PII: S0043-1354(00)00585-6

Wat. Res. Vol. 35, No. 12, pp. 28332840, 2001 # 2001 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/01/$ - see front matter

EVALUATION OF NEW METHODS FOR THE MONITORING OF ALKALINITY, DISSOLVED HYDROGEN AND THE MICROBIAL COMMUNITY IN ANAEROBIC DIGESTION
LOVISA BJORNSSON*, MARIKA MURTO, TOR GUNNAR JANTSCH and BO MATTIASSON
Department of Biotechnology, Lund University, P.O. Box 124, SE-221 00 Lund, Sweden (First received 30 June 2000; accepted in revised form 5 December 2000) Abstract}New methods for spectrophotometric alkalinity measurement, dissolved hydrogen monitoring and for obtaining a ngerprint of the microbial community were evaluated as tools for process monitoring in anaerobic digestion. The anaerobic digestion process was operated at organic loading rates of 1.5, 3.0 and 4.5 g volatile solids l1 d1 and subjected to pulse loads of carbohydrate, lipid, protein and a mixed sludge substrate. The spectrophotometric alkalinity monitoring method showed good agreement with traditional titrimetric alkalinity monitoring and has the advantage of being easy to modify to on-line monitoring applications. The on-line monitoring of dissolved hydrogen gave valuable information about approaching process overload and can be a good complement to the conventional monitoring of volatile fatty acids. Changing process conditions were also reected in the microbial ngerprint that could be achieved by partitioning in two-phase systems. The investigated methods showed potential for application in increasing our understanding of the anaerobic digestion process as well as for being applicable for monitoring in the complex environment of full-scale anaerobic digestion processes. # 2001 Elsevier Science Ltd. All rights reserved Key words}anaerobic digestion, monitoring, dissolved hydrogen, Pd-MOS, alkalinity, CCD

INTRODUCTION

The organic waste produced by municipalities, industry and agriculture has a potential energy value, which may be exploited by anaerobic digestion and methane production. However, instability during both the start-up and operation of the anaerobic degradation process can be problematic. Anaerobic digestion involves a complex series of microbially catalysed degradation steps. The instability is mainly due to the interdependence of the dierent microbial groups involved in the degradation (Gujer and Zehnder, 1983). Process instability can be avoided by operating the anaerobic digestion process far below maximum capacity. Another, more economically viable, way of circumventing this problem is to improve the monitoring and control of the process in order to allow waste treatment at a higher rate. Ideal monitoring methods should be on-line, robust and give early indications of imbalance in the microbial status of the system. The measured parameter could either be an indirect status indicator, such as the concentration of a metabolite in the anaerobic

*Author to whom all correspondence should be addressed. Tel.: +46-46-222-8324; fax: +46-46-222-4713; e-mail: lovisa.bjornsson@biotek.lu.se

degradation, or a direct indication of the microbial status of the system, such as the number or the metabolic activity of the microorganisms involved. Some of the most commonly used indirect process indicators include the gas production rate, the gas composition, pH, the alkalinity and the concentration of volatile fatty acids (VFAs) (Ahring et al., 1995; Moletta et al., 1994). The monitoring of gasphase parameters has, in several earlier investigations, proved insucient in indicating process status and was not considered in this investigation (Frigon and Guiot, 1995; Pauss et al., 1990a). This work was focused on liquid-phase parameters, since the conditions in the liquid reect the microbial environment where the important reactions in anaerobic digestion occur. pH as a process indicator is strongly dependent on the buering capacity, or alkalinity, of the system (Ahring et al., 1995). The main buering species in an anaerobic digester are the VFAs and the bicarbonate. Total alkalinity (TA) measured by titration to a pH endpoint of 4.3, as suggested by APHA et al. (1985), includes both these species. This has been considered as an insensitive parameter for indicating process instability since an increase in VFA concentration will cause a decrease in bicarbonate concentration, resulting in a fairly constant TA-value (Hill and Jenkins, 1989; Jenkins

2833

2834

Lovisa Bjornsson et al.

