Environ. Sci. Technol. 2009, 43, 1049–1054
Enhanced Biofiltration Using Cell Attachment Promotors
have been found in the literature dealing with the potential environmental applications of these polymeric cofactors and bioadhesives on the immobilization of biofilms for environmental purposes.
JUAN J. GONCALVES AND RAKESH GOVIND* Department of Chemical and Materials Engineering, 620 Rhodes Hall, University of Cincinnati, Cincinnati, Ohio 45221-0012
Materials and Methods
Received April 27, 2008. Revised manuscript received October 14, 2008. Accepted November 11, 2008.
H2S polluted airstreams were treated in two biotrickling filter columns packed with polyurethane (PU) foam cubes, one with cubes coated with a solution of 25 mg/L of polyethyleneimine (PEI, coated reactor) and the other containing just plain PU cubes (uncoated reactor) at empty bed residence times (EBRT) ranging from 6 to 60 s. and inlet H2S concentrations ranging from 30 to 235 ppmv (overall loads of up to 44 gH2S/m3bed/h), with overall removal efficiencies (RE) in the range of 90-100% over 125 days. The acclimatization characteristics of the coated reactor outperformed those of the uncoated one, and both the observed elimination capacity (EC) of 77 gH2S/m3bed/h and retention of volatile solids (VS) of 42 mgVS/cube were maxima in the coated reactor. Insights into the controlling removal mechanisms were also provided by means of dimensionless analysis of the experimental data. Denaturing gradient gel electrophoresis (DGGE) showed that the dominant surviving species in both units belonged to the genus Acidithiobacillus.
Submerged batch experiments were carried out using open pore, black PU foam cubes (McMaster Carr, Santa Fe Springs, CA) of 2.54 cm in side length, 8-12 pores-per-inch (PPI) and porosity of 98%, submerged in a beaker containing nutrient solution with the following composition (modified ATCC medium 290 with a pH of 6.6) (g compound per 1000 g of deionized water): Na2HPO4, 1.2; KH2PO4, 1.8; MgSO47H2O, 0.1; (NH4)2SO4, 0.1; CaCL2, 0.03; FeCl3, 0.02; MnSO4, 0.02 and agar, 1.5. The medium was inoculated with 2.50 g of a microbial blend used for BOD5 analysis (BOD Seed, Biosystems Int. IL). The batch was bubbled with 15 sccm of a 5% v/v H2S in nitrogen source (Matheson, IL) as well as air while stirring gently. Before submersion, six different cationic solutions were used to coat the PU foam cubes, eight cubes per solution, whereas the other eight cubes were submerged without any previous treatment. The aqueous solutions were polyethyleneimine (PEI) (Supelco, PA) at 25, 50, and 100 mg/L; poly-D-lysine (PDL) (Sigma, MO) at 50 and 100 mg/L; and collagen from calf skin (Sigma, MO) at 3 g/L. The coating was accomplished by submerging the cubes in a beaker containing the solution and stirring gently for 20 min. The cubes were then removed and allowed to dry by natural convection under a stream of room temperature air. Continuous experiments were carried out in the setup shown in Figure 1. Compressed air (room temperature and pressure, 0.1% relative humidity, 30 ppmv H2O) from the laboratory air piping system entered two transparent PVC packed bed reactors (90 cm length, 10 cm diameter) filled up to an initial height of 60 cm (initial bed porosity of 53%) with open-pore, white PU foam cubes of 1.25 cm in side length, with 8-12 PPI and a porosity of 98%. The EBRT of the air
Introduction Polyurethane (PU) foams have been successfully used for treating H2S in fouled airstreams (1, 2), wherein the biofilters have not suffered from clogging issues (3-8) due to low biomass yield in the absence of carbon source. However, emission of odors during the start-up (acclimation) period is still an issue with PU foam biofilters used for treating H2S from wastewater treatment plants (3). Attempts to improve the start-up performance of PU foam biofilters have included modifying their material properties by either coating or mixing them with adsorbents like activated carbon or zeolites (4, 5). In this paper, we explore a different strategy for acclimatization acceleration, and possibly higher removal capacities due to increased biomass contained in the media carriers by using positively charged polymers in PU foams. Indeed, anchorage-dependent cells such as bacteria colonizing a biofilter do not produce sufficient quantities of positively charged extra cellular matrix proteins during the initial startup, thereby adhering weakly to the plastic surface. Cowan et al. (9), for instance, showed that using a cationic polymer, poly-L-lysine (PLL), there was increased attachment of biofilms on solid surfaces. Therefore, by precoating the plastic surface of PU foams with extra cellular matrix proteins, such as collagen, fibronectin, laminim, etc., or with synthetic polymeric cations, improved cell attachment could potentially be accomplished. Interestingly, however, no studies * Corresponding author phone: (513) 673-3583; fax: (513) 984 5710; e-mail:
[email protected]. 10.1021/es801156x CCC: $40.75
Published on Web 01/15/2009
2009 American Chemical Society
FIGURE 1. Schematic of the system used in the biological uptake of H2S with the PU foam media (coated and uncoated PU foam reactors were run in parallel in duplicate systems). A: PVC reactor bed (diameter 10 cm and length 90 cm) equipped with four sample ports. Initial bed height was 60 cm in both coated and uncoated reactors; B: nutrients tank; C: H2S cylinder (5% v/v H2S, balance N2); D: H2S mass flow controller; E: electrochemical sensor for the detection of the inlet and outlet H2S concentrations; F: H2S concentration data acquisition system; G: laboratory compressed air (relative humidity 0.1%); H: nutrient solution recirculation pump. VOL. 43, NO. 4, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1049
FIGURE 2. H2S removal performance for the coated and uncoated reactors over the first 5 days of operation. EBRT was 60 s. in both reactors.
FIGURE 5. Removal efficiency (RE) for the bioreactor sections as a function of the dimensionless numbers Peclet (Pe) and Damkohler (Da). Clockwise from top left corner: Upper uncoated, Lower uncoated, Lower coated, Upper coated. Peclet numbers are: 2 × 106 (dotted circle), 4 × 106 (dotted triangle), 9 × 106 (half-full circle) and 11 × 106 (star).
FIGURE 3. Performance of uncoated reactor. Inlet loads, removal efficiencies, and EBRT are based on overall bed height. FIGURE 6. 16S rRNA bands separated through DGGE techniques from biomass samples taken from the bioreactors after 125 days. Bands 1.1 to 4.2 correlate with Acidithiobacillus gene databank in 0.900-1.000. Reactor sections: AT, upper uncoated; AB, lower uncoated; BT, upper coated; BB, lower coated.
FIGURE 4. Performance of coated reactor. Inlet loads, removal efficiencies, and EBRT are based on overall bed height. stream in the system was varied from 6 to 60 s. A liquid nutrient solution, as described above, was sprayed on the top surface counter-current with respect to the air stream at a flux of 0.061 GPM/ft2 (0.0025 L/min/cm2). Both reactors were equipped with four sampling ports separated 30 cm from each other, and each sampling port was connected to a solenoid valve operated by a computer program to allow controlled sampling. One of the columns was packed with PU foam cubes coated with the cofactor (PEI solution at 25 mg/L) selected from the batch experiments as described above (coated) and the other column was packed with plain PU foam (uncoated) for control and comparison purposes. The two columns were seeded with a 1:1 mixture of secondary sludge from Cincinnati’s Mill Creek WWTP and the nutrient solution. The inlet concentrations of H2S varied between 30 and 235 ppmv. The reactors were operated to achieve a treatment efficiency of at least 90% by changing either the 1050
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 4, 2009
inlet H2S concentration and/or the inlet gas flowrate. The inlet pollutant load was increased and decreased several times to simulate real operating conditions. H2S concentrations were measured in real time by means of an inline electrochemical sensor (iTrans, Industrial Scientific Corp., PA) able to read concentrations of up to 500 ppmv at an inlet flow of 0.8-1.0 STPL/h. Upon completion of the continuous experiments, volatile solids (VS) and total solids (TS) were determined in each half of both reactors by weighing three samples containing the aforementioned biomass after drying them at 105 °C overnight and 500 °C for 4 h, subsequently. Additionally, DGGE of 16S rRNA biomass fragments was performed on samples containing a 1:1 biomass and sterile nutrients solution mixture obtained from each half of both reactors, using the protocol described by Goncalves and Govind (10).
