Process Biochemistry, Vol. 31, No. 5, 477-483, 1996 pp. Copyright 0 1996 Elsevier Science Ltd

Printed in Great Britain. All rights reserved 0032-9592196 $15.00 +O.OO

0032-9592(95)00099-2 :LSEVIER

Influence of Ammonia Concentration on Thermophilic Anaerobic Digestion of Cattle Manure in Upflow Anaerobic Sludge Blanket (UASB) Reactors

Rafael Borja,a * Enrique Sgnchez’ & Peter Weiland” ’ Institute de la Grasa (C.S.I.C.). Avda. Padre Garcia Tejero 4, E-41012 Sevilla, Spain ‘Departamento de Estudios sobre Contaminaci6n Ambiental (DECA-CNIC), Centro National de Investigaciones Cientificas (CNIC), PO Box 6990, La Habana, Cuba Institute of Technology, Federal Research Center of Agriculture (FAL), Bundesallee 50, D-38116 Braunschweig,

Germany

{Received 20 September 1995; accepted 19 November 1995)

4mmonia concentrations of 5 g Nllitre or more inhibited thermophilic iznaerobic digestion of cattle manure in upflow anaerobic sludge blanket (UASB) reactors. A stable digestion of cattle manure could be maintained with ammonia concentrations up to 7g Nllitre after 6 months of operation. Howeve< !he methane yield was reduced and the concentration of volatile fatty acids increased from 1 to 3 gllitre as acetic acid, compared to controls with an ammonia concentration of 3 g Nllitre. The temporary strong inhibition following a one-step increase in ammonia concentration was reduced by applying a gradual increase. The specific methanogenic activity of ammonia-inhibited reactors (7g Nllitre) with acetate or hydrogen as substrate was reduced by 72 and 5670, respectively. Tests of ammonia toxicity on the acetate- and hydrogen- utilizing populations showed a higher sensitivity of the aceticlastic compared to the hydrogenotrophic methanogens; the specific growth rate for the aceticlastic Ynethanogens was halved at ammonia concentrations of 4 g Nllitre, compared (0 7.5 g Nilitre for the hydrogenotrophic methanogens.

INTRODUCTION

Inhibition during anaerobic digestion of cattle manure is often caused by high ammonia con- centration. In addition to ammonia (NH3 + NH:) cattle waste contains compounds that readily release ammonia when degraded, e.g. urea and proteins. For cattle manure

*To whom correspondence should be addressed.

especially, the total ammonia concentration is often higher than 4 g N/litre.’

Many investigations have dealt with the ammonia inhibition level, but results are con- flicting and have been obtained under different conditions, such as pH, temperature and inocu- lum used. McCarty* reported ammonia inhibition to occur at concentrations from 15 to 3-O g N/litre at pH levels above 7.4, whereas inhibition occurred for all concentrations higher

477

478 R. Borja et al.

than 3 g N/litre at all pH levels tested. Likewise Koster and Lettinga” reported ammonia inhibi- tion to occur at 1.7 g N/litre at pH 7.5.

Much higher inhibitory levels have however been reported by other authors. Van Velsen4 showed in batch experiments, with inoculum adapted to high concentrations of ammonia, that methanogenesis occurred after a lag phase at ammonia concentrations as high as 5 g N/ litre.

Only a few investigations have dealt with ammonia inhibition at thermophilic tempera- tures. Zeeman et aL5 reported an initial inhibition at 1.7 g N/litre at 50°C. Hashimoto6 found ammonia inhibition at about 2.5 g N/litre for both mesophilic and thermophilic reactors when the reactors were not previously acclima- tized to ammonia. However, the corresponding value was 4 g N/litre for thermophilic reactors previously acclimatized to ammonia concentra- tions between 1.4 and 3.3 g N/litre. In their experiments the effluent pH was approximately 7.2.