et al., 1983; Quimby and Whittle, 1992). Instead, a number of methods based on titration have been developed in order to discriminate between bicarbonate alkalinity and VFA alkalinity (de Haas and Adam, 1995; Moosbrugger et al., 1993). The partial alkalinity (PA) as measured by titration to an endpoint pH of 5.75 has been suggested, which indicates changes in the bicarbonate concentration (Jenkins et al., 1991; Ripley et al., 1986). Measuring the PA can be one way of indirectly measuring the VFA accumulation. On-line methods have been developed for the monitoring of bicarbonate concentrations, based on acidication of the sample with subsequent determination of the carbon dioxide ow or pressure (Hawkes et al., 1993; Rozzi and DiPinto, 1994). In the present study, a new approach to PA and TA monitoring was investigated, employing the spectral properties of pH indicators (Jantsch and Mattiasson, submitted). On-line methods for measuring the concentration of total VFAs have been suggested by Powell and Archer (1989), using a titration technique, and by Rozzi et al. (1997) employing a biosensor with denitrifying bacteria. On-line monitoring of the individual VFAs has been performed by von Zumbusch et al. (1994), who measured acetic and propionic acids. The method suered, however, from problems associated with membrane fouling. An interesting alternative is to monitor the dissolved hydrogen concentration, since this is known to be closely related to the accumulation of VFAs (CordRuwisch et al., 1997; Schink, 1997). Several methods have been developed for on-line monitoring of dissolved hydrogen in anaerobic digesters, including the hydrogen/air fuel cell detector (Pauss et al., 1990b), membrane-covered electrodes for direct liquid-phase hydrogen measurements (Kuroda et al., 1991; Strong and Cord-Ruwisch, 1995) and a trace reduction gas analyser (Cord-Ruwisch et al., 1997). In the present study, a new approach to dissolved hydrogen monitoring was investigated including hydrogen transfer from the liquid through a Teon membrane, and gas-phase detection with a Pd metal oxide semiconductor (Pd-MOS) sensor (Bjornsson et al., in press). Direct process indicators are concerned with the number or status of dierent microbial groups in anaerobic digestion. As reviewed by Oude Elferink et al. (1998), these methods can include direct enumeration or microscopic studies of the microorganisms. The molecular-based methods include the use of membrane lipids (Hedrick et al., 1991), genetic probes (Raskin et al., 1994) or immuno-techniques (Srensen and Ahring, 1997) for characterisation of microbial communities. In the present study, a method aimed at obtaining a ngerprint of the microbial community in the process was investigated. By performing multi-step partitioning of the microorganisms in an aqueous polymerpolymer two-phase system, the microorganisms will be frac-

tionated according to size and surface properties and the obtained distribution pattern will give a ngerprint of the culture (Murto et al., submitted). The aim of this study was to evaluate the potential of the new methods as indicators of process changes in anaerobic digestion. It was considered important to evaluate the methods upon process changes in a complex environment since many full-scale applications today include the co-digestion of dierent kinds of waste. The strategy was to use a laboratory-scale mixed culture fed with a base load of a mixed sludge substrate from a large-scale digester. Pulse loads of dierent organic composition were then added to simulate the variations in load and composition that might occur in large-scale co-digestion processes.
MATERIALS AND METHODS

