Anaya Chics At 954 (257) Contents liste available at ScienceDirect Analytica Chimica Acta ELSEVIER Journal homepage: www.clseviercom/locate/ace Review Chemical oxygen demand: Historical perspectives and future Downer challenges René Bernard Geerdink *", Ricardo Sebastiaan van den Hurk °, Onno Jacob Epema * + ewes, Mn of scare nd the Ere PO Boe 78208 AA Lt Te Neher ery of rtd Secs ne rie Soc Chany Hebron 7.584 eh Te Neronds TA tora view of the develope of the COD determination is ee) presented " + Dicomate has suvived sn standard -_ ss {jvamtaive procedures wo fo prover baden the env & —~ €&p-— & fonment with hazardous chemicals ie mercury a « Hecro- or photo oxidative methods cop methods. ARTICLE INFO ABSTRACT ‘rena ‘eterining the chemical oxygen demand (COD) scallenging beauze ofthe chemicals hat are sein Receved 5 uly 208 the current methods There an urgent ned to determine the plltion in srface waters from (hey) S juny 20:7 ected 7 jay 2017 Arable online 4 January 2017 industry va simple, reliable an fast method, However, the cient procedures use hazardous chem- teas, such as mercury, to suppress chloride interferences. This challenge as not been resolved The introduction of modern detection techniques, such a5 chemaluminescence and oxidation pro- cedures, eg microwave and ultrasound discolaton, has not resulted in a breakthrough, Curent the applicability of mereuy-ttee methods is limited to samples with a chloride content of approximately 3000 mgiL New elecz0- and photo-oxidative methods show potential as replacements forthe standard Dihuoate COD methods that requice toxic reagents. but further research is stl needed Pernganate In this pape, a lstorical view ofthe development of chemical oxygen demand determination methods Organic plu 4s presented. Citical notes on and outlines ofthe development of the frst through the cutent pro- horde nerterence cedures ae escussed. In addition, the Duteh situation is presented as an example other counties in Photocatalitieoidaton thet "strugele” to establish a COD procedure with two basic quai, reliably and enviconmental fiiendlines, (© 2017 Elsevier BV. All rights reserved Incroduction 2 2. Chemical and analy background 2 21, Equations 2 ‘mal odes ene gerdinkrs (RA. Geen5. Oxidation procedures Si. Permanganate oxidation 32, Diehromate oxidation 33, Chloride interference, ermal andlor corection BA. Mereury-fee 215. Oxidation procedures with a less chemical nature 4 Conclusion References eam os 1. Introduction 150 6060 states that the accepted definition of COD (chemical oxygen demand) is the mass concentration of oxygen that is equivalent fo the amount of dichromate consumed by dissolved and suspended matter when a (water or sludge) sample is treated ‘with that oxidant under defined conditions (1), ‘The earliest methods to determine COD were developed approximately 150 years ago and involved the observation of colour changes ofa permanganate solution mixed with water samples (2) ‘The earliest attempts to examine the oxygen consumption caused by pollution in water and wastewater used a permanganate test. The first attempts to use dichromate to oxidize organic matter were during the period from 1925 to 1930 and were not very promising. In 1949, Moore et al. [3] applied the dichromate pro- cedure to wastewater analysis. Quantitative oxidation of com- pounds was obtained by refluxing the sample for 2 h with dichromate in a 50% sulphuric acid solution at 145-150 °C. Using this procedure, approximately 90% of the organic matter present and most of the chloride were oxidized, but a correction was required for the chloride (3, n 1951, Moore etal, introduced a small amount of silver suiphate to a mixture to catalyse the oxidation, and this resulted in nearly quantitative results even with several relatively inert compounds, such as carboxylic acids and aliphatic alcohols, In 1963, Dobbs etal. /5] demonstrated that chloride oxidation during COD digestion was reduced substantially by the addition of mercuric sulphate to the sulphuric acidjdichromate reagent. which meant that a separate chloride determination was no longer necessary Chloride is the most common interference in a COD analysis. tis oxidized in acidic solutions by dichromate, but itis not oxidized by natural processes in the environment (2). Additives such as HgSO, ‘AgNO, and Cr{l), or their combinations, will minimize chloride interference but not completely remove it (5). At high tempera- tures, the masking of chloride is insufficient even with mercury sulphate, Additionally, these additives create additional hazardous Aeveloped that do not require toxic reagents. These new methods are based on either electracatalytic 9-15] or photocatalytic [16-22) oxidation principles. Although some of the methods have great potential, none of the methods have been widely adopted, Recently, photoelectrocatalytic COD determination using TiO, was introduced, and it may have a great future for dissolved COD. In the commercially available PeCOD® system. the chloride tolerance is approximately 350 mg L-!. and some authors [17.24] report a Chloride tolerance up to approximately 2000 mg L-! with a ‘modified system. ‘As outlined above, the current COD determination methods are not satisfactory because ofthe use of highly toxic substances such as CH(VD), Hg and Ag. This is true even when considering the pos- sibility of strongly reducing or completely recycling the spent chemicals. Inthe literature, there are few procedures that advocate an environmentally friendly method for COD determination, and the chance of successfully introducing such a method is sim aslo a5 the use of the current COD methods are embedded in legislation Sand taxation. 