L a b o r a t o r y - S c a l e I n v e s t i g a t i o n s o f A l g a l Toxin R e m o v a l by W a t e r T r e a t m e n t T. Hall, DPhil (Member)*, J. Hart, PhD**, B. Crol l , PhD (Fellow)*** and R . Gregory, MPhil, CEng, MI C hem E (Member)****
The paper describes some water-treatment processes which have been investigated on a laboratory and pilot- plant scale for their effectiveness in removing toxic algal cells and the dissolved toxins microcystin-LR and anatoxin-a. Oxidation with ozone or potassium perman- ganate, or treatment by biological activated carbon, were found to be the most effective processes for removal of the dissolved toxins. Chlorination was effective only for the removal of microcystin-LR. Toxins, contained within algal cells, could be removed effectively by coagulation, clarification and filtration under suitable conditions. Consideration of the structure and properties of micro- cystin variants suggests that treatments which are suitable for removal of microcystin-LR and anatoxin-a should be suitable for removal of other microcystins.
Key words: Algae; anatoxin; microcystin; toxin; water treatment.
In t roduct ion
Blooms of blue-green algae can give rise to the pro- duction of toxins which might contaminate sources abstracted for potable water('). Surveys in 1990 and 1991(2) in the U K showed that a high percentage of algal scums were toxic. Because of the potential risk to water supplies, suitable water-treatment strategies need to be identified. The UK water industry commissioned a series of research projects, funded initially by the Foundation for Water Research (FWR) and subsequently by U K Water Industry Research Limited (UKWIR), to investi- gate the effect of water treatment on both toxic algal cells and dissolved toxins. The objectives of the research were:
(i) To identify suitable treatment methods for the destruction and removal of dissolved toxins; and
(ii) To examine the effect of treatment processes on toxic algal cells and their associated toxins.
The following treatment processes were examined:
(a) Coagulation, settlement and filtration; (b) Oxidation; and (c) Activated carbon adsorption, using either granular
(GAC) or powdered (PAC) carbon.
*Principal Technical Specialist, WRc, Swindon, UK **Senior Development Engineer, Linnhoff March Ltd, Northwich, UK "'Consultant, WQM Ltd, Buckden, UK. ****Principal Technical Specialist, WRc, Swindon, UK
The laboratory and pilot-scale investigations were undertaken in two phases. Initially, the studies were conducted spiking only with dissolved toxins, because no suitably accurate analytical techniques were available to enable the quantitative assessment of both intra- and extra-cellular toxin concentrations. However, develop- ments and improvements in the analytical techniques subsequently enabled quantitative studies to be undertaken with toxic algal cells. Detailed descriptions and results of the research are given in several report^(^-^), and the work also has been summarized in various paped-l0).
Toxins o f I n t e r e s t
Two of the most common species of algae in the U K which can produce toxins are Microcystis aeruginosa and Anabaena Jos aquae. T h e former produces a group of compounds called microcystins which act as hepato- toxins, capable of producing severe liver damage, and one of the most commonly found microcystins is micro- cystin-LR (MLR). Anabaena species produce anatoxin-a (ANA) - a neurotoxin which can attack the central nervous system. The structure of ANA is different from that of the microcystins. I t is a much smaller compound, with a molecular weight of 165 compared to approxi- mately 1000 for the microcystins. It might, therefore, be expected to have different properties and behave differently during water treatment.
T r e a t m e n t Studies
Laboratory tests were used to examine the different treatment methods. For the granular activated-carbon tests, however, small-scale pilot tests were also under- taken, because it was not possible to simulate biological GAC treatment within the laboratory. For all experi- ments, samples of raw or treated waters were collected from suitable sources, and spiked with either dissolved toxin (commercially available) or with cell cultures of toxic Microcystas aeruginosa (strain CCAP 1450/ 10)t.
Equivalent studies were not possible with toxic Anabaena Jos aquae, due to culturing difficultie~(~). Samples were also obtained from three reservoirs with blue-green algal blooms, but none of the samples had high enough toxin concentrations to allow useful laboratory studies to be carried out, or to provide any reliable information on the performance of treatment for toxin removal. Studies included:
tProvided by the Institute of Freshwater Ecology, Ambleside, UK
0 J .C IWEM, 2000, 14, A p r i l 143
T. Ha l l , J. H a r t , 6 . C r o l l and R . Gregory on
(a) Jar tests to examine the effect of coagulant dose on
(b) Tests with a nanofiltration membrane; (c) Oxidation tests with ozone, chlorine, chlorine
dioxide, potassium permanganate, hydrogen per- oxide and ultraviolet radiation;
removal of toxin;
(d) Tests to examine the effect of PAC type and dose; (e) GAC rapid column tests to examine the effect of
carbon type and contact time; the results from these tests were then used to predict full-scale performance by modelling;
(f) GAC pilot column-tests to assess the potential of biological activity in toxin removal; and
(g) Prediction of removal of other toxin variants based on their chemical structure.