Laboratory-scale digester The laboratory-scale reactor had a working volume of 3 l and was maintained under mesophilic conditions (378C). A mixed culture from a full-scale biogas plant was used as inoculum. The contents of the reactor were continuously mixed with a mechanical stirrer (KeboLab, Spanga, Sweden) operating at 180 rpm. The substrate was kept in a stirred vessel at 48C. For feeding of the substrate, which contained particulate organic matter, a semi-continous feeding mode was used. A piston pump (Watson-Marlow Alitea AB, Stockholm, Sweden) fed 5 ml of the substrate to the reactor each time, and the loading rate was adjusted by changing the feeding interval. The euent was collected in a bottle, and gas from the reactor and the euent bottle was collected in gas-tight bags. Operating conditions and substrate The organic material was quantied as volatile solids (VS), and the organic loading rate (OLR) was calculated as gram VS per litre reactor per day (g VS l1 d1). The laboratory-scale process was run consecutively at three OLRs, 1.5, 3.0 and 4.5 g VS l1 d1. As substrate at these base loads, a mixed sludge from a full-scale anaerobic digester was used. This complex substrate was a mixture of sludge from a municipal wastewater treatment plant and a carbohydrate-rich, food-processing waste. The substrate had a VS concentration of 3%, of which on average 50% was starch and 20% fatty acids (4.2 g l1 lactic acid, 1.9 g l1 acetic acid and 0.30.4 g l1 of propionic-, butyric-, and valeric acid). At each base load level, the process was subjected to pulse loads of 3 g VS l1 using organic material of dierent chemical composition (Fig. 1). The sludge substrate was used also in pulse loads. d(+)glucose (KeboLab, Spanga, Sweden) was used for the carbohydrate pulses. Olive oil (LabKemi, Stockholm, Sweden) was used in the lipid pulses, and whey powder (90% proteins, Twin Laboratories Inc., New York, USA) was used for the protein pulses. Towards the end of the experimental period, the process was overloaded with a carbohydrate pulse of 9 g VS l1. Sampling Alkalinity, VFAs and pH were measured o-line. Samples were taken daily at the base load levels and every 1030 min during 24 h after pulse loads. Ammonia concentrations were analysed o-line in connection with the protein pulses. Hydrogen was continuously sampled on-line with readings at 7.5 min intervals. The samples for microbial ngerprinting were taken several times at base load levels to

Evaluation of new methods

2835

Fig. 1. The organic loading rate (OLR), the pulse loads and the response in volatile fatty acid (VFA) concentration during the experimental period. The mixed sludge substrate was used as the base load. The organic material in the pulse loads is indicated as follows; S=sludge substrate, C=carbohydrate, L=lipid, P=protein.

evaluate the reproducibility of the method, and 610 h after every pulse load.

Analytical methods Total solids (TS) and volatile solids (VS) were determined according to APHA Standard Methods (1985). The ammonia concentrations were measured with the Lange method LCK 303 (Dr. Bruno Lange GmbH, Dusseldorf, Germany). Samples for alkalinity measurements were centrifuged at 6000 rpm for 3 min and the supernatant was used for further analysis. The titrimetric alkalinity was evaluated as PA by titration to pH 5.75, and the TA by titration to pH 4.3 with standardised 0.1 M HCl using a TitraLabTM 80 titrator (Radiometer, Copenhagen, Denmark) and expressed as mg CaCO3 l1. For the measurements of PA and TA by photometry, the sample was added to a mixture of dilute HCl and a pH indicator; the absorbency was then measured (Jantsch and Mattiasson, submitted). The indicator was methyl red for PA measurements and bromphenol blue for TA measurements with colour change in the pH region 4.2 6.2 and 3.04.6, respectively. The initial sample to acid ratio or the acid concentration were adjusted to obtain a nal pH of 4.3 for the TA and 5.75 for the PA sample. Changes in the sample alkalinity will then inuence the nal pH of the mixture, which will be reected in the measured absorbance. PA and TA values expressed as mg CaCO3 l1 were calculated from calibration curves obtained with standard solutions containing known concentrations of NaHCO3 and NaCH3COO. Absorbance measurements were performed in 1-ml plastic cuvettes with a Pharmacia Biotech Ultrospec 1000 spectrophotometer. VFAs were analysed with HPLC, Varian Star 9000 (Varian, Walnut Creek, USA), with a Biorad column, 125 0115 (Hercules, USA). The column temperature was 658C, sulphuric acid (1 mM) was used as mobile phase and the liquid ow was 0.8 ml min1. Peak detection was achieved with UV absorption at 208 nm. The samples were acidied, centrifuged, stored at 208C and ltered before analysis. The dissolved hydrogen was sampled on-line by transfer from the liquid through a Teon membrane. The sample was injected to a semiconductor sensor, and the response given as a voltage output. No direct quantication of the dissolved hydrogen in the digesters was performed, but calibration of the method with dissolved hydrogen in water (Bjornsson et al., in press) gave detection limits for the method at 580 and 160 nM, before and after the sensitivity change on day 110.

