Monitoring Biodegradation of VOCs Using High ... - ACS Publications

A new sampling system has been designed and interfaced with high-speed gas chromatography (HSGC) to monitor and assess the performance of a trickle-be...
0 downloads 0 Views 80KB Size
Environ. Sci. Technol. 2001, 35, 1452-1457

Monitoring Biodegradation of VOCs Using High-Speed Gas Chromatography with a Dual-Point Sampling System ROBERT W. CURRENT, EVGUENII I. KOZLIAK, AND ANTHONY J. BORGERDING* Department of Chemistry, Box 9024, University of North Dakota, Grand Forks, North Dakota 58202

A new sampling system has been designed and interfaced with high-speed gas chromatography (HSGC) to monitor and assess the performance of a trickle-bed bioreactor designed for the removal of volatile organic compounds from air. A portion of a gas stream containing styrene and toluene was sampled both before and after passing through the bioreactor by means of a dual-loop sampling system. With a frequency of as high as 2 per minute, treated and untreated samples were alternately transferred on-line to the cryofocusing injection system of a HSGC and analyzed. This analytical system generated data with less than 2% relative standard deviations for standard samples, and residual contamination of subsequent analyses from a highly concentrated sample (2000 µg/L) was not observed. A benchscale bioreactor with a fiber mat support was used in these studies, resulting in residence times for analytes in the bioreactor of as little as 1 s. Rapid monitoring of this system detected subtle changes in the concentration of analytes with 30 s temporal resolution. Measurements showed a statistically significant increase in the removal of styrene from 22% to 27% when water was sprayed over the immobilized bacteria for 30 min. Overall, the bioreactor removed styrene from the air stream with a specific elimination capacity of 1700 g of styrene (m3 of biocatalyst)-1 h-1 at a space velocity of 3400 h-1.

Introduction High-speed gas chromatography (HSGC) has become a more highly developed technique in the past few years. In many ways, HSGC has the potential to be a very important technique for environmental studies, but this potential has been infrequently illustrated. HSGC instrumentation utilizing narrow injection pulses has led to separation times for mixtures of volatile organic compounds (VOCs) that are 1060 times faster than those of traditional GC techniques, with little or no loss of resolution (1-6). HSGC therefore allows rapid, highly selective analysis of gas phase samples. The technology for doing these rapid separations has been commercialized, and use of HSGC is now almost routine. Despite this, and although some of the early work in cryofocusing HSGC was directed toward instrumentation that could monitor environmental processes in near-real-time (7), few papers discussing applications of this type have been * Corresponding author email: [email protected]. und.nodak.edu; fax: (701)777-2331; phone: (701)777-2542. 1452