Coagulation, Settlement and Filtration Several studies have shown that conventional
coagulation and clarification treatment methods are ineffective at removing dissolved toxins(' ' ,I2), and the results of experiments with dissolved toxins confirmed these findings. Physico-chemical treatment has been shown to cause cell lysis and toxin relea~e('~1'~). However, experiments which were conducted with cultured Microcystis showed that mixing conditions and pH changes representative of those found in water treatment, did not cause cell lysis or toxin release(@.
Coagulation with aluminium sulphate was shown to be effective at reducing the total toxin concentration due to the removal of the algal cells, rather than the extra- cellular toxin (Fig. I)(@- Similar results were obtained with ferric sulphate and have been found with other studies(I5).
Oxidation of Dissolved Toxins Fig. 2 shows that when oxidising agents were
applied individually to raw water, only ozone and potassium permanganate were effective in removing the dissolved MLR in the range 3.5-8.6 pg/l, although an impracticable potassium permanganate dose of 10 mg/l was required to achieve a 50% reduction in the toxin concentration. When applied to treated water (Fig. 3 ) ,
8 1.5 I 1.0 5
\ I 0
some removal of the dissolved MLR was achieved with chlorine dioxide, but the doses which were required to achieve an effective reduction in the toxin concentration were in excess of 6 mg/l- a dose which cannot be applied in the UK because of the regulations concerning chlorite and chlorate. As with the raw-water tests, hydrogen peroxide was ineffective at reducing the MLR concen- tration. However, relatively low doses (2 mg/l) of ozone or potassium permanganate resulted in removal of the toxin to below limits of detection. Removal was greater for treated than raw water because of its lower oxidant demand.
For dissolved ANA (2.4-4.3 &I), as with MLR, oxidation was more effective when the oxidant was applied to treated water (as shown in Fig. 4). For both ozone and potassium permanganate, a dose of 2 mg/l was sufficient to remove the toxin to below limits of detection - a reduction in the toxin concentration of more than 90/o. A comparison between the removal of the two toxins by ozonation indicated that MLR was the more readily removed, as found elsewhere(16).
Potassium permanganate is a possible treatment method for both dissolved toxins, but unfortunately it was only effective at doses to treated waters which would be unacceptable in terms of final manganese concen- trations. This could be overcome by dosing to raw or clarified waters, prior to filtration, but higher doses would be required to satisfy the oxidant demand.
Chlorination has been found to be effective in reducing the toxicity of algal extracts, although chlor- amination was shown to be ineffective("). These results, however, were for a (high) chlorine concentration of 15 mg/l maintained for a contact time of 30 mins, with greater effectiveness found at lower pH.
In order to simulate the effect of chlorine at con- centrations similar to those used for final disinfection of potable water, experiments were performed using a treated water spiked with MLR and an applied chlorine dose of 1.7 mg/l, which gave a free chlorine residual of 0.7 mg/l after a contact time of 30 mins. Samples were taken after 30 mins and 22 h - the latter to simulate the impact of chlorine in distribution.
Fig. 7. Effect of coagulation with aluminium sulphate on concentration of intra- and extra-cellular MI R
144 0 J CIWEM, 2000, 1 4 , A p r i l
Laboratory-Scale Inves t iga t ions o f A lga l Toxin Removal b y Water Treatment
'1 0.8 E c 3 0.6 's
.a- - - - - - - -- - -, .. -- ', --..
-* -- --__ --- a- *.
-. .. r(
x . . . ) ( -. -chloriM,dioxide -*- H y d w peroxide - .I - P-im permanganate
--+.X + - ~ f +. + + ~+
0 1 2 3 4 5 6 7 8 9 10 Oxidant dose (IT@)
Fig. 3. Comparison of different oxidizing agents for removal of ML R from treated lowland river water
0.8 .- C '1 0.7 E
0.8 'i? : O5
0 1 2 3 4 5 e oddant do- (w)
Fig. 4. Comparison of different oxidants for removal of ANA from raw and treated lowland water
0 J CIWEM. 2000, 14, A p r i l 145
T. Ha l l , J. Hart, 8 . C r o l l and R. G r e g o r y o n
The results of these tests showed that the effective- ness of chlorination was highly dependent upon p H (Table as found in other studies(). I t was concluded that, for MLR only, chlorination under waterworks dis- infection conditions can cause substantial removal, which is reinforced by the effect of residual free chlorine in distribution. Similar chlorination tests were also under- taken with water spiked with ANA solution(4). However, as found elsewhere(7), no discernible removal of this toxin occurred.
