INTRODUCTION

The dominating role of particulate and colloidal forms of many chemical elements is the main feature of matter migration in river water [2, 12]. That is why the reliable differentiation between these and dissolved forms is a prerequisite of correct evaluation of river water quality, the specific features of the seasonal variability of their chemistry, and chemical element fluxes with river flow. Filtering through membrane or capsule filters with pore size of 0.45 or 0.22 µm is the most commonly used method for separation of dissolved and particulate forms [6, 35]. In this case, the “dissolved” fraction by definition contains the majority of colloid forms (0.001–0.45 µm), and the suspended one, contains a part of coarse colloids (0.45–1 µm).

The use of filtration as a method for separation suspended and colloidal/dissolved migration forms of elements is accompanied by some methodological problems. The problem of contamination would be better of the dissolved phase in the course of sampling, filtration, and analysis has been largely solved by the introduction of “clean ould be better” methods of water sampling, processing, and analyzing [9, 17, 20], though efforts are still being made to improve them [30]. The next problem, associated with the separation of colloid and dissolved forms, is successfully solved by the application of various forms of ultrafiltration, especially, in combination with dialysis [23, 24, 33].

The third problem is associated with the uncertainty in the separation of suspended particles and coarse colloids during filtering through filters with pore size of 0.45/0.22 µm due to pore clogging. In this case, the filtering capacity decreases and a part of coarse colloids with a size <0.45/0.22 µm is retained by the filter, and the concentration of chemical elements, associated with these colloids, in the filtrate decreases. This phenomenon was actively discussed more than 20 years ago [13, 16, 17, 20, 32, 35]; however, as can be seen from the most recent review [15], a final solution to the problem has not yet been found. This is due to the fact that the degree of clogging depends not only on the characteristics of the filter and filtering methods, which can be standardized, but also on the properties of the suspended and colloidal material, which can be different in different rivers and in the same river in different seasons. Capsule filters with a much greater filtering capacity were recommended to use to reduce the clogging effect [17, 35]. Tangential filtration [21] and the use of filter cells with mixing [23] are not accompanied by clogging; however, they are more complex technologically and their use can affect the proportions of colloidal forms [33]. Another solution is to minimize the volume of filtered water [25, 30]. It was shown that the concentrations of Fe, Cu, Zn, and U in filtrate did not change significantly at the passage of 50 mL of river water with suspension concentration 20–200 mg/L through a filter with pore size of 0.45 µm [25]. However, commonly 0.5–1.0 L of river water is filtered through filters 0.45/0.22 µm, and the first 100–300 mL portions of filtrate are used to rinse the filtering system and the vessels [22, 24, 33]. Therefore, it is not impossible that in that case the distribution of metals over colloidal fractions will be studied with the use of water in which the concentration of coarse colloids has been reduced. The objective of this study is to assess the effect of clogging on the concentrations in filtrate of a wide range of elements: Fe, Al, Ti, Pb, rare earth elements, Hf, Zr, Nb, Mn, Co, as well as Li, B, Si, Ni, Cu, As, Sr, Rb, Mo, Ba, U, and V in the case of filtration of river water samples 100–500 mL in volume through membrane filters with a pore size of 0.45 µm. Such estimate can contribute to the optimization of the scheme of separation of suspended and colloid/dissolved forms of chemical elements in the case of river water quality monitoring and assessment. The chemical composition of filtrates after membrane filters was compared with such after capsule filters with the same pore size, but larger filtration area, which serve as an alternative to membrane filters [35].

MATERIALS AND METHODS

Characteristic of the Examined Rivers

The study used water samples taken from typical medium-size rivers in the southern Primorski Krai: the Razdol’naya and Tumannaya, which suffer appreciable but not extreme anthropogenic load, and the rivers of Bikin and Bol’shaya Ussurka (B. Ussurka) (Fig. 1), which drain poorly developed taiga catchment.

Fig. 1.
figure 1

Schematic map of the study area. (1) River water sampling sites.