2. Chemical and analytical background ln this section, the chemistry behind the analytical procedures will be described, For details on the conditions used in standard COD analysis, refer to the standard proceduses. eg, 180 6060 (1). It is important to note that the conditions havea large impact on the respective methods, but that falls outside of the scope of this 2, Equations In the standard COD determination method, an oxidant (Cr0}), a catalyst (Ag*) and sulphuric acid are added to an aqueous sample that is then heated for 2 h, Potassium hydrogen Phthalate is often used to represent organic material when pz0- cedures are cested. The conversion of potassium hydrogen phtha- late by dichromate is as follows: 2KCpHSO4 + 10K,Cr20) + 414,504 16CO, + 46H,0 + 10Cr(S0,); + 1K,S0q a wastes [71 ‘Although COD methods that replace mercury with AgNO; and) or rfl) have been propased, they cannot be used for samples with variable salt loads [$}. Furthermore, silver ions are toxic and expensive, and the addition of Cr(ll) precludes colorimetric detection, Therefore, mercury containing COD reagents continue to prevail In recent years, some new “green” COD methods have been ‘When dichromate is replaced by oxygen, the equation becomes: 2KCH5O4 + 150, + Hy$04 16C02 ~ 6H:0 + KzS04 @ Eqs. (1) and (2) show that each mole of Cr:0} oxidizes the amouint of potassium hydrogen phthalate that is equivalent to 15 mol of Oo. ‘The excess dichromate is determined via titration with ferrousGein Chia eu 861207) 1-1 ammonium sulphate, which is also known as Moht's sat: GF 4 CraOF + 14H Oe 4 ace 4 7H,0 6) q,(3) shows that each mole of Cr,0} uses 6 mol of Mohr’s salt. Although dichromate is a powerful oxidant, it does not oxidize chloride in aqueous solutions, n the presence of sulphuric aci, however, chromyl chloride (CrO,Cl,) is formed: K,Ch 0} + ANaCl + GH,S0, * 2KHSO, + ANAHSO4 + 2Cr04C1, + 31,0 During the 2h of dichromate oxidation of organic compounds at 148 °C, chloride is oxidized quantitatively to chlorine: C0} + Gch + 4H = ach + 20° = 710 6) ‘The interference of chioride can be mitigated by adding mer- curyil) sulphate: Hg? 4 2c MgCl, © Dichromate is not stable at high temperatures (+155 °C) or under strongly acidic conditions, and this ean contribute toa false COD measurement: 2Cr20} + 16H! = Ace’! + 302 + BHO 0 Note that the amount of dichromate used can also be deter ‘mined using spectroscopic methods that follow the Cr(ll) forma- tion and CVD) depletion In the permanganate oxidation method samples are oxidized at an elevated temperature in an acidic solution. Reduction of the permanganate is performed by the oxidizing material in the sam- ple. The amount of permanganate consumed is determined by adding an excess of oxalate solution, followed by backetitration Wwith potassium permanganate, The corresponding reaction equa tions are 2KMn04 + SHyC04 + 34,504 = 2MnS04 + K;S04 + 10CO, + 81,0 ‘The corresponding hal-reactions are the following: BH! + MnO4 + Se = Mn?! + 44,0 (reduction) © x04 2C0; + 26° (oxidation) (10) ‘Above an oxidation temperature of 60 °C permanganate decomposes: 2KMn04 = KzMn04 ~ Mn02{s) + 02 ro) Note that manganese dioxide catalyses the decomposition of permanganate, which leads to a higher COD. Potassium petmanganate in an acidic medium can oxidize chloride ions to Ch, which leads to ahigher COD. The stoichiomettic ratio ofthe reaction is as follows: Mnog + 8H* 4 4c Mn? 4 2c + 44,0 «ny SH” + Mn0g + 4e- > Mn? + 4110 (reduction of permanganate) 3) @ 3. Oxidation procedures 31, Permanganate oxidation The first atempts to determine the amount of organic pollution in a water sample using permanganate as the oxidant were con- ducted by Forchamer in 1849 and Wyatt in 1893, The methods involved observing the length of time required for a reduction in the colour of the permanganate ion solution, which was deter- ‘mined by the quantity of organic matter present in the sample (25.25) “Moore [3 stated thatthe analytical procedure for permanganate oxidation in a hot acid solution was developed by Stamm in 1934, Until 1949, that test was the most widely used. Unfortunately, many «lasses of organic compounds are poorly oxidized by permanganate under those conditions. In addition, the reactions of permanganate with compounds that are readily attacked result in the formation of, manganese dioxide, which catalyses the decomposition of per- ‘manganate. This seemingly magnifies the consumption of per- manganate by organic matter (27), Potassium permanganate’s effectiveness in oxidizing organic compounds varies widely, and in ‘many cases, the biochemical oxygen demand (BOD) measurements are higher than the COD measurements. Lovett and Gardner (28) Stated that different values can be obtained depending on whether 0.125 N or 0.0125 N potassium permanganate was used. Although the permanganate test serves an important function in ®) waste water control, there has been opposition tothe procedure: “I think that most chemists would be glad to see the permanganate test become an historical curiosity, and assaults upon it such as Dr. Klein is ‘making, may well bring thar about” |29}. Additionally, Moore stated in 1948, “The amount of research carried out with the permanganate method shows that i not entirely satisfactory fr the determination of oxygen consumed values” [3}. Also during that period, an exten- sive comparison of the permanganate and dichromate methods as made by Klein [25], ‘Today. the permanganate test is stil used for relative mea- surements of organic pollution, mainly to check the quality of potable and surface water. These types of water generally have thloride concentrations less than 300 mg L-. At many waste water reclamation sites, the permanganate method is used to gain a general insight into the pollution level of the discharge. Theadaition of waste (water) to a boiling permanganate solution and ‘monitoring the rate of discoloration quickly shows the relative degree of pollution. However, ithas been shown that this method is rot suitable for quantitative measurements. A large number of organic substances are either not converted of ate only partially converted, Theoretically the oxidation efficiency should increase at higher sulphuric acd concentrations and higher temperatures, In practice, the oxidation temperature must be fairly low, approxi- ‘mately 60 °C, to prevent permanganate decomposition. This is a ‘much lower temperature than that used in the dichromate method and severely limits its applicability Sprenger (30) used both the permanganate and dichromate procedures. Using permanganate under alkaline conditions instead ff acidic conditions further reduces the oxidation potential and fewer substances are oxidized, In the 1980s, the permanganate method had a revival as a ‘modern, ie, automated and miniaturized. technique and appeared to overcome the struggles from the previous decades, Various publications on and performance characteristics of the method are summarized in Tablet. These examples show that applying highly instrumental tech- niques instead of manual techniques is a prerequisite to achieve progress in COD determination using permanganate. 