A ferricyanide-mediated activated sludge bioassay for fast determination of the biochemical oxygen demand of wastewaters

w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 8 1 e5 9 8 8

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

A ferricyanide-mediated activated sludge bioassay for fast determination of the biochemical oxygen demand of wastewaters Mark A. Jordan a,b, David T. Welsh a,*, Peter R. Teasdale a, Kylie Catterall c, Richard John b a

Environmental Futures Centre, Griffith University, Gold Coast campus, Qld 4222, Australia Griffith School of Environment, Griffith University, Gold Coast campus, Qld 4222, Australia c Gold Coast Water, PO Box 5042, Gold Coast Mail Centre, Qld 9729, Australia b

article info

abstract

Article history:

Activated sludge was successfully incorporated as the biocatalyst in the fast, ferricyanide-

Received 7 March 2010

mediated biochemical oxygen demand (FM-BOD) bioassay. Sludge preparation procedures

Received in revised form

were optimized for three potential biocatalysts; aeration basin mixed liquor, aerobic

25 June 2010

digester sludge and return activated sludge. Following a 24 h starving period, the return

Accepted 14 July 2010

activated sludge and mixed liquor sludges reported the highest oxidative degradation of

Available online 23 July 2010

a standard glucose/glutamic acid (GGA) mixture and the return activated sludge also recorded the lowest endogenous FM-respiration rate. Dynamic working ranges up to 170

Keywords:

mg BOD5 L
1 for OECD standard solutions and 300 mg BOD5 L
1 for GGA were obtained.

BOD5 assay

This is a considerable improvement upon the BOD5 standard assay and most other rapid

Wastewater treatment

BOD techniques. Time-series ferricyanide-mediated oxidation of the OECD170 standard

Fast BOD

approached that of the GGA198 standard after 3e6 h. This is noteworthy given the OECD

Ferricyanide-mediated respiration

standard is formulated as a synthetic sewage analogue. A highly significant correlation with the BOD5 standard method (n ¼ 35, p < 0.001, R ¼ 0.952) was observed for a wide diversity of real wastewater samples. The mean degradation efficiency was indistinguishable from that observed for the BOD5 assay. These results demonstrate that the activated sludge FM-BOD assay may be used for simple, same-day BOD analysis of wastewaters. ª 2010 Elsevier Ltd. All rights reserved.

1.

Introduction

Developed in 1936, the 5-day biochemical oxygen demand (BOD5) assay remains the only internationally recognized standard method for estimating the biodegradable organic loading of wastewaters (APHA, 1999). The BOD5 bioassay is a direct estimate of the amount of oxygen consumed by microbial respiration under standard incubation conditions.

Most determinations of BOD5 are made on samples of wastewater or treated effluent to meet regulatory requirements (Higgins et al. 2004). The assay commonly incorporates a mixed microbial seed activated sludge from a wastewater treatment plant (WWTP) as the biocatalyst. The incorporation of a biological component in the assay means that BOD5 measurements can be used to provide an estimate of the biodegradability and hence the oxygen demand of a sample.

* Corresponding author. Tel.: þ61 7 55529186; fax: þ61 7 55528067. E-mail address: [email protected] (D.T. Welsh). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.07.042

5982

w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 8 1 e5 9 8 8

This is a significant advantage of the assay compared to rapid surrogates such as chemical oxygen demand (COD) measurements, which cannot differentiate between the biodegradable and non-biodegradable fraction of organics (Zhao et al., 2004). However, due to the low solubility of oxygen in water (8.7 mg O2 L
1 at 25  C) and the limitations of dissolved oxygen electrodes, BOD5 measurements must be measured over multiple serial dilutions to ensure that changes in dissolved oxygen concentrations are able to be accurately quantified (APHA, 1999). The requirement for the assay to extend over 5 days is partly to ensure sufficient degradation of the waste material, 60.5% substrate oxidation is typical (APHA, 1999), and partly an historical idiosyncrasy (Baird and Smith, 2002). However, this duration also represents the most significant shortcoming of the assay, as it would not be uncommon for the results of assays to be obtained long after the wastewater or effluent being measured had to be released. Given the need to decrease analysis time and the costs of undertaking BOD5 analysis, BOD biosensors were developed (Karube et al., 1977) and have attracted considerable interest (D’Souza, 2001; Liu and Mattiasson, 2002; Rodriguez-Mozaz et al., 2005). BOD biosensors measure the bacterial respiration rate, at an oxygen electrode interface, as a function of biodegradable material in as little as 1e20 min (Liu and Mattiasson, 2002). However, as the analytical signal consists of monitoring the flux of oxygen to the electrode surface, BOD biosensors are also constrained by oxygen solubility. The biological component generally consists of an immobilized aerobic bacterial population (often grown as a biofilm) with a broad substrate range or preferably a consortium of heterotrophs (D’Souza, 2001; Liu and Mattiasson, 2002). Another issue with BOD biosensors is that generally <1% of available biodegradable organics are mineralized over the duration of the measurement (Morris et al., 2001). This means the most readily degradable substrate pools are over-represented in the analysis, which can lead to overestimation of the BOD of a sample. A small number of studies have sought to partially overcome this inherent limitation by successfully immobilizing activated sludge as the biocomponent (Liu et al., 2000; Matsuoka and Isoda, 2005; Sakai et al., 1995). However, practical application of the biosensors is limited, given the instability of the community composition over time and their tendency for biofouling (Liu and Mattiasson, 2002). In view of the intrinsic problems associated with utilizing oxygen as the terminal electron acceptor, Yoshida and coworkers developed a BOD biosensor replacing oxygen with a redox mediator, the ferricyanide ion (Yoshida et al., 2001, 2000). Results gained using this novel approach compared very favorably with existing oxygen-based biosensors, but still suffered from the limitations common to biosensors. Around the same time, Pasco et al. (Pasco et al., 2000) developed a bioassay, also using the artificial terminal electron acceptor ferricyanide in place of oxygen, but with a separate incubation and detection step, similar to the BOD5 method. Ferricyanide-mediated biodegradation proceeds via glycolysis and the tricarboxylic acid cycle (Morris et al., 2005) and is analogous to aerobic respiration (Eq. (1a)). However, in the absence of oxygen, the electrons produced via substrate oxidation are shuttled down the electron transport system to