Results and Discussion Submerged batch experiments performed as described in the materials and methods section showed that the dry weight gain of the 25 mg L-1 PEI coated cubes was the highest (7.5% relative) while 50 mg L-1 of collagen and PDL gave maximum weight gains of 5.6 and 4.6%, respectively. Higher concentrations of PEI, PDL, and collagen gave lower weight gains, due to increased lysis and toxicity (11). Based on these results, we selected the 25 mg/L PEI solution as coating cofactor for the subsequent continuous experiments. As shown in Figure 2 for the uncoated and coated reactors, during the start-up
VOL. 43, NO. 4, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1051
TABLE 1. Modeling Equations for the Performance of the Pu Foam Packed-Biofilter (13) and Derived in the Supporting Information
TABLE 2. Values Used in the Determination of Monod’s Velocity of Reaction for the Coated and Uncoated Reactors
phase of the first 5 days, the coated reactor achieved almost 100% RE after three days of operation, while the uncoated reactor exhibited intermittent periods of high treatment efficiency. Figures 3 and 4 show the performance of the uncoated and coated reactors, respectively, for all collected data over the duration of 125 days. The reactors were operated by randomly increasing and decreasing both the inlet H2S concentration in the range 30-235 ppmv, and EBRT in the range 6-60 s. providing that the observed RE values were in excess of 90%. The elimination capacity of the coated reactor was considerably higher than the uncoated one, since loads as high as 55 gH2S/m3bed/h could be maintained within the threshold of 90% RE as opposed to 35 gH2S/m3bed/h in the uncoated reactor. Due to fast accumulation of biomass in the foam, both beds compacted steadily from day 1 to day 70, after which the bed height remained approximately constant, attaining a reduction of the bed volume of 25% for
the uncoated reactor and 33% for the coated one. The highest compaction in the latter is a consequence of the higher accumulation and growth of biomass promoted by the coating factor and the higher amounts of substrate fed to the bioactive films. Representing biofilter performance in terms of removal efficiency and inlet loading does not clarify whether the performance is limited by mass transfer and/or degradation kinetics. To clarify the interplay of mass transfer and biodegradation kinetics, all experimental data collected in the present study was compared using two dimensionless numbers that account for the ratio of mass transfer controlling mechanisms, i.e., diffusion versus advection, specifically Peclet number (Pe), and the biomass ability to chemically oxidize the H2S, namely Damko¨hler (Da), as defined in Table 1. The Damko¨hler number includes the maximum velocity of reaction of the Monod’s type, which was determined by fitting all collected data to a mathematical model, given in Table 1, that quantifies the performance of a porous synthetic media biofilter degrading a single pollutant (12, 13). This mathematical model has been derived in the Supporting Information. Table 2 contains the values used for the quantification of the parameters in Table 1. The biofilter height was subdivided into two sections to better demonstrate the interplay of mass transfer and reaction kinetics, since the pollutant concentration varies with biofilter height. Figure 5 shows a compilation of these two dimensionless numbers for each section of the two reactors. The sections belong to, clockwise from the top left corner, the upper uncoated, lower uncoated, lower coated, and upper coated half-reactors. It is clear from Figure 5 that for small values of the Peclet number, the system is kinetically controlled. When the Peclet number is small, the contact time between the polluted stream and the biomass, the diffusion of the pollutant into the biomass, or both are maximized. This translates into low convective and diffusive mass transfer resistance, for which the pollutant removal would depend on the activity of the biomass toward the biodegradation of the H2S. When the Peclet number is small, a minor change on the Damko¨hler number increases the removal efficiency greatly, and this tendency is seen in all sections studied. Conversely, at higher Peclet numbers, both mass transfer and kinetics mechanisms are important in the biodegradation of the pollutant, and sharper changes in the