32, Dichromate oxidation ‘Adeney and Dawson |41) were the first to report on the use of, hexavalent chfomium (dichtomate) to estimate the amount of organic matter in water. In 1926, these authors published an acidic dichromate method for analysing waste and sewage water. In 1927, ‘Abbott (42) published a similar procedure using a lower digestion temperature. Both methods began by adding a small volume of sulphuric acid tothe sample, evaporating the sample to drynessin a water bath and further heating the sample to drive off the chlo- rides. The recoveries of pure organic substances were low because of the partial loss during the chloride evaporation step. Although dichromate is believed to be a less powerful oxidant than per- ‘manganate and Ce(IV), it has been considered the best oxidizing agent as itis relatively stable in a boiling concentrated sulphuric acid solution at temperatures below 150°C [43] ‘bie eam os ‘The standard dichromate solution is stable over long periods of time (4), The titration end point is sharp when the concentration of sulphuric acid in the mixture does not exceed 30%; hence, the re- action mixture must be diluted before titration. To achieve complete organic oxidation, the sulphuric acid con- centration must be as high as possible. This increases the boiling point and increases the oxidation efficiency of chromic acid. Un- fortunately, the increasing temperature also enhances terferences. When 50% by volume of sulphuric ac is used in the reflux mixture, chloride, bromide andjor iodide will consume di- chromate and produce elemental halogens |7| Initially, the COD values for industrial wastes containing high chloride concentrations were corrected for the contribution from chloride. However, applying a mercuric salt will reduce the chloride interference, Ifa 10:1 weight ratio of HgSO4:Cl-is maintained, in- terferences from chloride are suppressed up to a concentration of 2000 mg L (7), ‘The COD cannot be measured accurately in samples containing ‘more than 2000 mg L-' CI- [44]. However, this is determined by the relative concentration rather than the absolute concentration of| chloride. High COD levels can be reliably determined in high salinity solutions. In the Netherlands, the COD standard method NEN 6633 imposes a limit of 100 for the chloride to COD ratio to ensure accurate COD figures [45] Many modifications to the original dichromate method have been introduced over the years, often to correlate the results with ODS and permanganate oxidation |26.45~i3), ‘Today, COD determination is largely based on the work of Moore etal. (2-1) The use of a 50% sulphuric acid solution and silver sul- phate as a catalyst as proposed by Muers |49) effectively oxidizes ‘most organic compounds. The introduction of mercury sulphate by Dobbs in 1953 eliminated the need fora chloride correction, but it introduced the problem of an additional highly toxic and envi- ronmental unfriendly agent [5 Inaddition to mercury, the use of silver ions is subjectto dispute from an economic point of view. The use ofthe catalysts precluded in some cases |4]. Certain compounds tend to precipitate silver and either nuliy its effects or result in a lower COD value. This is also generally rue when the sample contains a high concentration of chloride. Significant errors (20-500%) have been reported for Reeence Reh DISSIRAR Reb Doon, - cemmesence Digestion 5o=100°C or Witton, 256 ne "Co Pato, 365 [271 Deron 525 am Inmnet03 51 Sampling cate + 4031] - 20,52) -4 (33) -30 [34] -3035) LW phaalysmina [3] Ssewel plate 25 amino ukesound assed dgestin/Flame AAS 0 4 fst) 27]~288 [23] 12 5) -24 00) Chl aecance me CI 1-* 150023) ~ 700 (32) - 18.000 (33) - 1003-60 3 S020) ~10ao027)- 175 38}- 100381" 100 (30) recon (502g COD o7p0[31) aanope 05,2052) 1331 Saito 3s) disops) dojoa? 23) 53] 271300] Linear rangelngL 20-5005 240005) oi90)22) eaao7 os [33 aie-1924 28) 05-50 34) esests tongs) $500 (4) Tins heGerd ea / Ant Chea Ac 861 (207) 1-11 5 samples containing, COD levels up to 500 mg L-? and chloride Concentrations of 6000 mg (50) ‘The amount of mercury sulphate used for chemical masking of chloride isnot critical aslong asthe mercuric ion concentration iat least a factor of 10 greater than tha of chloride. The extent of the Chloride oxidation does not appear to depend on the ratio ofthe thioride concentiation ofthe sample to the dichromate and acid wolumes, nor tothe amount of metcutic salt added; rather, Dears to vary withthe substrate [44]. Various authors have noted that both the chloride concentration and organic composition of the sample influence the extent of the choride oxidation (51) ‘Therefore the performance of a method cannot be qualified by 3 statement such as “suitable for samples containing up to 500 mg 1+ chloride This statement assumes equivalent in- terferences in samples with $00 mg L-chloride and zero COD and samples with 500 mg I~? chloride and 500 mg -~* COD, whichis incorrect [5] “Another often neglected interfering substance is ammonia, and its effect on COD was studied extensively by Kim (52). In the Dresence of halide ions (CI or Br) arnmonia contributes signif- Canty to the observed COD values (753), Additionally. in the Dresence of ehlosides and high concentrations of ammonia organic fsnine. oF nitrogenous matter, a continuous reduction of dichro- tate occurs (5) Another more serious problem arises (com the oxidation of inorganic substances such as Fe) Sor Mail ons, Which results in artificially high COD values [54.5 ‘The question remains whether these interferences will be dminished since oxidation by the natural processes of ammonia or organic nitrogen compounds will lead to oxygen depletion ofthe In the last few decades, the dichromate method has been adapted to modem techniques Canals eal [56] developed an ultrasound-asisted method for CoD determination in water samples. The oxidation was performed in 66% 11,804 using a sonication time of 2 min. Under these cit- cumstances, the oxidation temperature doesnot exceed 30°C, and a chloride on concentration up 7000 mg CI~L~ canbe tolerated For example. Kim etal. [57) successfully analysed samples cone taining 100-500 mg Lt potassium hydrogen phthalate (KAP) However, zeal wastewater samples had cilleences of 5202 compated to the standard procedure, and it was concluded that Sonication was not sufcient to digest all ofthe organics inthe Sample. In ultrasonic oxidation, the strength ofthe ultrasonic probe decreased with the operation hours, which caused larger errors Domini eta. (58) compared three optimized methods: closed i= crowave, open rnicrowave and ultrasound irradiation. Ina three Cases, the interference concentrations produced deviations greater than 10% inthe COD values at 13:4, 23.4, 211 and 2819 mg L-* for Ses, NOp and Cl respectively. The use of ultrasound energy for COD determination seems to be an interesting and promising altemative to conventional open reflux and microwave-asited digestion methods because the instrumentation is simpler, cheaper