Original article
Lead removal from wastewater using faujasite tuff K. M. Ibrahim Æ T. Akashah
Abstract The Jordanian chabazite-phillipsite tuff and faujasite-phillipsite tuff have suitable mineralogical and technical properties that enable them to be used for ion-exchange processes. These include suitable grain size and total cation exchange capacity, acceptable zeolite content, good attrition resistance and high packed-bed density. The chabazite-phillipsite tuff (ZE1 and ZE2) has an excellent efficiency to remove Pb and an acceptable efficiency to remove Fe, Cu, Zn and Ni from effluent wastewater of a battery factory and other industries. The faujasite-phillipsite tuff (ZE3) is 6.97 times more efficient than the ZE1 and ZE2. A bed of ZE3 (1,000 kg) loaded in a 1.17-m3 column is capable of cleaning about 2,456 m3 of the effluent from the factory at a cost of US $200/ton. The wastewater is contaminated with 20 ppm Pb in the presence of competing ions including Ca (210 ppm), Na (1,950 ppm) and Fe (169 ppm). This quantity of wastewater is equivalent to 120 working days of effluent discharge from the factory at a flow rate of 20 m3/day. Keywords Lead removal Æ Faujasite Æ Chabazite Æ Industrial wastewater Æ Jordan
Received: 19 January 2004 / Accepted: 8 April 2004 Published online: 22 June 2004 ª Springer-Verlag 2004 K. M. Ibrahim (&) Department of Earth and Environmental Sciences, Hashemite University, P.O. Box 330101, 13133 Zarqa, Jordan E-mail:
[email protected] Tel.: +962-5-3826600 ext 4332 Fax: +962-5-3826823 T. Akashah Department of Chemistry, Hashemite University, 13115 Zarqa, Jordan
Introduction Lead contamination in the environment has long been recognized as a cause of serious health problems, and increasing pressure is being placed on industries to reduce their lead wastes (US EPA 1993). Owing to its toxicity to animals and plants, lead is one of the most undesirable metals in the waste streams. According to the environmental quality standards, the maximum allowable concentration of Pb for drinking water is about 0.05 mg L)1 (Colella 1995). Current abatement and remediation procedures for wastes containing lead and other heavy metals include pH adjustment with hydroxides, other precipitation methods, reverse osmosis, coagulation-sedimentation and the use of organic ion-exchangers. Such processes, however, have drawbacks that limit their use (Water Treatment Handbook 1991). Natural zeolites have an excellent cation exchange capacity among other naturally occurring products, which enables them to be used as cation exchangers. They frequently display good selectivity for heavy metal cations, which makes them valuable for the purification of industrial wastewater (Colella 1996). Very interesting results were obtained about lead removal from water (Blanchard and others 1984; Colella and Pansini 1988; Pansini and Colella 1989, 1990; Groffman and others 1992; Grube and Herrmann 1993; Pansini and others 1996; Kazemiana and others 2001). Colella and Pansini (1988) and Pansini and Colella (1989) demonstrated the favorable behavior of Naloaded chabazite towards Pb. Selectivity and efficiency were as high as 91 and 76% respectively, provided that the concentration of competing ions is less than that of tap water. As far as Na-phillipsite is concerned, Pansini (1996) reported that all data give high values of working cation exchange capacity, selectivity and efficiency even in the presence of competing cations. Of the natural zeolites, chabazite, phillipsite and faujasite, which occur in Jordan in large deposits associated with the Quaternary Aritayn Volcaniclastic Formation (Ibrahim 1996), faujasite seems to be the most efficient ion-exchanger for the removal of heavy metal cations, including lead. Some ion-exchange properties of the zeolite-rich Aritayn Volcaniclastic Formation toward several different cations have been investigated in higher concentration and published elsewhere (Attili 1992; Shammout 1993; Ibrahim 1996, 2001; Naser Ed-Deen 1998; Ibrahim and others 2002). This work was carried out to examine the ability of the faujasite-phillip-
DOI 10.1007/s00254-004-1074-4 Environmental Geology (2004) 46:865–870
865
Original article
site tuff and chabazite-phillipsite tuff from the Aritayn Volcaniclastic Formation to remove lead from industrial wastewater using an ion-exchange column. Real effluent wastewater from a lead acid battery factory and simulated wastewater contaminated with Pb and other toxic cations were used in the experiments.
