Turbidity reduction in drinking water by coagulation-flocculation with chitosan polymers

Turbidity reduction by coagulation-flocculation in drinking water reduces microbes and organic matter, increasing effectiveness of downstream treatment. Chitosan is a promising household water coagulant, but needs parameters for use. This study tested the effects of chitosan dose, molecular weight (MW), degree of deacetylation (DD), and functional groups on bentonite and kaolinite turbidity reduction in model household drinking water. Higher MW or DD produced greater reductions. Highest reductions were at doses 1 and 3 mg/L by MW >50,000 or >70% DD (residual turbidity <5 NTU). Higher doses did not necessarily continually increase reduction. For functional groups, 3 mg/L produced the highest reductions by lactate, acetate, and HCl, and lower reductions of kaolinite than bentonite. Doses where the point of zero charge was observed clustered around 3 mg/L. Chitosan reduced clay turbidity in water; effectiveness was influenced by dose, clay type, MW, DD, and functional groups. Reduction did not necessarily increase with MW. Bentonite had a broader effective dose range and higher reduction at the optimal dose than kaolinite. Chitosans with and without functional groups performed similarly. The best of the studied doses was 3 mg/L. Chitosans are promising for turbidity reduction in low-resource settings if combined with sedimentation and/or filtration. This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/). doi: 10.2166/wh.2019.114 om http://iwaponline.com/jwh/article-pdf/17/2/204/623610/jwh0170204.pdf er 2020 Ampai Soros Mark D. Sobsey Department of Environmental Sciences and Engineering, Gillings School of Global Public Health, University of North Carolina Chapel Hill, CB 7431, Chapel Hill, NC 27599, USA James E. Amburgey Civil and Environmental Engineering Department, William States Lee College of Engineering, University of North Carolina Charlotte, 9201 University City Boulevard, Charlotte, NC 28223-0001, USA Christine E. Stauber Lisa M. Casanova (corresponding author) Division of Environmental Health, School of Public Health, Georgia State University, P.O. Box 3995, Atlanta, GA 30302, USA E-mail: lcasanova@gsu.edu This article has been made Open Access thanks to the generous support of a global network of libraries as part of the Knowledge Unlatched Select initiative.


INTRODUCTION
Effective reduction of turbidity is one of the primary goals in effective drinking water treatment because of potential interference with downstream treatment processes and negative effects on consumer acceptance. Turbidity might interfere with filtration by clogging the filter prematurely. It can interfere with chemical disinfection by creating oxidant demand, UV irradiation by blocking light transmission, and reduce the efficacy of both by providing protection to microbes in aggregates or internal to other particles. Turbidity also has negative impacts on consumer acceptance of water; visible cloudiness in finished water may create the perception for consumers that it is not clean or safe to drink. Turbidity is not necessarily a direct measure of microbial contamination, but microbes are often associated with particles in water. Therefore, removing turbidity serves a two-fold purpose in water treatment: it removes some microbes, while reducing the levels of organic matter and other particles, increasing the effectiveness of downstream treatment processes.
For drinking water, the World Health Organization has suggested <1 nephelometric turbidity unit (NTU) for water that will undergo disinfection and <4 NTU for water to be acceptable to the naked eye (World Health Organization ). The US Environmental Protection Agency sets the maximum level of turbidity in finished drinking water at 1 NTU and at no time >5 NTU; the vast majority of water treatment plants must be less than 0.3 NTU 95% of the time with a maximum of 1 NTU (United States Environmental Protection Agency ).
Turbidity reduction is one part of effective water treatment processes in large-scale centralized treatment plants, small community systems, and at the household level. In areas without water treatment systems or with impaired sources of drinking water, water may need treatment at the household level, or point of use (POU) to render it safe to drink. This household level treatment can include turbidity reduction, which should be followed by POU filtration and ideally disinfection. Turbidity removal removes some microorganisms, but most importantly prepares water for these downstream treatment processes.
Coagulation-flocculation, a treatment process where colloids in water are destabilized so they can aggregate and be physically removed, can effectively reduce turbidity when combined with sedimentation and/or filtration. An example of a combined POU system would be one where water is collected in a traditional container (such as a clay jar), coagulant is added, and turbidity can flocculate and settle. The water can then be decanted into a household level filter (e.g., a ceramic pot filter or biosand filter), after which, the filtrate can be disinfected and safely stored and the floc disposed of as waste.
Conventional coagulants used in large-scale water treatment are largely metal salts such as aluminum sulfate, ferric sulfate, and ferric chloride, which depend on the pH of water and precise dosing to produce consistently high coagulation efficiency (Yang et al. ). When coagulation using metal salts is done, the resulting sludge also contains residual metals that must be properly disposed of so that they do not pollute. These limitations of conventional coagulants make them less suitable for household level water treatment, where people need simple but robust and safe methods to treat their water at home. Organic polymer coagulants are an alternative to metal salts in the household setting.
Chitosan, a biopolymer of D-glucosamine and N-acetyl-D-glucosamine produced by deacetylation of chitin, has properties of a promising household-level water coagulant: positively charged when dissolved, non-toxic, and biodegradable. Based on its structure, chitosan could be an effective coagulant for negatively charged particles in water by charge neutralization, electrostatic patch, or interparticle bridging mechanisms. It has been used for reduction of contaminants in wastewaters (Chi & Cheng ; Rizzo et al. b; Renault et al. ). Studies have found that chitosan coagulation can remove turbidity at low doses (1-10 mg/L) (Divakaran & Pillai ; Rizzo et al. a; Brown & Emelko ). For effective treatment of household drinking water, there are parameters for its use that need to be established, including selection of alternative chitosan polymers and optimal dosing. To choose the optimal chitosan polymer, properties that influence coagulation performance, including molecular weight (MW), deacetylation (DD), and the addition of functional groups (Yang et al. ), need to be understood. Therefore, the purpose of this study was to determine the effect of dose, MW, DD, and the addition of functional groups on the efficacy of chitosans for turbidity reduction of two different clays, kaolinite (a 1:1 clay) and bentonite (a 22:1 clay), in artificial surface water used as a model for household drinking water.