Free ammonia (NH,) has been suggested as the active component causing ammonia inhibi- tion. A level of 80 mg N/litre of free ammonia has been proposed as the minimum inhibitory leve1.3Y7 McCarty and McKinney,8 and Braun et al.” found 150 mg N/litre to be the inhibitory free ammonia concentration. As the free ammonia fraction increases with temperature and pH, the ammonia level tolerated at high pH and thermophilic temperatures would be expected to be low. Biogas reactors operating with cattle waste often have a high pH (about 8) and, especially at thermophilic temperatures, the free ammonia concentration will be up to 10 times higher than the free ammonia concentra- tions reported as inhibitory.

The effects of addition of different ammonia concentrations and the possibility of adaptation to ammonia during anaerobic thermophilic digestion of cattle manure in UASB reactors were examined. The specific methanogenic activity (SMA) of an uninhibited and ammonia- inhibited reactor are reported. Finally, the effect of various ammonia concentrations on thermophilic aceticlastic and hydrogenotrophic methanogens in batch experiments were studied.

Table 1. Data on cattle manure used

Parameter

Total solids, TS (%)

Experiment I Experiment II

6.9 7.0

Volatile solids, vs (%)

4.8 5.0

Total nitrogen (g N/litre)

3.2 4.1

Ammonia nitrogen (g N/litre)

2.0 3.0

PH Volatile fatty acids,

VFA (glitre)

7.6 7.8 7.9 8.2

Values are averages of five determinations; the differ- ences between the observed values were less than 2% in all cases.

MATERIALS AND METHODS

Wastewater Cattle manure was used as substrate and was provided in one batch for each of two experi- ments. The ammonia concentration of the first batch was 2 g N/litre. The batch used for the second experiment had a higher ammonia con- tent, 3 g N/litre. Data from the two batches are given in Table 1.

Continuously fed reactor experiments: equipment The experiments were performed in six 4-litre UASB reactors with a working volume of 3 litres. Each reactor was equipped with a gas collector, gas-biomass-liquid separator and influent liquid distributor. A detailed descrip- tion of the equipment used is given elsewhere.” The reactors were fed continuously with a peri- staltic pump; the effluent left the reactor through a hydraulic seal with a 35 cm high liquid column to prevent the entrance of air into the reactor and the escape of biogas. Bio- gas produced from the reactor was collected by positive displacement of acidified water (pH 2-3) into 5-litre gasometers. The reactors were placed in a temperature controlled room at 55°C. The hydraulic retention time (HRT) was 15 days.

Inoculum The reactors were inoculated with 750 ml of well digested sludge [180 days digestion time from a laboratory scale thermophilic continuous

Ammonia inhibition of cattle manure digestion 479

stirred tank reactor (CSTR) that processes diluted cattle manure with an ammonia content $)f 0.5 g N/litre] with a total solids concentration f)f 4.9% and volatile solids concentration of G&3% on a dry weight basis. Distilled water was ,idded to the reactors until their working ‘Jolumes were completed.

Experimental design Two series of reactor experiments were carried .,ut. In the first experiment the effect of adding Gimmonia to a level of 5 and 7 g N/litre was compared to the performance of control reac- tors with an ammonia concentration of 2 g N/litre. In the second experiment the effect of 7 g N/litre was compared to the effect of a gradually increasing concentration of ammonia. The N concentration was increased at intervals of 30 days (corresponding to 2 HRT) and the levels were 3, 4, 5 and 6 g N/litre. The results obtained in this experiment were compared with those from control reactors with 3 g N/litre ammonia. In both experiments the extra ammonia was added to the feed as NH4Cl. Duplicate reactors were operated for each con- centration tested. As the variation between duplicate reactors was always small (less than 5%) mean values are reported.