Full size image

The Razdol’naya is among the major rivers of the Maritime Territory, Russian Federation, with average annual flow of 75 m3/s and drainage area of 16 830 km2 [4]. Sixty percent of its catchment lies in PRC territory. The relatively high level of economic development of the drainage area results in a higher concentration of nutrients [3] and some metals [7] in river water. The water regime of the Razdol’naya river shows a winter minimum discharge (5–10 m3/s) in freezing period, spring flood (250–350 m3/s) in April–May, summer monsoon floods (up to 440–510 m3/s) in August–September with a subsequent decrease in water discharge to 20–30 m3/s in late autumn. The seasonal variability of suspended matter concentration in the Razdol’naya river is also wide: from 3–6 mg/L in winter to 300–400 mg/L during floods at an average of 73 mg/L [7], which is close to the available Roshydromet estimate of 55 mg/L [4]. The samples from the Razdol’naya river used in this study were taken in its lower reaches (43°18.2′ N, 131°47.6′ E) in warm season at different water regimes (Table 1).

Table 1.   Hydrochemical parameters of the rivers of Razdol’naya (R-80, R-83, R-91), Tumannaya (T-66), B. Ussurka (U-2), and Bikin (U-3) during sampling for studying the role of clogging (TP is total dissolved phosphorus, DSi is dissolved Si, SS is suspended solids concentration; electrical conductivity EC, µС/cm, pH, other parameters in mg/L)
Full size table

The Tumannaya is the largest river on the western coast of the Sea of Japan with a catchment area (41 200 km2), lying in PRC (70%) and North Korea (30%). Only the last 17 km in the lower reach of the Tumannaya river lie on the border between Korea and RF; the samples were taken in this segment (42° 17.6′ N, 130° 41.8′ E) in July 2017 in the period of monsoon flood (Table 1). The water regime of the Tumannaya river is similar to that of the Razdol’naya. By the data of the gage nearest to the mouth, located 70 km from it, the normal annual water discharge in the Tumannaya is 215 m3/s [34]. Suspended matter concentration in the Tumannaya varies from 3–6 mg/L in winter to 40–50 mg/L in spring and to 180–540 mg/L during summer monsoon. The water of the Tumannaya examined in this study was taken in July 2019 in high-water period at suspension concentration of 341 mg/L.

The concentration of nitrogen and phosphorus compounds in water of the Tumannaya river is appreciably higher than that in the Razdol’naya (Table 1) because of the greater population density and more intense anthropogenic load, including agricultural. The population density in the Razdol’naya basin is 38.6 person/km2 in RF territory, and 29.8 person/km2 in the Chinese part of the basin; the same value in the Tumannaya river basin is 53.4 person/km2 [5, 10].

The rivers of Bol’shaya Ussurka (catchment area 29 500 km2, mean annual discharge 283 m3/s) and Bikin (catchment area 22 300 km2, mean annual discharge 238 m3/s) account for 2/3 of the flow of the Ussuri river, the main right tributary of the Amur river in RF territory. The water regime of the Bol’shaya Ussurka and Bikin rivers is similar to that of the Razdol’naya and Tumannaya rivers: winter low-water season, moderate spring flood, and summer rain freshets. Unlike the catchments of the Razdol’naya and Tumannaya rivers, the major portions of the catchments of the Bol’shaya Ussurka and Bikin are covered by taiga. The percentage of forests in the Bol’shaya Ussurka basin is >90%. The catchment of the lower reaches of the Bikin river also contains vast swampy floodplains (mari). The proportion of agricultural landscapes in the northern Primorskii krai, where the catchments of the Bol’shaya Ussurka and Bikin lie, is 5–7%. The population density in the basins of these rivers is on the average 0.8–1.6 person/km2 [5], i.e., very low. The samples were taken in the lower reaches of the rivers (Fig. 1) during summer flood; however, even under such conditions, the suspended solids content was 30–32 mg/L at lower concentration of macroions and nutrients (N, P, Si), but at higher concentration of dissolved organic carbon, especially, in the Bikin river.