Damko¨hler number are required for the reactors to attain similar removal efficiencies than when operating at smaller Peclet values. The advection and diffusion mass transfer resistances are more evident in the lower section of the uncoated reactor (top left) since at a Peclet number of 9 × 106 a considerable increase in the Damko¨hler number from 1 to 7 does not improve significantly the removal efficiency, increasing from around 20% to only 50%. In general, it can be said that the reactors are kinetically controlled when the Peclet number is 2 × 106 and both mass transfer and kinetically controlled for higher values. When comparing the uncoated and coated reactors (top versus bottom plots of Figure 5) both reactors show the same rate of change in the removal efficiency of H2S with respect to the Damko¨hler number for the same Peclet values; this is, the differences in the biomass activity and accumulation in both reactors due
TABLE 3. Distribution of Measured Volatile Solids for the Coated and Uncoated Reactors
VS per cube [mg/cube] TS per cube [mg/cube]
1052
9
uncoated upper
uncoated lower
coated upper
coated lower
23.59 ( 0.56 58.97 ( 0.57
23.32 ( 1.44 48.58 ( 1.45
30.36 ( 1.95 67.47 ( 1.96
41.71 ( 1.07 85.12 ( 1.08
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 4, 2009
to the polymeric cofactor did not influence the physical mechanism of pollutant removal. Samples of the biomass attached in each biofilter section were squeezed out from the foam for biomass quantification and bacterial speciation using 16S rRNA DGGE, upon stoppage of the biofilters after 125 days of operation. The detailed protocol for this analysis has been presented earlier (10). All samples were taken at the same time. Using 341F as the primer, products were then sequenced by the University of Tennessee, Knoxville, Molecular Biology Resource Facility. Chromatogram files that were received from the sequencing facility were then aligned to known DNA sequences The 200 basepair amplicon was aligned using the Riobosomal Database Project (17). Results of the bacterial speciation are shown in Figure 6 indicating the most prominent rRNA bands on the denaturing gel panel. The four samples are named AB, AT, BB, and BT. The first letter, A and B, denotes reactors A (uncoated) and B (coated), whereas the second letter, B and T, represents, respectively, reactor bottom and top. All bands correlate with more than 90% similarity stored fragments of bacteria belonging to the strain Acidithiobacillus. Even though a microbial consortia was used to seed the reactors, a predominant genus survived the rather specific combination of nutrients and sulfide substrate. Even though different amounts of biomass accumulation were observed in each half of each reactor, all sections were populated by the same dominant bacterial genus, suggesting that the PEI does not favor speciation of the bacterial consortia. Due to concerns regarding contamination and disruption of the ongoing distribution of biomass in the reactors, samples of biomass could not be retrieved during different stages of this study in order to determine the bacterial speciation evolution. The distribution of volatile solids (VS) and total solids (TS) in the upper and lower sections of the uncoated and coated reactors are shown in Table 3. The coated reactor had a significantly higher VSS and TS in both the upper and lower sections than the uncoated reactor, demonstrating that more biomass was attached on the coated media than the uncoated media. During the long-term experiments, the foul gas exhibited a pressure drop ranging from 42-59 Pa/m for both reactors, even though considerable accumulation of biomass occurred and compaction of about 33% was observed. These values compare satisfactorily to others found in the Literature. Hirai (18) reported pressure drops ranging from 60-304 Pa/m in biofilters treating H2S. Wani (19) reported a pressure drop of 149 Pa/m in airstreams treating odors in a biofilter using natural media.
Nomenclature As Ab Af b Cf Cg Da Df Dl RE Fpd g H kl Ks L Pe
Effective specific area [m-1] External foam specific area (m-1) Internal foam specific area (m-1) Decay coefficient (s-1) Substrate concentration in the biofilm (kg/m3) Substrate concentration in the gas (kg/m3) Damko¨hler number (-) Effective diffusivity in biofilm (m2/s) Substrate diffusivity in liquid (m2/s) Removal efficiency (%) Packing factor 1812 (ft-1) Gravity (10 m/s2) Dimensionless Henry’s law constant (-) Substrate liquid phase mass transfer coefficient (m/s) Half-saturation constant (kg/m3) Reactor length (m) Peclet number (-)