Specific methanogenic activity (SMA) test The SMA of the control reactors and the reac- tors receiving 7 g N/litre ammonia were compared. The tests were performed in 60 ml serum vials containing 20 ml BA medium” adjusted to pH 7.9 (corresponding to the pH of the reactors) with NaOH and a gas phase of NJ CO2 (90: 10) in order to keep the pH at 7.9-8.0 during the experimental period. Acetate (30 mM) or 200 kPa of HJCO* (80: 20) were applied as substrates and the methane produced was compared to vials without substrate added. The vials were inoculated anaerobically with 25% (v/v) reactor content of digested sludge (described earlier) from a laboratory CSTR reactor treating diluted cattle manure with an ammonia content of 0.5 g N/litre, and they were incubated in a shaking water bath at 55°C. The SMA was estimated as the initial methane pro- duction rate per gram biomass (volatile solids, VS). The mean activity found in control vials (without substrate addition) was subtracted from activities found in the experimental vials.

Effects of ammonia on methanogenic populations The effect of different concentrations of ammonia on aceticlastic and hydrogenotrophic methanogenic populations was tested in batch experiments, using 5% (v/v) digested manure as inoculum in BA medium (content of ammonia in BA medium was 0.25 g N/litre). The pH in these experiments was adjusted to 7.2, con- sidered to be the optimum pH for these bacterial groups. As substrate, 30 mM acetate or 200 kPa H2/C02 (80 : 20) was applied. For the acetate series, the range of concentrations tested was 0.3-13 g N/litre, and 0.3-20 g N/litre for the HJCO* series. The specific growth rates of the aceticlastic or hydrogenotrophic popula- tions were estimated by a semi-logarithmic plot of methane production versus time. Each experiment was run in triplicate and all experi- ments were repeated.

Analytical methods Total, volatile solids and pH were determined using standard methods.‘* Methane and CO2 were determined by gas chromatography with a stainless-steel column (200 x O-3 cm) packed with active carbon (30-60 mesh) using thermal- conductivity detection. Volatile fatty acids (VFA) were determined by gas chromatography using a 2 m x 4 mm glass column packed with Supelcopor (loo-120 mesh) coated with 10% Fluorad FC 431. The temperature of the col- umn, the injection port and the flame-ionization detector were 130, 220 and 24O”C, respectively. Nitrogen saturated with formic acid was used as the carrier gas at a flow rate of 50 ml/min.

Total ammonia content was determined by the Kjeldahl method. The free ammonia con- centration was calculated from the equilibrium relationship:

[NH,]=[T-NH,]/(I +[H+]lk,),

where [NH,] and [T -NH31 are the free and the total ammonia concentrations, respectively, and k, the dissociation constant, with the value of 38.3 x lo- ‘() at 55°C; during calculations, the appropriate pH values were used.

RESULTS

Addition of ammonia to a total concentration of 5 or 7 g N/litre to the feed of reactors fed

480 R. Borja et al.

Methane yield (litre methane/g VS added)

o’25----------------------1 0.2

“ia, I cl-F_ -:i -1

-- 1 ~_ _~~. m 0 10 20 Time3Pdays) 40 50 60

- CONTROL ++ 5 g Nllitre 7 g N/he ’

Fig. 1. Methane yield (litres methane/g VS added in a 5-day average) for the continuous reactor Experiment I. At day 8, ammonia concentration was changed from 2 g N/litre to 5 and 7 g N/litre.

with cattle manure containing 2 g N/litre (at day 8) resulted in a decrease in the methane yield after 3 weeks (Fig. 1). The methane yield decreased from 0.2 to 0.05 litres CHdg VS for both reactors after 48 days operation, corre- sponding to 25% of that of the control reactors receiving no extra ammonia. The decrease was faster for the reactors with 7 g N/litre ammonia (Fig. 1). The VFA concentration for both reac- tors receiving additional ammonia increased as the methane yield decreased (Fig. 2).

In the second experiment the basic level of ammonia was 3 g N/litre. The methane yield and the VFA concentration were the same as for the control reactors in the previous experi- ment, showing that 3 g N/litre ammonia had no apparent effect on the biogas process compared to manure with 2 g N/litre ammonia (Figs 3 and

,6yFA (g/litre)

I 14’

0 IO 20

Time (day:: 40 50

” CONTROL L 5 g N/We L 7 g N/lilre j I

Fig. 2. Volatile fatty acids (VFA) concentration (cal- culated as acetic acid) for the continuous reactor Experiment I. At day 8, ammonia concentration was changed from 2 g N/litre to 5 and 7 g N/litre.

