with mining of coal, ores and with discharge of municipal waste into river bodies.

The mass fractions of different elements in sediment indicate the level of geochemical or pollution related changes in a water body. Thus, the analysis of sediments can aid in reconstructing its history, in understanding human impact on the ecosystem [1, 2], and in suggesting possible remedial strategies. This information can also be used for modelling different processes [3]. In past, many elemental analysis techniques have been employed for the exploration of sediments. Among these, the X-ray fluorescence spectrometry [4, 5] and instrumental neutron activation analysis (INAA) have been widely applied [6, 7]. INAA is one of the most suitable methods for geological and environmental samples due to its multielement capability, milligram sample size requirement and freedom from contamination. INAA with semi-absolute methods adds another advantage whereby almost all radionuclides in a spectrum can be quantified; thus extending the range of elements. For this reason k 0-insturmental neutron activation analysis (k 0-INAA) has been used extensively for sediment analysis [8, 9]. Another semi-absolute method known as “activation constant (AC) method” [10, 11] also produces accurate results for reactors which show good neutron flux reproducibly such as miniaturized neutron source reactor (MNSR). In our study these two methods of standardization have been applied for the determination of elemental mass fraction in sediments collected along Hunza River in Central Karakorum, Pakistan. This river is an important source of water for Pakistan. To this date, no literature has been cited on the distribution of elements in the Karakoram river system. This paper attempts to collect this information for the first time.

Experimental

Sampling area

Hunza River basin lies in the high Karakoram Range within longitudinal from 74o 02′ to 73o 03′ and latitudinal from 35o 54′ to 37o 05′ in the northern territory of Pakistan [12]. In the west it borders by Gilgit River basin, in the north by Afghanistan and China, and in the south by Shigar and Indus River basins. The small portion in the northeast of this basin drains in the Chinese territory. Most of the terrain of the basin is in the elevation range of about 1,500 m above sea level (ASL) to more than 7,500 m ASL. The basin occupies an area of 16389.4 sq. km out of which 87.5 % contributes to Hunza River flow. The glacier cover of the basin is about 4677.3 sq.km [13]. Hunza River is one of the main contributor of water to the river system existed in Pakistan. Its water is assumed pollution free because very little human population inhabit along its banks. Most of the time, it flows in gorges at high speed and is characterized by rapids and falls.

Sampling and sample pre-treatment

In January 2007, nine sediment samples were collected from different locations along Hunza River. The sediments were collected from river side and the upper most layer of sediment from 0 to 10 cm depth was obtained with a stainless steel trowel. The sampling points covered almost 100 km along the river. The geographical parameters of the sampling sites are produced in Table 1 and sampling locations are depicted in Fig. 1. Among nine samples, five were obtained from the main stream, while four samples were taken from different tributaries entering into the river. Sampling along the river was difficult due to accessibility issues. All samples were collected in clean polyethylene bottles and kept tightly closed. Care was taken to avoid any contamination during and after the sampling [14]. All samples were first air dried and then kept in oven at 105 °C to remove any water content left. Later, all sample were grinded in ball mill for at least 1 hour.

Table 1 Geographical information of sampling locations
Full size table
Fig. 1
figure 1

Sample locations along Hunza River

Full size image

Instrumentation

Each sample weighing up to 150–250 mg was packed inside a polyethylene rabbit along with the Au/Zr monitors [15]. The irradiation was done at Pakistan Research Reactor (PARR-2), which is a MNSR having a nominal thermal neutron flux of 2 × 1012 cm−2 s−1. After irradiation the samples were transferred to pre-weighed clean polyethylene capsules for counting. The gamma-ray spectra were acquired using a p-type coaxial HPGe detector (Eurisys Measures, France). It is coupled through a 570 ORTEC made spectroscopy amplifier to Trump PCI, 8 k ADC/MCA card with GammaVision-32 ver. 6 Software (ORTEC, USA). The detector has 60 % relative efficiency and FWHM of 2.04 (measured) at 1,332 keV. In spectrum processing, the nuclide identification and quantification were performed using GammaLab software [16]. Full peak efficiency calibration of the detector was done for different detector to source geometries using 241Am, 133Ba, 137Cs, 60Co and 152Eu point calibration sources [17].