Water Sampling and Filtration

Water samples were taken from river bank into a polyethylene vessel attached to a 3 m long pole. The samples were filtered within several hours after their delivery into the laboratory in a laminar box with a flow of filtered air. Sample aliquot 500 mL in volume was filtered by 100-mL portions through a pre-weighed membrane Millipore Durapore PVDF (diameter 47 mm, pore size 0.45 µm, filtration area 15 cm2), placed into a Sartorius polycarbonate filter holder. Prewashed filtration system was twice rinsed by 50 mL highly purified water Milli-Q (18 M Mohm) immediately before filtration with application of moderate (0.2–0.3 bar) vacuum to remove excessive moisture. Each portion of filtrate was analyzed as a separate sample. The filtration time was fixed by a stopwatch to monitor the clogging by flow rate. Variations of pH in the course of experiment were monitored by pH-meter Mettler Toledo with a combination electrode. Thirty to fifty mL from each filtrate portion were used for triple rinsing of prewashed vessels for samples. The filtrate for determining macro-ions, nitrates, and dissolved organic carbon (DOC) were poured into tightly pluggable 30-mL glass bottles and stored in a refrigerator for no more than several days; the filtrates for determining chemical elements by ICP-MS were poured into pluggable polypropylene test tubes 50 mL in volume with subsequent acidification by double-distilled HNO3 to рН 2. Parallel to that, an aliquot of the sample was filtered through a capsule filter Pall GWV with a pore size of 0.45 µm and filtration area of 700 cm2 with the use of a peristaltic pump. The first 50–100 mL of this filtrate were used to rinse vessels for samples. Filtration of 100 mL of Milli-Q highly purified water was used to assess and control possible pollution during filtration.

All vessels used for sampling, processing, and storage of sample, as well as the filtration system were washed by distilled water, soaked by 0.1 N HCl, rinsed again by bidistilled water, by Milli-Q water, and dried in a laminar box. The filters were also washed by acid and Milli-Q water. The personnel involved in sample taking and filtration used expendable plastic gloves.

Methods of Chemical Analysis

The electric conductance as a measure of TDS was measured by YSI Pro Plus devise. In the course of filtration, pH was measured by pH meter Mettler Toledo Seven Compact. The accuracy of pH measurements was 0.02 pH units. DOC was determined by high-temperature catalytic oxidation in device Shimadzu TOC-V cpn with an accuracy of ±5% relative. Blank samples in DOC determination amount to ≤0.1 mg/L. The concentrations of anions \({\text{SO}}_{4}^{{2 - }},\) Cl–, and \({\text{NO}}_{3}^{ - }\) were determined by ion chromatography in Shimadzu LC-10 instrument with an accuracy of ±4–5%. The concentration of suspension SS was determined by weighing filters before and after filtration of 500 mL water samples and drying at 80°C. The particle size distribution of suspended matter in river water was determined by laser diffraction on a Fritsch Аnalysette-22. All analytical studies, except for ICP-MS analysis were carried out in Shared Research Facilities, PGI FEBRAS.

Acidified filtrates were analyzed to determine some microelements from Li to U by ICP-MS method on spectrometer “Аgilent 7700х” Shared Research Facilities, PGI FEBRAS with In as internal standard. The correctness of the results was controlled by regular analysis of standard samples of bottom deposits BCSS-1 and water CRM-TMDW-A. The difference between the obtained and standard data was ≤15%. The results of blank samples during filtration were <10% of the concentrations determined in samples for most microelements, except for Zn, for which blank samples were comparable with the concentration in filtered samples (0.2–0.4 µg/L). Because of this, Zn was excluded from further consideration.

The measure of the decrease in the concentration of element i due to clogging was taken to be the difference ΔCf between the concentration in filtrate after the filtration of the first 100 mL (Ci100) and after 500 mL (Ci500), as well as the proportion of losses Ki = ΔCf/Ci, the source concentration Ci was taken equal to the mean of the concentration Ci100 and the concentration in the filtrate after the capsule filter.

