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

Project Title: A System for Remediation of Polychlorinated Biphenyls in Sediments
Principal Investigator(s): Kevin H. Gardner, Warren Chesner, James Melrose
Additional Investigator(s): Emese Hadnagy, Irina Calante, Jean M. Spear
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

Tables

Table 1

• Complete the mass balance experiments and close the mass balance of the system.
• Complete processing pending samples on the GC/MS.
• Investigate the possibility of biphenyl degradation/complexation into organic molecules not detectable by the GC/MS.
• Further test the degradation potential of “aged” Mg/Pd.
• Continue PCB degradation experiments in pure solvents to identify congener-specific degradation pathways, relative resistance of positional isomers, changes in dioxin-like toxicity, and the reaction endpoints.
• Conduct experiments to determine the relative degradation potential of various PCB congeners by Mg/Pd
• Analyze whether better sediment homogenization improves replication of the results within the sediments.
• Continue to study the effect of solvents on PCB availability in actual contaminated sediments and compare findings with results from the desorption experiments. In particular, evaluate the effect of two biodegradable, citrus derived solvents, d-Limonene and Citrus Burst® 2 (the latter solvent contains a surfactant as well) on PCB availability. Investigate how all these factors influence PCB dechlorination in the sediment matrix.
• Present research findings in two conferences:
  1. 21st Annual International Conference on Soils, Sediments and Water, October 17-19, 2005, Amherst, Massachusetts.
  2. SETAC North America 26th Annual Meeting, November 13-17, 2005, Baltimore, Maryland.
• Publish research articles.

Tasks to meet objectives

• Run batch experiments in pure solvent systems with careful accounting for PCB dechlorination end products.
• Conduct mixing tests using the laboratory scale mixing device and fluorescent microsphere beads to evaluate time of mixing required.
• Run batch experiments with contaminated sediment and solvents.
• Measure extraction efficiency of biodegradable, natural solvents.
• Make sure that GC/MS operation is continuous and all QA/QC requirements are met.
• Conduct continuous literature search in major environmental research journals.
• Communicate and collaborate with other researchers and attend scientific conferences and workshops.
• Extract samples with Ultrasonic Extraction (EPA Method 3550B) and analyze them with a Varian Saturn 2200 Ion Trap Gas Chromatograph and Mass Spectrometer (GC/MS) following the manufacturer’s method (“Analysis of Semi-volatile Organic Compounds with Ion-Trap GC/MS”, developed based on EPA methods and requirements ­ Methods 8000B, modified 8270C, and 8290). Establish a calibration curve for congener-specific PCB analysis on GC/MS.

Progress on Tasks
(1) Batch Experiments in Pure Solvent:

• All experiments were conducted with NASA Mg/Pd as a reducing agent.
• Quality control experiments were conducted with Aroclor 1260 in 10% methanol (in distilled water) solution to investigate PCB extraction efficiency by liquid-liquid extraction into hexane. The effects of batch volume, initial PCB concentration and mixing time were tested. Based on the results the optimum liquid-liquid extraction method was determined for future experiments.
• A kinetic study was performed with Aroclor 1260 in 10% methanol solution using Mg/Pd. Samples were taken at 1, 2, 3, 4, 5, 10, 15, 20, 30, and 60 min. Rapid degradation was observed (Figure 1) with 58% of the initial PCBs removed in 4 min and the % removal not changing considerably in the next 1 hour. The 5 min sample was a suspected outlier and therefore its result was not plotted on the graph. No byproducts were observed in any of the samples.
• A revised mass balance experiments were conducted with Aroclor 1260 and biphenyl in 10 % methanol solution with Mg/Pd. The dissolved, volatile and adsorbed (onto the Mg/Pd surface) fractions of all PCBs and biphenyl, as well as chloride in the dissolved phase in the Aroclor experiments were measured after 2 hours reaction time. Results indicated that volatilization was not significant. A higher chloride concentration was detected in the Aroclor experiments in the treated sample relative to the control indicating that dechlorination was occurring. No significant amount of biphenyl was detected in the same experiments in any of the measurements. A significant amount of PCBs (or biphenyl) were extracted from the filtered Mg/Pd material suggesting that PCBs (or biphenyl) first adsorb to the surface of the bimetal and then dechlorination occurs. The mass balance still could not be closed because of the reaction of the biphenyl (Figure 2).
• Terphenyls, quaterphenyls, cyclohexylbenzene and bicyclohexyl were shown to be potential by-products of PCB dechlorination in a mechanochemical treatment by Nomura et al (2005). These compounds were formed from biphenyl either through polymerization or hydrogenation reactions in the mentioned study. For the present project all the above compounds except quaterphenyl were run and their peaks/retention times were successfully identified on the GC/MS. Chromatograms of all sample runs in each experiment are carefully inspected for any potential by-product peaks.
• Kinetic study was conducted with biphenyl and Mg/Pd in pure methanol to be able to increase the initial concentration of the contaminant (due to higher solubility in methanol as solvent than water or only 10% methanol) and thus potentially increase the amount of the forming by-products. Based on the literature, it was expected that the dechlorination reaction would slow down from the order of minutes to maximum 1 day (Halle et al, 2005). Samples were taken after 15, 30, 60, 90 and 120 min and 1, 2 and 5 days. No significant biphenyl degradation was observed in these experiments.
• After this, kinetic studies were conducted with biphenyl and Mg/Pd in 10% methanol solution (in distilled water) with three different contaminant-loading rates to establish a stable curve. Samples were taken at 0.5, 5, 10, 15, 30, 60 and 120 min. About 80% of the initial biphenyl was degraded in 2 hours (Figure 3).
• Kinetic studies were also performed with single PCB congeners BZ 170 and BZ 204 and Mg/Pd in 10% methanol solution. Samples were taken after 0.5, 5, 10, 15, 30, 60 and 120 min in the BZ 170 experiments and after 0.5, 5, 10, and 30 min in the BZ 204 runs. An increased loading rate resulted in slower dechlorination of BZ 170 and a lower overall %removal was observed in the same case over the course of 2 hours (Figure 4 and Figure 5). About 80% of the initial BZ 170 was dechlorinated in 2 hours with 4 mg/g (mg BZ 170/g Mg/Pd) loading rate (Figure 4), whereas 98% BZ 170 was degraded over the same time period when the loading rate was decreased to 1.3 mg/g (Figure 5). The initial BZ 170 concentration was also lower in the second case, 13 ppm vs. about 40 ppm in the higher loading rate experiments. No by-product peaks were observed in these runs. Results of the BZ 204 experiments indicated 62% degradation of the congener in 30 min (Figure 6). Significantly higher chloride concentrations were measured in the shorter-term treated runs (0.5 to 30 min depending on the experiments) in both the BZ 170 and BZ 204 experiments indicating that dechlorination was occurring. Blank runs with Mg/Pd in 10% methanol (with no PCB added) indicated that there was a significant chloride release from the Mg/Pd that seemed to vary over time and that was potentially influenced by the presence of PCBs and the dechlorination reactions as well. The chloride from the Mg/Pd might have been released from the Pd (typically available in a chlorinated form commercially, e.g. potassium hexachloropalladate) during the dechlorination reaction, although the manufacturing process and the exact composition of the Mg/Pd powder are not known. On the other hand, as the reaction time increased, the dissolved chloride concentration decreased, indicating that chloride ions might have reacted with the Mg/Pd material through adsorption or surface complexation. Current efforts are focusing on resolving the previously mentioned issues in order to close the chloride mass balance.

(2) Sediment Homogenization:

• Previous experiments sometimes showed large variations in PCB concentration between sample duplicates. It was hypothesized that the PCB distribution throughout the sediment was not uniform. Even after the sediment batch was manual mixed with a spatula some samples sometimes had distinctively higher PCB concentrations, which are referred to as “hot spots”. This was especially true for Hudson River (Hud.) and New Bedford Harbor (NBH) sediments, which consist mostly of silts and clays and have high organic contents. The Housatonic River (Hous.) sediment is mostly sand which made manual homogenization easier to perform and therefore had less PCB concentration variation between duplicates. In order to decrease the PCB variation within the sediment an experiment was conducted where three sediments, Hous., Hud., and NBH, were homogenized sequentially by three different mixing methods and then sampled. First the sediment was mixed in the sediment’s container with a stainless steel paint mixer attached to a power drill for 10 minutes. A smaller sediment batch was obtained and placed inside a stainless steel bowl to be homogenized with a Hobart mixer for 5 minutes. The Hobart mixer was set at the lowest mixing level of 1 to prevent the sediment from ejecting out of the bowl. The mixed sediment was then placed inside a glass jar and mixed manual before extracting triplicate samples. Overall, the sequenced homogenization of the sediments still showed some variation between the triplicate samples (Figure 7). Housatonic River sediment had the highest standard deviation of 62.7ppm (error bars in Figure 7) due mostly to one outlier sample that had 156.8 ppm versus the average 3.2 ppm. Otherwise the standard deviation would have been 2.2ppm. Hudson river followed with the second highest standard deviation of 14.3 ppm and New Bedford Harbor was last with 10.1 ppm. Overall, the sequenced mixing of the sediment lowered the sediment PCB variation.