Laboratory-Scale Investigations of Algal Toxin Removal b y Water Treatment
Table 2. Effect of chlorination contact time and dose on MI R concentration for a treated lowland water spiked with Microcystis cells
Chlorine MLR concentration Contact time
7.7 7.5 7.5 8.5 8.5
7.7 7.7 7.7 7.7
0 1.5 2.5 1.5 2.5
0 1 2 3
lntra Total f
T. Ha l l , J. Hart, B. C r o l l a n d R . Gregory on
-Influent - - 7.5 min6 -7.6 mins (-I) - * * 15mkrs -15 mins (model) - I
0 2 4 6 8 10 12 Time of operation (w)
Fig. 6. Pilot-scale results for MI R removal by biological GAC compared with modelled results for ordinary GAC
T r e a t a b i l i t y o f M i c r o c y s t i n s other t h a n MLR
The studies used MLR because the analytical method was developed for this application. However, a range of other microcystins exists - many of which have been isolated from natural waters - and information was needed on their likely significance in water treatment. Computer models have been used to predict the proper- ties of the microcystin variants, based on their chemical structures, relevant to their removal by water treat- ment(). The modelling suggested that most variants would be adsorbed by activated carbon similarly to, or better than, MLR. Therefore any strategy for using activated carbon, based upon the available data for MLR, would probably remove most of the other microcystins.
Attempts to model the reactivity of the microcystin variants with oxidants were unsuccessful because of the complexity of their molecular structures. A principal mechanism of action of oxidants, particularly ozone and chlorine, on organic compounds is the breakdown of double bonds. Any modifications to the basic microcystin structure which increases the degree of double bonding in the molecule would therefore be expected to enhance its reaction with ozone or chlorine. It was concluded (from consideration of the other functional groups in the variants) that some variants would be expected to be more reactive with oxidants than MLR, although the effect may not be important in practical terms because the basic molecular structure is not changed radically. For the same season, the other variants would not be expected to be much less reactive than MLR with oxidants. There- fore any strategy for oxidant application based on MLR data is expected to be equally as effective for the other microcystins.
The modelling approach for biodegradability can only class compounds as biodegradable or non- biodegradable and cannot provide any further quantifi- cation to the degree of biodegradability. The model classed MLR as biodegradable and the differences between microcystin variants did not change this classifi-
cation. Therefore all the variants would be expected to show similar biodegradability to MLR. This is significant in relation to the performance of biological GAC processes.
lmpl i c a t i o n s f o r W a t e r T r e a t rn ent
The investigations were laboratory based, using spiked solutions, and this should be taken into account when assessing the efficacy of full-scale treatment. There are a number of key findings from the treatment studies and those reported in the literature with respect to efficient removal of intra- and extra-cellular microtoxins. These are:
(i) Chemical doses and their application should be optimised; under-dosing of coagulant and non- optimal pH will result in poor removal of algae in clarification and problematical filtration, with the risk of breakthrough of algal cells containing toxin;
(ii) Some oxidants can be usefully dosed before clarification and filtration, but only with care, to avoid cell lysis and to limit problems with T H M formation;
(iii) Ozone-GAC facilities which are installed to remove pesticides should be robust enough to remove toxins effectively, especially if the GAC supports substantial biological activity;
(iv) The effectiveness of treatment plants without ozone but with GAC, will depend upon the EBCT, degree of biological activity on the GAC; extent of exhaustion of the GAC and magnitude and duration of the toxin occurrence;
(v) Treatment plants without ozone and GAC might be expected to be effective in removing algae and dissolved toxins only if coagulation, clarification, filtration and superchlorination-dechlorination (with ct >15 mg.min/l) are carried out effectively; and
148 0 J CIWEM, 2000, 14, A p r i l
Laboratory-Scale Invest igat ions of Algal Toxin Removal by Water Treatment
(vi) Provided that slow sand filter plants are managed well, they should remove algal cells effectively. Because of the biological activity in slow sand filters and long contact times, some removal of dissolved toxins should be expected, but this capability has not been evaluated. However, modern slow sand- filter plants with pre-ozonation and sand-GAC sandwich should be effective.