The sample for microbial ngerprinting was pretreated with sonication to disintegrate ocs. The two-phase system used for separation contained 6% Dx T500 (Pharmacia AB, Uppsala, Sweden), 6% PEG 4000 (Merck-Schuchardt, Hohenbrunn, Germany), 5 mM Li2SO4 solution and 5 mM phosphate solution at pH 7.0, and was prepared as described by Murto et al. (submitted). The counter current distribution (CCD) apparatus consisted of a rotor with 60 cavities. The cell suspension was mixed with the bottom phase and added to the rst cavity while two-phase system was added in the other cavities. The bottom phase volume was 0.6 ml and the top phase volume was 0.8 ml. The volume of the cavity in the lower part of the rotor was 0.8 ml. The bottom phases and interfaces were thus kept stationary in the lower half of the rotor while the top phases were moved in the upper half at each transfer. A cycle involved shaking (30 s), settlement (10 min) and transfer. Fifty-eight transfers were carried out in a CCD-run. After a complete CCD-run, the separate CCD-fractions were diluted with 0.1 M glycinesodium hydroxide buer, pH 9.0, in order to dissolve the two-phase system. The distribution of cells in the dierent fractions was analysed by measuring the cell density at an absorbance of 620 nm. By plotting the absorbance as a function of the fraction number, a cell distribution diagram was obtained.

RESULTS

The values of the monitored parameters at the three levels of OLR prior to the pulse load sequences are shown in Table 1. The response in VFA concentration during the experimental period is shown in Fig. 1. At overload the VFA concentration increased to 3500 mg l1 (not shown). The response to the pulse loads at an OLR of 1.5 g VS l1 d1 is shown in Fig. 2. The addition of pulse loads as outlined in Fig. 1 is indicated by the vertical lines, and there is a 4-day interval between each pulse load. The spectrophotometric TA measurement initially deviated from the titration values due to an erroneous callibration curve. This was corrected at day 108. After this modication the titrimetric and spectrophotometric alkalinity measurements showed good agreement. The sensitivity of the dissolved hydrogen method was too low at the

2836

Lovisa Bjornsson et al.

Table 1. Mean values and standard deviations for the monitored parameters at dierent organic loading rate before each pulse load sequence. The only volatile fatty acid (VFA) detected during steady-state operation was acetic acid (b.d.=below detection limit) Organic loading rate (g VS l1 d1) 1.5 3.0 4.5 pH 7.2 0.1 7.0 0.1 7.0 0.1 Partial alkalinity (mg CaCO3/l1) 2050 40 1900 40 1910 20 Total alkalinity (mg CaCO3/l1) 2670 40 2410 40 2390 30 VFA Conc. (mg l1) 27 2 34 4 36 8 Dissolved hydrogen (mV) b.d. b.d. b.d.

Fig. 2. The response in dissolved hydrogen, alkalinity and pH to the pulse loads at a base load of 1.5 g VS l1 d1. TA=total alkalinity, PA=partial alkalinity, spectr=spectrophotometric method, titr=titrimetric method.

beginning of the experimental period, and at day 110 the sensitivity was increased by allowing longer time for the equilibration in the liquid-to-gas hydrogen transfer. The response to the pulse loads at the base load of 3.0 g VS l1 d1 is shown in Fig. 3. The vertical lines indicate the addition of pulse loads, and there is a 4day interval between each pulse load. Change in pH and alkalinity was similar to that at the lower OLR level, but the VFA accumulation was signicantly higher as shown in Fig. 1. The spectrophotometric method did not respond to alkalinity below 1500 mg CaCO3 l1 and at day 111, the acid concentration was lowered to accommodate PA measurements in a lower range. After modication it showed good agreement with the titrimetric method. Comparing VFA and dissolved hydrogen data showed that VFA accumulation was similar both for the substrate and carbohydrate pulses whereas the dissolved hydrogen response diered with a factor of 4.