reduce the terminal electron acceptor ferricyanide to ferrocyanide (Eq. (1b)). The overall process of ferricyanide-mediated (FM) respiration is displayed in Eq. (1c) (Morris et al., 2005). Glycolysis=TCA

CH2 O þ H2 O




! CO2 þ 4Hþ þ 4e


(1a)

 3
4
ETS  FeðCNÞ6 þe

! FeðCNÞ6

(1b)

  3
Biocatalyst 4
CH2 O þ H2 O þ 4 FeðCNÞ6





! CO2 þ 4Hþ þ 4 FeðCNÞ6 (1c) Since its inception in 2000, considerable research into commercialization and optimization of the FM-BOD method and biocomponent has been undertaken (Catterall et al., 2001, 2003; Morris, 2005; Morris et al., 2001, 2003; Pasco et al., 2004). Early studies employing Escherichia coli generated a great deal of interest in the FM-BOD assay as reported glucose-glutamic acid (GGA) degradation efficiencies were comparable with the BOD5 standard method and yet the total analysis time was a little over an hour (Morris et al., 2001; Pasco et al., 2000). Achieving degradation efficiencies of this magnitude over such a short space of time is possible by virtue of the high solubility of ferricyanide in water compared with oxygen. Ferricyanide is w10,000 fold more soluble in water than O2 and consequently a considerably larger population of microorganisms may be used as the biocomponent without significant depletion of the electron acceptor (i.e. ferricyanide). Unlike the BOD5 assay and oxygen-based biosensors, which are limited by the electron acceptor concentration, the rate-limiting reactant in the FMBOD assay is the substrate, which is characteristically consumed according to the Michaelis-Menten model (Lehninger, 1970). Thus, the reaction rate is quantitatively related to the substrate concentration. The choice of biocomponent in any bioassay is of critical importance. E. coli represented a good biocatalyst for the early proof of principle studies (Morris et al., 2001; Pasco et al., 2000) chiefly because ferricyanide-mediated respiration is well characterised for the species (Emde et al., 1989; Patchett et al., 1989). In particular, an important recent study (Catterall et al., 2010) has clearly demonstrated that FM-respiration responds quantitatively to both stimulation and inhibition of E. coli activity, validating the analytical technique for the first time and confirming its suitability as a BOD assay. Nevertheless, in spite of excellent degradation efficiencies over short time scales, E. coli is not recognized as a suitable species for rapid BOD analysis, given its limited ability to assimilate a wide range of organic substrates. In order to extend the substrate spectrum, a defined consortium of pure culture microorganisms from diverse heterotrophic groups was employed as the FM-BOD biocatalyst (Catterall et al., 2003; Morris et al., 2003). Using the GGA standard as a calibrating solution, excellent agreement was observed between FM-BOD and BOD5 measurements for a large number of industrial effluent samples (n ¼ 27e30) in as little as three hours. However, in some cases relatively recalcitrant samples were grossly underestimated (Catterall et al., 2003; Morris et al., 2003). It became apparent that if the FM-BOD assay was ever to be considered for widespread and facile BOD determination of industrial wastewater samples, varying greatly in terms of

w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 8 1 e5 9 8 8

BOD5 value and biodegradability, a more diverse biocatalyst was needed. Successful integration of activated sludge as the biocomponent in an FM-BOD assay would overcome the timeconsuming and expensive requirement of culturing several strains of bacteria to produce a mixed consortia biocatalyst. Activated sludge also represents the ideal biocatalyst for the BOD5 assay as it contains a diverse community of heterotrophs from a number of functional groups that will ensure a high proportion of biodegradation occurs (Dias and Bhat, 1964, 1965; Hiraishi et al., 1989). In addition, activated sludge can be readily sampled from most WWTPs and an activated sludge FM-BOD assay would be far more relevant to the wastewater industry, as measurements of the industrial effluent BODs, would be representative of the BOD loading exerted upon the plant activated sludge ultimately treating the same wastewater. The primary objective of this study was therefore to develop an FM-BOD assay utilizing activated sludge as the biocatalyst. A number of experimental conditions were optimized with regards to increasing net FM-respiration of the sludge. The rate and pattern of substrate degradation over time was investigated using both the GGA198 and OECD170 standard solutions. These standards were also used for calibration of the assay and determining the working range, which is based upon adherence to the Michaelis-Menten model. Finally, the utility of the assay for rapidly predicting BOD5 values was investigated by correlation with the BOD5 assay, using a range of real industrial effluents of varying complexity and BOD5 value.