safer and faster Despite the advantages of microwave and ‘leasound digestion over conventional COD methods, these pro- cedures stil requite the use of expensive and toxic chemicals for Sample digestion 59-62), Bellet al. (63| developed a microwave assisted COD assay using colorimetric determination of Cll) at 590 nm, and they reported that without adding meseury salts thioride concentrations up to 8000 mg did not interfere. This result was likely de tothe unrealistically high COD concentration of 7000 mg L. Cuesta etal. [5 used a stopped flow FIA injection procedure with microwave asisted oxidation and flame atomic absorption detection, The chloride interference could be compensated for up {0 5500 mg LCI using 011 M HgS0q,Dharmadhikari etal. (95| used a closed microwave system witha reaction time of 15 min, nd the eblorie ion interference was removed up to 5000 mg i Using toxic reagents. Gomez etal. [95] reported tha the microwave digestion procedure was more effective using different and grad- ually higher powers than a igh fixed power. Del Valle et al (07) evaluated an open miciowave digestion method. There was good agreement between the values obtained by the standard method dnd their values, andthe microwave procedure was 25 times faster. hen et al. [58 used a focused microwave system and concluded that the iolerable concentration of chloride is -20,000 mg with mercury sulphate compared to 10,000 mg L! without mercury sulphate when the COD is 150 mg I. The tolerable concentration is lower when the COD is lower, which shows that a chloride Concentration af 100 relative to COD s feasible with the dichromate methods. Finally, an interesting. “compromise solution” to the toxicity of the chemicals used in COD determinations was pub- lished by Carbaja-Palacios eta. (69). These authors described a diownscaled COD procedure and obtained a reduction of upto 80% in the use and generation of toxic residues and atime reduction of ‘p10 67% compared to the standard method. Using 2500 yL sample ‘lume and a eaction time of 40 min, an average relative eror of 1.x for the standard KHP of 150-400 mg L-" COD was abtained, whereas the average petcentage standard deviation forthe same Samples was 211 using the standard method. The linea vas 30-600 mg I~ COD, andthe detection limit was 36 “The chloride tolerance was good for standards of 200 mg L COD (KHP). Chioride concentrations up to 2000 mail. did not have a significant interference, but higher Cl concentrations caused large interferences. 33. Chloride interference, removal andor correction From the beginning of the development of COD determination ‘methods, it was clear that not only chloride but also bromide and iodide contsibute significantly to the results. In 1932, Bach found that the oxygen consumption value of raw sewage increased from 258 to 291 ppm when the sodium chloride content increased from 20 to-2000 ppm (70), Eqs. (3) ana (5) explain why this interference fccurs. Asa result, effort was exerted to overcome the interference. In their first attempt, Adeney and Abbott drove off the chlorides before digesting the sample with dichromate by heating the sample to dryness (41.12), ‘Another approach by Ingols and Murray |26] determined the chlorides in the sample separately to correct the apparent COD result, Baumann || collected liberated chlorine from highly saline 2m 20052 Chloride load mg L- @<500 1500-1000 21000-3000 123000-10,000 1 10.000-20.000 @>20,000 Fig. 1 chonide leaés (mg 1) Im about 100000 wastenater samples in the Netenes 2008.‘ B,Cerdnk ea Anat Chee ea 861 207) 1-1 aualary electrode Toomer creuitond GR Sey eeeree| —®@ OH on Fig. 2 Schematic istration of 10, server apeation ‘wastes and determined the concentration via titvation with standatd sodium thiosulphate solution. Consideration was given to the removal of chloride via precip- itation and filtration before digestion [71], However, experiments ‘demonstrated that co-precipitation occurred in the presence of ‘colloidal matter, eg, fatty acids and proteins. In standard methods, this approach is rejected because substantial error is caused by the 00 (mat) ‘occlusion and carry down of COD matter from heterogeneous samples (7) In 1951, Medalia 27) reported that the chloride removal step ‘can be omitted by binding the chloride ion in a soluble undissoci- ated mercuric chloride complex by adding mercuric nitrate, which does not interfere with any stage of the determination and resists ‘oxidation under the conditions of the COD test. However, it was ig 3. COD in tere (045 ym) an non-itere water samples (eve meteRE. Ceri Anyi Chi ta 961207) 1-1 another 12 years before mercury addition became common prac- tice, Dobbs (5 is known to be the fist to include mercury in the COD method, but his work is based on the wark of Glew |72] and Medalia (271, Theit work was confined to samples with chloride contents of 2000 mg Lor less. The errors were pronounced when high-chlorideflow-COD samples and high-chloridejhigh-cOD samples were analysed. Burns [72] stated that the assumption thatthe COD of high chloride samples can be measured successfully using stepwise diluting was not valid. The COD results increased markedly with dilution, sometimes by a factor of two or more for dilutions of 10 o 1. Areasoa for this behaviour was not given but it could be due tothe low levels of COD after dilution, which prevent reliable COD determination. ‘The excess mercury ion required to prevent the oxidation of chloride was initially determined to be a 5:1 ratio of Hg50s to Cl Cripps and Jenkins |78) reported an appreciable chloride contri- bution to the COD value even with 10 mg HgS0s per mg Cl- using a |AgaSOg catalyst. They suggested the use of a correction factor (0014 mg COD L- per mg Cl” L"! for a2 h reflux) to compensate for thisinterference nthe Dutch COD standaed from that time (75), a correction factor of 0226 mg COD per mg CI- L-* is used, and the consumption of dichromate is not to exceed SOX Currently, a 10:1 weight ratio is generally accepted as sufficient to mitigate CI- concentrations up to 2000 mg L~* 8). This does not imply that corrections are unnecessary [3]. Even with the addition of mercury, there is stil a significant positive bias at chloride con- centrations over 1000 mg L | This bias is greater for low COD values. An apparent COD value of 59 mg 1 was reported for a 2000 mg L. "blank chloride solution [3 Currently, the COD cannot be accurately measured in samples containing more than 2000 mg L* of CI 44) Soto etal. (55) re- ported that in a semi-micro closed reflux system, HgSO¢ ratios of 10:1 can lead to erroneous determinations, especially