RESULTS AND DISCUSSION

Variations of Flow Rates and Concentrations of Macroions, Nitrates, and DOC in Filtrates

The rate of flow of river water samples 0.5–5 L in volume through capsule filters did not changed, thus suggesting that no appreciable clogging was taking place. However, the flow rate through membrane filters (47 mm, 0.45 µm) started decreasing as early as after 100–200 mL of the samples and dropped by 1–2 orders of magnitude at the filtration of 500 mL (Fig. 2), which can be explained by nothing but the clogging of membrane pores. The flow rate of the first 100 mL was maximal for water from rivers in the Ussuri basin with a relatively low (30–32 mg/L) suspension content. The flow rate for water from more turbid rivers of Razdol’naya and Tumannaya is much lower (Fig. 2). For an individual river, e.g., Razdol’naya, the higher the concentration of suspension, the less the flow rate, i.e., the faster the clogging. However, the process, no doubt, depends not only on the amount, but also on the quality of the suspension. The suspension in the Tumannaya River was coarser and contained only 75% of pelite particles <0.01 mm in size, compared with 92–98% in the suspension of the Razdol’naya River. Therefore, the dynamics of clogging for the Tumannaya River (sample T-66, SS 341 mg/L) was comparable with the dynamics for the Razdol’naya River (sample R-83, SS 23 mg/L), notwithstanding the much greater concentration of suspended matter (Fig. 2).

Fig. 2.
figure 2

Flow rate (V) through the membrane 47 mm with a pore size of 0.45 µm depending on the volume of filtered water (V) of the rivers of Bol’shaya Ussurka (U-2), Bikin (U-3), Razdol’naya (R-80, 83, 91), and Tumannaya (T‑66) with different concentrations of suspension (U-2 is for 32 mg/L; U-3, 30 mg/L; R-83, 23 mg/L; R-91, 110 mg/L; R-80, 270 mg/L, T-66 – 341 mg/L).

Full size image

pH measurement in the filtrates showed minor alkalization by 0.1 pH unit (from 7.55–7.87 to 7.69–7.97) in the process of filtration, which is likely due to degassing even at weak evacuation, which was used in the course of filtration.

The concentrations of anions (sulfates and chlorides), as well as nitrates and DOC, in the course of filtration showed no considerable variation trends, whatever the type of the river, water regime, and suspension concentration, or the type of the filters used (Fig. 3). As it has been shown before, the concentrations of the major cations (K+, Na+, Ca2+, Mg2+) in river water in Primor’e also do not depend on the type of filters [29]. This suggests the predominance in the water of the examined rivers of truly dissolved forms of macroions and nitrates, as well as fine colloids for DOC, which are not caught even by clogged filters. The predominance of colloids <1 kDa (<0.0013 µm) in DOC transport in river water was established by detail studies in rivers in the boreal zone of RF, in particular, in Karelia [23, 27].

Fig. 3.
figure 3

Concentrations of chlorides (Cl–, mg/L) and sulfates (\({\text{SO}}_{4}^{{2 - }}\), mg/L), nitrates (mg N-NO3 mg/L), and dissolved organic carbon (DOC, mg/L) in filtrates after a capsule filter GWV and after a membrane filter at filtration of 100, 200, 300, 400, and 500 mL sample.

Full size image

Thus, the concentration of main ions, nutrients, and DOC in river water practically does not depend on the type of filters used (capsule or membrane) and their clogging degree, which is in agreement with many published data [17, 22]. At the same time, in many other rivers some DOC can be associated with large enough colloid fractions, which are retained by clogged filters [20].

The Effect of Clogging on the Concentrations of Trace Elements in Filtrates

Among trace metals, several groups of elements can be distinguished by different character of changes in their concentrations in filtrates in the course of filter clogging.

The first group includes Li, B, Si, Ni, Cu, Cd, As, Sr, Rb, Mo, Ba, U, V, Cr, and Th, the concentrations of which in filtrates remain relatively stable at an increase in the volume of filtered water (Fig. 4), i.e., the concentrations do not depend on clogging at filtration of 0.5 L. This group includes elements with different chemical properties, but their common feature seems to be the minor role of coarse colloids in the balance of migration forms in the examined rivers. For most elements of this group, the dominance of ionic forms and aquacomplexes with different charges in river water is in agreement with their chemical properties. However, in the case of Cu and Ni, we probably have complexes with organic compounds, including colloids, but having small size and passing even through clogged filters. The migration of Cu and Ni predominantly in the form of dissolved complexes or organic colloids <1 kDa in size has been repeatedly identified in rivers of the boreal zone [22].

