CICEET Progress Report for the period 3/15/05 Through 9/15/05

Project Title: Autotrophic Biological Denitrification with Hydrogen or Thiosulfate for Complete Removal of Nitrogen from a Septic System Wastewater
Principal Investigator(s): Sukalyan Sengupta, Sarina Ergas
Project Start Date: 1 September 2003

Figures

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Figure 21

Tables

Table 1

Table 2

Table 3

Table 4

Tasks to meet objectives
1. Sulfur oxidizing denitrification field-scale bioreactors:
Testing of the two field-scale bioreactors continued from March 15 through July 28, 2005. The details of the assembly and conditions of testing are available in the Progress Report submitted in March 2005. Both bioreactors were monitored for pH, Total Alkalinity, Nitrate-Nitrogen, Nitrite-Nitrogen, Sulfate, Chemical Oxygen Demand, Biochemical Oxygen Demand, and Total Kjeldahl Nitrogen. The bioreactors were stopped on July 28, 2005. In late August, the bioreactors were disassembled and samples of the solid-phase media (marble chips, crushed oyster shell, and sulfur pellets) were taken for examination.

2. Titration tests on solid-phase buffer:
Kinetics of dissolution of the two solid-phase buffers (marble chips and crushed oyster shells) and their buffering rates were investigated in laboratory-scale tests. Two identical stirred reactors, each containing a buffer (with a mass such that it could buffer the addition of an acid for many cycles) and a synthetic solution, were employed. An external source of H⁺ was added to each reactor in a cycle. The equivalents of H⁺ added corresponded to the alkalinity that would be destroyed on a stoichiometric basis in one 24-hour time period (after normalization of volumes) by a wastewater entering the field-scale bioreactors at a flowrate of 330 gpd and containing 40 mg/L NO₃^- - N. The pH profile as a function of time after addition of H⁺ was created. When the pH value stabilized, another cycle of H⁺ was added; a total of 5 cycles of H⁺ addition were conducted.

3. Hydrogenotrophic denitrification using a hollow fiber membrane:

3.1 Enrichment cultures
Denitrifying bacteria were cultured from wastewater samples obtained from Berkshire wastewater treatment plant (Lanesboro, MA) and Belchertown wastewater plant (Belchertown, MA). Batch cultures were set up in 1000 mL Erlenmeyer flask and sealed to ensure anoxic conditions. The flasks were placed on a shaker table at 150 rpm in a dark room at a constant temperature of 20°C. Each flask had 250 mL of the supernatant from the sludge and 250 mL of synthetic wastewater with the following composition: 100 mg/L NO₃^- - N, 0.5 g/L NaHCO3, 8.5 mg/L KH₂PO₄, 21.75 mg/L KH₂PO₄, 33.4 mg/L Na₂HPO₄.7H₂O, 2.5 mg/L MgSO₄.7H₂O, 0.25 mg/L FeCl₃.6H₂O, 27.5 mg/L CaCl₂.

Groundwater from a nearby farm in Amherst, MA was used to prepare synthetic wastewater. The groundwater was found to have 0.416 mg/L NO₂^- - N and 0.088 mg/L NO₃^- - N. The flasks were flushed with H₂ gas (Merriam Graves, Springfield, MA) to remove all the gases present in the headspace; excess H₂ was collected in a 1 L tedler bag (SKC) attached to the flask. Carbon dioxide gas was flushed when required to maintain a pH range of 7 and 8.

The cultures were monitored for nitrate reduction by drawing out 10 mL of the solution and filtering through 47 mm Millipore glass fiber filter paper with a pore size of 0.45 mm. The pH of the solution was measured using Orion 720A pH meter and tested for NO₃^- - N using NitraVer HACH reagents (Loveland, CO. Range 0-30 mg/L NO₃^-N) using a Thermo Spectronic (Genesys) spectrophotometer at 507 nm UV absorbance. On observing the reduction of nitrate, 250 mL of fresh synthetic wastewater was added and the flasks were flushed with hydrogen gas. In case of a rise in pH, carbon dioxide gas was added to adjust the pH in the range of 7-8.