Methods

In this study, two semi-absolute standardization methods of NAA were employed. These include the k 0-INAA and the AC method. In k 0-INAA, the thermal to epithermal neutron flux ratio (f) and epithermal flux shape factor (α) are determined for the irradiation channel. The f and α values were determined using Al-0.1 % Au wire (IRMM-530RC, Belgium, Geel) and ZrO2 (99.99 %, Aldrich, Wisconsin) powder. The f and α were determined for A2 irradiation channel of PARR-2 using “bare triple monitor method” and their values were found as 21.5 and −0.0321 respectively [17, 18]. All the recommended k 0,Au factors and nuclear data were taken from the literature [19].

Using k 0 method, the concentration (ρ) of an element “a” is calculated as:

$$ \rho_{a} = \frac{{A_{sp,a} }}{{A_{sp,Au} }}\frac{1}{{k_{0,Au} }}\frac{{f + Q_{0,Au} \left( \alpha \right)}}{{f + Q_{0,a} \left( \alpha \right)}}\frac{{\varepsilon_{p,Au} }}{{\varepsilon_{p,a} }} $$
(1)

where \( A_{sp} \) is the specific activity and is calculated as

$$ A_{sp} = \frac{{N_{p} }}{{wt_{m} SDC}} $$
(2)

where N p is the peak area, w the mass of element, t m the measurement time, S the saturation factor \( \left[ {1 - e^{{ - \lambda t_{irr} }} } \right] \) for \( t_{irr} \) irradiation time, D the decay factor \( e^{{ - \lambda t_{d} }} \) for t d decay time and C is the counting factor \( \left[ {{{\left( {1 - e^{{ - \lambda t_{m} }} } \right)} \mathord{\left/ {\vphantom {{\left( {1 - e^{{ - \lambda t_{m} }} } \right)} {\lambda t_{m} }}} \right. \kern-0pt} {\lambda t_{m} }}} \right] \). Further details about the k 0-INAA implementation can be found in our previous work [17]. Similarly, activation constants were obtained from our previous publication [11].

In spectrum processing, several types of interferences are encountered. Usually, geological materials require correction for nuclear reaction by fast neutrons such as (n,p) and (n,α) and fission products produced by 235U [20]. Both types of corrections were applied on the final mass fraction data.

Results and discussion

In this study, three irradiations for 3 min, 10 min and 5 h were required for the determination of 38 elements in sediment samples and two reference materials i.e., IAEA-S-7 (soil) and IAEA-SL-1 (lake sediment). The nuclear data for these determinations can be found in our previous studies [15, 21]. Reference materials were analysed for the assessment of experimental accuracy. The results of IAEA-S-7 and IAEA-SL-1 are presented in the form of Z-score in Fig. 2. In this figure all elements are within ±3 Z-score except for Zn in IAEA-SL-1, which is due to error introduced by the multiplet at 1,115 keV. The Z-score was calculated as

$$ Z = \frac{{\left( {x_{i} - x_{ref} } \right)}}{{\sigma_{ref} }} $$
(3)
Fig. 2
figure 2

Z-score plots of IAEA-S-7 and IAEA-SL-1, showing that all elements are within ±3 Z-score except Zn in IAEA-SL-1

Full size image

where x i is the measured mass fraction, x ref is the reference mass fraction and σ ref is the reference uncertainty in reference mass fraction for element i. Z-scores for IAEA-S-7 were equally distributed in positive and negative values, while for IAEA-SL-1 there were slightly more negative values. This could be due to variation in flux parameters during the irradiation of IAEA-SL-1.

In this study, Ca and Eu were quantified by the AC method, while rest of the elements was measured by the k 0-INAA standardization. Many elements have multiple stable isotopes which may produce more than one radionuclide upon irradiation. In our study, eight elements were determined using two different radionuclides including Ba (131Ba, 139Ba), Br (80Br, 82Br), Ce (141Ce, 143Ce), Eu (152Eu, 152m1Eu), Sb (122Sb, 124Sb), Sr (85Sr, 87mSr), Yb (169Yb, 175Yb) and Zn (65Zn, 69mZn). The resultant mass fraction was calculated using simple arithmetic average of the two values from two radionuclides. The final mass fractions are based on the average of 3–7 independent measurements i.e., irradiating samples 3–7 times.