2.

Experimental

2.1.

Reagents

All reagents used in this study were of analytical reagent grade and all dilutions were made using deionized (Milli-Q, Element) water, unless otherwise stated. The BOD198 GGA and BOD170 OECD synthetic wastewater standard solutions were prepared according to the APHA standard methods (APHA, 1999) and OECD-209, activated sludge respiration inhibition assay (following a 100-fold dilution) (OECD-209, 1984) respectively. Potassium ferricyanide (Ajax) solutions were prepared in phosphate buffer (PB) (0.08 M KH2PO4/0.12 M K2HPO4, pH 7, Chem-supply).

2.2.

Activated sludge preparation

The type of sludge used in the assay, the absorbance of the sludge and the starvation period were optimized in this study. Further details on the sludge characteristics are in the supporting information (Table S-1) and in Section 3.1. All sludge was collected fresh as required from Coombabah WWTP, Queensland, Australia. In the laboratory, the sludge was vigorously shaken with glass beads for w1 min, wet sieved through a 500 mm sieve with PB and washed, centrifuged (1 min at 1300 rcf) and resuspended in PB. This step was repeated once. The final sludge/PB slurry was then aerobically incubated on an orbital shaker (100 rpm) at room temperature for the stated starvation period. The sludge microorganisms

5983

were then harvested by centrifugation and the absorbance of a diluted suspension of sludge was determined spectrophotometrically at 600 nm. The sludge was then diluted to the stated absorbance (optical density) using PB.

2.3.

Sample preparation and analysis

Solutions to be incubated contained mixtures of activated sludge, ferricyanide and the sample/standard to be analyzed. Each 6.0 mL incubation solution was prepared with the following volumes: 3.0 mL of activated sludge, 2.0 mL of sample/standard and 1.0 mL of 0.36 M potassium ferricyanide (final concentration 0.06 M). Controls to determine endogenous metabolism (baseline metabolic rate in the absence of added substrate) were prepared by replacing the sample/ standard with Milli-Q water and sample interference blanks were prepared by substituting the activated sludge with a 10 mM potassium ferrocyanide/ferricyanide solution in PB. The optimal incubation time of the assay will depend upon the nature of the samples to be analyzed. Further details are provided in the results and discussion section. Samples were incubated in 10 mL plastic vials at 20  1  C (consistent with the BOD5 assay) on an orbital shaker. Approximately 1.5 mL aliquots were subsampled at 3 and 6 h into 2 mL microtubes and the reaction was terminated by centrifugation at 13,700 rcf for 3 min (Catterall et al., 2003). The supernatant was analyzed for ferrocyanide using chronoamperometry at a 25 mm platinum working electrode, as described previously (Catterall et al., 2001; Morris et al., 2001). The analytical signal, the diffusion limiting current, is proportional to the flux of ferrocyanide to the electrode interface (Eq. (2)) and is therefore proportional to the concentration of ferrocyanide in the bulk solution, according to the equation (Catterall et al., 2003): ilim ¼ 4nFrDC

(2)

Where ilim ¼ the limiting current (A); n ¼ the number of electrons transferred per mole of ferrocyanide (¼ 1); F ¼ the Faraday constant (C mol
1); r ¼ the microelectrode radius (cm); D ¼ the diffusion coefficient of ferrocyanide (cm2 mol
1 s) and; C ¼ the concentration of ferrocyanide in the bulk solution (mol cm
3).

2.4.

Calculation of FM-BOD equivalent values

As with the BOD5 assay, either the GGA198 or OECD170 standard solutions were used as a standard check to determine microbial seed effectiveness, or in this case, standardize sludge activity. Calculation of FM-BOD equivalent values for real samples relies upon a four-point calibration curve that is based upon adherence to the Michaelis-Menten model (Eq. (3)). The four calibration standards (three standards plus one blank) are incubated and analyzed alongside real samples. Therefore, amperometric limiting current (ilim) values derived from a real sample can be easily converted to an equivalent FM-BOD value, relative to either of the BOD5 standard solutions, once the coefficients a and b are empirically defined (Eq. (3)). In all cases, the limiting current for the control endogenous metabolism incubation was subtracted from the

5984

w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 8 1 e5 9 8 8

sample/standard gross limiting current value prior to calculation of FM-BOD values. ½FM
BOD ¼

b$ilim ða
ilim Þ

(3)

Where ilim ¼ the corrected analytical signal (sample e endogenous control) and represents the stimulation of the rate of the enzymatic reaction; a ¼ the maximum reaction rate and; b ¼ the Michaelis-Menten rate constant (Lehninger, 1970).

2.5.

BOD5 and statistical analysis

BOD5 values were derived according to the APHA 5210B standard method (APHA, 1999). The microbial seed used for the assay was return activated sludge (RAS) from a local WWTP (Coombabah, Gold Coast Water). Following a 3000-fold dilution, the seed was prepared as described (APHA, 1999). The slope of the principal axis of the correlation ellipse was determined as described previously (Catterall et al., 2003; Morris et al., 2003). All assumptions were met and subsequently the data was not transformed.