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 7, 2001

published. Instead, most recent publications involving rapid gas chromatographic separations are centered on instrumental advances and theory (8-18). GC is generally considered too slow a technique to be used as a monitoring instrument, but some examples in the literature illustrate that it can be done when measurement systems are improved. Mitra and co-workers have monitored VOC emissions of various systems using a microtrap GC system (19-23). Vas et al. and Matz et al. have used GC systems to monitor toxic VOCs in fermentation processes (24, 25). Measurements in these studies were taken every 5-10 min, necessitated by sampling and separation times. Williams and Pappas have used a simple GC system interfaced with a surface acoustic wave (SAW) detector for near-realtime monitoring of explosives and chemical warfare agents (26, 27). More recently, Gaffney et al. used rapid gas chromatography to separate nitrogen dioxide from a group of unresolved peroxyacyl nitrates in atmospheric studies performed in an airplane. Both of these examples were based on low-resolution separations of simple mixtures, with measurements recorded every 1-2 min. In this paper, we examine the use of cryofocusing HSGC as a tool for studying the bioremediation of VOCs in air streams. By making rapid measurements on bioreactors, we can monitor inlet and outlet concentrations for unwanted changes. Perhaps more importantly, we can study with statistical accuracy the changes in bioreactor performance that occur over short periods of time. Trickle-bed bioreactors, with synthetic materials as matrixes for bacterial immobilization, were sampled in this work. As discussed below, rapid measurements using HSGC can improve studies of transient effects on the bioreactors (i.e., the effect of changing influent concentrations or water trickling rates). Even at steady state, however, the inlet/outlet pollutant concentration measurements usually have poor reproducibility as a result of large variations in both the VOC evaporation into the air stream and the bioreactor’s performance (28-33). Frequent sampling and analysis of the process would detect more precisely any fluctuations in the steady-state inlet chemistry and the resulting changes in the performance of the bioreactor. In addition to the steady-state effects, transient phenomena have been observed as a result of various treatments of the bacteria, causing significant variations in the performance of the bioreactor over small periods of time. For instance, changing the physical parameters of the inlet air stream, such as empty-bed residence time (EBRT) and pollutant mass loading, may result in unexpected lag periods or fluctuations in the pollutant removal efficiency (29, 30, 32-36). These effects are most pronounced when a mixture of VOCs is treated by a biofilter; changes in the inlet concentrations of some pollutants may affect the biodegradation of the rest of the VOCs due to catabolite repression and co-metabolism (32, 34, 35, 37). Also, various biomass growth control treatments (addition/removal of mineral nutrients supplied with an aqueous medium, wetting/drying the biofilter, or periodic backwashing) result in significant “swings” in the pollutant removal efficiency (31, 38-44). The time span over which these changes occur depends on the EBRT. Literature in this area shows that bioreactor systems respond to transient effects within approximately 100 residence times (30-32, 35, 39-43). For industrial biofilters, EBRT is of an order of 1 min, corresponding to behavior changes within 1-2 h. For small but efficient devices, such as low-biomass granular and fiber-based trickle-bed bioreactors, residence times may be as short as 1-3 s (44, 45), 10.1021/es001708g CCC: $20.00

 2001 American Chemical Society Published on Web 02/10/2001

corresponding to changes in 10 min or less. Detection and explanation of these transient phenomena would lead to a better understanding of factors influencing the performance of the bioreactor. In previous studies of bioreactors, VOC concentrations at the inlet and outlet were measured using conventional GC, which required at least 5 min to analyze each sample of airborne pollutants (28-35, 38-51). Reducing this analysis time to 0.5-1.0 min would be very helpful to understanding the transient processes described in the previous paragraph. For systems with low residence times, frequent measurements are required in order to statistically validate observed changes occurring over the approximately 10 min time span. For example, using conventional GC, a maximum of 1 inlet and 1 outlet measurement would be possible over the course of the change, while using HSGC 5-10 measurements of inlet and outlet concentrations could be recorded. HSGC is also potentially advantageous for studying transient effects on systems with longer residence times. By taking frequent measurements, it would be possible to more precisely determine the time required for changes to occur, and to reliably monitor (with statistical verification) the system’s performance during the change. Such data would allow development of models to describe fast transient processes such as changes in mass loading and water trickling. Because of measurement limitations, the dynamic models that have been developed can only describe behavior during biomass accumulation resulting from bacterial growth, a much slower process (28, 29, 34, 47-49). The objective of our study was to develop a HSGC-based monitoring device suitable for measuring the inlet and outlet VOC concentrations in near-real-time, and to use those measurements to accurately assess rapid changes in bioreactor performance. Samples from an air stream are collected continuously through two loops. One loop collects the gases prior to the bioreactor, and the other samples the treated air. The samples collected in these loops are alternately transported on-line onto a cryofocusing HSGC injector and analyzed in less than 30 s. The technique is simple, and should be suitable for improved studies of many environmental processes involving rapid concentration changes of gas-phase pollutants.

