Indigenous bacteria are essential for the performance of bio-filters for drinking water treatment. Yet it is slow and difficult to develop biofilm in a granular activated carbon (GAC) filter with low nutritional levels in the influent, especially during winter. In this study, the biofilm development in three laboratory-scale GAC columns with different types of influent was investigated in southeast China during winter. The results indicated that nitrogen was the limiting factor for biofilm development in GAC columns for this source water. The biomass density in the column with ammonia nitrogen addition was much higher than those of the other two filters, while its microbial diversity and biological activity were lower. Moreover, the ammonia-feeding column also showed the highest removal of organic contaminants during the stable operating periods, i.e. chemical oxygen demand (CODMn), assimilable organic carbon as well as biodegradable dissolved organic carbon. Therefore, nitrogen amendment favors the formation of biofilm. It could shorten the start-up time of a GAC filter and enhance the bio-stability of its effluent. This might add some new insights towards the operation of GAC filters with low nutritional levels in the influent during winter.
As granular activated carbon (GAC) filters play an important role in improving water quality, especially in removal of natural organic matter (NOM), they are widely employed in drinking water treatment plants (DWTPs) all over the world (Simpson 2008). In a pristine GAC filter, excellent organic matter removal could be expected by adsorption. However, active biofilm will cover the GAC media and the GAC filters will change to biological activated carbon (BAC) filters after several months’ operation. Then, the adsorption capacity will gradually be depleted, and NOM and artificial organic pollutants will be removed mainly via biodegradation instead of adsorption. Significant functions of biomass have been recognized in BAC filters (Velten et al. 2011; Gibert et al. 2013; Liao et al. 2015), which are reflected in three aspects. First, biodegradation biofilm can extend the service life of the BAC media and it does not require frequently expensive regeneration (Aktas & Cecen 2007). Secondly, biodegradation could ensure the bio-stability of the effluent via removal of the biodegradable organic compounds (BDOC) and assimilable organic carbon (AOC) (Chien et al. 2008). Thirdly, biofilms also benefit the removal of non-biodegradable compounds as bio-regeneration increases the adsorptive capacity of GAC (Seredynska-Sobecka et al. 2006).
Biofilm formation was found to closely relate to the fluctuations of water temperature. Low temperature generally has an adverse effect on the growth rates of microorganisms and biofilm development (Emelko et al. 2006). Velten found that the growth rates at 7 ± 0.7 °C (0.0001–0.0043 h−1) were much lower than that at 9–22 °C (0.038–0.16 h−1) (Velten et al. 2011). The transition from a GAC to a BAC filter may be a time-consuming process in winter, especially where there is an oligotrophic source water. As elevating the temperature of the influent is impractical for DWTPs, other factors that could boost biofilm development should be studied (Marzorati et al. 2008).
Biofilm formation in the GAC filter was also found to be sensitive to environmental perturbations. The rates and extents of biofilm formation were influenced by biomass (Yu et al. 2002) and water quality (particularly the concentration of growth promoting substrates) (Zhu et al. 2010). In addition, it could be affected by dissolved oxygen (DO) (Feng et al. 2013) as well as backwashing (Chu & Lu 2004). Of these factors, a microbial growth-promoting substrate containing organic carbon, nitrogen and phosphorus is one of the most important. The molar ratio required for bacterial growth is approximately 100C:10N:1P (Pelmont 1993). Organic carbon, especially AOC, has been considered as the main limiting factor for microbial growth (Zhang & Huck 1996). However, the impact of nitrogen has rarely been covered (Li et al. 2015). As nitrogen is also an essential element for bacterial growth, it is necessary to study its impact on the bacterial community and biomass density, especially for waters containing low concentrations of ammonia.
Hence, the aim of this study was to determine the factors that affect the biofilm development in GAC filters receiving oligotrophic source water in winter. For this purpose, three laboratory-scale GAC columns receiving various types of influents were constructed. The development of biomass as well as organic matter removal was monitored over time in the three columns. Moreover, the biological activity of the biomass, the microbial communities, and other factors (AOC/BDOC) that affect the bio-stability of the effluent were assessed during the steady-state phase.