2.6.

FM-BOD limit of detection

Ten separate samples, each containing 10 mg BOD5 L
1 OECD standard solution were incubated for 3 and 6 h and analyzed for microbially produced ferrocyanide, less the endogenous contribution. The method detection limit (3  S of the limiting current values) was converted to an FM-BOD detection limit concentration as described above (Eq. (3)) and equaled 13.7 and 9.78 mg FM-BOD L
1 for the 3 and 6 h incubations, respectively.

3.

Results and discussion

3.1.

Biocatalyst optimization

Return activated sludge, aerobic digester activated sludge (ADS) and aeration basin mixed liquor (ML) were trialed as possible biocatalysts for use in the rapid FM-BOD assay, using the GGA198 standard as the sample. Preliminary assessments revealed that, with no starvation period prior to sludge washing and incubation, the contribution of endogenous metabolism to total ferrocyanide production for all sludges was unacceptably high, at w80%, thus decreasing the sensitivity of the assay (Fig. 1); treatment (a). Two treatments were investigated to reduce this effect, both incorporating a 24 h starvation period prior to incubation, but with the sludge either washed after (b) or before starving (c). Fig. 1 shows that the contribution of endogenous respiration to the analytical signal was reduced most effectively by washing the sludge prior to the 24 h starving period (treatment c), with the endogenous contribution representing 64, 69 and 67% of the total for the RAS, ADS and ML sludges respectively. This proportion is still relatively high compared to those reported for pure culture microorganisms (Catterall et al., 2001; Morris et al., 2001; Pasco et al., 2004), although not unexpected. In addition to living microorganisms, the flocs present in activated sewage sludges contain particulate and soluble organic

Fig. 1 e Optimization of net ferrocyanide production from three different activated sludge seeds (RAS, ADS and ML) and as a factor of three different sludge preparation procedures prior to FM-BOD incubation, a) no starvation period, sludge washed only; b) sludge starved for 24 h, then washed; and c) sludge washed and subsequently starved for 24 h. All sample solutions were incubated in the presence of the GGA198 standard for 3 h. The black shaded bars represent the endogenous respiration and the open bars represent oxidation of the added GGA (total e endogenous). Final sludge absorbance [ 7; final ferricyanide concentration [ 60 mM. Limiting currents determined by chronoamperometry at Eapp [ D450 mV (vs. Ag/AgCl).

matter derived from the wastewater, microbially produced polysaccharides and dead cells (Bura et al., 1998), all of which could fuel respiration in the absence of exogenously added organic carbon. In addition to the improved endogenous response, treatment c also recorded the highest net FMrespiration rate, i.e. GGA substrate oxidation only, representing an increase of 242% for the RAS, 180% for the ADS and 162% for the ML compared with treatment a (Fig. 1), thereby greatly improving the sensitivity of the assay. This disparity between sludges is influenced, at least in part, by the relative population density of heterotrophs in each sludge and their metabolic state. The aeration basins, where the ML and ultimately the RAS are sampled from, select for predominantly heterotrophic bacteria, while the ADS sludge consists of both heterotrophs and autotrophic nitrifying bacteria, selecting for the latter due to carbon limitations. A follow-up study investigated the optimal starvation period for each sludge (i.e. 12, 24, 48 and 72 h). 24 h represented the best compromise between maximizing sensitivity and minimizing both the endogenous response and time constraints, which would limit WWTP applications (Fig. S-1). Results gained from both optimization studies suggest that the RAS and ML have a superior net sensitivity and a lower endogenous contribution than the ADS. The RAS was chosen as the biocatalyst for all subsequent FM-BOD incubations, in part due to the lower endogenous respiration, but also the RAS is maintained at a total suspended solids concentration 3 fold that of the ML (Table S-1) and therefore in practical terms, the volume of RAS requiring centrifugation/washing is a third that of the ML.

5985

w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 8 1 e5 9 8 8

3.2.

Microorganism concentration

The optical density of the RAS was optimized with regard to maximizing sensitivity and the dynamic working range of the method and yet maintaining a sludge density that could be easily prepared. Three replicate incubations investigated FMrespiration at 12 different sludge absorbances ranging from 1e15 (final concentration). Mean net ferrocyanide production remained linear up to optical density (OD) 13 (n ¼ 10, R2 ¼ 0.997, p < 0.001), at which point the additional ferrocyanide production started to decrease with increasing OD. However, beyond OD 10 the sludge was too thick to be quantitatively pipetted and instead required individual weighing into each sample tube. Moreover, a choice of OD 10 required no compromise between sensitivity and the dynamic working range of the assay (data not shown). Thus, a final OD of 10 was chosen for all subsequent incubations.

3.3.

Time-series substrate oxidation

5

GGA

4

OECD

3 2

Dynamic working range

The working range of the assay was investigated using various dilutions of the GGA and OECD standard solutions, over 3 (Fig. S-2) and 6 h (Fig. 3) incubations and on three separate occasions using the OECD standard (i.e. using sludges

30 25 20 15 10 5 0

1 0

3.4.