when the organic load ofthe waters relatively ow. It was aso shown that at Hg:CI" ratio over 15:1, the addition of AgzS0, dissolved in 150, caused precipitation of HgS04, which might be due to the increase in the SO{- concentration and non-formation of HgCh. The chloride interference becomes a more serious issue when the measured sample does not contain substantial amounts of organic material since more unreacted dichromate enhances the oxidation of the chloride ion [5). The halogen interference is reduced as the organic matter concentration increases 6.5. These differences reflect an interaction between the organic matter and the chloride content of the sample when the sample is oxidized in the standard procedure for COD determination. The error in the COD determination caused by the presence of Cl ions increases with the organic matter’s resistance to oxidation. This should be considered when analysing waters with a high suspended solids content. In addition tothe required HgSO4:CI- ratio, the significance and Importance of the CI {COD ratio increased (55), Greater HgSO, concentrations allow for the analysis of samples with CI-{COD ra tios up to 30, but for waters with higher CI [COD ratios, efficient ‘masking of CI using #g5Ogis impossible. This implies tha for high salinity water, eg. seawater, with a concentration over 18,000 mg GL, the COD values must be over 600 mg L-! to allow for reliable determination, which is almost never the ease. The Dutch national standard NEN 6633 of 2008 allows a Cl-[COD ratio of 100, and matrix matching (by raising the chloride concentration in the bilan to the level in the sample) is prescribed in case this ratio is, exceeded 45]. The accuracy is lower at the high ratios, but governmental institutes require that laboratories analyse all samples. ‘The removal of excess chloride from samples prior to COD determination has been incorporated in the German standard |75) (Chloride i tobe driven out as hydrochloric acid and captured with calcium hydroxide toa level below 1000 mg Lin order to allow the masking by mercury to take place. Vaidya (77 and Domini [78], used a similar approach. They removed chloride as hydrochloric acid gas from acidified sample solutions in the range of 1001065 mg CI~ L-! at 150 °C in a closed reaction tube. HCl was captured using 2 bismuth-based adsorbent suspended above the heated solution. Miller [2) modified the procedure using an offline chloride removal cartridge with sodium bismuthate while simal- taneously replacing dichromate with a trivalent manganese oxidant. A silver catalyst was not required, which made the pro- cedure relatively environmentally friendly. In general, chloride removal procedures are cumbersome and rot very practical. Masking the chloride interference also has drawbacks because it generally requires substantial amounts of toxic chemicals. Typically, COD determinations for samples with to COD ratios greater than 50 are unteliabe. 34. Mercury-fiee Silver, hexavalent chromium and mercury salts used in the COD determinations create hazardous waste, Phasing out mercury has been a major challenge over the last few decades (3). the lter- ature, the most effort has been directed towards the use of silver as an alternative to mercury 51,7982), The addition ofa silver ion Solution a the first step in the procedure, preferably while cooling, was very promising Ballinger et al. (51) tested the use ofa silver nitrate solution to suppress the chloride interference, This procedure would be attractive because of the comparatively low toxicity of silver com- pounds in biological systems and the ease of recovery of silver from waste solutions. Using the standard procedure, relative standard deviations of approximately 11% were obtained for samples con- taining up to 1000 mg L-" chloride and with chloride to COD ratios 0f 3:1 or less and approximately 15% for ratios up to 5:1. Above this ratio, the standard and proposed procedures delivered results that were significantly alfected by chloride. Silver sulphate addition was reported by several authors (74,82,83). A substantial reduction in chloride oxidation was achieved with a ten-fold increase in the silver sulphate concentration |74). Unfortunately, this procedure had high costs and poor precision, Zietz |79) described the use of silver nitrate in the COD analysis fof pure organic compounds in saline solutions, The order of ok Fig 4. Recovery of COD" rele te mate method eave meth) eerrinstin of COD In wasewater simpleaddition for the silver nitrate solution and the time delay before adding other reagents were important [51,73]. No significant sup- pression of the chloride interference was observed unless the silver nitrate solution was added frst. The suppression level was reduced if the dichromate solution and sulphuric acid were added within 5 min of the silver nitrate solution addition. Thompson [5] concluded that these “mercury-free” techniques still suffer from chloride interference bias at low sample COD values. Chloride interference usualy decreases with an increasing COD. because chloride is complexed by the Cri! formed. The addition of Ce(}) is the basis of a Hg-free method (5.8. The chromium(il) added is approximately equivalent to half of the original chro- ‘mium(V1) present in the digest (3). With this amount of chromiu ‘M(H, the chloride interference was effectively suppressed for chloride levels up to approximately 3500 mg, L-' Geerdink eta. [81] questioned whether Cr(l}) favours chloride complexation. No Cr) effect was observed unless the amount of silver added was sufficient Morrison [3] reviewed mercury-fiee procedures. Because mercury is only used to suppress chloride, the oxidation circum- stances can be optimized for chloride such that almost no oxidation of chloride occurs. but the organic compounds are stil oxidized. Chloride oxidation can be avoided using mild oxidizing conditions, but at the expense of the incomplete oxidation of organic matter (aay Recently, Geerdink et al. 81] published a mercury-free COD ‘method in which they lowered the digestion temperature to selectively inhibit the oxidation of chloride. The results for undi~ luted samples with up to 3000 mg Cl L! oxidized at 120 °C showed a good correlation with the results from the standard procedures. The chloride interference was further reduced by the addition of silver ions and was practically absent at a molar ratio of ‘Ag'/CI" over 17. In Fig 1, six chloride ranges are shown. These data were derived from approximately 100,000 wastewater analyses in 2008. The figure shows that 95% of chese samples contained