Fig. 4.
figure 4

Concentrations of Sr, Ni, As, Cu (µg/L) in filtrates after a capsule filter GWV and after a membrane filter at filtration of 100, 200, 300, 400, and 500 mL.

Full size image

The next group includes rare earth elements (REE), as well as elements-hydrolysates Y, Zr, Nb, Hf. The behavior of these elements in the case of filter clogging depends on the type of the river. In the Razdol’naya and Tumannaya rivers, an increase in suspension concentration was accompanied by a significant trend towards a decrease in the concentrations of these metals in filtrates by 40–60% at an increase in the volume of filtered water from 100 to 500 mL. In the case of heavy REE (e.g., Dy), as well as Hf, the general level of concentration degree due to clogging was 30–40% (Figs. 5b, 5d), against 50–60% for La, Ce, Zr (Figs. 5a, 5c). This suggests that an appreciable part of REE and some other elements-hydrolysates occur in water of the Razdol’naya and Tumannaya rivers as components of coarse colloids. At the same time, in the Bol’shaya Ussurka and Bikin rivers, the effect of clogging was observed only for Ce and La with a decrease in their concentrations in the filtrate by 30–40%, i.e., it is lesser than that in the Razdol’naya and Tumannaya rivers (Fig. 5а), while the concentrations of Zr and Hf in the filtrates of the Bol’shaya Ussurka and Bikin rivers show no effect of clogging (Figs. 5c, 5d). It is likely that the higher concentration of DOC in the rivers of the Ussuri basin (Fig. 3) is accompanied by a more uniform distribution of elements-hydrolysates and rare-earth elements (REE) between colloidal fractions and an increase in the proportion of relatively small organic complexes, which are less sensitive to clogging. This is in agreement with the data of detail studies of dissolved and colloid REE forms in rivers in boreal zone [18, 19, 28].

Fig. 5.
figure 5

Concentrations of Ce, Dy, Zr, Hf (µg/L) in filtrates after a capsule filter GWV at filtration of 100, 200, 300, 400, and 500 mL samples.

Full size image

The more pronounced response to the clogging of lighter REE (La, Ce) and Y is also explainable, because they are known to have a closer affinity to coarse colloids than that in the case of heavier REE [31]. It was shown that lighter REE tend to associate with colloids of Fe hydroxides while heavier REE are more often bound with low-molecular fractions of organic substances [22].

The concentrations of Mn and Co in filtrates gradually decreases in the process of clogging of membrane filters (Fig. 6), maybe because of the more uniform distribution of these metals over colloidal fractions of different sizes compared with REE or Fe and Al. The gradual decrease of the concentration at cascade filtration through a filter 0.22 µm and ultrafilters 0.025 µm, 100, 10, 1 kD, established for rivers in the White Sea basin [23], also shows a uniform distribution of Mn and Co over colloid fractions. In this case, in all experiments with water from the Razdol’naya and Tumannaya rivers with a sufficiently high suspension content, the concentration of Mn in filtrates decreased to 0.4–0.6 µg/L as soon as after 400 mL, irrespective of variations from 4 to 42 µg/L in the first 100 mL (Fig. 6a). The dynamics of Mn decrease in water filtrate from the Razdol’naya River was different at a low concentration of suspension during autumn low-water season (R-83), when the initially high Mn concentration gradually decreased from 48 to 29 µg/L after filtering 500 mL of a sample. A similar gradual decrease from 7–17 µg/L in the first 100 mL of filtrates to 4 µg/L after 500 mL was observed for the water of the Bol’shaya Ussurka and Bikin rivers (Fig. 6a), containing ~30 mg/L of suspension (Table 1).

Fig. 6.
figure 6

Concentrations of Mn and Co (µg/L) in filtrates after a capsule filter GWV and a membrane filter at filtration of 100, 200, 300, 400, and 500 mL sample.

Full size image

Additional studies are required to find out whether these differences are due to the specific features of Mn distribution over colloid fractions in river water at different water regimes and river types or they are the results of sorption processes in the course of filtration [26].