On neither of the two rst baseline levels did any of the methods respond to the lipid pulses. The ammonia concentration in the reactor increased by 305 and 311 mg l1 (results not shown), respectively, after the two protein pulses. The increase in ammonia concentration was accompanied by increases in pH and alkalinity. The time required to reach the maximum ammonia concentration was 22 h at the OLR of 1.5 g VS l1 d1 and 15.5 h at the OLR of 3.0 g VS l1 d1. A comparison of VFA and hydrogen data for the protein pulses shows that the VFA response is signicantly higher at the higher OLR, whereas the hydrogen response is in the same range at both levels. At the OLR level of 4.5 g VS l1 d1 the process was subjected to a carbohydrate pulse of 3 g VS l1 followed, after 11 h, by another carbohydrate pulse of 9 g VS l1, to study the eects of process overload. The response of pH, alkalinity and dissolved hydrogen is shown in Fig. 4. The spectrophotometric

Evaluation of new methods

2837

Fig. 3. The response in dissolved hydrogen, alkalinity and pH to the pulse loads at a base load of 3.0 g VS l1 d1. TA=total alkalinity, PA=partial alkalinity, spectr=spectrophotometric method, titr=titrimetric method.

method was not in operation during this last part of the experiment. The change in the microbial ngerprint at dierent base loads, after the pulse loads at a base load of 3.0 g VS l1 d1, and at overload is shown in Fig. 5.

DISCUSSION

investigated methods to monitor the eects of the pulse loads was thus dependent both on the total OLR and the substrate used. There is still a lack of sensors for monitoring the degradation of compounds where the initial hydrolysis is rate limiting. Undegraded lipids can give clogging problems in the digesters, and have been shown to have a bactericidal eect on the anaerobic digestion (Rinzema et al., 1994). Alkalinity measurements TA measurements have earlier proved unreliable for process monitoring (Hill and Jenkins, 1989; Jenkins et al., 1983), which was conrmed in this study. It was also shown that the correlation between PA and the accumulation of VFAs is strongly dependent on the organic composition of the substrate. For sludge substrate and carbohydrate pulses the correlation between VFAs and PA was good. Following protein pulses, the VFA accumulation was in the same range, but the alkalinity and pH both increased due to the formation of ammonia during anaerobic protein degradation. The method of process monitoring by PA measurement is highly empirical and is only useful if the history of the reactor is known and the inuent is of a relatively stable composition. Within these limitations the spectrophotometric alkalinity measurement method was as good as the conventional titration method at detecting process disturbances. The new

The reason for using the sludge substrate for pulse loads was to study the eects of a mixed substrate to which the process was adapted. Carbohydrate was used to give pulses with an easily degradable substrate, where the last steps in the degradation, acetogenesis and methanogenesis, would be expected to be limiting. Lipid was used to study the eects of a material where the initial hydrolysis was the limiting step in the degradation, and proteins were used to study the eects of the degradation product ammonia. As expected, all pulse loads except the lipid pulses gave corresponding increases in the concentrations of VFA. At a higher OLR, the VFA accumulation was higher and faster due to higher metabolic activity of some of the microbial groups involved in the degradation. This illustrates the often observed limitations in the last steps of the anaerobic degradation process, the acetogenic degradation of propionate and butyrate and the methanogenesis from hydrogen and acetate. The ammonia produced by protein degradation increased the buering capacity in the reactor. The potential of the

2838

Lovisa Bjornsson et al.

Fig. 4. The response in dissolved hydrogen, alkalinity and pH to the pulse loads at a base load of 4.5 g VS l1 d1. TA=total alkalinity, PA=partial alkalinity.