0.25 Methane yield (litre methane/g VS added)

,I :

0.2 Viii’~ 0 5 r

,I-

‘I ,. I 0.15 r\

01 __L _~1_ _.~~ 1 / .A 0 20 40

T%e (da;) 100 120 140

* CONTROL INCREASING AMMONIA ‘- 7 g Nllitre

Fig. 3. Methane yield (1 methane/g VS added in a 5-day average) for the continuous reactor Experiment II. Ammonia was introduced at day 8 at a concentration of 7 g N/litre or stepwise from 4 (at day 8) to 5 (at day 38) and 6 g N/litre (at day 68) in the reactors receiving increasing ammonia concentrations.

4). As found in the first experiment, addition of 7 g N/litre resulted in a decreased methane yield, to 0.06 litres CH$g VS. The VFA concen- tration increased from about 1 to above 5 g/litre as acetic acid. However, after 2 HRT (30 days) the methane yield increased again to ca O-1 litres CHdg VS and the VFA level decreased to 3 g/litre as acetic acid (Figs 3 and 4). In the reactors receiving substrate with a gradually increasing concentration of ammonia, 4 g N/ litre did not result in any changes in methane yield and VFA concentration compared to the controls. Addition of 5 g N/litre, however, resulted in an increase in the VFA concentra- tion, followed by a decrease in methane yield.

VFA (g/litre) 61- ~ ~.

()_- .~ A

0 20 40 T%e (da::) 100 120 140

~’ CONTROL “ INCREASING AMMONIA - 7 g Nllitre ’ _1

Fig. 4. Volatile fatty acids (VFA) concentration (cal- culated as acetic acid) for the continuous reactor Experiment 11. Ammonia was introduced at day 8 at a concentration of 7 g N/litre or stepwise from 4 (at day 8) to 5 (at day 38) and 6 g N/litre (at day 68) in the reactors receiving increasing ammonia concentrations.

Ammonia inhibition of cattle manure digestion 481

After 2 HRT the methane yield partially recovered although the VFA level still increased.

When 6 g N/litre was introduced, serious pro- cess failure occured and the methane yield tfropped to 0.1 litres CHJg VS, i.e. the same level as the reactors receiving 7 g N/litre from clay 8. At the end of the experiment (180 days) reactors with both strategies of ammonia increase (instant and gradual) stabilized at a methane yield of O-15 litres CHJg VS and a ‘JFA concentration of 3 g/litre (not shown in l;igs 3 and 4).

In both experiments the basic pH level in the reactors ranged between 7.90 and 7.95. Accu- :nulation of VFA in the inhibited reactors resulted in a lowering of pH to 7.5.

SMA test ‘The activity of the aceticlastic and the hydro- ,::enotrophic methanogenic populations was ignificantly lower in the reactors receiving 7 g ‘\J/litre ammonia. The decrease in the activity jvas higher (72%) for the aceticlastic population I han for the hydrogenotrophic methanogens ! 56%) (Table 2).

.bnmonia toxicity experiment ‘The maximum growth rates were 0.65 day-’ .md 0.12 hh’ for the aceticlastic and hydro- :;enotrophic methanogens, respectively. The nhibitory effect of ammonia was in general

*,tronger for the aceticlastic than for the hydro- ,;enotrophic methanogens with initial inhibition ‘rccuring at an ammonia concentration of 1.8 g V/litre for the aceticlastic and 3.3 g N/litre for

rable 2. Specific methanogenic activity (SMA) of reactors *standard deviation) under different states of ammonia

nhibition

?eactors” Substrate Reduction h (%)

Zontrol Acetate 25.3 f 2.0 72

Yrnmonia Acetate 7.2 & 0.4 30ntrol HJCOz 24.4 t_ 1.2

56 9mmonia WCOz 11.1 kO.8

‘The control reactor received 3 g N/litre ammonia intro- &ced with the cattle waste while the ammonia reactor received ammonia to a total level of 7 g N/litre. ‘Reduction of the SMA of the ammonia reactor com- pared to the uninhibited reactor.