MATERIALS AND METHODS
Pilot plant layout and operation
The water samples were taken every other day and GAC particles were collected every month. Dissolved organic carbon (DOC) was measured with a Shimadzu 5000A TOC analyzer. Chemical oxygen demand (CODMn) and ammonia nitrogen (NH4+-N) were determined according to the standard methods described by the China Environmental Protection Agency (2002). The organic matter in water was oxidized by potassium permanganate; the amount of oxidant consumed was proportional to the value of CODMn. The ammonia nitrogen was detected by a colorimetric method using Nessler's reagent. A pH meter (PHS-3C, Leici, China) was used to measure the temperature and pH.
AOC and BDOC measurements were conducted as previously described (Liu et al. 2002; Escobar et al. 2000). A phospholipid extraction method was used to measure the biomass on the activated carbon (Yu et al. 2002). The method of bacterial community detection was the same as in our previous paper (Liao et al. 2013). The biological activity of the bacteria in the three BAC filters was estimated by measuring the specific oxygen uptake rate (Moussa et al. 2005).
RESULTS AND DISCUSSION
Water quality parameters
CODMn values in the influents of columns I and III were 2.72–4.38 mg/L and that of Column II was 3.44–5.72 mg/L. The ammonia concentration of the lake water was between 0.05–0.11 mg/L. Ammonia chloride was added to keep the nitrogen concentration at a level of 1.05–1.11 mg/L for column III. The total phosphorus contents of the influents were about 0.08–0.11 mg/L for the three columns. As for the biodegradable organics, the AOC and BDOC of untreated water were 298 μg/L and 1.89 mg/L, while those of ozonated water rose to 372 μg/L and 2.25 mg/L, with a 24.8% increase of AOC and a 19% increase of BDOC. Microbial biomass in the influent raw water of column II (275 CFU/ml) was nearly 10 times as much as that of columns I and III (30 CFU/ml).
The residual ozone concentrations in the influents of columns I and III changed with temperature, with an average concentration of 0.05 mg/L, which indicated that most of the ozone was consumed in the ozone contact tank.
Removal efficiency of CODMn of the three bio-filters
During the first start-up period (days 0–20), as the GAC media are pristine and barely colonized and the role of adsorption predominated, so high organic matter removal efficiencies were achieved. The large decrease in CODMn removal was observed during days 20–80. At this stage, the adsorption capacity was reduced while biodegradation occurred (Simpson 2008; Velten et al. 2011). After 80 days, biodegradation played the leading role in CODMn removal (Boon et al. 2011).
AOC/BDOC removal by the three bio-filters
The removal of biodegradable fractions (AOC/BDOC) is essential to guarantee the bio-stability of the finished water. Water samples from the influent/effluent of the three columns were collected for AOC/BDOC analysis.
Slightly higher AOC concentrations were observed for the influents of column I and III than for column II, indicating an increase of biodegradable organics after the ozonation process. This may result from oxidation of organic matter to biodegradable and assimilable compounds such as oxalic acid (Chien et al. 2007). The AOC removal efficiencies were about 70–80% for the three columns, which is similar to Yang et al. (2011), reporting that the AOC removal efficiencies in the BAC and GAC columns were 70% and 82%, respectively.
Biomass density of the three bio-filters
GAC is an ideal medium for the development of attached microorganisms.
Bacterial attachment is an initial step in biofilm formation. However, the bacterial adherence rate was very low in the early stage of colonization as most microorganisms present on the GAC were not adapted to permanent attachment (Servais et al. 1994). Initial biofilm development proceeded at the highest rate in the third month, which was probably associated with the slight increase in water temperature. Column II had a higher biomass density than column I, which may be due to three factors. (1) A limited effect of ozone in increasing the biodegradable organics was observed in our studies, as the raw water of the lake was dominated by small molecular organics (DOC in the lake with MW < 3 kDa accounting for 98.2%) (Liao et al. 2014). This result was the reverse of other researchers’ results, which found that ozonation can significantly increase microbial nutrient concentrations and stimulate biofilm formation in the GAC filter (Thayanukul et al. 2013). (2) There were sufficient biodegradable compounds (shown in Table 1) for biofilm growth, as AOC accounted for about 15.7–16.5% of the BDOC, which is a much higher percentage than that reported by Velten (3%) (Velten et al. 2011). According to Monod kinetics, bacteria with a similar maximal growth yield but distinct substrate affinity (Ks) would be less sensitive to substrate concentration fluctuations when higher nutrient concentrations were present (Boon et al. 2011). (3) The biomass in the influent of column II was much higher (275 CFU/ml) than that of column I (only 30 CFU/ml), which was negatively impacted by residual ozone (Hammes et al. 2008). So once the nutrient conditions were similar, the biomass in the influent was the determinant factor.