1. Dissolved toxins can be removed effectively to less than 1 pg/l under conditions which are normally used in water treatment by biologically active GAC, ozone, potassium permanganate and chlorine (microcystin-LR only).
2. Coagulation with clarification and filtration removes algal cells with intra-cellular toxin. These processes do not release toxin from the cells, but sufficient cells could penetrate particulate removal processes (except membranes) to be of toxicological concern.
3. Intra-cellular microcystin-LR can be removed by ozone and chlorine at doses which are normally used in drinking water treatment.
4. Physical and chemical properties of microcystin variants were assessed to predict their treatability by GAC, oxidation and biodegradation. The results suggested that any treatment strategies which are suitable for removal of microcystin-LR should be suitable for other microcystin variants.
5. The results of the investigations provide the water industry with a high degree of confidence that dissolved and intra-cellular microcystin-LR, and dissolved anatoxin-a, can be effectively removed by conventional water-treatment processes.
The work referred to in this paper was funded initially by the Foundation for Water Research and subsequently by UK Water Industry Research.
(1) Toxic Blue-Green Algae. A report by the National Rivers Authority. NRA Water Quality Series No. 2, September 1990.
(2) COOD, G. A. AND BELL, S. G. The Occurrence and Fate of Blue-Green Algal Toxins in Freshwaters. R&D Report 29, Dundee University, HMSO 1996.
(3) HART, J. ANO STOTT, P. Microcystin-LR Removal from Water. FWRReport No. FR 0367,1993.
(4) CARLILE, P. R. Further Studies in Investigate Microcystin-LR and Anatoxin-a Removal from Water. FWR Report No. 0458, 1994.
(5) UKWIR. Pilot-scale GAC Tests to Evaluate Algal Toxin Removal. Report Ref 96/DW/07/1, 1995.
(6) UKWIR. The Fate of lntracellular Microcystin-LR during Water Treatment. Report Ref. 96/DW/07/4,1996.
(7) UKWIR. Algal Toxins: Occurrence and Treatability of Anatoxin and Microcystins. Report Ref 97/DW/07/E, 1997.
(8) FAWELL, J. K., HART, J., JAMES, H. A. ANO PARR, W. Blue-green algae and their toxins - analysis, toxicity, treatment and environmental control.
(9) CROLL, B. ANO HART, J. Algal toxins and customers. In Proc. of UKW/R- AWWARF Technology Transfer Conf, Philadelphia, October, 1996.
(10) HART, J., FAWELL, J. K. AND CROLL, B. The fate of both intra- and extra- cellular toxins during drinking water treatment. SpecialSubjectNo. 18, SSl8-1-6 IWSA World Congress, 1997.
(11) HOFFMAN, J. R. H. Removal of microcystis toxins in purification processes. Wafer SA, 1976,2, (2), 58.
(12) LAHTI, K. AND HUSTVERTA, L. Removal of cyanobacterial toxins in water treatment processes: Review of studies conducted in Finland. Waf. Suppk, 1989,7, (4), 149.
(13) DICKENS, C. W. S. ANO GRAHAM, P. M. The rupture of algae during abstraction from a reservoir and the effects on water quality. J. Wat. Serv. Res. & Techno/., 1995,44, (I), 29.
(14) FOUNDATION FOR WATER RESEARCH. Detection and Removal of Cyano- bacterial Toxins from Freshwaters. FWR Report No. FRO211,1991.
(15) LAM, A. K-Y., PREPAS, E. E., SPRINK, D. AND HURDEY, S. E. Chemical control of hepatotoxic phytotoxic btooms: implications for human health. Wat. Res., 1995,29, (8), 1845.
(16) ROSSITANO, J. AND NICHOLSON, 6. C. Destruction of cyanobacterial toxins by ozone. Paper presented at lnternational Ozone Association meeting, Sydney, Australia, February, 1996.
cyanobacterial peptide hepatotoxins by chlorine and chloramine. Wat. Res., 1994,28, (6), 1297.
(18) CHORUS, I. AND BARTRAM, J. (Eds). Toxic Cyanobacteria in Water. f& fN Spon on behalf of WHO, 1999.
Wat. Supph, 1993,11, (3/4), 109.
(17) NICHOLSON, B. c. , ROSITANO, J. AND BURCH, M. D. Destruction of
0 J.CIWEM, 2000, 14, Apr i l 149