Sampling and laboratory techniques Natural Quaternary zeolitic tuff occurs in several localities in northeast Jordan, including Tell Rimah and Jabal Hannoun volcanoes (Ibrahim and Hall 1996). Two representative bulk samples weighing about 50 kg each were collected from the Tell Rimah volcano (ZE1 and ZE2) and from the eastern flank of Jabal Hannoun volcano (ZE3). Sample ZE1 was subjected to grinding and sieving, whereas samples ZE2 and ZE3 were additionally beneficiated by magnetic separation. Following the mineral processing, the samples were subjected to characterization following the procedures of Ibrahim (1996). Zeolite content was calculated using X-ray diffraction techniques (Ibrahim and Inglethorpe 1996). Cation exchange capacity (CEC) was determined according to Mercer and Ames (1978). The wet attrition and packed-bed density tests were determined as described by Ibrahim (2001). Characterization of the three samples indicates that they have suitable zeolites content and CEC, and are characterized by good attrition resistance and high packed-bed density. Samples ZE1 and ZE2 are characterized by a good stability under acid attack for 24 h at pH 5, and that they lose about 4% of the original crystallinity, whereas sample ZE3 is more stable under the same conditions and loses less than 2% during a contact time of 24 h and less than 5% during a contact time of 48 h. At pH 3, the ZE3 resists the acidic attack in the first 4 h, and then it starts to lose part of its crystallinity. The other technological parameters of the three samples are given in Table 1. To increase the efficiency of the zeolites, the samples were loaded with Na and converted to their Na-form by soaking in 1 M NaCl at 80 C for 2 weeks with continuous stirring prior to purification experiments following recommendations of Bremner and Schultze (1993). A composite industrial wastewater sample (about 150 L) was collected from effluents of the International Storage
Table 1 Technological parameters of the studied zeolitic tuff
Sample no. Zeolite grade (wt%) Zeolite type (wt%) Total CEC (meq/100 g) Grain size (mm) Attrition resistant wt% loss Wet packed-bed density g/cm3
866
Environmental Geology (2004) 46:865–870
Battery Company in the Sahab Industrial City/Amman. Analysis of heavy metals was carried out using an ICP-AES model Optima Perkin Elemer-3000. Three columns were used in this experiment filled with zeolite bed (Table 2). The bottom of each column contains a plug of glass wool to support the zeolite bed and to distribute backwash water and promote uniform collection of the treated water. In the column system, a typical breakthrough curve is plotted as a function of effluent concentration and the volume of effluent (bed volume, BV). The dynamic parameters of a column operating at a constant flow rate have been calculated from the breakthrough curves according to Michaels (1952) and Pansini (1996).
Results and discussion The battery factory The chemical characteristics of the untreated wastewater from the factory are given in Table 3. In this experiment, the efficiency of ZE1 on the removal of Pb from wastewater was examined using ion-exchange columns A and B. The operation conditions are given in Table 2. The electrical conductivity (EC) and pH of the initial solution is equal to 7,400 lS/cm and pH 4, respectively. At the beginning of the experiment, the EC and pH values have increased due to the exchange of Na by the heavy metals and incorporation of H+ in the zeolite structure. Afterwards, EC and pH show diminishing values until reaching the initial values. Figure 1 shows the equivalent amount of Na+ compared with Fe+2 and Pb+2 concentrations, through eluting Naexchanged zeolitic tuff bed with wastewater solution. The Na-curve illustrates the continuous decay of Na concenTable 2 Operational conditions of the columns
Internal radius (cm) Bed length (cm) Bed volume (cm3) Flow rate (mL/min)
Column A
Column B
Column C
2.25
0.70
0.50
25
20
6
400
20
5
15
15
12
ZE1
ZE2
ZE3
48 Chabazite Phillipsite 143 )1.0+0.5 8.9
87 50 Chabazite 42 Phillipsite 341 )0.5+0.25 6.3
84 53 Faujasite 31 Phillipsite 289 )0.5+0.25 7.1
1.09
1.09
0.98
Original article
Table 3 The composition of the wastewater effluent from the studied battery factory Cations