Limiting current (nA)

[Ferrocyanide] (mM)

The pattern of substrate degradation over time was investigated using the GGA198 and OECD170 standard solutions (Fig. 2). The net rate of GGA oxidation surpassed that observed for the OECD standard over the first three incubation hours. This was not unexpected given the higher BOD5 value of the GGA standard compared to the OECD170 standard, which is formulated as a synthetic sewage analogue and is typically not as readily biodegradable as the GGA198 standard (Kwok et al., 2005; Morris, 2005; Pang et al., 2007). Oxidation of both standards appears to follow the same trend; a somewhat 3-phase respiration curve that increases between 90 and 180 min. This is possibly explained by a substrate oxidation lag phase, as the starved bacteria increase their metabolic rate and return to an active growth phase. By 6 h, mineralisation of the OECD standard is indistinguishable from that of the GGA standard. Once much of the readily degradable glucose is exhausted (GGA standard; w3 h), oxidation of the more recalcitrant

glutamic acid proceeds at a slower rate (3e6 h). This is not uncommon as a number of microorganisms, such as E. coli, are known to preferentially and rapidly oxidize glucose before all other substrates, as glucose causes catabolic repression of other carbon and energy source use (Catterall et al., 2001; Morris et al., 2001; Pasco et al., 2004; Tortora et al., 1997). These findings are significant given the complexity of the OECD standard, which is composed of a range of substrates of varying lability and therefore better represents WWTP influents that the activated sludge community is typically exposed to. For these reasons, a number of other rapid BOD studies have used the OECD standard in preference to the GGA standard (Kumlanghan et al., 2008; Liu et al., 2000; Velling and Tenno, 2009; Yoshida et al., 2001, 2000). Although not shown in Fig. 2, since the endogenous control values have been subtracted, the assay is restricted to a minimum incubation time of 3 h, below which the endogenous contribution to the overall signal becomes excessively large, as high as 89  5.7 and 91  2.8% for the 15 min incubation with the GGA198 and OECD170 standards respectively. In comparison, for a 3 h incubation the endogenous proportion fell to 68  5.0 and 70  5.9% for the GGA198 and OECD170 standards respectively and further declined to 66  5.0% for both standards after 6 h incubation. Further validation of the calibration standards and incubation time for the activated sludge FM-BOD assay will be investigated in the following two sections.

0

60

120

180

240

300

360

Incubation time (mins) Fig. 2 e Time-series oxidation of the GGA198 and OECD170 standard solutions as measured by ferrocyanide production. Each point represents the mean response from three replicate incubations. Endogenous control values have been subtracted. RAS was used as the biocatalyst; final sludge absorbance [ 10; final ferricyanide concentration [ 60 mM. Limiting currents determined by chronoamperometry at Eapp [ D450 mV (vs. Ag/AgCl).

0

100

200

300

400

[BOD5] (mg L-1 ) Fig. 3 e Determination of the 6 h activated sludge FM-BOD dynamic working range, using the GGA and OECD standards as the test substrates. The OECD working range was replicated on three separate occasions (closed symbols; solid lines). The GGA working range is displayed also (open symbols; dotted line). Endogenous control values have been subtracted. RAS was used as the biocatalyst; final sludge absorbance [ 10; final ferricyanide concentration [ 60 mM. Limiting currents determined by chronoamperometry at Eapp [ D450 mV (vs. Ag/AgCl).

5986

w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 8 1 e5 9 8 8

3.5.

FM-BOD and BOD5 comparison

The relationship between FM-BOD and standard BOD5 measurements was investigated for 35 industrial wastewater effluents of high diversity (with BOD5:COD ratios ranging from 0.16e0.76; mean 0.52  0.13) and variable BOD5 values (Table S-2). Calibration of FM-BOD equivalent values employing the OECD standard at longer incubation times (i.e. 6 h), provided the most robust correlation (Fig. 4). The strength of this relationship is noteworthy, given the high correlation coefficient and the slope of the principal axis of the correlation ellipse (1.07) approaches unity. Although on average the FM-BOD measurements are slightly underestimated, the 95% confidence intervals of the correlation ellipse range between 0.96e1.19. Based upon this, the mean variability of the FM-BOD/BOD5 relationship is 10.7%, which represents an improvement upon the recognized variability associated with the BOD5 assay alone of 15% (APHA, 1999). It is therefore likely that a greater proportion of the variability evident in Fig. 4 arises from the BOD5 measurements, especially considering the FM-BOD variability (i.e. R2 values) observed in Fig. 3.