Selection and preparation of chitosans
A total of 17 chitosans were tested (11 acid-soluble and six water-soluble modified) (Table 1). MW and DD were obtained from the vendor (Table 1). To study the effects of MW on turbidity reduction, five chitosan polymers with different MW and similar DD (∼90%) were compared: 5,000, 50,000, 100,000, 600,000, and 1,000,000 Da. To test the effects of DD on turbidity reduction, a set of chitosan polymers with different DD and approximately the same viscosity and MW (≈50,000 Da), were used: 70%, 75%, 80%, 85%, 90%, and 95% DD. Six chitosans modified with functional groups to increase water solubility were tested: chitosan acetate, chitosan lactate, chitosan HCl, carboxymethyl chitosan, and two commercially available coagulants made of proprietary formulations of chitosan acetate (acetate-SK) and chitosan lactate (lactate-SK).
Stock solutions of chitosan were made for all polymers at 10,000 mg/L (1%). Chitosan powder was dissolved in 0.5% acetic acid (Roussy et al. ) and stirred at room temperature until totally dissolved. Stock solutions were stored at room temperature (25 C). The stock solutions of modified (water-soluble) chitosans were prepared similarly, using deionized water instead of acetic acid. The pH of all stock solutions was 3.5-4.5 with the exception of carboxymethyl chitosan, which was pH 7.5.

Preparation of test waters
Two mineral clays, kaolinite, a 1:1 clay, and bentonite, Sodium chloride (NaCl) was used as an adjustment material for TDS and tannic acid (University Lake, Chapel Hill, NC) was used as an adjustment material for TOC. Water pH was not adjusted after adding clay, NaCl, and tannic acid and water pH ranged from 7 to 7.5. Turbidimeter, Hach, Loveland, CO). Water pH was also measured before and after the jar test experiment using a pH meter.
Doses of chitosans were 1, 3, 10, and 30 mg/L. These doses were selected because they are in the same ranges as optimum doses of conventional coagulants (2-5 mg/L for aluminum and 4-10 mg/L iron salt coagulants) (WHO ) and in the ranges of chitosan effective doses for turbidity reduction in preliminary studies (data not shown). Three replicates, plus one control (no chitosan) for natural settling, were performed for each set of experimental conditions.
The parameters tested for effects on coagulation performance were chitosan type and dose (1, 3, 10, and 30 mg/L) and turbidity type (bentonite or kaolinite).

Data analysis
Turbidity reduction was calculated as percent turbidity reduction relative to the natural settling control:

Effects of molecular weight
The effects of chitosan polymer MW on reduction of bentonite turbidity at varying chitosan doses are shown in Figure 1.  The effects of MW on reduction of kaolinite turbidity at varying chitosan doses are shown in Figure 2. Overall, kaolinite reduction differed significantly by MW (p < 0.0001).
The highest reductions in kaolinite turbidity were achieved at doses of 1 and 3 mg/L (like bentonite), and higher doses did not produce higher turbidity reduction, except MW 5,000 Da at dose 30 mg/L. Higher MW chitosans produced greater turbidity reductions, with poor reductions at the lowest MW.
MW 50,000, 100,000, and 1,000,000 Da chitosans had kaolinite turbidity reduction ranging from 87% to 90% at 1 and 3 mg/L dose. At 1 and 3 mg/L dose, there was no significant difference in kaolinite reduction between 50,000, 100,000, and 1,000,000 Da (p > 0.05). MW 600,000 was less effective than the other MWs, with <25% reduction.
The lowest MW, 5,000 Da, performed very poorly (<1% reduction). All MWs, even those that showed reduction at 1 and 3 mg/L, performed poorly for kaolinite reduction at 10 mg/L and 30 mg/L doses (<5% reduction for all MWs).
The exception was 5,000 Da, which increased from <1% at lower doses to ∼80% reduction at 30 mg/L.
For the same MW at the same dose, reduction of bentonite was significantly better than reduction of kaolinite at both 1 and 3 mg/L dose (one-way ANOVA, Tukey's posttest, p < 0.05). The exception was 50,000 Da at 1 mg/L, which showed similar reductions of bentonite and kaolinite (p > 0.05). At 10 mg/L, reductions of bentonite were significantly better than reductions of kaolinite for the same MW (one-way ANOVA, Tukey's post-test, p < 0.05). Reduction of bentonite was significantly higher than reduction of kaolinite for the same MW at 30 mg/L dose (one-way ANOVA, Tukey's post-test, p < 0.05) except for 5,000 (kaolinite reduction ∼80%, bentonite reduction ∼41%). There were three MW chitosans that effectively removed both kaolinite and bentonite at 1 and 3 mg/L: 50,000, 100,000, and  1,000,000 Da. These chitosans could achieve >90% turbidity reduction, and bring residual turbidity from 30 to 70 NTU to <3 NTU for bentonite and <5 NTU for kaolinite.
Overall, chitosans >50,000 Da at doses 1 and 3 mg/L brought kaolinite and bentonite turbidity to the <5 NTU standard, and bentonite turbidity to the <1 NTU standard ( Table 2).
At 3 mg/L, 70% and 75% DD had significantly higher kaolinite reductions than 90% and 95% DD (p < 0.05), although all of them had reductions >91%. At 1 mg/L, kaolinite For the same DD, reduction of bentonite was significantly higher than reduction of kaolinite at 10 and 30 mg/L doses (one-way ANOVA, Tukey's post-test, p < 0.05), but still poor compared to 1 and 3 mg/L doses. For the same DD, reduction of kaolinite was significantly lower than reduction of bentonite at 1 and 3 mg/L (one-way ANOVA, Tukey's post-test, p < 0.05). Overall turbidity reduction was lower for kaolinite than bentonite, but the highest reductions of both were at 3 mg/L for all six DDs. For 80% DD at 1 mg/L, reductions of bentonite (90.5% (±3.9)) and kaolinite (68.5% (±37.9)) were not statistically different (p > 0.05). Based on these results, 3 mg/L was the optimum dose that exhibited the highest reduction of both bentonite and kaolinite turbidity. At this dose, reduction of bentonite by different DD chitosans was similar (99%), and resulted in residual bentonite turbidity <1 NTU (starting turbidity 32-98 NTU). The reduction of kaolinite at 3 mg/L was also similar across DDs (90%) and brought residual kaolinite turbidity to <5 NTU (Table 3).

Effect of modified functional groups
Six chitosans modified with functional groups were tested:  Except for carboxymethyl, residual bentonite turbidity was lower than 1 NTU at 3 mg/L ( Table 4).
Reduction of kaolinite turbidity differed significantly by functional group and dose (p < 0.0001) (Figure 6). At 1 mg/L kaolinite, reduction was low for all functional groups except acetate-SK (82%). At 10 mg/L, HCl had 85% reduction, but other functional groups had <3%.
At 30 mg/L, all functional groups had <3% reduction. As with bentonite, the highest kaolinite reductions were at   As with unmodified chitosans, chitosan polymers with functional groups demonstrated poorer reductions of kaolinite than bentonite turbidity. At 1, 10, and 30 mg/L, reduction of bentonite was significantly higher than reduction of kaolinite (one-way ANOVA, Tukey's post-test, p < 0.05). For the same functional groups, reduction of bentonite was significantly higher than reduction of kaolinite at a dose of 3 mg/L (one-way ANOVA, Tukey's post-test, p < 0.05). The exception was HCl; reductions of bentonite and kaolinite at 3 mg/L were not statistically significantly different (p > 0.05).