less than 3000 mgL~) CI and could be analysed with the proposed Hg- fiee COD method, Esakku etal. [$4] used antimony(il) sulphate in a Sb:CI~ ratio of 5:1 to eflectively mask chlorides up to 2800 mg 1" for COD values Lup t0 2000 mg L Microwave and ultrasound-assisted mercury-free digestion techniques have also shown promising results 43). Microwave- assisted dichromate oxidation allows for rapid digestion in 2 min [5] Ina microwave, the samples are heated evenly, whereas the ‘pen reflux method! and closed tube methods rely on heating the sides ofthe flask, The sample probably remains in a more dispersed and readily oxidizable form in a microwave. Consequently, ess concentrated sulphuric acid is required for full sample oxidation, ‘which reduces the chloride interference to an acceptable level However, the maximum chloride concentration tolerated was only 250 mg L Balconi et al. [85] used a flow injection method based on mi- rowave oxidation without mercury and_spectrophotometric detection for well and fiesh water. The results abtained were in good agreement with the standard procedures. Liu etal. (85) described a micro-tuminol-H,0, chemoluminescence-dichromate system with a lower limit of detection of 100 mg L | No chloride interference was observed because the short reaction time between the oxidizing reagent and sample did not allow the side reaction of the reagent with the chloride ions to occur. Chloride levels were tolerable up to a C:COD ratio of 500, eam ‘These observations show that itis possible to proceed without the addition of mercury, but the applicability of mercury-free ‘methods is limited to samples with a chloride content of approxi- mately 3000 mg L- 25, Oxidation procedures with a less chemical nature Inthe last decade, the application of electro- and photo-catalysis in COD determination has grown exponentially [18,19 3487-89), Electrochemical determination of COD, also known as the amperometric method, does not produce hazardous waste mate- Hials and is environmentally friendly. Organic pollutants are oxidized by hydroxyl radicals (COH) generated at the working electrode, Tis leads tothe generation ofan electrical signa, which is proportional to the COD concentration. Based on this detection ‘mechanism, the key inthe electrochemical determination of COD is finding the optimal working electrode to determine the quantity of “OH generated, Carbajal-Palacios et a. [90] used hydrogen peroxide instead of dichromate to oxidize samples. Inthe range of 80800 mg! COD, good results were obtained. The detection was conducted at 240 nm. and heat was not required during the 120 min oxidation of the samples. The authors concluded thatthe results were statisti- cally comparable with the cuvette test of Hach-Lange. Recently, Silva et al. |15) introduced a copper electrode as an electrocatalytic sensor. The linear range was 502800 mg Land the procedure was applicable for wastewater samples from the soft drink industry. No toxic reagents were used in this method, Yang et al. [9| used nanocopper particles electrodeposited on the surface ofa copper disk electrode for an enhanced surface area and increased oxidation activity This resulted in a more sensitive ‘method to determine the COD using a copper electrode. A detection limit of 3.6 mg L-’ was achieved with a linear range of 48-600 mg L- Chloride concentrations of 700 mg 1 did not affect che COD result Zhou et al. [10] used a NiCu alloy electrode as an electrocatalytic sensor The NiCu was electrodeposited on a glassy carbon electrode (GO) The linear range was 101500 mg L”. High chloride con- centrations (1 mol L) had no influence on the current response, Whereas electrodes based on metals and alloys usualy lose their activity because of chloride ion poisoning. etal, [11] optimized the conditions for a F-PbO, modified Pt dise electrode. A linear range of 100-1200 mg L' COD with a detection limit of 15 mg. was achieved. Chloride to COD ratios of 5:1 Jed to a deviation of +4.4% and 10.5% at a rato of 100. Ma et al, (12) fabricated a novel Ti/Sb-Sn0,|Pb0, composite electrode for COD determination. This new electrode configuration improved the sensitivity of the amperometric method. A linear range of 05~200 mg L-' COD and a detection limit of 03 mg L was achieved under optimal conditions. However, the procedure used Ag” to precipitate chloride, and according to standard ‘methods, this approach must be rejected because ofthe substantial extors due to the occlusion and carry down of COD matter from heterogeneous samples [7]. In 2007, Yu etal. [13] developed and optimized a boron-doped diamond (BDD) thin lm on titanium as a sensor for the amperometric determination of COD. A wide linear range of 20-8000 mg L- COD and a detection limit of 75 mg L were achieved, Over 400 measurements could be performed without any loss in the performance. However, no indication of chloride interference was presented Bogdanowie7 etal, 91) used BDD on silicon and glassy ca bon electrodes instead of titanitim, Their papers show low detec tion limits of 09 and 11 mg L-" COD for silicon and glassy carbon,RE Ceri a Anyi hia te 861 (207) 1-1 respectively. The linear range and chloride interferences were not presented, Photocatalysts can convert light into chemical energy through a seties of electronic processes and surface reactions. TiO, is used ‘most frequently. Ina review paper by Hernandez-Alonso etal (92), the pros and cons of TiO, and alternative photocatalysts were described. Hydroxyl radicals ("OH) are the key intermediates inthe photo oxidation reactions with TiO. These entities readily react With organic molecules, which results in oxidative degradation: OW + organics -+ CO; + HO a4) In Fig. 2, a schematic ifustration is presented for the operation of the Ti, sensor Zhang et al, [16] used ion chromatography to determine the COD. In a photocatalytic oxidation system including nano- particulate Ti0,-K,S,0,, al the organic compounds degraded while 5,0} was reduced to SO}, which was measured via ion chroma- tography. Chloride was separated in the chromatographic system, bout st was still present in the oxidation step; no indication of chloride interference was given in the paper. 1 and Song [17] used nanoparticle TiO; as a photocatalyst and ceriumi{V) sulphate a5 the oxidizing agent. The fluorescence in- tensity of Ce?! was detected, The method was linear in the 0-100 mg L-' COD range, Nochloride interference was found upto 2000 mg L-*in this rapid and accurate method. One problem was hhow fine the nanoparticle powder was: the powder was so fine that catalyst recovery was difficult. Kim etal. [18] compared a photocatalytic sensor combined with TO, and an oxygen electrode with the permanganate and dichro- ‘mate method. The response values of the sensor were extremely 00d forall 20 compounds investigated. The sensor was described ina separate paper [151 Zhao et al.