|Column .||pH .||Turbidity (NTU) .||CODMn (mg/L) .||Ammonia (mg/L) .||AOC (μg/L) .||BDOC (mg/L) .||TP (mg/L) .||Biomass (CFU/ml) .|
|Column .||pH .||Turbidity (NTU) .||CODMn (mg/L) .||Ammonia (mg/L) .||AOC (μg/L) .||BDOC (mg/L) .||TP (mg/L) .||Biomass (CFU/ml) .|
Note: the residual ozone concentrations in the influents of columns I and III changed with temperature, the averaged value was about 0.05 mg/L, which indicated that most of the ozone was absorbed by the water.
The discrepancies in biomass density between column I and column III were evident; the latter had a much higher biomass density than the former. The difference between them lay only in the ammonia concentration of the influent, which was consistent with the biomass density. It may due to the fact that the molar ratio of nutrients required for bacterial growth is approximately 100C:10N:1P (Pelmont 1993), and the low level of ammonia in the influent of column I was the limiting factor for bacterial growth and biofilm formation. In addition, the biomass density of column II was lower than that of column III. Although the biomass in the influent of the former was nearly 10 times higher than that of the latter, the ammonia concentration of the former was lower. If both nutrients and biomass differ in the influents, the nutrients play a more important role than the biomass. Hence, for this source water, the pivotal factor in the development of biofilm was nitrogen.
To summarize, the biomass formation is susceptible to the quantities of nutrients in the influent as well as the indigenous microbial population (Boe-Hansen et al. 2002).
Biological activity of the three bio-filters
As can be seen in Figure 6, the biological activity of the bio-filters varied.
The consumption rate of DO for column I was 14.8 × 10−2 mg/g/h, while the rates for columns II and III were 11.9 × 10−2 mg/g/h and 11.1 × 10−2 mg/g/h, respectively. The order of biological activity was column I > column II ≈ column III.
Bacterial communities of the three bio-filters
Figure 7 shows that the community richness of column I was 2003 operational taxonomic units (OTUs) and the diversity index (Shannon index) was 6.17. The sample collected from column II had the highest OTU score (2637 OTUs) and the highest Shannon index (7.06). Column III, with the addition of ammonia nitrogen, has the lowest OTU score (1620 OTUs) and Shannon index (5.84).
With abundant indigenous bacteria in the raw water influent, column II had the richest species diversity. Column III had the lowest microbial diversity, which might be due to two factors. (1) Some species of bacteria were killed by the ozonation process. (2) Nitrogen addition facilitated the growth of some bacterial species (such as Alpha proteobacteria and Gamma proteobacteria) that adapted to high nitrogen (Liao et al. 2013), which would restrain the growth of other species, and thus reduce the Shannon index. Therefore, high levels of ammonia nitrogen reduced the bacterial diversity while increasing the biomass density in the bio-filters. Zou also found that increasing ammonia concentration could significantly change the bacterioplankton community composition (Zou 2011).
Hence, levels of nutrients in the influents played a major role in biofilm formation. The nutrient levels could not only impair the biomass density but also affect the bacterial diversity and community composition in the bio-filters, as reported in a previous paper (Liao et al. 2013). Velten et al. also found that the striking differences in species richness and dynamics in BAC filters might be due to the concentration of available nutrients (Velten et al. 2011). Both nutrients and native bacteria affected the biomass densities as well as the microbial communities. Untreated lake water with a low nitrogen concentration led to greater microbial dynamics and richness but lower biomass density. The addition of ammonia benefited biofilm development in the GAC filter, but reduced bacterial diversity.