06/05/09

4000

-1

[BOD5 ] (mg L )

collected on three different days). Close examination of the data reveals that each of the 6 h OECD standard curves remain linear to w80 mg BOD5 L
1 and the 6 h GGA standard curve remains linear to ∼100 mg BOD5 L
1. The linear working range for the 3 h incubation is approximately half that reported above for both standards. These reportable linear ranges compare favorably with the BOD5 assay and rapid O2 based BOD biosensors but poorly with previous FM-BOD studies using pure cultures of microorganisms (e0e300 mg BOD5 L
1) (Morris et al., 2005). It is apparent that the trend of substrate degradation remains similar, albeit at different magnitudes depending upon the specific sludge activity for the given day (Fig. 3). As expected, these patterns conform well with the MichaelisMenten model (Fig. 3; Eq. (3)) (Lehninger, 1970). At low substrate concentrations, the reaction rate is first-order, beyond which point the reaction rate starts to decrease and is no longer linearly proportional to the substrate concentration. By basing the dynamic working range of the assay upon adherence to the Michaelis-Menten model, the working range can be extended beyond linearity and compares well with the previously published FM-BOD literature. This is a more relevant approach for deriving FM-BOD equivalent values when the linear range is small rather than reliance upon a single point calibration and the assumption of linearity as previous FM-BOD studies have done, especially given the innate variability of activated sludge activity (Fig. 3) compared to pure cultures grown under standard conditions in the laboratory. For the OECD and GGA standards respectively, the range extends to approximately between 10e170 and 10e300 mg BOD5 L
1 for the 6 h incubation (Fig. 3) and 14e80 and 14e200 mg BOD5 L
1 for the 3 h incubation (Fig. S-2). The lower limit of the assay working range is derived from the FM-BOD limit of detection. Derivation of real sample FM-BOD equivalent values using Eq. (3) requires dilution to within these ranges and at least a four-point calibration curve (including zero).

28/05/09 03/06/09

3000

1

17/06/09

2000

1

R = 0.952 Slope = 1.07 p <0.001

1000 0

1

0

1000

2000

3000

4000

-1

[FM-BOD] (mg L ) Fig. 4 e Relationship between the activated sludge FM-BOD equivalent values and BOD5 measurements for 35 industrial effluent samples (Table S-1), using RAS sludge seeds collected on four different days. FM-BOD equivalent values were calibrated with the OECD standard; industrial effluent samples have been diluted to within the working range of the assay; incubation time [ 6 h. The solid line represents the principal axis of the correlation ellipse; the dotted line represents the ideal slope of 1. Samples marked with 1 have been divided by three so as not to skew the relationship.

The mode of deriving FM-BOD equivalent values (i.e. calibration based upon the Michaelis-Menten model) was validated by comparison with BOD5 measurements. Given the logistical restraints imposed primarily by the BOD5 method, the 35 industrial effluent samples were incubated and analyzed in four separate batches (on separate days), thus using four different individually calibrated, activated sludge seeds (Fig. 4). Yet, in spite of intrinsic temporal variability with sludge activity (Fig. 3) a strong relationship between the two bioassays was still observed. Closer scrutiny of the four separate data sets in Fig. 4 reveals that no one data set deviates markedly from unity or the other three sets. This finding is significant given the range of BOD5 values measured, the diverse nature of the samples analyzed, the different activated sludges employed and the inherent variability of the BOD5 assay. Incubation time and the relative degradability of the FMBOD calibrating standard were integral in predicting BOD5 values. Irrespective of the standard used for FM-BOD calibration, the relationship with BOD5 data was improved using a 6 h incubation (Fig. 4, Fig. S-3) compared with a 3 h incubation (Fig. S-4). Assays employing the OECD standard were superior predictors of BOD5 values for the majority of industrial effluent samples analyzed in this study. The 3 h GGA incubation was the poorest predictor of BOD5 values. This was in keeping with the readily biodegradable nature of the GGA standard, particularly over the first 3 h of incubation, relative to the OECD synthetic wastewater standard (Fig. 2). Calibration with GGA typically resulted in overestimating the BOD by a factor of almost 2, particularly for readily biodegradable, high range samples (Fig. S-4b). This effect is also evident with the 6 h GGA incubation, although to a lesser extent, as it is ameliorated by the increased incubation time (Fig. S-3). The suitability of

w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 8 1 e5 9 8 8

which standard to use in an industrial application will depend upon the nature of the majority of wastewaters that would normally be analyzed for BOD5. For example, if the BOD5 values of industrial effluents are known to be relatively low (i.e. ∼ <1500 mg BOD5 L
1) and unchanging, a 3 h incubation using the OECD solution as the calibrating standard would be sufficient (Fig. S-4a). If on the other hand, BOD5 values are relatively unknown, variable and/or high range, a 6 h incubation would be far more suitable, calibrated with the OECD standard for relatively biodegradable samples or with the GGA standard for relatively recalcitrant samples. For best results, application of the activated sludge FM-BOD assay in WWTPs requires some knowledge of the nature of the industrial effluent samples to be analyzed and ideally some optimization of the method to customize it to the given WWTP sludge. The BOD5 assay has a number of strengths and weaknesses. Most notably it boasts a high degree of substrate oxidation (e60.5%) and yet achieving this requires five days (APHA, 1999). The strength of the relationships outlined above indicate that the activated sludge FM-BOD assay is capable of oxidizing a wide range of organic substrates in complex matrices, much the same as the BOD5 assay does over a fiveday period, but over a considerably shorter time-frame. Comparing the extent of biooxidation of both methods confirms this. Mean FM-BOD substrate oxidation of all 35 industrial effluent samples compares very favorably to the amount of biooxidation observed in the BOD5 assay, for the 3 h (80  22%; ranging from 43 to 130%) and 6 h (96  23%; ranging from 62 to 167%) incubations. These values were calculated by comparing the amount of electrons that were biochemically transferred to ferricyanide during the FM-BOD incubation (by measuring ferrocyanide production) with the amount of electrons biochemically transferred to O2 in the BOD5 assay (by measuring O2 consumption) (Catterall et al., 2003; Morris et al., 2003). This high level of substrate oxidation is possible given that essentially the BOD5 and FM-BOD assays use the same diverse microbial community. However, the FM-BOD incubation time is greatly reduced, since the terminal electron acceptor is no longer rate limiting and much higher microbial concentrations can be employed. Maximizing the % oxidation of a wide diversity of real samples is integral to confident prediction of BOD5 values using any rapid BOD5 surrogate assay. It is this fundamental aspect of rapid BOD estimation that inherently limits BOD biosensors. While BOD biosensors are very rapid, typically the % substrate oxidation is <1% (Morris et al., 2001), therefore considerably overestimating the readily biodegradable fraction of industrial wastewaters. Previous FM-BOD studies (Catterall et al., 2003; Morris et al., 2003) have for the most part, avoided this problem, by integrating a consortium (4e5 species) of pure microorganism cultures from diverse heterotrophic groups as the biocomponent. Not unlike this study, these studies reported excellent relationships with BOD5 measurements for a large number of industrial effluents, although the relative extent of biochemical oxidation in this study was superior. To our knowledge, no other rapid BOD assay has achieved relative degradation efficiencies comparable with this study. The impetus for optimizing the FM-BOD assay with activated sludge as the biocatalyst was with a view toward its implementation in WWTPs, while still providing rapid results within

