Determining the presence of organic compounds in foundry waste leachates.

Determining the Presence of Organic Compounds in Foundry Waste Leachates Research to determine the extent of contamination of groundwater by organic matter in ferrous foundry wastes is presented. The first of two reports on this significant work focuses on a laboratory study to identify the organic compounds that leach from ferrous foundry waste.

At the request of the American Foundrymen's Society, the Dept of Civil and Environmental Engineering at the Univ of Wisconsin-Madison undertook a research program to determine the potential for and extent of contamination of groundwater adjacent to foundry waste landfills by organic matter arising from the wastes.

The original impetus for this project and a recently completed similar project emphasizing inorganic constituents[1] were to justify to the U.S. Environmental Protection Agency (EPA) exemption from hazardous waste regulations by delisting. This would justify the development of special landfill design requirements for foundry process solid wastes based on the specific characteristics of the wastes.

At the conclusion of the inorganics study, EPA appeared to be satisfied that the leaching of inorganic substances warranted the development of mono-landfill regulations; however, there was concern that the leaching of organic constituents had never been evaluated. This study was designed to provide such an evaluation.

The program was divided into two parts. The laboratory study[3] described this month identified organic contaminants that may be released to the environment from ferrous foundry process wastes after disposal in a landfill. The second part was a field study[4] to monitor groundwater quality adjacent to foundry waste landfills to determine whether any organic contamination of groundwater could be attributed to leaching of organics from the waste. Details of the field study, as well as discussion of the results of both studies, will be the focus of next month's article.

Organics in Foundry Binders

Most core binders used in iron foundries are synthetic organic polymers. The binders used in molding sands are also in some cases synthetic. However, the most common molding sand is still green sand, which uses clay as its binder. Possible contamination of the waste sands from the organic binders could come from unreacted materials, catalysts, solvents, soluble polymers or decomposition products.

The organic resins used in foundries are varied and constantly changing. The exact mixture of a binder system, including the resin, catalyst, solvent, etc, is often proprietary information. Therefore, descriptions of binder systems in these articles will be general and may be out of date. The following description of resins and binders is from a report on organic sand binder chemistry.[2]

The first organic binders used in foundries were the core oils. These were drying oils (unsaturated fats) that react with oxygen to form a film much like the skin that forms on paint. Heat is usually applied to speed up the reaction.

Other oxygen sources are sometimes added to reduce the baking time. These may be perborates, percarbonates, permanganates or peroxides. Other resins can also be added to improve the rigidity; and solvents such as turpentine, kerosene or mineral spirits are usually needed to improve flowability.

Other binders are based on resins formed by reactions of formaldehyde, phenol, isocyanates, furans or combinations of these. In addition to these basic chemicals, catalysts, releasing agents, cross-linking agents, solvents and additives to adjust flowing and wetability characteristics are added. Thus, a wide variety of organic compounds is necessary for manufacturing and using synthetic organic binders.

Laboratory Study Methods

To best serve the ferrous foundry industry, it was necessary to test wastes that are representative of the majority of foundries. AFS identified nine core binder systems as having the broadest usage in the metalcasting industry. These are: phenol formaldehyde, phenolic urethane, furan hotbox, furan nobake, phenolic ester, core oil, phenolic isocyanate, alkyd isocyanate and furan warmbox.

Analysis of the material flow at a typical foundry suggests that excess system sand, core butts and coreroom sweepings can be expected to contain most of the organics that originate in the binders. Earlier mass balances[5] indicate that these fractions account for about 70% of the materials going to the landfill from the typical foundry.

All the foundries sampled are gray or ductile iron foundries except a steel foundry using the phenol formaldehyde system. Two foundries were sampled twice to help identify the extent of sampling error.

The binder system used to identify the sample refers to the binder used for core-making. Various binders also are used for the system or molding sand.

Sampling at the foundries was performed by foundry personnel. The point of sampling in the foundry could influence the results.

For initial testing, the samples were composited in the rations suggested by Santa Maria's work.[5] This resulted in samples consisting of 82% excess system sand, 14% core butts and 4% core room sweepings. EPA's proposed Toxicity Characteristic Leaching Procedure (TCLP) was chosen to leach the samples because it is likely to be adopted by the EPA for determining the toxicity characteristic and allows the sample to be leached without the loss of volatile organics.

For leaching and analysis, the organic compounds are divided into volatile and semi-volatile compounds. Volatile compounds, which include most solvents and dissolved gases, can be removed from water by heating or by purging with an inert gas stream.

Semi-volatile compounds include large molecules such as the polycyclic aromatic hydrocarbons as well as polar molecules which are not easily purged from water, such as pyridine, phenols and phthalates. These can be extracted by an organic solvent and are not lost when the extract is concentrated, but can be volatized at the temperatures of the gas chromatograph column.

Upon receipt at the university, samples were transferred to Teflon capped one-quart canning jars and stored at 4C (39.2F). One set of samples was composited at the required ratios and extracted for volatile organics within a week of their receipt.

No special technique was employed to obtain representative samples because exposure to air during riffling, for example, would result in a loss of volatile compounds.

The TCLP requires a particle size for extraction of 9.5 mm or less. When necessary, core butts and so forth were rubbed together prior to weighing to produce sand-sized particles. A Millipore Zero Headspace Extractor (ZHE) was used to perform the extractions in accordance with TCLP methodology.