, 2. rlative growth rate (%)

i 80 ;

1

I I: I

60

40 - q. ‘I

I: 20

0’ 0 2 4

Ammonia (g N/litre)

u ACETATE -” HYDROGEN

Fig. 5. Reduction in the specific growth rate as a func- tion of added ammonia. Bars are the standard deviations of the means.

the hydrogenotrophic methanogens (Fig. 5). Growth rates were reduced to 50% of the unin- hibited value at 4 and 7.5 g N/litre (280 and 520 mg N/litre free ammonia) for the acetate- utilizing and hydrogenotrophic methanogenic bacteria, respectively.

DISCUSSION

Inhibition of the biogas process was observed when the ammonia concentration was increased to 5 g N/litre or more in the continuously fed biogas reactors. The methane yield decreased to 25%, with both 5 and 7 g N/litre added, com- pared to the controls with 2 g N/litre ammonia. When ammonia was introduced gradually, the process was unaffected up to 4 g N/litre and only slightly affected at 5 g N/litre, with signs of recovery after 1 HRT. At a concentration of 6 g N/litre, process performance was seriously affected and reached the same reduced level as the reactors fed with 7 g N/litre from the start of the experiments. After prolonged exposure to ammonia the reactors with ammonia concentra- tion of 6 or 7 g N/litre stabilized at a level of 0.15 litres CH4/g VS and 3 g/litre VFA (Figs 3 and 4).

482 R. Borja et al.

The experiments clearly demonstrate that it will be possible to obtain a stable digestion of manure with ammonia concentrations exceeding 5 g N/litre after an initial adaptation period. However, the methane yield will be lower (25% lower than for uninhibited reactors) and the VFA level will be higher than in reactors with a lower ammonia load.

Growth of methanogenic bacteria was inhibited by ammonia levels above 1.8 g N/litre (Fig. 5), while a concentration of 5 g N/litre was needed to affect the performance of the contin- uously fed biogas reactors. In a continuously fed reactor, inhibition is only detected when the reduction of the growth rates of the active bio- mass approaches the dilution rate used. In contrast a reduced growth rate will directly affect the outcome of a batch experiment.

The concentrations of free ammonia cal- culated for the reactor experiments are high. At 3 g N/litre, corresponding to the controls of the second experiment, the calculated free ammonia concentration (pH 7.9) was 590 mg/ litre. At 5 g N/litre ammonia, where the first signs of inhibition occurred, the calculated free ammonia concentration was 995 mg N/litre (pH 7.9). However, the actual pH of the reactors dropped to 7.6, due to the accumulation of VFA, resulting in a free ammonia concentration of 685 mg N/litre. This concentration of free ammonia resulted in a reduction in the growth rates of the aceticlastic methanogens to 18%, in the experiments with ammonia toxicity (Fig. 5). Growth rates at this level are still sufficient to retain the active biomass within the reactor at the HRT used in the reactor experiments. How- ever, ammonia concentrations higher than 4 g N/litre resulted in growth rates of the acetic- lastic methanogens close to the HRT, resulting in a decreased methane yield and an increased VFA concentration in the reactors.

Process instability due to ammonia resulted in VFA accumulation, which again led to a lowering of the pH and thereby decreased the concentration of free ammonia in the reactor. This decrease in free ammonia could explain the observed ability of the process to stabilize even with high ammonia concentration and with a lower but stable methane yield.