Fe+2
Pb+2
Na+
Ca+2
Concentration (ppm)
169
20
1,950
210
tration during the experiment due to release of Na through the exchange with heavy metals. The values of Ca during this experiment varied between 280 and 345 ppm, which is higher than the initial Ca content shown in Table 2. This increase may be attributed to the leaching of calcite from the zeolitic tuff into solution. As shown in Table 4, the average selectivity of iron using column A is 25.4% and the efficiency is 12.6%, while lead is hardly present in the effluent water. Apparently, the length of mass transfer zone of iron in this experiment (16.6 cm) is lower than the bed length (25.0 cm). A regeneration experiment has been performed by passing 4 M NaCl solution through the column A with a constant flow at a rate of 4 ml/min. Figure 2 shows the elution curves of Fe and lead attained during bed regeneration. The zeolite bed is not 100% regenerated, as traces of heavy metals remain combined within the zeolite structure. In detail, the lead concentration reaches the highest concentration at the beginning of the experiment almost greater
than the initial concentration. This is presumably due to the reverse exchange between the Na solution and lead within the cavities and channels of the zeolites. However, the case with Fe is different. The iron concentration of the treated solution in the experiment is less than 1 ppm compared with 169 ppm iron concentration in the battery effluent wastewater. This indicates that iron is almost irreversibly adsorbed and only minute amounts of the iron that enters the zeolite structure is back-exchanged. As indicated in Fig. 1, lead is absent in the effluent water. The experimental conditions using column B are the same as in the previous one. The dynamic data for column B are given in Table 4. Figure 3 illustrates results of the removal of lead (20 ppm) and iron (169 ppm) from wastewater by using column B, which is loaded with the ZE1 sample. In this experiment, iron is much more preferentially adsorbed than lead, probably because iron is found in the higher concentration, and it has a better opportunity to exchange the sodium ions on the zeolite. The zeolite selectivity for iron using column B is 23.3%, while lead has a selectivity of 12.2% (Table 4). The efficiency of iron is 13.7%, while lead has an efficiency of 4.7%, whereby in this case the total efficiency of the column is 18.4%. The removal efficiency of heavy metal by using column methods decreases with the presence of other alkaline cations like Na, Ca and K. In this column experiment, the small efficiency and selectivity is due to the higher concentrations of Ca and Na, reaching 210 and 1,950 ppm, respectively.
Simulated industrial wastewater Copper, zinc and nickel were also investigated as competing ions with lead to perform the experiment in column
Fig. 1 Relationship between effluent volume and concentrations of Pb, Fe and Na by using ZE1 in column A Table 4 Dynamic data for Fe and Pb by using ZE1 sample with columns A and Fig. 2 B. Vt Total volume at exhaustion point expressed in bed volume (BV); Regeneration of column A after exhaustion Vb Volume at breakthrough point expressed in BV; MTZ Mass transfer zone; CEC* CEC of individual cation (Pb or Fe) at exhaustion point; WEC CEC for the cation at the breakthrough point Column A Dynamic parameters Vb (BV) Vt (BV) MTZ (cm) CEC* (meq/100 g) WEC (meq/100 g) Efficiency (%)=100·(WEC)/ (total CEC)% Selectivity (%)=100·(CEC*)/ (total CEC)%
Column B
Fe 30 60 16.6 36.3 18.1 12.6
Fe 32.5 55 9.6 33.2 19.6 13.7
Pb 350 900 11.4 17.4 6.8 4.7
25.4
23.3
12.2
Fig. 3 Relationship between effluent volume and concentration of Pb and Fe in column B
Environmental Geology (2004) 46:865–870
867
Original article