The decrease in the concentrations of metals in the filtrates of water from the Tumannaya and Razdol’naya rivers in the course of clogging is maximal in the case of Fe and Al. The losses at filtration of 500 mL of water compared with the first 100 mL or the filtrate after a capsule filter were 82–97% (ΔCf/Сi = 0.82–0.97) (Table 2). At the same time, the decrease was maximal after the filtration of the first 200 mL of the sample, after which the rate of changes decreased considerably (Figs. 7a, 7b). The predominance of coarse colloid forms of Fe and Al, which are effectively retained by clogged filters, is the most plausible explanation of such picture [17, 20]. In the case of rivers in the Ussuri basin, which carry lesser amounts of suspension but are richer in DOC, the decrease in Fe concentration in filtrates caused by clogging is less pronounced (Fig. 7b), and the share of losses was 58–67%. Studying Fe distribution over colloidal fractions confirms the predominance of large colloids with sizes of 0.2–0.8 and 0.006–0.2 µm in the balance of “dissolved” Fe passing through the filter 0.2–0.45 µm [11, 22]. The Lena River water was shown to contain many authigenic colloidal (0.02–0.03 µm) particles of Fe hydroxide, which form aggregates 0.2–1 µm in size [14], which are clearly able to effectively clog filters.

Table 2.   Comparison of the initial concentration Сi, µg/L, of dissolved/colloidal forms of trace metals and losses due to clogging ΔCf, µg/L, at filtering 0.5 L of river water (cont. is contamination)
Full size table
Fig. 7.
figure 7

Concentrations of Al, Fe, Ti, and Pb (µg/L) after filtering through a capsule filter GWV and membrane filter at filtration of 100, 200, 300, 400, and 500 mL sample. Note the logarithmic scale on the concentration axes for Al and Fe.

Full size image

Notwithstanding the wide differences between the initial Al concentrations in the waters of the Razdol’naya and Tumannaya rivers (334 and 30 µg/L, respectively), the concentration in all fitlrates after 200–300 mL of the sample, levels at 4–6 µg/L (Fig. 7a), and the fraction of losses due to clogging is 79–99%. This demonstrates the predominant occurrence of Al in the composition of sufficiently large colloidal clay particles [14], which are effectively retained by clogged filters. In the samples from the B. Ussurka and Bikin rivers, which contain 38–114 µg/L Al in the first 100 mL of filtrate, the decrease is significant (Fig. 7a), but not too high, amounting to 56–67%. The changes in Ti and Pb concentration in filtrates at increasing volumes were similar, but showed greater variations of the data. At the same time, the general trend toward a decrease in Ti and Pb concentrations at filtration through a membrane filter >100 mL of the sample is obvious (Figs. 7c, 7d).

Fe hydroxides and clayey colloids are the most likely mineral forms that support the effect of clogging in the studied rivers [14, 22]. Perhaps this is why the effect of clogging is greater in the Razdol’naya and Tumannaya rivers with higher concentration of mineral suspension compared with rivers in the Ussuri basin with lesser amount of suspension, though with higher DOC concentration. In other regions, the effect of clogging was also recorded in rivers with low suspension but high concentration of colloidal DOC (the so-called “black” rivers) [20].

A consequence of clogging is the possibility to underestimate the concentrations in river water of “dissolved” forms of metals such as Fe and Al, as well as Ti, Pb, and REE, which gravitate to coarse colloids. The authors’ studies determine the limits of the losses at filtration of 0.5-L water sample (Table 2) and suggest the high probability of underestimation of the concentrations of some metals in the filtrate when a water sample >200 mL in volume is filtered.

The absolute value of the losses of “dissolved” forms of metals in the studied rivers due to clogging at filtration of 0.5 L of water varies widely even within one river at different water regimes as well as between different rivers (Figs. 5–7; Table 2). Despite this, chemical elements form a stable series by the proportion of losses at clogging Ki (Fig. 8). The maximal effect of clogging can be seen at the determination of dissolved/colloidal forms of Al, Fe, and Ti, the average proportion of losses for which was 70%, never dropping below 50%. In the case of Mn and Pb, such proportion was also 70%, but it decreased to 10–28% for rivers with low suspension concentration. In the case of other metals-hydrolysates, primarily REE, the average proportion of losses due to clogging was 30–40%; however, it varied widely from complete absence to 70–90%, depending on the type of the river and/or its water regime, which controls suspended matter concentration.