Relationships among the removal of CODMn/AOC/BDOC/, biomass densities and biological activity, as well as bacterial communities, are discussed here.
The order of average CODMn removal by the three columns was: column III > column II > column I, which coincided with the order of biomass density, as shown in Figure 5. Hence, the high biomass density facilitated the organic removal. The result was consistent with that of Ma (Ma et al. 2010), which found that adding NH4C1 as a nitrogen source in the O3-BAC process enhanced the CODMn removal by 5–6%. However, the CODMn removal has no direct relationship with the bacterial diversity. It is not the specific community composition but the biomass density inside the BAC that determines the removal of organics. The CODMn removal by BAC filters also depended on many other factors, such as the biological activity and the characteristics of the organic matter.
AOC is the part of DOC that can be easily assimilated by bacteria, the amount of AOC removed should be consistent with the amount of biomass formed. However, the order of AOC removal (column III > column I > column II), does not align perfectly with that of biomass formation (column III > column II > column I). It may due to the fact that AOC removal by BAC was calculated by the gap between the BAC influent and effluent, and AOC is measured by only two kinds of bacterial species (Pseudomonas fluorescens strain P17 and Spirillum strain NOx) (Liu et al. 2002), while the microorganisms in the BAC filter included bacteria (rods, cocci, filamentous bacteria), fungi and protozoa which all contribute to AOC removal (Chien et al. 2008).
The amounts of AOC removed in the three columns were 280 μg/L, 224 μg/L and 287 μg/L, respectively. The biomass density of the three columns were 32.5, 45 and 58.5 nmol-P/cm3 (1 nmol-P/cm3 is equal to 108 CFU). Thus, the normalized values of AOC removal per unit of biomass in columns I, II and III were 8.61, 5.81, and 4.91 μg/108 CFU. The order was completely consistent with that of biological activity, indicating that the AOC removal depended on the biological activity. It might also depend on many other factors, such as the amount of biomass and the bacterial community.
The amount of BDOC removal by BAC filters is associated with the biomass density. It might be the microorganisms growing on the GAC media that played the critical role in BDOC removal during steady state operation. In addition, about 16% BDOC was assimilated as biomass (measured by AOC), which illustrated that there are sufficient carbon sources in the influent of the three columns.
Overall, the amounts of AOC removed by column III (287 μg/L) were higher than those of the other two columns (280 and 224 μg/L). Moreover, the BDOC removed by column III (1.86 mg/L) was also greater than that of the other two columns (1.72 and 1.55 mg/L). Meanwhile, the average removal efficiency of CODMn for column III (61.6%) was higher than that of column I (49%) and column II (53%). Hence, the addition of nitrogen (column III) enhanced the bio-stability of the effluent for this source water with low ammonia concentration.
The effects of diverse influent nutrients on biomass growth and biofilm development in GAC filters in winter were studied. Results indicated that nitrogen addition to low-nitrogen containing source water benefited the biofilm formation in a GAC filter. A nitrogen-rich environment was favorable to biological growth. It could shorten the start-up time, stimulate the bacteria to remove more organic matter and enhance the bio-stability of the effluent. Carbon was not a limiting factor when the organic matter in raw water was predominantly composed of small molecular organics. The indigenous bacteria source was also an important factor under similar nutrient conditions. It is the biomass density, not the specific community composition, inside the BAC which determined the organic removal efficiencies in this study.
This work was financially supported by the National Natural Science Foundation of China (No. 51508209; No. 51578250), the China Postdoctoral Foundation (First Prize, 145723), the Natural Science Foundation of Fujian province (No. 2015J05102; No. 2014J01196), the Scientific Research Project for Young Teachers supported by the Education Department of Fujian Province (No. JA15033), and the Science Research Foundation of Huaqiao University (No. 15BS105). The authors are also greatly appreciative of the financial support from the Key Project of International Cooperation of Science and Technology of Fujian (No. 2014I0013).