In this study, three types of interferences were observed including spectral interferences, nuclear reaction based interferences and interferences by fission products of 235U. All these interferences were handled by our software GammaLab [16]. Spectral interferences are resolved during nuclide identification step, while nuclear reaction and fission products are corrected after the analysis is completed. In the analysis of sediment samples, the most significant correction was required for the 27Al(n,p)27Mg reaction. The other nuclear reactions with more than 0.1 % correction included 54Fe(n,α)51Cr and 88Sr(n,p)88Rb reactions. The peak areas of 141Ce, 140La, 147Nd, 103Ru and 95Zr were corrected for the contribution of fission products. The fission correction factors in μg/μg of uranium for 141Ce, 140La, 147Nd, 103Ru and 95Zr were 0.28, 0.0025, 0.22, 0.123 and 9.3 respectively. It should be noted that fission correction factors for 140La, 103Ru are decay time dependent. After making appropriate corrections, the mass fractions of different elements obtained for 9 samples are presented in Table 2.

Table 2 Elemental mass fraction (μg g−1) in 9 sediment samples with uncertainties at ±1σ
Full size table

In k 0-INAA, the important sources of uncertainty included the k 0-factor, Q 0-factor, efficiency, weight and counting statistics. In our study, all the spectra were acquired with less than 5 % dead time. Pulse pile-up corrections were applied to all of them. True coincidence correction was not required because the sample to-detector distance was more than 15 cm in all countings. All mass fractions were based on arithmetic average of individual values. Uncertainties were estimated following the “guide to the expression of uncertainties in measurement” (GUM) [22]. According to the GUM, the total expanded uncertainty contains statistically evaluated uncertainties (type A uncertainty) and uncertainties based on relevant information available (type B uncertainty). In our study, the type A uncertainties were from detector efficiency, weight of sample, peak areas and f value. The type B uncertainties included uncertainties in half-life, Q 0-factor, k 0-factor, α value and gold mass fraction (1.2 % from IRMM certificate). Uncertainty in α was taken as 10 % after the recommendations of Robouch et al. [23]. The uncertainties in k 0- and Q 0-factors for gold were zero by definition. Before combining, all uncertainties were converted to standard uncertainties.

The elemental mass fraction distribution in sediment samples can be grouped as;

Elements with mass fraction >0.1 %: Al, Ca, Fe, K, Mg, and Na

Elements with mass fraction <0.1 % but >1 mg kg−1: As, Ba, Ce, Co, Cr, Cs, Dy, Eu, Ga, Hf, La, Mn, Nd, Rb, Ru, Sc, Sm, Sr, Th, U, V, Yb and Zn

Elements with mass fraction <1 mg kg−1: Sb, Ta and Tb

The results of elemental mass fraction were explored further by applying principal component analysis (PCA) [24, 25]. It is a multivariate data exploration technique, which is usually carried out by exploring scores and loading plots. Figure 3 is the scores plot of principal component 3 (PC3) versus PC1. It shows samples in four distinct clusters:

Fig. 3
figure 3

Scores plot of PC3 versus PC1, showing distribution of samples in four clusters

Full size image

  • Cluster 1: sample 2

  • Cluster 2: samples 1, 3, 4 and 5

  • Cluster 3: samples 6 and 7

  • Cluster 4: samples 8 and 9

Similarly, Fig. 4 is the loadings plot of PC3 versus PC1, which shows distribution of elements. It illustrates that elements are distributed in different groups. Elements close to each other presents high correlation coefficient and indicate a common origin. On comparing Fig. 3 with Fig. 4, it reveals that the sediment samples 1–5, all collected from the main stream are having high mass fractions of rare earth elements (REE). In fact the highest quantities of REE in sample 2 indicate a possible source of rare earth mineral in the vicinity. All the samples collected from tributaries (samples 6–9) show high mass fractions of Al, Fe, Mg, Sc, V and Zn, which indicates their source as crustal elements.