Fig. 8.
figure 8

Shares of metal losses at filtration of 500 mL of river water; the values of losses are dots within boxes and at the ends of whiskers; the vertical size of the boxes denotes 25 and 75% percentiles; the horizontal segments in the boxes show the mean losses of the element.

Full size image

The initial concentration of dissolved and colloidal forms of Fe, Al, and other metals that respond to clogging, as well as the amount and composition of the suspension, appear to be the main factors that control metal losses during filtration. Of importance also is the concentration and composition of dissolved organic compounds [20, 23, 24]. Estimating the contributions of individual natural factors that have their effect on clogging can be a focus of further studies; however, a more important objective is the development of methods for taking into account, minimizing, and, in an ideal case, eliminating the effect of clogging.

The Possibilities to Minimize the Role of Clogging in Determining River Water Chemistry

The inevitability of clogging and the likelihood of its effect on the concentrations of some elements in the filtrate have been mentioned since long ago [17]. The use of a matter that clogs a membrane filter as a filtering medium is the basis of the method of continuous filtration [1, 8], which allows one to obtain practically continuous distribution of chemical elements over dimensional colloid fractions. However, the need to filter 10–30 L of river water limits the practical use of this method for routine monitoring or studying the spatial variations of river water chemistry. As mentioned above, an alternative for river water quality monitoring can be the use of capsule filters [7, 17, 35], tolerant to clogging, and minimization of the volume of filtration to 50 mL [25]. The drawback of the first variant is the high cost of capsule filters [17], and the drawback of the second variant is the small volume of filtrate, which is not enough for studying the proportions of different colloidal and dissolved form.

CONCLUSIONS

Successive filtration of water samples taken from typical rivers of boreal zones in the southern Far East in the Russian Federation through membrane filters with a pore size of 0.45 µm confirmed their appreciable clogging even after the passage of 100–200 mL of water. Clogging was observed in waters with suspension content of 23–340 mg/L, and the effect increased with increasing concentration.

Clogging does not affect the concentration of the principal ions, nutrients, and DOC in the filtrates. Many trace elements (Ni, Cu, Cd, As, Sr, Rb, Mo, Ba, U, V, and Cr) also show no significant changes in their concentrations in filtrates in the course of clogging, which suggests the predominance of dissolved and/or fine colloidal migration forms in the examined rivers for all these components.

REE and some elements-hydrolysates (Zr, Hf, Nb) show a decrease in concentrations in filtrates on the average by 30–40% as the sample volume increases from 100 to 500 mL with an accompanying increase in pore clogging, thus suggesting the important role of large colloids in their migration. However, these metals showed considerable variations in the shares of losses at clogging, depending on river type and water regime.

The largest effect of clogging is observed in the concentration of Fe, Al, and Ti in filtrates. As soon as after filtration of 200 mL of the sample, the concentration decreases by 62–90% relative to the first 100 mL or filtrates after capsule filters. This is a result of the predominance of large colloidal particles of Fe hydroxides and clay aluminosilicates in the composition of dissolved forms of these metals in the examined rivers.

Unlike other elements, which respond to clogging, the concentration of Mn and, partly, Co in the filtrates decreases gradually in proportion with the volume of filtered water, thus suggesting a uniform distribution of these metals over various colloidal fractions.

The decrease in the concentration due to clogging in the process of filtration of 200–500 mL of river water can lead to a considerable underestimation of the results for dissolved/colloidal forms of Fe, Al, Mn, Co, REE and some other metals-hydrolysates in river water when the analysis is made with filtrate portions other than first ones. Accordingly, errors are possible in water quality assessment. The most obvious methods for minimizing the effect of clogging can be the use of capsule filters and minimizing the volume of filtered water. However, further studies are required to optimize methods for separating dissolved/colloidal and suspended migration forms of chemical elements in river water.