C to simulate industrial effluent wastewater other than the battery industry. To reduce the effect of the anionic groups and to prevent formation of precipitates in the simulated influent, the standard solutions of the metals were selected in the nitrate form. The pH of the simulated influent was kept acidic (pH=3.5) throughout the experiments to keep the cations soluble in the solution. The samples ZE2 and ZE3 were selected to conduct the experiments by feeding simulated influent containing 10 ppm initial concentrations of each Ni2+, Zn2+ and Cu2+, whereas the initial concentration of Pb2+ is 20 ppm (Table 5). The maximum allowable concentration of the cations was selected as 0.5 ppm to accomplish the evaluation. Table 5 and Figs. 4 and 5 show that the two samples are efficient in the removal of trace amounts of metal ions from the simulated wastewater. Using 20 ppm Pb2+, the samples eliminated the cation from the collected effluent, an evidence of the high selectivity of the products towards this pollutant. The only difference is that the sample ZE3 cleaned all of the initial influent solution used in the experiment which is 7,670 ml (equal to 1,534 BV) and produced effluent free of Pb2+ pollution throughout the experiment, whereas the sample ZE2 produced only 1,100 ml of treated solution equal to 220 BV at a concentration of 0.5 ppm. Therefore, the relative efficiency of ZE3 to ZE2 is 6.97 times. Cu2+, the second selected cation, was removed from 52 BV and 325 BV of the treated effluent by using the samples ZE2 and ZE3 respectively. The sample ZE3 also exhibits higher capability in removing the other toxic cations from the simulated wastewater if compared with the ZE2 (Table 5 and Fig. 5). This may be attributed Table 5 Dynamic data of experiments using column C ZE 2 Dynamic parameters Breakthrough point (ppm) Vb of Pb in (BV) Vb of Cu in (BV) Vb of Ni in (BV) Vb of Zn in (BV) Relative efficiency of ZE3/ZE2 for Pb
0.5 220 52 45 48
ZE 3
>1,534 325 270 290 6.97
Fig. 4 The efficiency of sample ZE 3 towards Pb and the other competing toxic cations
868
Environmental Geology (2004) 46:865–870
Fig. 5 The efficiency of sample ZE 2 towards Pb and the other competing toxic cations
to the presence of faujasite in sample ZE3 as the main zeolite phase instead of chabazite. The selectivity of the samples ZE2 and ZE3 towards the cations used in this experiment is as follows: Pb2+ >>Cu2+ >Zn2+=Ni2+ (Figs. 4 and 5).
Discussion Zeolites may be used to remove heavy metals from solutions in two ways: the zeolite may either be mixed with the solution to be treated and then separated by filtration and/or sedimentation, or the solution may be passed through a zeolite packed column. In a column system, the zeolite is more effectively used because it approaches an equilibrium state with the influent solution. From the data presented in Tables 4 and 5, the efficiency and cation exchange capacity always take the following series ZE3> ZE2> ZE1. This is most probably related to the difference in zeolite types (faujasite-phillipsite in sample ZE3 or chabazite-phillipsite in samples ZE1 and ZE2), zeolite grades and grain size. It may well be noted that the zeolite grade and the total CEC of the two ZE2 and ZE1samples take the following trend ZE2>ZE1 (Table 1). This study revealed that the zeolite selectivity and efficiency for the different heavy metal cations employed are proportional to their concentration in the effluent solution, where the higher the concentration of the heavy metals the higher the efficiency of zeolitic tuff. For example, in the battery industry effluent, the efficiency and selectivity of the ZE1 for Fe is greater than Pb; and in the simulated wastewater, Pb is greater than Cu and the other cations. Among the other factors that influence selectivity and efficiency are ionic size, ionic valance, as well as hydration energy. According to Sherry (1969) and Colella (1996), Na-loaded natural zeolites such as ZE1, ZE2 and ZE3 are fairly selective to monovalent heavy metal cations with low charge density. Whereas, selectivity for divalent cations is predominantly determined by their hydration energies. That is, it tends to prefer cations with lower hydration energy. If the ions have similar valence, the selectivity increases with the decrease of hydrated ionic
Original article
Table 6 Evaluation of Vb and Vt for the treated wastewater from the battery factory Using ZE1 Vb=(1,000 kg·WEC)/(Con. meq/l·100); Vt=(1,000 kg·CEC)/(Con. meq/l·100) Cation Fe2+ Pb2+
Con. (ppm)
Con. (meq/l)
WEC (meq/100 g)
CEC (meq/100 g)
Vb (m3)
Vt (m3)
169 20
6.05 0.19
19.6 6.8
33.2 17.4
32.4 352.4
54.8 901.6