Measurement of zeta potential
All water samples exhibited negative zeta potential ( Zeta potential values and PZC of chitosan coagulation were measured by the titration method ( Table 6). Chitosans that provided high and low turbidity reduction (MW 100,000 Da, 70% DD, 95% DD, and modified chitosan HCl) were selected as representatives to: (1) observe zeta potential over the course of the coagulation process and (2) determine the dose at which the water/coagulant mixture reached the PZC (Figure 7). These doses at which the PZC was observed clustered around 3 mg/L, the dose that resulted in the highest turbidity reductions in jar test experiments. The dose at PZC of chitosan MW 100,000 Da for kaolinite (4.61 mg/L) was higher than that    reduction at doses other than the optimum were variable.
There was better kaolinite reduction by lower DD at a dose of 1 mg/L, however at the higher dose of 10 mg/L, higher DD were more effective than lower DD.
Above about 80% DD, further increases in the DD may not greatly affect coagulation ( This is similar to the optimal dose range for bentonite observed in previous studies, although they found the optimum dose range of water-soluble chitosans was broader than that of acid-soluble chitosans (Chen & Chung ).
As observed for acid-soluble chitosan, water-soluble modified chitosans demonstrated better reduction of bentonite compared to kaolinite turbidity. However, the turbidity reduction of each functional group varied; HCl, acetate, and lactate were more effective than carboxymethyl chitosans. Chitosan HCl produced high bentonite reductions between 1 and 30 mg/L and high kaolinite reduction at 3 mg/L. Acetate-SK was best at the lowest dose and worst at the highest; acetate also had lower removal at the highest dose. Lactate, a slightly stronger acid, was also less effective at the highest dose. The carboxymethyl group had the poorest turbidity reduction among water-soluble chitosans.
When test water was titrated with carboxymethyl chitosan to determine the PZC, it was not reached even at doses up to 50 mg/L; it was the only chitosan where this was observed. Carboxymethyl was worst at the lowest dose and kept improving at higher dosages. Also, simply increasing the solubility of the polymer in water does not necessarily increase turbidity reduction efficacy.
The best of the studied chitosan dosages for reduction of both bentonite and kaolinite turbidity was 3 mg/L, although there was a broader effective dose range for bentonite than kaolinite. A dose range of 1-10 mg/L was effective for bentonite reduction (80-99% reduction with residual turbid- Other studies (Huang et al. ) have also observed differential bentonite and kaolinite reduction by chitosans.
The observed differences in reduction of bentonite and kaolinite turbidity may be related to the cationic exchange capacity (CEC) properties of these clays and suggest that charge neutralization is one of the mechanisms underlying coagulation by chitosans, but probably not the exclusive or dominant mechanism. CEC is the ability of a soil particle to retain and exchange positively charged ions; the higher the CEC, the greater the capacity of clay particles to attract positively charged molecules. Bentonite has a CEC ranging between 0.8 and 1.2 meq/g, which is much higher than that of kaolinite (CEC 0.03-0.15 meq/g) (Kahr & Madsen ; Meier & Kahr ). Higher CEC may lead bentonite to react more rapidly with a cationic polymer like chitosan while kaolinite reacts slowly, as observed in this study.
Higher CEC may also improve coagulation by causing bentonite to attach more effectively to the positively charged chitosan polymer. Bentonite, as an expandable 2:1 clay, also has a higher specific surface area (40-800 m 2 /g) than kaolinite (5-40 m 2 /g), a 1:1 non-expanding clay ( Coagulation-flocculation can work via multiple mechanisms; depending on the situation, they may occur together, or one mechanism may dominate over others (Bratby  Charge neutralization is also stoichiometric and susceptible to overdose as coagulant amount increases. Therefore, it can be inferred that charge neutralization is a contributing but not primary mechanism for coagulation by chitosans in water at near neutral pH. In a system where a charged coagulant is applied to dispersed particles of opposite charge, the bridging model may explain part but not all of coagulant behavior (Bratby ).
In this study, MW influenced turbidity reduction more than

CONCLUSIONS
• Chitosan polymer MW affected bentonite and kaolinite turbidity reduction, with higher MW more effective than lower MW.
• DD of chitosans had less impact on bentonite and kaolinite turbidity removal than did MW.
• Low doses of chitosan (1-10 mg/L) were effective for removing up to 93% of kaolinite and 99% of bentonite turbidity.
• Of the doses tested, 3 mg/L gave the highest removal of both bentonite and kaolinite turbidity.
• Acid-soluble chitosans were as effective as water-soluble chitosans for bentonite and kaolinite turbidity removal.
• The optimum dose range for effective bentonite and kaolinite turbidity removal was similar for acid-and water-soluble chitosans.
• Interparticle bridging and charge neutralization played a role in bentonite and kaolinite turbidity coagulation, but the electrostatic patch model may explain observed coagulation behavior.
• Measured points of zero charge of chitosans during bentonite and kaolinite turbidity coagulation were close to the optimum chitosan doses obtained from jar test experiments.
• Chitosans have the potential to serve as effective alternative coagulants for the removal of turbidity from water.
• Bentonite, a 2:1 clay, and kaolinite, a 1:1 clay, responded somewhat differently to chitosan coagulation-flocculation and sedimentation for turbidity reduction, perhaps due to their differences in structure, surface charge distribution, and reactivity with water and dissolved ions in water.