Kinetics of the denitrifying cultures was determined over a period of 10 hours. At the start of the experiment, fresh synthetic wastewater was added to the flasks, the cultures were flushed with H₂ gas and placed on a shaker table at 150 rpm. At intervals of 2 hours, 10 mL of the solution was withdrawn and tested for volatile suspended solids, total solids, pH, and NO₃^- - N.

3.2 Membrane materials:
Hollow fiber flow through membrane bundles potted in polysulphone fittings were used in this research manufactured by Mitsubishi Rayon, NY. The membranes were a composite laminated hollow fiber with thin dense polyurethane layer sandwiched between two microporous polyolefin layers. The bundle consisted of 228 fibers, each 50 cm in length with an internal diameter of 200 mm, and an external diameter of 280 mm. Porosity was about 50% with a nominal pore size of 0.03 mm. The surface area was 0.198 m².

3.3 Reactor configuration:
A 500 mL glass reactor sealed at both ends by rubber O rings was used as the HFMB. A schematic of the reactor system is shown in Figure 1. The gases flowed through the lumen side of the membrane and the liquid flows through the shell side. The liquid and the gases flowed from the bottom to the top of the reactor. The liquid side was equipped with a Parker valve (Jacksonville, AL) to maintain the liquid pressure above bubble point. The inlet gas to the HFMB used a Magnehelic pressure gauge (0-6 inches of water) to measure the inlet pressure to the system. The gas line from the top of the reactor used a glycerin filled pressure gauge (Ametek 0-30 psi) to measure the exit gas pressure. The exit gas line was connected to a needle valve to maintain the pressure within the system.

The circulation rate is maintained by a Cole Palmer pump (Model 7553-70, 6-600rpm). A sampling port consisting of a glass cylinder with a septum cap was located in the recirculation line. A helium bubble was maintained in the sampling port to allow gases to partition from the liquid side.

The synthetic wastewater was stored in a 25 L Nalgene carboy and supplied to the bioreactor using a Masterflex C/L 77120-40 peristaltic pump having a flow range of 0-5 mL/min. 0.059 inch ID Tygon microbore tubing was used for the influent line liquid flow. During the initial stages of the experiments, the carboy was flushed with nitrogen gas to maintain anoxic conditions and to drive out any other gases present in the headspace. Ultra pure H₂ and bone dry CO₂ (Merriam Graves, Springfield, MA) were used in this research. The tubing for all gas lines were FEP lined to have least gas permeating through it, the tubes are 0.18 inch OD with 0.25 inch thickness. The flowrate was measured using a GFM17 mass flowmeter (Aalborg, Orangeburg, NY) with a flow range of 0-10 ml/min. Automatic timer (ChronTrol, CA) was used to control the switching of the influent pumps (on/off at different times of the day) to create varying flows to the HFMB.

3.4 Confocal Microscopy
Biofilms attached on the hollow fiber membranes were observed using an LSM-510 Meta Confocal System equipped with X10 multi-immersion lenses (Zeiss, Jena, Germany). Two fluorescent probes were used to investigate specific chemical compounds. Lectin-FITC from Canavalia ensiformis (Sigma-Aldrich) was used to stain _-D-Mannose and _-D-Glucose of polysaccharides in EPS. SyproOrange (Invitrogen, Molecular Probes, Eugene, OR) was used to stain proteins in extracellular polymeric substances (EPS).

Lectin staining was conducted at the concentration of 100 µg lectin/cm² membrane surface area as described in Lawrence et al (2003). The stained membrane was incubated for 20 minutes at room temperature in humidity chamber, and carefully rinsed with phosphate buffer solution. The SyproOrange stock solution was prepared by 5000 times dilution with 7.5 % (v/v) acetic acid solution (Invitrogen, Molecular Probes, Eugene, OR), and then 1 cm² of the membrane was stained with 100 µl of the stock solution. The membrane was incubated at room temperature in dark chamber for 40 minutes, and carefully rinsed with 7.5% (v/v) acetic acid.

Progress on Tasks
1. Sulfur oxidizing denitrification field-scale bioreactors:
Composite samples of influent to the bioreactors and the effluent from each bioreactor are collected twice every week and analyzed for:
pH (Figure 2),
Total Alkalinity (Figure 3),
NO₃^- - N (Figure 4),
NO₂^- - N (Figure 5),
SO42- (Figure 6), and
COD (Figure 7)
BOD5 (Figure 8) and
TKN (Figure 9) are monitored weekly.