| 19] developed a photoclectrochemical method based fn the direct quantitation of the electron transfer extent at a TiO> ‘nanoporous film electrode in athin layer photoelectrochemical cell “he linear range was 0-200 mg L-!, and the procedure required, 5 min per sample and very little reagent. Excellent agreement with a standard dichromate method was achieved. The method was stated to overcome matrix interferences, but no indication of chloride interference was presented. Currently, the photo- electrochemical oxidative degradation principle is commercially available as the PeCOD® system. Its not clear whether this system can replace the standard COD procedure, and the main problem ‘might be the small detection cell with a volume of 3 jl. which implies filtration of samples, , only the dissolved COD is analysed. Particles might be the most important COD cartier, and no comparability with standard procedures exists unless the samples ate filtered. In an attempt to investigate the effect of the COD in waste water particles, 17 samples were analysed in duplicate with and without filtration by a 0.45 um filter. These analyses were Conducted using a small sealed tube test (LC1 500, Hach-Lange). Fig. 3 shows thatthe COD concentration in the filtered samples is less than that in the non-filtered samples, which implies that the results are not comparable. The mean recovery was 84% (range 16-17132). In Fig, 4, the recovery percentages for the PeCOD® and COD. results for 74 of 109 wastewater samples are displayed and compared with the cuvette analysis. The average recovery was 73.3% 28.1%. These results are in good agreement with the results from the unfiltered and filtered samples. Thirty-five samples were below the detection limit because ofthe necessary dilutions for the chloride content. ‘Another distinction between the methods is the sample pl, which is much higher in the photoanalysis method. Preservation of the samples at alow pH and analysis of the samples at ahigh(er) pH requires additional laboratory handling that is difficult to automate. ‘According to Han et al. [20] a Ti03(BDD heterojunction photo- anode can effectively handle some PeCOD® problems, the BDD sensor has a significantly longer life time, a wider application range (including low pH samples), strong acid resistance and improved photocatalytic activity in comparison to the ITO and Ti electrode substrates. Zhang and Zhao [21] proposed an organic addition method to overcame the chloride interference in PeCOD® systems. ‘This method works well for (diluted) samples provided the Cl / CCOD ratio of 5:1 is not exceeded, ‘Wang et al, [22] developed 2 TiO, nanorod array electrode as a sensor to determine COD based on the photoelectrochemical degradation principle. This method has a practical detection limit of 183 mg L and a linear range of 20-280 mg L-' The method was tested with synthetic and real samples, and the results were in good agreement with a standard dichromate method. However, the tolerance for chloride was nat specified, “Mut al. [23 used TiO nanofibres fixed in a microfluidic device as sensot for PeCOD® measurements. The use of nanafibres instead of nanoparticles leads to better adsorption of organic compounds ‘on the TiO surface and is wel sited for the efficient separation of| photogenerated holes and electrons. The procedure resulted in an increased photocatalytic activity with a working range of (0-250 mg L * COD and a practical detection limit of 0.95 mg L. No indication of the chloride interference was presented. Zhang etal, [24] developed and optimized a preparation method with a highly ordered Tid nanotube array electrode. Ths electrode was used in PeCOD® systems and has a wider working range compared to other methods using nanoparticles or nanafilms. A working range of 0-850 mg L-" was achieved with this method with a chloride tolerance up to 2000 mg L- U et al, [95] developed a visible lightassisted photocatalytic system using 5-sulphosalicylic acid-Ti0,-KMn0,, The dissolved COD could be determined via depletion of Mn03. Under optimized conditions (pli 3.0, 80°C, 30 min radiation time), linear range of| (03-400 mg 1! was achieved. The chloride interference up to 180 mg 1- was less than 5% and could be eliminated. 4. Conclusion ‘The substitution ofthe lengthy BOD by a faster COD procedure has resulted in a number of investigations on the two methods. It has been a struggle to obtain a reliable method that delivers reproducible results in all types of matrices, Over the years, four primary oxidation chemicals have been introduced, Permanganate and dichromate are stil n use today, and dichromate is still used in standard quantitative procedures. The contribution of chloride, Which is not oxidized under natural conditions to the COD figure has been a challenge to chemists. Standard COD reagent formula- tions use toxic chromium(V1). mereury(l) and silver) sats to mitigate chloride interferences. Regrettably, these toxic reagents contribute to environmental pollution. In environmental work, stich methods afe not recommended. Governments play an important role in this matter as it is necessary to legislatively regulate wastewater discharges using reliable and robust methods. In addition, governments benefit from the accompanying tax ‘measures, so phasing out the mercury-assisted dichromate method is not achieved easily, Modem techniques do not use these sub- stances anymore, and good correlations compared to standard» Cerda Anat hrc a 861 207) 1-1 methods have been obsained. However, legislation is not easily amended, Real progress from an environmental, technological and analytical-chemical point of view has been reported over the last ‘decade, and four procedutes show great promise. (1) A mercury-free method based on international standards in ‘hich the reaction temperature is the main alteration. AC- cording to Geerdink et al [$1], approximately 95% of the ‘samples in the Netherlands could be analysed without ‘mercury by reducing the reaction temperature to 120°C. (2) Microwave and ultrasound oxidation in which the duration fof the analysis is reduced (53,65,56,58]. However, in these procedures, the toxic substances are still present. The chlo- Fide ion interference can be inhibited up to 6000 mg Cl” L (3) Flow injection and permanganate oxidation with luminol- HpO;-chemiluminescence