Experimental Section Reagents/Preparation of Standards. Compounds used for analysis included styrene (Fisher Scientific, Fair Lawn, NJ), toluene (Eastman Kodak Co., Rochester, NY), and ethyl acetate (Fisher Scientific). Gas standards were prepared by injecting 4.4 µL of the neat liquids into 2 L Tedlar gas sample bags (Supelco, Bellefonte, PA) filled with zero-grade nitrogen (Airgas, Rador, PA). The analytes were allowed to volatilize and equilibrate in the sample bags for 2-3 h. The resulting standard samples contained 2000 µg/L styrene and ethyl acetate, and 1900 µg/L toluene; these concentrations were similar to those of the air stream treated by the bioreactor. Bioreactor Apparatus. Figure 1 is a schematic diagram of the bioreactor apparatus, which has been described in detail in an earlier publication (44). The main body of the apparatus was 20 cm in height, and had an inner diameter of 4 cm. A toluene-specific strain of bacteria, Pseudomonas putida TOD, was selected and grown as shown by Kozliak et al. (44). Bacteria were then immobilized on a polyaramide fiber mat by trickling a bacterial suspension over the mat for 2 days. The fiber mat had a thickness of 1.1 cm and was positioned between the gas inlet and outlet points. Both gas and an aqueous solution containing 2.0 g/L ammonium hydrophosphate (Sigma, St. Louis, MO) entered the bioreactor concurrently from the top. A plastic top cap covering the top of the bioreactor sprayed this solution (hereafter referred to as simply “water”) evenly over the fiber mat containing the

FIGURE 1. Schematic of the bioreactor used in these studies.

FIGURE 2. Diagram of the dual-point sampling system. immobilized bacteria. The fiber mat was sprayed with water in a cycle of 30 min on and 30 min off. The water flow provided by a Liquid Metronics model A141-155 systolic pump was adjusted to approximately 4 mL/min, and was controlled using a standard timer. The water was contained in a 1 L glass flask which enclosed the bottom of the bioreactor. The bioreactor was operated at steady state after 45 days of continuous use (see ref 44 for details). A flow of medical-grade air through the system was controlled to a constant 800 mL/min using an electronic gas flow regulator (model 810C-DR-1, Sierra Instruments, Monterey, CA), and monitored at the exit vent with an AMD3000 gas flow meter (J&W Scientific, Folsom, CA) to ensure the system was airtight. After leaving the flow regulator, the air passed through an analyte introduction chamber containing Parafilm-covered vials of toluene, styrene, and ethyl acetate. The volume of this chamber was 250 mL, and the vials were 1 cm in diameter and 7.5 cm high. Analyte concentrations were governed by evaporation, and adjusted by the number of vials of each analyte and the layers of Parafilm covering the vials. Specific concentration levels were calculated by comparison to signals generated by the gas standards. A portion of the analytes from inside the system was collected continuously through septa at two sampling points. Samples collected from point 1 in Figure 1 were used to determine the inlet concentration of analytes coming from the analyte introduction chamber. Samples collected from point 2 in Figure 1 were used to determine the concentration of analytes after the gas flow had passed through the fiber mat with immobilized bacteria; point 2 is henceforth referred to as the outlet. Sample Collection Apparatus. Figure 2 shows a schematic diagram of the sampling apparatus. The system was constructed entirely of commercially available stainless-steel components. The apparatus used two six-port two-position valves (Valco, Houston, TX) in order to create the sample loops. Sample loops with a volume of 0.785 mL were made from a 1 m length of 1/16 in. stainless-steel tubing (Valco) VOL. 35, NO. 7, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1453