higher OLR. This pattern of hydrogen accumulation gives additional information about the microbial limitations and degradation pattern that cannot be concluded from the VFA accumulation alone. A dissolved hydrogen concentration of more than 40 nM has been said to be crucial in the regulation of the ow of carbon during mineralisation of organic matter (Pauss et al., 1990b). Higher hydrogen concentrations will direct the electron ow from methane production to the production of electron sinks such as butyrate, propionate, lactate or ethanol (Mosey, 1983; Schink, 1997). Furthermore, degradation of propionate and butyrate by the obligate syntrophic acetogenic bacteria is hydrogen dependent and will be inhibited at elevated hydrogen concentrations (Harper and Pohland, 1986). Thus, hydrogen accumulation is a good indicator of imbalance between some of the most sensitive microbial groups in the anaerobic digestion process. One advantage of this method, compared to other on-line methods for dissolved hydrogen monitoring is that the semiconductor sensor responds specically to hydrogen. This means that after transfer from the liquid, no further treatment of the sample is necessary. The Pd-MOS sensor used can be poisoned by hydrogen sulphide (Hornsten et al., 1991), which is often present in anaerobic digesters. However, the stability of the sensor during the 3 months of on-line operation during this study indicated that the hydrogen sulphide could not penetrate the Teon membrane. The main problem of using the on-line method was the microbial growth on the liquid-togas hydrogen transfer membrane, which was circumvented by changing the membrane every 48 h, as described by Bjornsson et al. (2000). The microbial ngerprints The method of partitioning in aqueous two-phase system is highly sensitive for the separation and fractionation of biological particulates (cells, organelles) on the basis of surface properties (Walter et al., 1985). The method has successfully been used, e.g. in fractionation of rat bone marrow cells (Garcia-Perez et al., 1990). Another advantage is that the conditions in a two-phase system are mild and cells that has been subjected to partitioning in a CCD are viable and can, for example, be cultured (Walter and Johansson, 1994). Repeated measurements at the dierent OLR levels showed good reproducibility in the distribution pattern, but the absorbance in the dierent CCD-runs varied (data not shown). Under changing culture conditions, the CCD proles were found to change accordingly. A broader prole was observed for the overloaded culture (Fig. 5(c)) than when the culture had been run at a constant OLR of 1.5 g VS l1 d1 (Fig. 5(a)) and 3.0 g VS l1 d1 (Fig. 5(b)). A broader cell distribution reects heterogeneity in cell surface properties. Furthermore, dierent CCD-proles were observed when the

spectrophotometric method described in this paper avoids any titration procedure and the potential errors that are introduced by the use of a pH electrode. The method could be automated if a suitable sample pre-treatment method could be found. A continuous ow injection analysis (FIA) system that can minimize analysis errors by producing many measurements within a short period of time could be used. Determination of dissolved hydrogen The pulse of sludge substrate at the OLR of 3.0 g VS l1 d1 gave a fast accumulation of hydrogen whereas the accumulation of VFAs reached its maximum after 7 h. The metabolisation of pure glucose caused a successive build-up of hydrogen and VFAs, both reaching the maximum concentrations after 4 h. The glucose pulse at an OLR of 4.5 g VS l1 d1 gave a 4 times higher hydrogen accumulation, but a VFA accumulation in the same range, both reaching the maximum after 2.5 h. For the protein pulses the hydrogen accumulation was in the same range irrespective of the OLR level, but the conversion rate of the protein was higher at the

Evaluation of new methods

2839

Fig. 5. Fingerprint of the microorganisms. The organic material in the pulse loads is indicated as follows: S=sludge substrate; C=carbohydrate; L=lipid; P=protein. (a) Organic loading rate (OLR) 1.5 g VS l1 d1, (b) OLR 3.0 g VS l1 d1, (c) after overloading the process, (d) after the carbohydrate pulse, (e) after the lipid pulse, (f) after the protein pulse.

reactor had been subjected to dierent carbon sources (Fig. 5(d)(f )). When the reactor was subjected to lipid and protein perturbations a rougher prole was observed than when subjected to sludge substrate and glucose perturbations. The patterns observed are rather complicated since they reect both dierences in surface structure and, as a consequence of this, dierences in aggregation behaviour. During sample pretreatment, the complete disintegration of ocs could not be achieved and aggregates could be found in the rst CCD fractions giving high absorbance. Whether these ocs are the result of very strong binding, or ecient reassociation, remains to be elucidated.

understanding of the metabolism in the anaerobic digestion process. It was shown that CCD proles could be used to obtain a ngerprint of the microbial population in mixed cultures and that the partition patterns showed an interesting versatility upon changes in the process condition. This opens up a new eld of applications for the CCD method.
Acknowledgements}This work was supported by the Swedish National Energy Administration, the Nordic Energy Research Programme and Borregaard Industries Ltd. We thank Gisela Wendt-Jonsson for skilful laboratory assistance. The friendly assistance of the sta at Ellinge Wastewater Treatment Plant is gratefully acknowledged.