2. Titration tests on solid-phase buffer:
Results of the titration tests on marble chips and crushed oyster shells are shown in Figure 10. It can be noted from Figure 10 that:

• The drop in pH upon addition of H⁺ is sharper for marble chips than for crushed oyster shells.
• Crushed oyster shell system rebounds from the pH drop faster and to a final pH value closer to the original value compared to the marble chip system.

These two observations clearly demonstrate that crushed oyster shells are better suited to act as a solid-phase buffer compared to marble chips in denitrification systems using sulfur as electron donor.

3. Hydrogenotrophic denitrification using hollow fiber membrane

3.1 Batch culture enrichment:
Denitrifying cultures were obtained from Berkshire Mall wastewater treatment plant (Lanesboro, MA). Synthetic feed water for the cultures was prepared using non-chlorinated groundwater. Groundwater contains trace elements such as nickel (Ni²⁺) and selenium (Se²⁺) that are needed for growth of microorganisms. Nickel is essential for growth of hydrogenotrophic bacteria because all hydrogenase (enzymes for growth) contain Ni²⁺ as their cofactors (Madigan et al, 2003). Also, hydrogen-oxidizing bacteria need mineral salts such as Fe²⁺ for enrichment. Methanol was initially added to the cultures to verify that there was an active denitrifying population present, since the cultures obtained from the wastewater treatment plant used methanol as an electron donor. On observing denitrification, the conditions were shifted to H₂ as the electron donor. Acclimatization to the new conditions occurred over days 15-50 (Figure 11). On day 60, regular sparging of the flasks with CO₂ was initiated to control pH, resulting in increased denitrification rates. Stable denitrification has been observed in the flasks for approximately 1 year.

3.2 Batch culture kinetics:
The results of batch culture kinetic experiments are shown in Figure 12. A first order denitrification rate with a kinetic coefficient of 0.0621 hr-1 was observed in this study. A comparison between denitrification rates using different substrates is shown in Table 1 (Rezania et al, 2004, Haugen et al, 2002). A zero order kinetic model was proposed for nitrate reduction at two temperatures by Rezania et al, 2004. The values obtained in this study are close to the values obtained in other studies.

3.3 HFMB Performance:
Influent and effluent NO₃^- - N concentrations over time for the HFMB are shown in Figure 13. The HFMB was initially operated with a 12.6-hour HRT and a shell side recirculation rate of 286 mL/min. The influent NO₃^- - N concentration was maintained at approximately 50 mg NO₃^- - N/L while H₂ was supplied in excess of the stoichiometric requirements. The acclimation time for complete denitrification was approximately 10 days. Effluent nitrite concentrations increased during the acclimation period from 3 to 33 mg NO₂^- - N /L but decreased on day 10 to 0.2 mg NO₂^- - N /L, as shown in Figure 13. This transient increase in NO₂^- - N concentration has been observed in other studies (Ergas and Reuss, 2001) and has been attributed to the need for organic carbon for nitrite reduction, which is provided either by decaying cells or soluble organic matter from the densely populated biofilm (Liessens et al, 1992).

Between days 4 and 41 effluent NO₃^- - N and NO₂^- - N concentrations were less than 5 and 0.2 mg/L, respectively. The World Health Organization (WHO) has set a safe limit of 11 mg NO₃^- - N /L (Mateju et al, 1992). The U.S. drinking water standard for NO₂^- - N is set at 1 mg/L (EPA, 2002). The effluent pH during this period increased from 7.8 to 9.0 (Figure 14). Studies have shown a decrease in nitrate and nitrite concentrations when pH increased from 7.2 to 8.4 but an increase in nitrate and nitrite concentrations when pH rose to 9.5 (Lee and Rittmann, 2003). This phenomenon was not observed in this study. The optimum pH was found to be 9.0. At this pH nitrate and nitrate concentrations were below the EPA drinking water standards.