5987

a single working day. For practical purposes, the use of activated sludge as the biocomponent requires a fraction of the timeconsuming and expensive preparation of pure culture microorganisms, as it can be easily collected from the plant at any time. Perhaps more importantly, by using the same microbial assemblage in the FM-BOD assay as the WWTP activated sludge community, the FM-BOD assay provides a completely relevant and representative measurement of the BOD loading to the plant. Such inferences are impossible using pure culture biocatalysts or even judiciously selected consortia of pure cultures, as their metabolic capacities will be different to those of the activated sludge community in any given WWTP.

4.

Conclusions

Although the FM-BOD approach is not new, this is the first time that activated sludge has been successfully demonstrated to be a suitable biocatalyst. This has several important implications. The BOD of complex wastewaters can be analyzed in a single working day, with a similar extent of organic biodegradation compared to the standard BOD5 assay. The use of Michaelis-Menten equations for the calibration curve was also a novel development. Application of the FMBOD assay in the wastewater industry would assist operators of treatment plants by providing timely data on problems with wastewater or effluent quality, allowing corrective actions to be made much sooner. By using activated sludge from each individual treatment plant, the FM-BOD assay provides a highly representative and practical result. Furthermore, as the composition of the activated sludge community changes over time, so will the FM-BOD response. The FM-BOD assay requires substantially less use of dilutions as the measurement working range covers three orders of magnitude. These features make the fast FM-BOD assay highly suitable for application in the extensive and diverse wastewater industry.

Acknowledgements This research was supported by an Australian Research Council Linkage Grant (LP0882894) co-funded by Gold Coast Water. The authors also gratefully acknowledge technical assistance from Coombabah and Logan wastewater treatment plants.

Appendix A. The supplementary data associated with this article can be found in the on-line version at doi:10.1016/j.watres.2010.07. 042.

references

APHA, 1999. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, Washington, DC.

5988

w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 8 1 e5 9 8 8

Baird, R.B., Smith, R.-K., 2002. Third Century of Biochemical Oxygen Demand. Water Environment Federation, Alexandria VA, USA. Bura, R., Cheung, M., Liao, B., Finlayson, J., Lee, B.C., Droppo, I.G., Leppard, G.G., Liss, S.N., 1998. Composition of extracellular polymeric substances in the activated sludge floc matrix. Water Science & Technology 37 (4), 325e333. Catterall, K., Morris, K., Gladman, C., Zhao, H., Pasco, N., John, R., 2001. The use of microorganisms with broad range substrate utilisation for the ferricyanide-mediated rapid determination of biochemical oxygen demand. Talanta 55 (6), 1187e1194. Catterall, K., Robertson, D., Teasdale, P.R., Welsh, D.T., John, R., 2010. Evaluating use of ferricyanide-mediated respiration bioassays to quantify stimulatory and inhibitory effects on Escherichia coli populations. Talanta 80 (5), 1980e1985. Catterall, K., Zhao, H., Pasco, N., John, R., 2003. Development of a rapid ferricyanide-mediated assay for biochemical oxygen demand using a mixed microbial consortium. Analytical Chemistry 75, 2584e2590. D’Souza, S.F., 2001. Microbial biosensors. Biosensors and Bioelectronics 16 (6), 337e353. Dias, F.F., Bhat, J.V., 1964. Microbial ecology of activated sludge. I Dominant bacteria. Applied Microbiology 12, 412e417. Dias, F.F., Bhat, J.V., 1965. Microbial ecology of activated sludge II bacteriophages, Bdellovibrio, coliforms and other organisms. Applied Microbiology 13, 257e261. Emde, R., Swain, A., Schink, B., 1989. Anaerobic oxidation of glycerol by Escherichia coli in an amperometric poised-potential culture system. Applied Microbiology and Biotechnology 32 (2), 170e175. Higgins, J., Warnken, J., Teasdale, P.R., 2004. Comparison of water quality criteria in Australian recycling guidelines and effluent discharge licences. Australian Journal of Environmental Management 11, 227e236. Hiraishi, A., Masamune, K., Kitamura, H., 1989. Characterization of the bacterial population structure in an anaerobic-aerobic activated sludge system on the basis of respiratory quinone profiles. Applied and Environmental Microbiology 55, 897e901. Karube, I., Matsunaga, T., Suzuki, S., 1977. A new microbial electrode for BOD estimation. Applied Biochemistry and Biotechnology 2 (2), 97e104. Kumlanghan, A., Kanatharana, P., Asawatreratanakul, P., Mattiasson, B., Thavarungkul, P., 2008. Microbial BOD sensor for monitoring treatment of wastewater from a rubber latex industry. Enzyme and Microbial Technology 42 (6), 483e491. Kwok, N.-Y., Dong, S., Lo, W., Wong, K.-Y., 2005. An optical biosensor for multi-sample determination of biochemical oxygen demand (BOD). Sensors and Actuators B: Chemical 110 (2), 289e298. Lehninger, A.L., 1970. Biochemistry. Worth Publishers, Inc., New York. Liu, J., Bjo¨rnsson, L., Mattiasson, B., 2000. Immobilised activated sludge based biosensor for biochemical oxygen demand measurement. Biosensors and Bioelectronics 14 (12), 883e893. Liu, J., Mattiasson, B., 2002. Microbial BOD sensors for wastewater analysis. Water Research 36 (15), 3786e3802. Matsuoka, Y., Isoda, Y., 2005. Application of BOD biosensor to immobilize activated sludge. Environmental Conservation 34 (8), 585e593.