• Different batch solvent experiments were run using the NBH sediment treated with NASA palladized magnesium (Mg/Pd; 99.9%/0.1% by weight; 4 mm Mg particle size). The solvents used included n-hexane (hex), dLimonene (dLim) and Citrus Burst 2 (CB2). The last two solvents were obtained from Florida Chemical Company, Inc. Both are derived from orange terpenes, which are a by-product from the orange juicing process, and CB2 also contains a plant-based surfactant. These products were selected to be tested with sediments for future implementation with an insitu PCB dechlorination technology because both are naturally derived and biodegradable. The hypothesis behind using biodegradable solvents with PCB dechlorination in sediments is that the solvents will aid in desorbing the PCBs from the sediment thus making the PCBs available for dechlorination with Mg/Pd and not be harmful if left in the environment. It is also hypothesized that CB2 will perform better than dLim at desorbing PCBs from sediment because the surfactant may facilitate desorption of the PCBs from the organic content in the sediment to the aqueous phase where it can react with the Mg/Pd.
• The above hypotheses were investigated by a series of experiments using NBH sediment treated with 10% Mg/Pd (1g) in addition to either water, dLim or CB2. All experiments included control samples that constituted of just sediment and were used as the theoretical initial average concentration of the treated samples for the percent dechlorination calculations. The samples with water were used to compare the dechlorination performance of the two citrus solvents. Each sample consisted of an individual beaker so that there was no subsampling only replication. All samples, except for the CB2 samples, were run for 2 and 16 hours. Sediments with water performed the best at an average dechlorination of 21.3% after 2 hours and 28.4% after 16 hours (Figure 8).
• dLim had the poorest performance with negative dechlorination values indicating no dechlorination. CB2 achieved 13.5% dechlorination after 16 hours. It was observed that the sediment in the samples with solvents dried out very quickly. This was problematic because in order for complete dechlorination to occur the sediment must be in an aqueous environment. It was hypothesized that the drying was consequence of the exothermic reaction between the Mg/Pd and the sediment, which caused the solvents to volatilize. This could explain why dLim had no dechlorination.
• In order to keep the solvents in the sediment the next experiment was conducted by adding either additional water or the initial solvent at intervals throughout the experiment. NBH sediment was first mixed with 10% Mg/Pd (1g) and then added 5mL of either deionized water, dLim or hexane. The water samples were used to evaluate dechlorination performance whereas the hexane samples were to compare the performance of the dLim as a solvent in extracting the PCBs from the sediment. Hexane is a much stronger solvent than dLim and thus it was hypothesized that it would perform better at dechlorinating the sediment. For each set of solvent samples half were added water and half their initial solvent throughout the experiment. Overall samples with water had the highest dechlorination at both 24 and 48 hours (Figure 9). Hexane performed well with additional water but showed no dechlorination with additional hexane. dLim had no dechlorination at 24 hours but performed better after 48 hours. This variation in results could be explained by that some samples had PCB concentrations greater than the average of the controls. For example the average PCB concentrations for the samples of hexane with additional hexane after 24 and 48 hours were 118.97 ppm and 120.99 ppm respectively, whereas the average concentration of controls was 115.28 ppm (Table 1). Since the percent dechlorination calculation uses the average of the controls as the theoretical initial concentration for all the treated samples, if a treated sample has a higher concentration this would yield a negative percent dechlorination.
• In the next solvent dechlorination experiment several changes were made in the structure of the experiment. This experiment consisted of treating NBH sediment with 10%Mg/Pd (1g), 10mL of water and 2.5 mL, 5.0 mL, 7.5 mL and 10.0 mL of dLim or hexane for 48 hours. Firstly, instead of only adding the solvent, dLim or hexane, to the sediment the various volumes of solvent were combined with 10.0 mL of water. In previous experiments samples with just solvent added always dried out very quickly whereas those with water remained as slurries throughout the experiment. The addition of water to the solvent helped keep the sediment moist enough for dechlorination to occur and this combination of solvent with water attained better dechlorination rates than experiments with solvent alone (Figure 10).
• Secondly, there was better mixing of the Mg/Pd with the sediment. In previous experiment the Mg/Pd was added directly to the sediment, manually mixed with a spatula and then added the solvent or water. The Mg/Pd powder would always fall on top of the sediment but manual mixing did not truly distribute the Mg/Pd throughout the. Here, Mg/Pd was first added to the solvent/water mixture, shaken on a rotary shaker and then all poured together onto the sediment and manually mixed. By adding the Mg/Pd to the solvent/water mixture and shaking it the Mg/Pd powder was better distributed within the liquids, which infiltrate better into the sediment particles and thus bring the Mg/Pd in contact with the PCBs. The liquid was able to disperse the Mg/Pd throughout sediment better that manual mixing of the Mg/Pd powder with the sediment alone.
• Lastly, each sample consisted of two beakers, a control and treated, in order to better estimate the true initial concentration of each sample. Rather than using the average of the control samples, each sample had its individual control with no added Mg/Pd, solvent or water. Due to the irregular distribution of the PCBs in sediments at times control samples had lower PCB concentrations that treated ones indicating that perhaps the treated samples had much higher initial concentrations than the controls. To resolve this problem instead of using the average of two or three controls as the estimated measurement of the initial PCB concentration of all the samples, each sample was composed of two 250mL beakers. Before treating the sediment with 10% Mg/Pd, 20 g of the sediment was placed in one beaker and homogenized manually with a spatula. Half the sediment was then placed in a second beaker and treated with the Mg/Pd-solvent-water mixture. The untreated beaker was used as the initial concentration for that particular sample and the treated beaker as the final PCB concentration in the sediment. This allowed to obtained a more accurate percent dechlorination for each sample as it was based on its individual initial concentration rather than an average of control samples.
• The best dechlorination performance was achieved by samples with just water and no solvent. Although in previous experiment samples with water always performed better they only achieved approximately 40% dechlorination after 48 hours versus an average of 75.6% achieved in this experiment (Figure 10). dLim dechlorination rates were greatly improved when compared to the negative dechlorination values from previous experiments. The 7.5 mL dLim with 10 mL water sample achieved a slightly higher dechlorination that the average water with 76.6%. When compared to the hexane samples dLim generally performed better a 2 mL and 5 mL of dLim, but slightly less at the 7.5 mL and 10.0 mL dLim (Figure 11).
• Overall, this experiment showed higher dechlorination than previous experiments with water or solvent. This may be attributed to adding water to the solvent, better distribution of the Mg/Pd within the sediment and using an individual control for each sample.