radius compared with the free dimensions of faujasite, phillipsite and chabazite channels. Hydrated ions can pass readily through the channels if the radii are smaller than the channels. Therefore, the selectivity sequence obtained from ZE2 and ZE3 is Pb+2 >>Cu+2 >Zn+2 >Ni+2. This preference increases with the decrease in the total ionic concentration of the solution. The presence of competing ions, such as Na+, K+ and Ca+2, is another major factor controlling the selectivity of the zeolite. According to Colella and others (2001), Na-loaded phillipsite is more selective for K+ than any other cation. On the other hand, only half of the cation sites in phillipsite structure are accessible by Ca+2. In general, as the concentration of the interfering cations increases, less capacity is available for the heavy metal removal. It is evident from the experiments of the battery factory that the presence of a high concentration of sodium decreased the efficiency and selectivity of zeolite to remove iron and lead from wastewater. The most important result of this work is the constancy of the efficiency and selectivity of the zeolites used towards the different cations. The quantity of effluent wastewater from the battery factory, which could be purified by the chabazite-phillipsite tuff using ZE1, is calculated based on the given results in Table 6 as breakthrough volume (Vb) and effluent volume at exhaustion point (Vt). If an effluent wastewater passes at a constant flow rate through a 1,000kg zeolitic tuff bed packed in a 1.17-m3 column (length=1.5 m, radius=0.5 m and density of zeolite is about 1 g/cm3); the values of Vt and Vb could be calculated as in Table 6. These results indicate that the suggested bed of chabazite-phillipsite tuff (1,000 kg) has a high efficiency to remove Pb completely from about 352 m3 effluent wastewater of the battery factory. This means that the factory needs to regenerate the bed every 17 working days because the flow rate of the effluent wastewater from that factory is 20 m3/day. If the chabazite-phillipsite tuff bed is replaced by a faujasite-phillipsite tuff bed, which is 6.97 times more efficient (Table 5), the factory will not regenerate the bed before 120 working days. It is important to mention here that the cost of 1 ton of faujasite-phillipsite tuff from the local market is less than US $200.
Conclusion The Jordanian chabazite-phillipsite tuffs from Tell Rimah volcano and faujasite-phillipsite tuff from Jabal Hannoun
volcano have suitable mineralogical and technical properties that enable them to be used as ion-exchangers. These include suitable grain size and total cation exchange capacity, acceptable zeolite content, as well as good attrition resistance and high packed-bed density. The chabazite-phillipsite tuff (ZE1 and ZE2) has an excellent efficiency to remove Pb from effluent wastewater of the battery factory and other industries, and has an acceptable efficiency to remove Fe, Cu, Zn and Ni. The faujasite-phillipsite tuff (ZE3) is 6.97 times more efficient in removing Pb from the wastewater compared with the ZE1 and ZE2. This study revealed that zeolite selectivity and efficiency for the different heavy metal cations employed are consistent and proportional to their concentration in the effluent solution; where the higher the concentration of the heavy metals the higher the efficiency of zeolitic tuff. The presence of competing ions, such as Na+, K+ and Ca+2, is another major factor in controlling the selectivity of the zeolite. The results indicate that a chabazite-phillipsite tuff bed (1,000 kg), loaded in a 1.17 m3 column, has a high efficiency to remove Pb (20 ppm) completely from about 352 m3 of the effluent wastewater from the battery factory, in the presence of competing ions. Whereas, a bed of similar weight made of faujasite-phillipsite tuff is capable of cleaning up about 2,456 m3 of the wastewater. This quantity of wastewater is equivalent to 120 working days of effluent discharge at a flow rate of 20 m3/day. The expected cost of one ton of faujasite-phillipsite tuff from the local market is less than US $200. Acknowledgments The results presented here are from the project ‘‘Ion-exchange properties of the Neogene faujasite tuff of Jordan’’, which is financially supported by a grant from the Deanship of Research and Graduate Studies, of the Hashemite University.