FIGURE 3. Diagram of the modified cryofocusing injection system for HSGC. with an i.d. of 1.0 mm. The samples were collected through 24 gauge syringe needles (Hamilton, Reno, NV) connected to the six-port valves with 1/16 in. stainless-steel tubing (Minnesota Valve and Fitting, Eden Prairie, MN) using a 1/16 in. stainless-steel Swagelok union (Minnesota Valve and Fitting) and a graphite ferrule. The sample was pulled into the sample loops using the vacuum pump from the cryointegrating injection system (Chromatofast, Ann Arbor, MI). The vacuum line was connected to each of the six-port valves using a 1/16 in. tee fitting (Minnesota Valve and Fitting) and 1/16 in. stainless-steel tubing. A 1/8 in. needle valve (Valco) in the vacuum line was used to regulate the flow rates. The sample was transferred to the cryointegrating system using a flow of zero-grade nitrogen (Airgas, Rador, PA) regulated with a 1/8 in. needle valve (Valco). A 1/16 in. tee fitting (Minnesota Valve and Fitting) allowed the nitrogen to flow to both sample collection loops. A two position three-port valve (Valco) directed the flow of nitrogen. Sample Collection. The design of the sample collection apparatus allowed for the alternate collection of samples from the two different locations. The collected samples were then transferred to the HSGC instrument (see Figure 2). Note that, when the valves are in the position with the flow paths indicated by solid lines, the contents of the top sample loop (sample 1) are transported to the HSGC while the bottom sample loop is being filled with sample 2. In the opposite valve positions (indicated by flow through the paths shown by the dotted lines), the contents of the bottom loop are transferred to the HSGC while the top loop is being filled with sample 1. A needle valve was used to regulate the flow sampled by the vacuum pump so that 25 mL/min was continuously transferred from the bioreactor by the vacuum pump (see Figure 2). This corresponds to a flow of 12.5 mL/min from both the inlet and the outlet sample ports. Another needle valve (not shown in Figure 2) was used to regulate the sample transfer gas flow to a constant 50 mL/min. A Nafion dryer was placed between this sampling system and the cryofocusing injector to remove water vapor from the samples and prevent freezing of the cryotrap. Chromatographic Apparatus. HSGC for these analyses was performed using a Varian 3350 GC (Sugarland, TX) equipped with a modified cryointegrating injection system (Chromatofast, Ann Arbor, MI) and a flame ionization detector (FID). The cryointegrating injection system was modified using an eight-port two-position valve (Valco) to create a sample loop which allowed isolation of the GC carrier gas flow from the sample collection gas flow. A schematic diagram of the modified injection system is shown in Figure 3. The gas flow through the cryointegrating system was alternated between the GC carrier gas flow and the sample gas flow by changing the position of this eight-port valve. With the eight-port valve in the “sample” position (see Figure 3), the GC carrier gas flowed through a needle valve used to 1454

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 7, 2001

FIGURE 4. Sequence of chromatograms showing the analysis of a standard sample containing toluene, ethyl acetate, and styrene, with blanks sampled and analyzed in between. create a restriction that matched the restriction of the cryointegrating system, and the sample gas from the bioreactor flowed through the cryointegrating system to a vent. With the two position eight-port valve in the “inject” position, the GC carrier gas was directed through the cryointegrating system to the GC column, and the sample gas from the biofiltration apparatus bypassed the cryointegrating system and flowed directly to the vent. Valves were operated manually. When monitoring with a new measurement every 30 s, the general procedure was to switch the valve to the “sample” position for a 25 s period. The valve was then switched to the “inject” position for 5 s, during which time the cryotrap was rapidly heated using a capacitive discharge (1-5 ms) and the vaporized analytes transferred into the GC column with a short pulse duration. The vaporization temperature was held for 1.5 s, which was sufficient to completely flush the trap. Because the trap was small (∼10 cm of a metal capillary), it could be cooled back to trapping temperatures in less than 2 s while the valve remained in the inject position. The valve was then switched back to “sample”, and the recooled cryotrap began collecting analytes from the next sample while the previous ones were being separated on the column. Isolating the HSGC carrier and sample gas streams with the eight-port valve served several purposes. First, it allowed the entire flow from the sample loop to be directed through the cryofocusing injector, which maximized the sensitivity of the technique. Also, isolation of the sample stream allowed very high gas flow rates through the chamber at relatively low pressure. Finally, it allowed separation and sampling to occur simultaneously, minimizing analysis time, particularly for analysis of multiple samples. Ultra-high-purity hydrogen flowing at 2.5 mL/min was used as a carrier gas for GC separations. The GC column was a 1.5 m length of 0.250 mm i.d., DB-5 column with coating thickness of 0.250 mm (J&W Scientific, Folsom, CA). Separations were performed isothermally at 40 °C with a detector temperature of 200 °C. The cryointegrator was held at -50 °C.

Results and Discussion System Characterization. A primary concern with using this dual-point sampling system is the potential for residual analyte from one sample (for example, the analyte stream before biodegradation) to carry over into subsequent samples (e.g., the stream after treatment with the bioreactor). This is particularly important given the high concentrations used in this study. To assess the potential for this kind of contamination to occur, the system was used to sample and analyze an air standard containing high concentrations of analytes (∼2000 µg/L each of toluene, ethyl acetate, and styrene) alternately with a blank sample containing only nitrogen. The results are shown in Figure 4. Despite these high concentrations, only an insignificant amount (