(∆P)/(L)|f (∆P)/(L)|b vf vg vl Xf Y Greeks δf δl εb εf µg µl µmax Fg
Gas stream pressure drop within the foam (kg/m2/s2) Gas stream pressure drop in the bed (kg/m2/s2) Superficial gas velocity through foam packing (m/s) Superficial gas velocity (m/s) Superficial liquid velocity (m/s) Biofilm density (kg/m3) Yield coefficient (-) Biofilm thickness (m) Liquid thickness (m) Bed porosity with or without biofilm and water layer [-] Operation foam porosity with biofilm, water layer or both [-] Gas viscoty [kg/m/s] Liquid viscosity [kg/m/s] Maximum Monod’s velocity of reaction [s-1] Gas density [kg/m3]
Subscripts i Inlet o Outlet
Supporting Information Available Biotrickling filter packed with open pore media: performance prediction model. This information is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Aizpuru, A.; Khammar, N.; Malhautier, L.; Fanlo, J. L. Biofiltration for the treatment of complex mixtures of VOC. Influence of the packing material. Acta Biotechnol. 2003, 23 (2-3), 211–226. (2) Govind R.; Narayanan, S. Selection of bioreactor media for odor control. In Biotechnology for Odor and Air Pollution Control; Shareefdeen Z.; Singh, A., Eds.;Springer Verlag: Germany, 2005. (3) Gabriel, D.; Deshusses, M. A. Performance of a full-scale biotrickling filter treating H2S at a gas contact time of 1.6 to 2.2 seconds. Environ. Prog. 2003, 22 (2), 111–118. (4) Li, C.; Moe, W. M. Sequencing batch biofilter operation for treatment of methyl ethyl ketone (MEK) contaminated air. Environ. Technol. 2003, 24 (5), 531–544. (5) Kim, H.; Xie, Q.; Kim, Y. J.; Chung, J. S. Biofiltration of ammonia gas with sponge cubes coated with mixtures of activated carbon and zeolite. Environ. Technol. 2002, 23 (8), 839–847. (6) Van Groenestijn, J. W.; Liu, J. X. Removal of alpha-pinene from gases using biofilters containing fungi. Atmos. Environ. 2002, 36 (35), 5501–5508. (7) Moe, W. M.; Irvine, R. L. Polyurethane foam based biofilter media for toluene removal. Water Sci. Technol. 2001, 43 (11), 35–42. (8) Park, S. J.; Nam, S. I.; Choi, E. S. Removal of odor emitted from composting facilities using a porous ceramic biofilter. Water Sci. Technol. 2001, 44 (9), 301–308. (9) Cowan, S. E.; Liepmann, D.; Keasling, J. D. Development of engineered biofilms on poly-L-lysine patterned surfaces. Biotechnol. Lett. 2001, 23 (15), 1235–1241. (10) Goncalves, J. J.; Govind, R. H2S abatement in a biotrickling filter using iron(III) foam media. Chemosphere 2008, doi: 10.1016/ j.chemosphere.2008.07.048. (11) Helander, I. M.; Alakomi, H. L.; Latva-Kala, K.; Koski, P. Polyethyleneimine is an effective permeabilizer of gram-negative bacteria. Microbiology 1997, 143, 3193–3199. (12) Martin, R. W.; Mihelcic, J. R.; Crittenden, J. C. Design and performance characterization strategy using modeling for biofiltration control of odorous hydrogen sulfide. J. Air Waste Manage. Assoc. 2004, 54 (7), 834–844. (13) GoncalvesJ. J. GovindR. Simulation of Biotrickling Filters using Novel Foams for Treating Odors and Volatile Compounds. Proceedings of the 2005 annual meeting of AIChE , Cincinnati, Ohio, 2006. VOL. 43, NO. 4, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1053
(14) Rittmann, B. E.; McCarty, P. L. Model of steady-state biofilm kinetics. Biotechnol. Bioeng. 1980, 22 (11), 2343–2357. (15) Robbins, L. A. Improved pressure drop prediction with a new correlation. Chem. Eng. Prog. 1991, 87 (5), 87–91. (16) Li, H. B.; Mihelcic, J. R.; Crittenden, J. C.; Anderson, K. A. Field measurements and modeling of two-stage biofilter that treats odorous sulfur air emissions. J. Environ. Eng. 2003, 129 (8), 684– 692. (17) Ribosomal Database Project, National Center for Biotechnology Information. Available at http://rdp.cme.msu.edu and http:// www.ncbi.nlm.nih.gov.
1054
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 4, 2009
(18) Hirai, M.; Kamamoto, M.; Yani, M.; Shoda, M. Comparison of the biological H2S removal characteristics among four inorganic packing materials. J. Biosci. Bioeng. 2001, 91 (4), 396–402. (19) Wani, A. H.; Lau, A. K.; Branion, R. M. R. Biofiltration control of pulping odors - hydrogen sulfide: performance, makrokinetics and coexistence effects of organo-sulfur species. J. Chem. Technol. Biotechnol. 1999, 74 (1), 9–16.
ES801156X