(4) GC/MS Analysis:

• A new Auto Sampler unit was installed on the instrument.
• A new GC column manufactured with a guard column was also installed. All 209 PCB congeners were identified on the new column and a new calibration curve was established.
• After the new GC column was installed the amount of sample put in the GC vial was lower from 1.0 mL to 0.4 mL due to that many samples were out of calibration.
• A total of 3 PCB congener calibration curves were established and edited during the reporting period in MS/MS mode. Biphenyl and several PCB congener calibration curves were also established in EI/MS mode.
• The calibration curve for congener-specific PCB analysis was rerun and reedited after major interventions in the GC/MS, following the procedures described by the manufacturer.
• Routine maintenance of the instrument was performed and run methods were checked and corrected if needed on a regular basis.
• Some smaller operational and maintenance issues were resolved and quality control samples were run.
• Routine analysis of samples described above was performed.

(5) Literature Review:
A continuous literature search was conducted on PCB dechlorination by Mg and Mg/Pd in various matrices, with special emphasis on the detected reaction end-products. Literature search was done on some potential by-product chemicals and on solubility issues as well. Accomplishments
Three calibration curves were established during the reporting period for the GC/MS for congener-specific PCB analysis in MS/MS mode including a new method for the newly installed GC column. Several PCB congener and biphenyl calibration curves were established in EI/MS mode as well. All QA/QC requirements were sufficiently met regarding the operation of the instrument. A total of 113 batches were run in pure solvent (10% methanol in distilled water or pure methanol) systems using purchased PCB or biphenyl standards that resulted in 149 samples run on the GC/MS. Replicate samples were taken from the batches as required.