CONCLUSIONS REFERENCES

The methods evaluated in this investigation showed good potential for use as monitoring tools for anaerobic digestion processes. The spectrophotometric alkalinity method proved to be a method as valuable as the traditional titration method for process monitoring. The potential of this method lies in the ease with which it can be modied to on-line sampling and monitoring. The hydrogen monitoring method was used as an on-line measurement system giving valuable information as dissolved hydrogen accumulated in the process. This method can be used as a process monitoring tool as well as for increasing the basic

Ahring B., Sandberg M. and Angelidaki I. (1995) Volatile fatty acids as indicators of process imbalance in anaerobic digestors. Appl. Microbiol. Biotechnol. 43(3), 559565. APHA, AWWA, WPCF (1985) Standard Methods for the Examination of Water and Wastewater, 16th ed. American Public Health Organisation, Washington DC, USA. Bjornsson L., Hornsten E. G. and Mattiasson B. (in press) Utilization of a Pd-MOS sensor for on-line monitoring of dissolved hydrogen in anaerobic digestion. Biotechnol. Bioengng. Cord-Ruwisch R., Mercz T. I., Hoh C.-Y. and Strong G. E. (1997) Dissolved hydrogen concentration as an on-line control parameter for the automated operation and optimization of anaerobic digesters. Biotechnol. Bioengng. 56(6), 626634.

2840

Lovisa Bjornsson et al. in an aqueous two-phase system. bo.mattiasson@biotek.lu.se. Oude Elferink S. J. W. H., van Lis R., Heilig H. G. H. J., Akkermans A. D. L. and Stams A. J. M. (1998) Detection and quantication of microorganisms in anaerobic bioreactors. Biodegradation 9, 169177. Pauss A., Andre G., Perrier M. and Guiot S. R. (1990a) Liquid-to-gas mass transfer in anaerobic processes: inevitable transfer limitations of methane and hydrogen in the biomethanation process. Appl. Environ. Microbiol. 56(6), 16361644. Pauss A., Samson R. and Guiot S. (1990b) Continuous measurement of dissolved H2 in an anaerobic reactor using a new hydrogen/air fuel cell detector. Biotechnol. Bioengng. 35, 492501. Powell G. E. and Archer D. B. (1989) On-line titration method for monitoring buer capacity and total volatile fatty acid levels in anaerobic digesters. Biotechnol. Bioengng. 33, 570577. Quimby J. L. and Whittle T. E. (1992) Overview and case histories of the intermediate/partial alkalinity monitoring technique for anaerobic process control. Hazardous Ind. Wastes 24, 602609. Raskin L., Poulsen L. K., Noguera D. R., Rittman B. E. and Stahl D. A. (1994) Quantication of methanogenic groups in anaerobic biological reactors by oligonucleotide probe hybridization. Appli. Environ. Microbiol. 60, 12411248. Rinzema A., Boone M., van Knippenberg K. and Lettinga G. (1994) Bactericidal eect of long chain fatty acids in anaerobic digestion. Water Environ. Res. 66(1), 4049. Ripley L. E., Boyle W. C. and Converse J. C. (1986) Improved alkalimetric monitoring for anaerobic digestion of high-strength wastes. J. WPCF 58(5), 406411. Rozzi A. and DiPinto A. C. (1994) Start-up and automation of anaerobic digesters with automatic bicarbonate control. Bioresource Technol. 48, 215219. Rozzi A., Massone A. and Antonelli M. (1997) A VFA measuring biosensor based on nitrate reduction. Water Sci. Technol. 36(67), 183189. Schink B. (1997) Energetics of syntrophic cooperation in methanogenic degradation. Microbiol. Mole. Biol. Rev. 61(2), 262280. Srensen A. H. and Ahring B. K. (1997) An improved enzyme-linked immunosorbent assay for whole-cell determination of methanogens in samples from anaerobic reactors. Appl. Environ. Microbiol. 63(5), 20012006. Strong G. E. and Cord-Ruwisch R. (1995) An in situ dissolved-hydrogen probe for monitoring anaerobic digesters under overload conditions. Biotechnol. Bioengng. 45(1), 6368. von Zumbusch P., Meyer-Jens T., Brunner G. and Markl H. (1994) On-line monitoring of organic substances with high-pressure liquid chromatography (HPLC) during the anaerobic fermentation of waste-water. Appl. Microbiol. Biotechnol. 42, 140146. Walter H., Brooks D. E. and Fisher D. (1985) Partitioning in Aqueous Two-phase Systems: Theory, Methods, Uses and Applications to Biotechnology. Academic Press, Orlando, Florida, USA. Walter H. and Johansson G. (1994) Methods in Enzymology}Aqueous Two-phase Systems, vol. 228. Academic Press, San Diego, California, USA.