References Attili S (1992) Evaluation of Jordanian phillipsite tuff with special regard to its applications in water softening, removal of ammonium and toxic heavy metal ions. MSc Thesis, Yarmouk University, Jordan Blanchard G, Maunaye M, Martin G (1984) Removal of heavy metals from water by means of natural zeolites J Water Res 18:1501–1507 Bremner PR, Schultze LE (1993) Exchange characteristics of metal cations commonly found in acidic drainage. In: Zeolite ‘93, 4th Int Conf on the Occurrences, Properties and Utilization of Natural Zeolites (program and abstract), International Committee on Natural Zeolite, Boise, Idaho, pp 55–57
Environmental Geology (2004) 46:865–870
869
Original article
Colella C (1995) Use of Italian chabazite and phillipsite for the removal of heavy metals from wastewaters: a review. In: Ming DW, Mumpton FA (eds) Natural zeolite ‘93, International Committee on Natural Zeolites, Brockport, New York, pp 363– 375 Colella C (1996) Ion exchange equilibria in zeolite minerals. Miner Deposita 31:554–562 Colella C, Pansini M (1988) Lead removal from wastewater using chabazite tuff. In: Flank WH, Whyte TE (eds) Perspectives in molecular sieve science, ACS Symp Series 368, Washington, DC, pp 500–510 Colella C, Torracca E, Colella A, de Gennaro B, Caputo D, de’ Gennaro M (2001) Ion-exchange features of intermediate-silica sedimentary phillipsite. In: Galarneau A, Di Renzo F, Fajula F, Vedrine J (eds) Studies in surface science and catalysis 135: Zeolites and mesoporous materials at the dawn of the 21st century. Elsevier, Amsterdam 01-O-05 Groffman A, Peterson S, Brookins D (1992) Removing lead from wastewater using zeolites. Water Environ Technol 4:54–49 Grube Jr WE, Herrmann JG (1993) Heavy metal pollutant fixation by natural zeolites In: Zeolite ‘93, 4thSuperscript> Int Conf on the Occurrences, Properties and Utilization of Natural Zeolites (program and abstract), International Committee on Natural Zeolite, Boise, Idaho, pp 112–114 Ibrahim KM (1996) Geology, mineralogy, chemistry, origin and uses of the zeolites associated with Quaternary tuffs of northeast Jordan. PhD Thesis, Royal Holloway, University of London Ibrahim KM (2001) Evaluation of Jordanian faujasite tuff by comparison with other natural and synthetic zeolites. Environ Geol 40:440–445 Ibrahim KM, Hall A (1996) The authigenic zeolites of the Aritayn Volcaniclastic Formation, northeast Jordan. Miner Deposita 31:514–522 Ibrahim KM, Inglethorpe SDJ (1996) Mineral processing characteristics of natural zeolites from the Aritayn Formation of northeast Jordan. Miner Deposita 31:589–596
870
Environmental Geology (2004) 46:865–870
Ibrahim KM, Nasser Ed-Deen T, Khoury H (2002) Use of natural chabazite-phillipsite tuff in wastewater treatment from electroplating factories in Jordan. Environ Geol 41:547–551 Kazemiana H, Rajecb P, Macasekb F, Orechovska Kufcakovab J (2001) Investigation of lead removal from wastewater by Iranian natural zeolites using 212Pb as a radiotracer. In: Galarneau A, Di Renzo F, Fajula F, Vedrine J (eds) Studies in surface science and catalysis 135: Zeolites and mesoporous materials at the dawn of the 21st century. Elsevier, Amsterdam, 31-P-06 Mercer BW, Ames LL (1978) Zeolite ion exchange in radioactive and municipal wastewater treatment. In: Sand LB, Mumpton FA (eds) Natural zeolites; occurrences, properties, uses. Pergamon Press, Oxford, pp 451–462 Michaels AS (1952) Simplified method of interpreting kinetic data in fixed-bed ion exchange. J Ind Eng Chem 44:1922–1930 Nasser Ed-Deen T (1998) Zeolite from Tell Rimah and its use for industrial wastewater treatment. MSc Thesis, University of Jordan, Amman, Jordan Pansini M (1996) Natural zeolites as cation exchangers for environmental protection. Miner Deposita 31:563–575 Pansini M, Colella C (1989) Lead pollution control by zeolites. Mater Eng Modena, Italy 1: pp 623–630 Pansini M, Colella C (1990) Dynamic data on lead uptake from water by chabazite. Desalination 78:287–295 Pansini M, Colella C, Caputo D, de’Gennaro M, Langella A (1996) Evaluation of phillipsite as cation exchanger in lead removal from water. Microspores Mater 5:357–364 Shammout N (1993) Removal of selected heavy metals using Jordanian zeolite. MSc Thesis, University of Jordan, 128 pp Sherry JA (1969) The ion-exchange properties of the zeolites. In: Marinsky JA (ed) Ion exchange and solvent extraction: a series of advances. Marcel Dekker, New York, pp 89–133 US EPA (1993) Drinking water regulations and health advisories. US Environmental Protection Agency, Washington, DC Water treatment handbook (1991) Anonymous water treatment handbook. Degremont, Rueil-Malmaison, France