One of the tasks of this study was to operate the bioreactor at minimum HRT for economic efficiency. Complete denitrification was observed at a HRT of 12.6 hour between days 4 and 32. Both NO₃^- - N and NO₂^- - N concentrations were within safe drinking water limits (EPA, 2002). On day 33, the HRT was reduced to 10 hours and a small accumulation of NO₂^- - N of 5 mg/L was observed for 2 days. On further reduction of the HRT to 5.7 hours, the effluent NO₃^- - N concentration increased to 26 mg NO₃^-N /L and a gradual increase in NO₂^- - N concentration was also observed. Since nitrite inhibits the denitrification process the reactor was flushed with nitrate free influent for four days and the HRT was increased to 6.5 hours. An optimum HRT of 8.7 hours was established after day 60.

Between days 66 and 80, an increase in effluent nitrite concentration was observed (Figure 13), accompanied by a decrease in effluent pH to 7.47. In previous studies (Lee and Rittmann, 2003; Haugen et al, 2002), nitrite accumulation was shown to be a sensitive indicator of process stress due to high pH. This is contrary to what was seen in this study, where nitrite accumulation was observed at both low and high pH. In the batch cultures, nitrite accumulation was observed at a pH of 9.5, therefore for further experiments a target pH of approximately 9.0 was used. On day 80, 0.5 g/L of sodium bicarbonate was added to the influent feed water and periodic sparging of carbon dioxide to the HFMB was stopped, to more consistently control pH in the system. After the addition of bicarbonate to the feed water, the effluent pH stabilized between 8.6 and 9.3 and the nitrite concentration decreased to less than 0.5 mg NO₂^- - N /L.

One of the objectives of this research was to test the robustness of the system when dissolved oxygen was present in the feed water. On day 107, sparging of the influent feed with nitrogen gas was discontinued. The effluent nitrate concentration was initially 12 to 14 mg/L NO₃^- - N and the nitrite concentration was below 1 mg/L NO₂^- - N. The nitrate concentration gradually decreased to below 10 mg/L NO₃^- - N. It is known that denitrification rates decrease with increase in dissolved oxygen concentrations and the process ceases with oxygen concentration greater than or equal to 1.0 mg/L (Whitmyer et al, 1991, Metcalf and Eddy, 2004). However, denitrification was observed in presence of dissolved oxygen in the feed water. In the presence of oxygen denitrifying bacteria prefer oxygen as an electron acceptor over nitrate (Madigan et al, 2003). Many hyrogeneotrophic (hydrogen oxidizing bacteria) grow best microaerobically and are most successful in oxic/anoxic interfaces. Typically, oxygen levels of about 5-10% support best growth for hydrogen bacteria (Madigan et al, 2003).

Membrane flux for the system was calculated to be 0.4 g NO₃^- - N /m² d. Table 2 summarizes the nitrate removal rates in other studies. The system was further evaluated using real wastewater. Composite samples of non-chlorinated nitrified wastewater were obtained from the Amherst, MA wastewater treatment plant. Although the facility uses an extended aeration to nitrify their wastewater and does not carry out denitrification, the nitrate concentrations were less than 5mg/L. Typical Chemical Oxygen Demand (COD) levels of the wastewater were 35 mg/L and a pH of 6.25. The flow rate for the HFMB was modified from the current setting to mimic a typical onsite flow system. The percentage of flow distribution over the period of 24 hours was determined on a typical day to an onsite system (Table 3).

Influent, effluent nitrate and nitrite concentrations, pH, turbidity, COD were measured during the start of the influent flow and just at the end of the times when the flow would stop. This protocol was initially continued for about 2 weeks and data of only single day has been presented here.

The average effluent nitrate concentration was 0.45 mg/L NO₃^- - N and 0.0067 mg/L NO₂^- - N for a period of 24 hours (Figure 15). Effluent turbidity was between 0.93 to 1.5 NTU while influent turbidity of the wastewater was 0.72 NTU. Influent pH was 7.68 and a rise in pH to 8.5 was observed on denitrification. A similar trend has been observed on other days for two consecutive weeks (data not shown).

Sodium bicarbonate was added to the influent of the HFMB because the composite sample obtained from wastewater treatment plant was in the range of 6.25 to 6.8. The HFMB was operated with an influent pH range of 7.5 and an increase in pH was required for the stability of the system. Change in pH could cause stress to the system as reported in previous studies (Lee and Rittmann, 2003). The system showed an average removal rate of 80 % over a period of 24 hours. On one particular day sodium bicarbonate was not added into the influent to the HFMB and effluent COD values were below the influent COD values suggesting the consumption of COD for carbon requirements. The system on this day also showed 78.23% removal rate.