de Haas D. W. and Adam N. (1995) Use of a simple titration procedure to determine H2CO3 alkalinity and volatile fatty acids for process control in waste-water treatment. Water SA 21(4), 307318. Frigon J. C. and Guiot S. R. (1995) Impact of liquid-to-gas hydrogen mass transfer on substrate conversion eciency of an upow anaerobic sludge bed and lter reactor. Enzyme Microbial Technol. 17(12), 10801086. Garcia-Perez A. -I., Sancho P. and Luque J. (1990) Settlingtime dependence of rat bone marrow cell partition and counter-current distribution in charge-sensitive aqueous two-phase systems. J. Chromatogr. 504, 7988. Gujer W. and Zehnder A. J. B. (1983) Conversion processes in anaerobic digestion. Water Sci. Technol. 15, 127167. Harper S. R. and Pohland F. G. (1986) Recent developments in hydrogen management during anaerobic biological wastewater treatment. Biotechnol. Bioengng. 28, 585602. Hawkes F. R., Guwy A. J., Rozzi A. G. and Hawkes D. L. (1993) A new instrument for on-line measurement of bicarbonate alkalinity. Water Res. 27(1), 167170. Hedrick D. B., Richards B., Jewell W., Guckert J. B. and White D. C. (1991) Disturbance, starvation, and overfeeding stresses detected by microbial lipid biomarkers in high-solids high-yield methanogenic reactors. J. Ind. Microbiol. 8, 9198. Hill D. T. and Jenkins S. R. (1989) Measuring alkalinity accurately in aqueous systems containing high organic acid concentrations. Trans. ASAE 32(6), 21752178. Hornsten E. G., Lundstrom I., Nordberg A. and Mathisen B. (1991) The use of palladium metal oxide semiconductor structures in quantitative studies of H2 and H2S in processes related to biogas production. Bioprocess Engng. 6, 235240. Jantsch T. G. and Mattiasson B. (submitted) A simple spectrophotometric method for monitoring partial and total alkalinity in anaerobic processes based on pHindicators. bo.mattiasson@biotek.lu.se. Jenkins S. R., Morgan J. M. and Sawyer C. L. (1983) Measuring anaerobic sludge digestion and growth by a simple alkalimetric titration. J. WPCF 55(5), 448453. Jenkins S. R., Morgan J. M. and Zhang X. (1991) Measuring the usable carbonate alkalinity of operating anaerobic digesters. Res. J. WPCF 63(1), 2834. Kuroda K., Silveira R. G., Nishio N., Sunahara H. and Nagai S. (1991) Measurement of dissolved hydrogen in an anaerobic digestion process by a membrane-covered electrode. J. Fermentation Bioengng. 71(6), 418423. Moletta R., Escoer Y., Ehlinger F., Coudert J.-P. and Leyris J.-P. (1994) On-line automatic control system for monitoring an anaerobic uidized-bed reactor: response to organic overload. Water Sci. Technol. 30(12), 1120. Moosbrugger R. E., Wentzel M. C., Ekame G. A. and Marais G. R. (1993) A 5 pH point titration method for determining the carbonate and SCFA weak acid/bases in anaerobic systems. Water Sci. Technol. 28(2), 237245. Mosey F. E. (1983) Mathematical modelling of the anaerobic digestion process: regulatory mechanisms for the formation of short-chain volatile acids from glucose. Water Sci. Technol. 15, 209232. Murto M., Dissing U. and Mattiasson B. (submitted) Characterisation of anaerobic mixed cultures by partition