Fig. 4
figure 4

Loadings plot of PC3 versus PC1, showing relationship among elements

Full size image

Average mass fractions of major elements (Al, Ca, Fe, K, Mg and Na) in samples were compared with those from other places of the world and the comparison is presented in Table 3. It can be observed that Ca and Na are highest in the Hunza River sediment than the average mass fraction for the Indian, world and surficial rocks. The average mass fraction of Al, Fe and Mg were higher in the Hunza River sediment than those in the Indian and the surficial rocks but lower than the world average mass fractions. The average mass fraction of K in Hunza River sediments was similar to those found at other locations. The data exhibits an enrichment of As at sites 1, 2, 8 and 9. Arsenic is usually associated with lead and copper ores from trace to 2–3 % and with gold ores up to 11 % [26]. The presence of As in these samples need more investigation to ascertain its source properly. The enrichment of Hf, I, Th and Zr at site 2 and Se at sites 8 and 9 was also observed.

Table 3 Mass fraction of major elements (in wt%) in the Hunza River sediments in comparison to the Indian [26] and World river sediment averages [27] and surficial rocks [27]
Full size table

REE are characterized by a single oxidation state (+3) except for Ce and Eu, which have additional valences. REE are important geological tracers for the modelling of geochemical processes. Our samples have been characterised with eight REE having mass fractions in the order: Ce > La > Nd > Sm > Dy > Yb > Eu > Tb. A comparison of mass fractions of REE found in our study, after excluding sample 2, with the world average crustal composition [27] shows that La, Ce, Nd and Sm were higher in the Hunza River main stream samples. In tributary samples, La, Ce, Nd, Sm and Yb were lower than the world crustal values. The mass fractions of Eu and Dy were higher in the tributary samples, while mixed trend was shown by these elements in the main stream samples. Elements Tb and Yb exhibited mixed trend in the main stream and tributary samples. The ratio of the light rare earth elements (LREE) to the higher rare earth elements (HREE) was about 20 for the main stream samples and about 8 for the sediments collected from tributaries samples. A Chondrite normalised distributions of REE found in the mainstream and tributary samples are presented in Figs. 5 and 6 respectively. Reference values for REE in Chondrite were taken from Haskin et al. [28, 29] as La: 0.332, Ce: 0.876, Nd: 0.6, Sm: 0.183, Eu: 0.0685, Tb: 0.047, Dy: 0.317 and Yb: 0.207.

Fig. 5
figure 5

Chondrite normalized mass fractions of REE in the main stream sediments (locations 1, 3, 4, 5). Sample 2 exhibits the similar trend but not shown here due to scale problem

Full size image
Fig. 6
figure 6

Chondrite normalized mass fractions of REE in the tributaries sediments showing Eu anomaly (samples: 6, 7, 8, 9)

Full size image

The profiles in Fig. 5 show enrichment in the LREE (La, Ce, Nd, Sm) and a depletion in HREE (Eu, Tb, Dy, Yb), which indicates the origin of REEs in the sediment samples from the lithosphere and not from anthropogenic activities. Tributary samples (Fig. 6) exhibited a positive europium anomaly. The Eu in +2 oxidation state has size similar to Ca+2 and in reducing magma it substitutes for calcium. Europium enrichment or depletion is attributed towards its tendency to be incorporated into plagioclase over other minerals. If a magma crystallizes stable plagioclase, most of the Eu will be incorporated into this mineral causing a higher than expected concentration of Eu in the mineral versus other REE in that mineral. A comparison of REE in sample 2 shows ratio of mass fractions in our sample to the world average crustal values as: 8.3 (La), 3.4 (Ce), 3.0 (Nd), 2.4 (Sm), 0.9 (Eu), 2.9 (Tb), 0.9 (Dy) and 3.4 (Yb). It indicates an enrichment of La, Ce, Nd, Sm, Tb and Yb at site 2.

Conclusions

In this study the application of INAA with k 0-standardization and AC methods has determined the mass fractions of 34 elements. This proves semi-absolute standardizations suitable for the characterization of sediment samples. These elements can be quantified using the analysis scheme with three irradiations (3 min, 10 min and 5 h) at a low power research reactor. Spectrum processing requires corrections for spectral interferences and for the contribution from nuclear reactions and fission products. With the help of PCA it was found that samples from different origin can be differentiated on the basis of elemental concentrations. A comparison of REE with the world average crustal mass fraction shows that at one location, the REE in Hunza River was quite high, which indicates the possible presence of REE mineral in the area. Since there is not much anthropogenic activity and inhabitants along the Hunza River, the physical weathering process is considered as a major factor for the distribution of trace and REE in the sediment.