Since the influent nitrate concentrations from the wastewater treatment plant were very low (< 10mg/L NO₃^- - N), the robustness of the bench scale system for treatment of real wastewater was not challenged. Hence, after two weeks of continuous operation, the composite samples obtained from the Amherst wastewater treatment plant were spiked to about 50mg/L NO₃^- - N by addition of sodium nitrate. No nutrients were added; however, 0.5 mg/L of sodium bicarbonate was added to maintain a starting pH of 7.5. Effluent nitrate concentrations were less than 10 mg/L NO₃^- - N on all sampled times but were reported higher on two occasions (16:00 hrs and 22:00 hrs) (Figure 16). This rise in nitrate concentrations was observed when the reactor was shifted to higher nitrate loading rates. But after the flow was cut off from 16:00 hrs to 18:00 hrs the system responded to effluent concentrations of approximately 8 mg/L NO₃^- - N suggesting that the no flow time period was just enough for the microorganisms to denitrify the accumulated nitrate in the HFMB. Influent pH to the system was 7.9 and a rise in pH to 9.4 was observed on denitrification. Mass removal rate was calculated to be 82.4%.

The system was operated for one additional week with spiked wastewater before shutdown and results for the second week are shown in Figure 17 and Figure 18. A typical trend of higher nitrate concentrations (>10 mg/L NO₃^- - N) was observed from 16:00hrs for two different non-consecutive days. Again, this could be attributed to higher flowrates in shorter time intervals during the day (60% of the total flow before 16:00 hrs) which puts the HFMB system in a shock load of high nitrate loadings to denitrify in shorter periods. Eventually we see the low nitrate concentration during the early morning periods at 8:00 hrs (Figure 18), suggesting that sufficient time was needed to undergo denitrification (<10 mg/L NO₃^-N). Effluent nitrite concentration was below 0.65 mg/L NO₂^-N for all times for both days. In Figure 18, effluent COD values were below influent COD values. There was considerable biomass accumulation observed visually suggesting higher growth rates during this period. Mass removal percentages were calculated to be 78.8 % and 73.5 % for data shown in Figure 17 and Figure 18 if respectively.

There is a need for documentation on onsite nitrogen removal system performances (Whitmyer et al, 1991). Addition of real wastewater to a bench scale system using hydrogen has not been reported in literature. Also, change in different nitrate loading rates at shorter intervals typical of that of an onsite system has also not been reported.

3.4 Biofilm thickness:
Biofilm growth was observed along the complete length of the membrane module. The thickness was uniform because the system was equipped with a flow through membrane. Non uniformity of biofilm thickness has been reported in various studies (Lee and Rittmann, 2000, Ergas and Reuss, 2001). The membranes used in both these studies were dead end membranes.

After the system was shutdown the membranes were removed from the system and the threads were cut and observed under the microscope for biomass thickness. Figure 19 shows the biomass accumulation on the hydrophobic membrane. The biomass developed was greater than 1mm on each side of the membrane. However, confocal laser scanning microscopy (CSLM) was conducted to increase the understanding of the biofilm structure on the membrane surface (Figure 20 and Figure 21).

Biofilms constitutes extracellular polymeric substances (EPS), multivalent cations, biogenic and inorganic particles as well as colloidal and dissolved compounds (Lawrence et al, 2003). EPS are mainly responsible for the structural and functional integrity of biofilms and are considered as the key components for the physicochemical and biological properties of biofilms (Madigan et al, 2003). Confocal laser scanning microscopy (CSLM) in conjunction with fluorescent chemical probes enabled examination of the three-dimensional structure of fully hydrated and intact bacterial biofilms (Wingender et al, 1999).

In this study, biofilm attached on the hollow fiber membrane surface was observed using the CSLM. The biofilm community is seen to be homogenous in nature (Figure 20) and not attached to the membranes because the membranes are hydrophobic in nature. Studies have shown that biofilms attached to hydrophilic membranes (Min et al, 2005) are denser; however, we did not observe this in the above pictures. The lectin stain reveals glycoconjugate, sugar and carbohydrates and the SyproOrange stain reveals the proteins present in the biofilm (Figure 21).