detection still requires more research. The concentration of permanganate or other re- agents such as luminol, H02 andjor acidity have to be ‘optimized. In addition, the flow-rate of the sample must be adjusted, and interferences from metals such as Cu(l) Co) ‘and Cll have to be overcome. Anather drawback ofthe FA method is the high consumption of reagents. (4) The latest development for COD analysis is photocatalytic and electrocatalytic oxidation, which most frequently use ‘Ti; [9-23]. These procedures do not use any of the afore- mentioned toxic reagents: they are simple and rapid and ‘may be the solution for dissolved COD analysis in the future. ‘The challenge for electrochemical detection is finding a good ‘working electrode to determine the quantity of “OH gener- ated, For the commercially available photocatalytic PeCOD®, ‘the challenge might be the small cell volume, Additionally, this technique relies on the measurement of currents from electrochemical reactions, which depends strongly on the Individual compounds in the case of organic. matter ‘oxidation. ‘The views expressed in this paper are on behalf ofthe authors and not necessarily the Dutch Ministry of Infrastructure and the Environment. References [n} 1506060, Determination of she Chemica Oxygen Berane. neratonl OF {anton fr staneardiation, Geneve, Swisetane 1285 fa] Be. hier Sv Srayton, Wt hoyle Water Fnuron. Ret 73 2001) 62-7 BB] WA Moore Ke Kconer, Ruch Anal chem 2111949" 953-9). Hal WA Moore Re Keener Ce fuetne, Anal chem 231981; 1297-130, [5] RA Dobbr kT Wiliams, Anal chem 35 155) Toe4-1057 To] F Asen GM Morison intern) Environ Anal Chem 59 (1585) 59-76, Il standard esas foe Famintan of Wate: ae Waser tenth 1a Keon. Mend. Bet KE Cate fea 981 (1986) Ia} vane 1 chen ¥. Zou, K Wa, Sens AesustrsB 153 (2011 7662, Ho] ¥.2hos Jing. Ham Zou 5 Mey lero Aeta 742012) 165-170. HMI] S.C ts Cheng ¥ Sean Ai Le Anal Chum. Acta 5462003) 199-200 [ta] Cia fTan tuo. 5 Che, x Quan Seas Actuators 95 2011) 1€—130 TAL Wang GaSe ng, Htc, Conan 9 (207 (1 Bega caponak. ML cay Osews, ML Sobek [5] CR Silva, CDC Conceigae. VG. Bonfécie, OF. Fho, MES. Teixeira, |. Solid Ste leche 1 2008) 569-039 Ine] 22H zhang Hoc Dong Yo) ane YZ Xan. Cin, Chem 252007 In7] Ce Song. Sens Actstrs 8 137 (2008) 432-836, Tiel ¥-c' kim Kat Lee 5 soeaie K thstimete, cebu, L Karube An lia) Hzno, 0. ang. 2ang. Jobs Aba Che, 76 (2004) 155-160. Hol ef). ih ng thane, Aa eh 3201 fat) 5 2hang zhao, Anat 133 208) 1684—160 TALE tg. gol, an ln. ta 6 (20 faa] Gea ¥L.@ zing Wang Sens. Actn 8 1552001 04-200, foal f2mang, ou © eng) at Wn, Wr ak Water Ree (009 (25) Fat, Eng. 3.56089) 16 FasL Rigs ray, Water Se Works 951548 far] Al Meda Ava chen 951950) B3Re1370. [ae ti lve carne) st sev Pr (1938) 29, Pas iin Poe tse Sew Punt ast te faa fy Sprenger vom Wasser a6 (1979) 15-198 [otf Lec Tan 534 Al chi eta 36 (199) 201s, faa € oreonn hat et 131900) 001-tont [hal Maz aoe Teekes Ke Myng-loon Ball Koran Cem Se. 19(1988) {a4 9 Da RC. Sandor, Wf, siya 120 (205) 227-22. Eis] MZ 5. Fjwura 7 Yonoyama foal Se 22209) 77-0 [a6 Bist thang’) Wang € RTs 6 (200) eS, fal ¥ Sic Cen ou Meche 47 207) 3 fae] Yao. 8c Hr Gu ¥cheng, Atal Chi, Acta 6292009) 760 18] Re Tn do, NA Cana 38 (988 (4a) 2-0 thang. ot. Ya LY, pet 25 2004 191196, Ua Ww adeay, i Dawson. coo. Dub Su 181826) 199-20. tea] WE Aor iad ng them. 19 a a9 Hea Et Monson, van 3a. 265-257. (ea sauman, Anal hem 48 (1974) 1338-1298, [45] NEN 9533 Determination of Chemist Oem Demand (Bepaing van bet Shea sanstverbie, 2008-200 (4s) AT Store Morgan Enon Sc ieehol 8 (1984) 450-456 U4 $0 tying Sarre Water Re 1177 9-05, tat] Xun Yang Wang’ Meng tovirn, Mont ses 10 (2005) 185-20 Us| hi Morn} Soc Chem Ind fond 5817950 70-7 Uo] Dic trew,SoncAona rs vito Tes 10 [148 1242—124. I51fB tanger A ioyd Alin gs 10 (a) 1047-105 tsa] Baxi) rer 611500 ola-ot ‘ Ua Seth arene Abo, Water Re 26 (1982 1577—158 Ul te Sto ne Vou R Mendes} Lea ion Teck i T0989 (56] Aca, del Reni Hennes. An. Res Chem. 374 (2002 {59 Hf i ME, cosine 422007) 1665~167 [5H ces Donia Haig Fitter. Can, Aaa Cn Aca 367 006 [5 Howser, Enon. Eng. S20 (2003 617—526. To Ktaos vo, teva, Anal Chin Aca 43 (203) 9-169, Ist WE Juri. JR Rotweader Water es 25 ows) 158-1071 Usa] Ne Sister # Aborough, ter S18 138) 185-14. ISL A Berra) sent rat ave V Monti A ld A. Can Ioa) fers Tce) er. Canal ACh A 372 (198 399-40, 1651 Ba harman AP Vanesa No, Bahve. Coote. Sr Teena 5 (anos st9e—ns 1} ae re Pe A Sera | Na.) Ma Me {67] Mul Vite Mt Foch}. Alonse | fatoh. vren, Tene Iss) Teng, Tien, Lf Wi Ana. Sei 17 (2001) $51-$53, U9] Falta 1 Tanes, &Rwire (70) {i bach, 2 Anal, Chem 891934) 29-44 [rt] Wat easton, tm Moore, Anlst 2 (1857) 677-582. fra] chew 2 Anal Chem 95 (1955) 305-310 (3) ex suas © Macs Weer 37 (965) 176172 [Pal Joncas Jens, wrcr 36 (1968) 1240—1248" [73] NEN 3150, Deermusion of cheniel Oxygen Desa, (Bepaing van e oevelbeldchemisch verte rast 263, [75] DIN 38409, summary Acton abé Material Characters values (Summa She Wirkungs und SttKenngossen) (cruppe Te 43), Geran Istute for Stncares I771 Bivscys SW. Wation,S). olden. M, Pores, Anal. chin, Acta 357 (1997) [74] CE Dorin Vil A Canals, Ukraton Sonachem, 16 (2009) $86-601, [79] U2et, Gv WateecAbwater 117 (1976) 18-184 THol & Coy. Anat To? (1982) 115-1018Gerd eo Any Chea Ac 861 (207) 1-11 Isto ce nO per Al Mec 2am) 18-116 HL Ee Gee Behe War Ft Canal Fe [ea] 1 Gadernaseh, W. Dopp 26 Waser Abwasser Foneh. § (1975 (oa 5 Ea Rag alsin Pasi aCe ag B (sy SEL Bao Bezel Alcina 1882 [as] Wu 2 Zhang Yang. Meroe Acta 160 (2008) 141145 [a2] 8 zhang tt Zhan Dang John, Anal hun. Aca 514 (2004) 8-37 [aa] § Zhang 0. jane HL zo, Enon sc. Technol 40 (2005) 2585-2358, 9] § zhang ULL H Zo, cu, Sens Actatrs 3141 2009) 634-640. [30] P.CabuaPlacion Pastderastemandes [ane Reveals, Water Se Technol. 65 (2012) 1059-1073 [pt] & bozeanowie | Ceupymiak. Gaya, yl Osoms, M.Sobazek Daromees Poe Eng 47 (2012) 11171130. [82] Mb Hemdndes-flono, Frese, Sule JM. Crensee nergy Environ. Se. 2 (2009) 1251-125) Iss] SLL F zheng. cau W. Wenge Lang, VU Sens. Atators 8 185 (2015) dao, Sone gcd Pi ene et of ge a Sth he as a pied nae ha Se Sposa Tr the last 30 yeas be has een worn st - dhulyea ebewonnenl hens ate ykswattstat inborn suface ad waste wate: eel ick van den Hark is «stent at Unversity f Appi ‘Rie ct it Senn ety Oc ‘Onna Hpema war dusted in chemise Universy sondern tay terpenes the sens share evo fhe Univers 9f Warr tuued a PRD in envortental cherish ont sped of Serengeti gp sey es Sbecuenly wok was dae a the nse far Ene tne tue fasted). Fr the hat 1s yess he has bees wong nana eoveenietal chery the jbowaterstant abate an ssface aod waste wate ‘suit