K. Thomas Klasson, Debra T. Bostick, Paul A. Taylor, and Linda L. Farr
Oak Ridge National Laboratory
Oak Ridge, Tennessee 37831-6044
(423) 574-2210
ABSTRACT
The removal of mercury from slip streams of water from a creek and from a wastewater treatment plant has been studied in the field. Two types of column studies were performed with different mercury sorbents. A short-term study that yielded information about the kinetic behavior of different sorbents in a column configuration and a long-term study that is still under way were used to obtain information about long-term performance and treatment capacity. The results to date indicate that sorbents containing thiol functional groups gave the best performance in both the short- and long-term studies. The results also indicate that water pretreatment may be an effective option to ensure that the mercury is free and oxidized. This approach may be a necessity when target levels of ng/L are desired.
INTRODUCTION
Mercury removal from water to trace levels is being tested at two U.S. Department of Energy (DOE) sites: the Oak Ridge National Laboratory (ORNL) and the Oak Ridge Y-12 Plant. Many sorbents have been developed for removing mercury and heavy metals from water (1); however, most of the data published thus far do not address the extremely low mercury concentrations represented in this project. In this work we are focusing on removal of mercury from microgram-per-liter (parts per billion) levels to low nanogram-per-liter (parts per trillion) levels.
The Oak Ridge Y-12 Plant has several large water streams with low concentrations of mercury. The Upper East Fork Poplar Creek (UEFPC) Characterization Area at the Oak Ridge Y-12 Plant is regulated by a National Pollutant Discharge Elimination System (NPDES) permit from the state of Tennessee. The flow, which is 1500 gal/min at the upper portion of the creek, contains levels of mercury in the range of 1000 ng/L.
The limit for mercury in the effluent from the Nonradiological Wastewater Treatment Plant (NRWTP) at ORNL was reduced to 19 ng/L for the monthly average by a 1997 NPDES permit, which is well below the current effluent concentration. This limit is being appealed, and is currently not being enforced; however, experimental work is being done to determine the feasibility of meeting such a low limit. The NRWTP flowsheet includes filtration using dual-media columns, an air stripper, and granular activated carbon (GAC) columns. The activated carbon columns at the NRWTP provide most of the mercury removal. The water flow rate at the NRWTP is typically 350 gal/min, and the mercury concentration prior to the GAC columns is about 200 ng/L, but fluctuates widely.
Mercury has been removed in batch experiments from contaminated water from both locations via adsorption onto Ionac SR-4 (Sybron Chemicals, Inc.), Keyle:X (SolmeteX), and Mersorb (Nucon International, Inc.) resins to levels below the target goal of 12-19 ng/L (2). Sorbents with thiouronium, thiol, amine, sulfur, and proprietary functional groups were selected for column studies.
METHODS
Several commercial macroporous polystyrene/divinylbenzene resins with (iso)thiouronium functional groups are available for mercury adsorption. These types of sorbents have been found by Sybron Chemicals, Inc., to bind ionic mercury(II), methyl mercury, and elemental mercury, but cannot be regenerated. They perform best at pH values between 0 and 6. At higher pH levels, the thiouronium group is converted to a thiol group. Examples of such sorbents are Ionac SR-3 (Sybron Chemicals, Inc.), SIR-400 (ResinTech, Inc.), and Purolite S-920 (Purolite Company).
Macroporous, weakly acidic, polystyrene/divinylbenzene, cation resins with thiol functional groups are also commercially available for mercury adsorption. The sorbents can be used at pH values ranging from 1 to 14. Materials with this type of functionality are Duolite GT-73 (Rohm and Haas), Ionac SR-4, SIR-200 (ResinTech, Inc.), and SAMMS (Self-Assembled Mercaptan on Mesoporous Silica), a new inorganic thiol resin developed by Pacific Northwest National Laboratory (PNNL). PNNL has presented data indicating that elemental and organic mercury may be adsorbed by SAMMS (3). In general, the thiol resins can sorb only mercury(II) ions.
Keyle:X is a polystyrene/divinylbenzene resin coated with a proprietary polymer with sulfur-based functional groups, making it mercury specific. The manufacturer recommends that the water be pretreated with chlorine, hypochlorite, or chlorine dioxide to 1-2 mg/L to ensure that the mercury is in the ionic state. Other polystyrene/divinylbenzene resin manufacturers specifically recommend removing these types of oxidizers prior to contact.
Mersorb LW, Mersorb 3, and Mersorb 1.5 mm are commercial carbons impregnated with sulfur. The only difference between the three carbons is particle size. Since the sorption is dependent on precipitation of the sulfide, competing metal ions are those that form strong sulfide bonds.
Noble metal-impregnated carbon (e.g., C-E100-Alpha1Mix by ADA Technologies, Inc.) is a carbon-based sorbent infused with noble metal(s). According to the manufacturer, the sorbent can be regenerated by heating to 370°C.
The Forager Sponge (Dynaphore, Inc.) sorbent is an open-cell cellulose sponge incorporating an amine-containing polymer that has selectivity for both cationic and anionic species of mercury as well as other heavy metals. It has the advantage of being a very porous compactible sponge; thus, column capacity can easily be adjusted by altering the packing density. The sponge can be pretreated with acid or base to change functionality (anionic versus cationic).
Granular activated carbon (GAC) is used at the NRWTP for removal of mercury and semivolatile organics. It is also used in two mercury treatment systems located at the Y-12 Plant for treatment of streams collected from building sumps. Filtrasorb 300 (Calgon Carbon Corporation) was chosen as a base sorbent for comparison purposes.