Morris, K., 2005. Optimisation of the biocatalytic component in a ferricyanide mediated approach to rapid biochemical oxygen demand analysis. PhD Thesis, Griffith University, Gold Coast, Australia. Morris, K., Catterall, K., Zhao, H., Pasco, N., John, R., 2001. Ferricyanide mediated biochemical oxygen demandedevelopment of a rapid biochemical oxygen demand assay. Analytica Chimica Acta 442 (1), 129e139. Morris, K., Zhao, H., John, R., 2003. In: Brebbia, C.A., Almorza, D., Sales, D. (Eds.), Water Pollution VII: Modelling, Measuring and Prediction. WIT Press, Boston, pp. 379e389. Morris, K., Zhao, H., John, R., 2005. Ferricyanide-mediated microbial reactions for environmental monitoring. Australian Journal of Chemistry 58 (4), 237e245. OECD-209, 1984. Activated Sludge, Respiration Inhibition Test. Organisation for Economic Cooperation and Development. Pang, H.-L., Kwok, N.-Y., Chan, P.-H., Yeung, C.-H., Lo, W., Wong, K.-Y., 2007. High-throughput determination of biochemical oxygen demand (BOD) by a microplate-based biosensor. Environmental Science & Technology 41 (11), 4038e4044. Pasco, N., Baronian, K., Jeffries, C., Hay, J., 2000. Biochemical mediator demand-a novel rapid alternative for measuring biochemical oxygen demand. Applied Microbiology and Biotechnology 53 (5), 613e618. Pasco, N., Baronian, K., Jeffries, C., Webber, J., Hay, J., 2004. MICREDOX e development of a ferricyanide-mediated rapid biochemical oxygen demand method using an immobilised Proteus vulgaris biocomponent. Biosensors and Bioelectronics 20 (3), 524e532. Patchett, R.A., Kelly, A.F., Kroll, R.G., 1989. Rapid detection of bacteria by an amperometric electrode system e a comparison of some redox mediators. Food Microbiology 6 (3), 159e169. Rodriguez-Mozaz, S., Alda, M.J.L., Marco, M.P., Barcelo´, D., 2005. Biosensors for environmental monitoring: a global perspective. Talanta 65 (2), 291e297. Sakai, Y., Abe, N., Takeuchi, S., Takahashi, F., 1995. BOD sensor using magnetic activated sludge. Journal of Fermentation and Bioengineering 80 (3), 300e303. Tortora, G.J., Funke, B.R., Case, C.L., 1997. Microbiology: An Introduction. Benjamin/Cummings, California. Velling, S., Tenno, T., 2009. Different calibration methods of a microbial BOD sensor for analysis of municipal wastewaters. Sensors and Actuators B: Chemical 141 (1), 233e238. Yoshida, N., Hoashi, J., Morita, T., McNiven, S.J., Nakamura, H., Karube, I., 2001. Improvement of a mediator-type biochemical oxygen demand sensor for on-site measurement. Journal of Biotechnology 88 (3), 269e275. Yoshida, N., Yano, K., Morita, T., McNiven, S.J., Nakamuraa, H., Karube, I., 2000. A mediator-type biosensor as a new approach to biochemical oxygen demand estimation. The Analyst 125 (12), 2280e2284. Zhao, H., Jiang, D., Zhang, S., Catterall, K., John, R., 2004. Development of a direct photoelectrochemical method for determination of chemical oxygen demand. Analytical Chemistry 76 (1), 155e160.