Difficulties
No significant difficulties have been encountered.

Project Objectives for Next Reporting Period

Objectives
None since the project end date is 9/1/05.

Tasks to Meet Objectives
Does not apply since the project end date is 9/1/05.

Work Plan for Next Reporting Period
Does not apply since the project end date is 9/1/05.

Anticipated Success in Meeting Project Objectives
We do not anticipate problems in meeting the project objectives for the next six months. Factors beyond our control - poor weather and soil conditions for field experiments, long equilibration times post-disturbance for the pilot-scale experiment - could put us behind schedule, but we have tried to plan for these eventualities.

Overall Project Timeline Update
We are on schedule vis-ŕ-vis the timeline specified in the proposal.

Preliminary Data
Please refer to Figures 1-21.

Dissemination

• Presentation was made at a general workshop on the topic of onsite denitrification systems hosted by the Waquoit Bay NERR on March 15, 2005.
• Presentation was made at a meeting of Water Alliance of Martha’s Vineyard on June 15, 2005.
• Two abstracts have been submitted and accepted for oral presentation at the annual meeting of the Water Environment Federation (WEFTEC 05) in Washington D.C., October 30 ­ November 1, 2005.

Outreach:
A high school student, Ye Moe, has been working as an intern on the project through the Amherst Regional High School “work based learning” program. Ye has done a wonderful job maintaining the batch culture experiments and assisting with bioreactor operation. He will end her internship with us at the end of December.

References
Environmental Protection Agency (EPA) (2002) EPA 816-F-02-013. List of drinking water contaminants and MCLs. url: http://www.epa.gov/safewater/mcl.html

Ergas, S. and Rheinheimer, D.E. (2004) Drinking water denitrification using a membrane bioreactor. Water Research. 38: 3225-3232.

Ergas, S.J. and Reuss, A.F. (2001) Hydrogenotrophic denitrification of drinking water using hollow fiber membrane bioreactor. Journal of water supply: research and technology-Aqua. 50(3):161-171.

Haugen, K.S., Semmens, M.J., and Novak, P.J. (2002) A novel in situ technology for the treatment of nitrate contaminated groundwater. Water Research. 36:3497-3506.

Lee, K.C. and Rittmann, B.E. (2003) Effects of pH and precipitation on autohydrogenotrophic denitrification using a hollow fiber membrane biofilm reactor. Water Research. 37:1551-1556.

Lee, K.C. and Rittmann, B.E. (2000) A novel hollow fiber membrane biofilm reactor for autohydrogenotrophic denitrification of drinking water. Water Science and Technology. 14(4-5):219-226.

Liessens, J., Vanbrabant, J., Vos, Paul de, Kersters, K and Verstraete, W. (1992) Mixed culture hydrogenotrophic nitrate Reduction in Drinking Water. Microbial Ecology.24:271-290.

Madigan, M., Martinko, J.M., and Parker, J. (2003) Brock Biology of Microorganisms. 10th edition, Prentice Hall, NJ.

Mansell, B.O. and Schroeder, E.D. (1999) Biological denitrification in a continuous flow membrane reactor. Water Research 33(8):4683-4690.

Mateju, V., Cizinska, S., Krejci, J.and Janoch, T. (1992) Biological water denitrification- A review. Enzyme Microb. Technol. 14: 170-183.

Metcalf and Eddy. (2004) Wastewater Engineering Treatment and Reuse. 4th edition, McGraw Hill, Boston.

Rezania, B., Oleszkiewicz, J.A. and Cicek, N. (2004) Kinetics of hydrogen dependent denitrification under varying environmental conditions. Proceedings at the 77th annual Water Environment Federation Technical Exhibition and Conference (WEFTEC), New Orleans, Louisiana.

Whitmyer, R.W., Apfel, R.A., Otis R.J., and Meyer, R.L. (1991) Overview of Individual onsite nitrogen removal systems. American Society of Agricultural Engineers. 10:143-154.

Wingender, J.; Neu, T. R.; Flemming, H. C. (1999), Microbial extracellular polymeric substances, Springer-Verlag Berlin Heidelberg