The SMA of the acetate-utilizing methano- gens of the ammonia-inhibited reactor decreased more than the hydrogenotrophic populations. Both the SMA test and the

ammonia toxicity experiment showed that it is the aceticlastic methanogens that are primarily affected by ammonia. This result is in accord- ance with other reports for mesophilic methanogens.‘3-‘5

Inhibition of the aceticlastic populations showed a sigmoidal pattern. The same pattern of inhibition of the aceticlastic populations was observed by Poggi-Varaldo et ~l.,‘~ who found that the bacterial growth rate and the specific acetate-uptake rate were affected by the free ammonia concentration in a three-stage pattern: initial inhibition, plateau and final inhibition. This inhibition pattern could indicate that two inhibition mechanisms are involved, acting at different concentration levels. The hydrogeno- trophic populations however, exhibited a more linear pattern of inhibition.

ACKNOWLEDGEMENTS

The authors would like to acknowledge the support of the Alexander Von Humboldt Foun- dation and Program of Scientific Cooperation with Iberomerica (Spanish Foreign Ministry) to develop this work.

REFERENCES

1.

2.

3.

4.

5.

6.

7.

8.

Angelidaki, I. & Ahring, B. K., Ammonia inhibition during anaerobic thermophilic degradation of animal waste. In Proceedings of the International Symposium on Environmental Biotechnology, ed. H. Verachtert & W. Verstraete. Royal Flemish Society of Engineers, Ostend, Belgium, 1991, pp. 389-92. McCarty, P. L., Anaerobic waste treatment fundamen- tals III. Public Works, 95 (1964) 91-4. Koster, I. W. & Lettinga, G., The influence of ammonia-nitrogen on the specific activity of pelletized methanogenic sludge. Agric. Wastes, 9 (1984) 205-16. Van Velsen, A. F. M., Adaptation of methanogenic sludge to high ammonia-nitrogen concentrations. Water Res., 13 (1979) 995-9. Zeeman, G., Wiegant, W. M., Koster-Treffers, M. E. & Lettinga, G., The influence of the total ammonia concentration on thermophilic digestion of cow manure. Agric. Wastes, 14 (1985) 19-35. Hashimoto, G., Ammonia inhibition of methanogene- sis from cattle wastes. Agn’c. Wastes, 17 (1986) 241-61. De Baere, L. A., Devocht, M., Assche, P. Van & Verstraete, W., Influence of high NaCl and NH4Cl salt levels on methanogenic associations. Water Res., 18 (1984) 543-8. McCarty, P. L. & McKinney, R. E., Salt toxicity in anaerobic digestion. J. Water Control Fed., 33 (1961) 399-415.

9. Braun, R., Huber, P. & Meyrath, J., Ammonia toxicity

Ammonia inhibition of cattle manure digestion 483

in liquid piggery manure digestion. Biotechnol. Lett., 3 (1981) 159-64.

0. Borja, R. & Banks, C. J., Kinetics of an upflow anae- robic sludge blanket reactor treating ice-cream wastewater. Environm. Technol., 15 (1994) 219-32.

1. Angelidaki, I., Petersen, S. P. & Ahring, B. K., Effects of lipids on thermophilic anaerobic digestion and reduction of lipid inhibition upon addition of bento- nite. Appl. Microbial. Biotechnol., 33 (1990) 469-72.

2. American Public Health Association, Standard Methods for the Examination of Waste and Wastewater, 16th edn. APHA, Washington, DC, USA, 1985.

3. Robbins, J. E., Gerhard, S. A. & Kappel, T. J., Effects

of ammonia in anaerobic digestion and an example of digestor performance from cattle manure protein mix- tures. Biol. Wastes, 27 (1989) l-14.

14. Bhattacharya, S. K. & Parking, G. F., The effect of ammonia on methane fermentation processes. .I. Water Polk Control Fed., 61 (1989) 55-9.

15. Sprott, G. D. & Patek G. B., Ammonia toxicity in pure cultures of methanogenic bacteria. System. Appl. Microbial., 7 (1986) 358-63.

16. Poggi-Varaldo, H. M., Tingley, J. & Oleszkiewicz, J. A., Inhibition of growth and acetate uptake by ammonia in batch anaerobic digestion. .I. Chem. Tech- nol. Biotechnol., 52 (1991) 135-43.