The mercury test system operated at ORNL and the Y-12 Plant consisted of a 1-µm (nominal pore size) filter cartridge, a set of flow meters, and a corresponding set of columns. Each of the columns had an inside diameter of 25 mm and was packed with 70 mL of sorbent to a sorbent of approximately 125 mm. A constant-pressure water-supply was available at both sites -- at ORNL as part of the wastewater treatment plant (just before existing GAC columns), and at the Y-12 Plant via a submersible pump in the creek.
Two types of studies were performed, the first of which was a short-term study in which the experiment was initially at a flow rate of about 7 mL/min [0.1 bed volume (BV)/min]. Two samples were taken (4 h apart) after 20 h, and the flow rate was then increased to a new setting. This sampling procedure was repeated, and the flow rate was again increased. After samples had been collected at the highest flow rate, the flow rate was decreased to 7 mL/min and samples were collected again. The long-term studies utilized the same column setup as in the short-term studies, but a single flow rate was studied: 0.3 BV/min at NRWTP and 1 BV/min at UEFPC. Data are still being obtained from these experiments.
Samples were collected in clean 125-mL I-Chem bottles prior to refrigeration, shipment, and analyses. Atomic fluorescence (4) was used to measure total mercury concentrations in the samples at an off-site laboratory (Frontier Geosciences, Inc., Seattle, WA). Unless noted, all samples were collected unfiltered.
RESULTS
The results of the short-term studies conducted at the UEFPC are shown in Figs. 1 and 2. The materials SIR-200, Keyle:X, and Duolite GT-73 were found to give the best performance for removal of low-level mercury. A large deviation in the measured mercury concentration in the column inlet indicates a wide variability in mercury levels. Less than 40% of the measured mercury was attached to particulates larger than 1 µm (data not shown). None of the tested sorbents reduced the mercury concentration to less than the targeted 12-ng/L NPDES limit. In these short tests, SIR-200, Keyle:X, and GT-73 were able to reduce the mercury concentration to less than the proposed 51-ng/L NPDES limit for water flow rates of less than or equal to 1 BV/min. Detailed results of the short-term column study at UEFPC have been published elsewhere (5).
Fig. 1. Results of the first set of short-term column studies at UEFPC. Granular activated carbon was used to compare performance with baseline technology. Solid lines indicate empirical data trends. Data collection was done over a week. The Forager Sponge was included in the study, but performed very poorly (data not shown).
Fig. 2. Results of the second set of column studies. Granular activated carbon was used to compare performance with baseline technology. Solid lines indicate empirical data trends. Data collection was done over a week.
The short-term column studies conducted at NRWTP initially showed excellent mercury removal by the Forager Sponge even at high flow rates. Later column testing showed that the mercury removal efficiency of the sorbents varied considerably over time, probably associated with changes in the speciation of the mercury in the wastewater. Samples of the NRWTP wastewater were analyzed for chemically active (ionic and elemental) mercury to estimate the fraction of the mercury that was easily accessible to the sorbents. Samples were taken at different points in the treatment process at the NRWTP, and some samples were filtered through 0.45-µm- and 0.02-µm-pore-size filters. The 0.02-µm filter will remove colloidal particles. For the unfiltered samples, the chemically active mercury was only about 30 to 45% of the total mercury. For one sample, the total mercury concentration was 194 ng/L unfiltered, 119 ng/L after a 0.45-µm filter, and 46 ng/L after a 0.02-µm filter. The active mercury in the same samples was 56, 53, and 29 ng/L (29, 44, and 63%), respectively. These results suggest that a significant fraction of the mercury in the wastewater is associated with particulates. It is likely that the change in performance of the sorbents at different times was associated with changes in the fraction of total mercury that was in a chemically active form. Chlorination of the water, which could convert the bound mercury to an ionic form, improved the performance of some of the sorbents. Overall, the Keyle:X resin performed the best with chlorinated water, followed by SIR-200 in the short-term studies at NRWTP (data not shown).
Based on the results from the short-term column studies, two thiol-based sorbents (SIR-200 and Keyle:X) were selected for use in the long-term studies. In addition to the previously tested sorbents, SAMMS was included in the study. The results of the long-term column studies conducted at UEFPC and NRWTP are shown in Figs. 3 and 4, respectively. Two sorbents, Keyle:X and SIR-200, have performed well at the UEFPC for several months. As of December 1998, 175,000 to 200,000 BV (3,500 gal) have passed through each of the columns. A Keyle:X sorbent bed, processing water that is pretreated with hypochlorite followed by SAMMS, has performed the best at NRWTP; 50,000 to 62,000 BV (1000 gal) have passed through each of the columns.
Fig. 3. Effluent concentrations from the columns used in the long-term study at UEFPC. The flow rate is approximately 1 BV/min, and the influent concentration to the columns has been in the range of 325-715 ng/L after the prefilter.
Fig. 4. Effluent concentrations from the columns used in the long-term study at NRWTP. The flow rate is approximately 0.3 BV/min, and the influent concentration to the columns has been in the range of 40-625 ng/L after the prefilter.
In conclusion, thiol-based resins have shown the best performance when removing low concentrations of mercury in water at two DOE facilities in field experiments. Effluent concentrations below 19 ng/L have been obtained at NRWTP when using hypochlorite-treated water and a Keyle:X resin. Other resins (SIR-200 and SAMMS) have performed well, and filtration tests have shown that a significant portion of mercury at both test locations is particle bound or associated with colloidal matter. Future studies will focus on lower flow rates and, possibly, water pretreatment at UEFPC.
ACKNOWLEDGMENTS
Funding for this work was provided by the DOE Mixed Waste Focus Area and by local waste management and environmental restoration organizations. ORNL is managed by Lockheed Martin Energy Research Corp. for the U.S. Department of Energy under contract DE-AC05-96OR22464.
REFERENCES