Kinetic curves were established for Aroclor 1260 and biphenyl in pure solvent. Kinetic studies were also run for both congener 170 and congener 204 in pure solvent and chloride was measured in the same experiments. Progress has been made in understanding the dechlorination reaction processes, although the mass balance still could not be closed because of the degradation of biphenyl, potentially to a number of low-molecular-weight organic compounds undetectable by the GC/MS.

Dechlorination performance in sediments treated with Mg/Pd and additional water, hexane or dLimonene were improved. These were mostly due to a better distribution of the Mg/Pd within the sediment, adding the solvent mixed with water, and calculating percent dechlorination based on each sample’s individual initial concentration.

Difficulties
In GC/MS Operation: Periodical difficulties were faced with GC/MS operation, including too low response in chromatograms and faulting of the Auto Sampler. The manufacturer’s (Varian) service representatives came out to service the instrument several times but some of the problems still persisted. These downtimes in the instrument operation caused a delay in the analysis of the experimental samples.

Project Objectives for Next Reporting Period

Objectives

• Continue working on understanding the reaction processes involved in the catalytic dechlorination of PCBs in pure solvents and closing the mass balance of the system.
• Continue the search for potential biphenyl degradation/complexation by-product molecules (including ones not detectable by the GC/MS).
• Investigate the surface of the Mg/Pd powder after reaction with water and PCBs.
• Continue PCB degradation experiments in pure solvents to identify congener-specific degradation pathways, relative resistance of positional isomers, changes in dioxin-like toxicity, and the reaction endpoints.
• Continue experiments to determine the relative degradation potential of various PCB congeners by Mg/Pd.
• Continue solvent experiments to determine if solvents help improve dechlorination of PCBs in sediments with Mg/Pd by making PCBs more available for reaction.
• Use other solvents in PCB dechlorination in sediment and analyze which is best.
• Investigate whether solvents also evaporate when added together with water due to the exothermic reaction between the sediment and Mg/Pd, or if they remain in the sediment.
• Publish research articles.

Tasks to Meet Objectives

• Run kinetic and mass balance batch experiments in pure solvent systems with careful accounting for PCB dechlorination end products and measuring chloride.
• Compare the extraction efficiency of biphenyl in hexane, methylene chloride and toluene and look for potential by-products.
• Extract chloride from the filtered Mg/Pd material.
• Use the “x-ray diffraction” method to investigate the surface of the reacted Mg/Pd material.
• Conduct an experiment that allows the solvent/water mixture to soak the sediment for some time before starting dechlorination with Mg/Pd in order to for the solvent to desorb the PCBs from the sediment and make them available for dechlorination.
• Replicating the last experiment only using a 7.5 mL of CB2-water-Mg/Pd instead to see whether the surfactant in this solvent is better at desorbing the PCBs from the sediment and thus have higher dechlorination rates.
• Continue a literary research on solvents in order to better understand the mechanisms of the solvent in the sediment and as well as to search for other solvents that could be tested with dechlorination.
• Make sure that GC/MS operation is continuous and all QA/QC requirements are met.
• Conduct continuous literature search in major environmental research journals.
• Communicate and collaborate with other researchers and attend scientific conferences and workshops.

Anticipated Success in Meeting Project Objectives
It is anticipated that the above objectives will be successfully met over the next reporting period.

Overall Project Timeline Update
Project is anticipated to be completed by the original time proposed.

Preliminary Data
See Figures.

Expenditures
Expenditures have been in the range anticipated for the work accomplished to date.

References
Halle, B.R., K.M. Carvalho-Knighton, C.L. Geiger, and C.A. Clausen. 2005. Dechlorination of PCBs in Solution with Pd/Mg Bimetallic Systems. 229th American Chemical Society National Meeting, Extended Abstracts, Mar 2005, San Diego, CA. pp 544-547

Nomura, Y., S. Nakai, and M. Hosomi. 2005. Elucidation of Degradation Mechanisms of Dioxins during Mechanochemical Treatment. Environ. Sci. Technol., 39: 3799-3804