Leachates were collected in a 50 ml gastight syringe and transferred to 40 ml glass screw cap VOA vials with Teflon-faced silicone septa for storage. At least four vials of each leachate were collected. One vial was delivered within two days to the Wisconsin State Laboratory of Hygiene for gas chromatography/mass spectroscopy (GC/MS) analysis. The State Laboratory of Hygiene uses a headspace method for the GC/MS analysis of volatile organic compounds. The remaining vials were stored at 4C for later quantitation.

A second set of foundry waste samples was prepared for leaching of semi-volatile compounds. Since there was no need to be concerned with the loss of volatiles from these samples, a riffler was used to obtain representative samples of each component.

The remaining composite sample was weighed and transferred to the extraction vessel and a volume of extraction fluid (in milliliters) equal to 20 times the mass of the sample (in grams) was added. The extraction vessels were one-gallon glass bottles with Teflon-lined caps.

After leaching and filtering through a glass fiber filter, one liter of each leachate was extracted with methylene chloride according to EPA method 3510. These extractions resulted in a base/neutral fraction and an acid fraction of the semi-volatile compounds. Both fractions were analyzed by GC/MS and stored at 4C for later quantitation.

Before extraction, the leachates were spiked with known concentrations of surrogate standards, which are later used to determine extraction efficiency.

GC/MS analysis uses a gas chromatograph to separate compounds, after which the effluent is ionized by electron bombardment and the ions are separated by mass and counted. This results in a mass spectrum for the compound, which can be compared to a library of spectra for known compounds. The State Laboratory of Hygiene library contains some 45,000 compounds from National Bureau of Standard files, priority pollutant files and files developed within the laboratory.

Of the compounds identified by GC/MS analysis, those available as pure compounds or ready-made standards were run as calibration standards on a capillary column gas chromatograph with a flame ionization detector (GC/FID). The analysis for volatile compounds used a purge and trap method for sample concentration. The analysis for semi-volatiles used direct injection of the methylene chloride extracts.

Quality Control

Quality control measures were undertaken to establish calibration curves, determine the method quantitation limits, identify sources of error and to determine the reliability of the results. Calibration curves for the GC/FID analysis were based on peak area and were determined from a minimum of five runs of each standard at different concentrations.

Method quantitation limits were determined from replicate analysis of the sample 2B leachate. The quantitation limit is defined as five times the standard deviation of the replicate analysis in accordance with EPA procedures. Seven replicates of Sample 2B leachate were analyzed on the GC/FID. The standard deviation of the concentrations for each compound present was calculated and used to calculate the quantitation limit.

This led to quantitation limits which were in most cases very low--lower, in fact, than was practical or necessary. Therefore, a minimum identifiable peak area was chosen. The concentration corresponding to this area was calculated for each compound. When this concentration was higher than the previously determined quantitation limit, the new value became the limit. The quantitation limits for all compounds are presented with the results for each sample in the results section.

The standard deviation can also be used to quantify the precision of the measurements. Relative standard deviations for the seven runs varied from less than 1.0% for m/p-xylene to about 30% for methyl isobutyl ketone, benzene and acetone. The relative standard deviation for all other compounds was in the range of 1-10%.

Using the assumption that the errors are normally distributed about the mean and that the relative standard deviation is constant, the 95% confidence interval for a single measurements is [+ or -] 1.96 times the relative standard deviation, or [+ or -] 2-20%.

To determine the effect of the extractions on the precision of the results, replicate extractions must be performed. Matrix effects are usually quantified by spiking samples with a known concentration of a compound and analyzing the sample. EPA has established control limits for the recovery of surrogate spikes from various sample matrices. Generally, recoveries are within these limits.

Up to this point, only errors introduced after the samples were leached have been considered. However, sampling errors also could have occurred when a portion was taken for analysis or when the samples were collected at the foundry. To evaluate error introduced by sample selection, leaching and analysis replicates were again used. The entire leaching procedure for volatiles was carried out four times on composites of Sample Six. The relative standard deviations ranged from about 4% to almost 30%; however, most are below 20%.

Results

The list of analytes for each fraction (volatile, semi-volatile acid and semi-volatile base/neutral) are all the compounds tentatively identified in at least one of the samples. Those compounds in the 45,000 compound library that were not found in any samples are not listed. The sample numbers represent the different binder systems.

When a calibration standard was available for the compound, an A or a B will be present. An A means that the compound also was found in the GC/FID analysis above the compound's quantitation limit. A B means it was below the quantitation limit of the GC/FID for that compound.

For those compounds where no calibration standard was run, a C or a D will be present when the compound was identifiable by GC/MS and an E when it was not identified. A C means that the concentration of the compound is at or near the detection limit of the GC/MS system. A D indicates that the concentration of the compound is well above the detection limit and the concentration was estimated from the GC/MS data. An entry of E occurs when the compound was not identified in that sample by GC/MS and no calibration standard was run.

Several inconsistencies exist in the data which need to be explained. In some cases, a compound will be identified by GC/MS but not found above the quantitation limit in the GC/FID or vice versa. The first case results in a symbol combination of Y,B; the second N,A.

The most likely reason for such results is that GC/MS is more sensitive to this compound than GC/FID or vice versa. Other reasons for such inconsistencies include misidentification from the mass sprectral data, interferences and coeluting compounds. Finally, for binder systems 1 and 2, two samples were collected; however, only one was subject to the GC/MS analysis.

In summary, the compounds and samples of most concern are those identified by Y,A or Y,D, which mean that the compound was found in that sample at a measurable concentration. Conversely, compounds identified with N,E or N,B were not detected in that sample by either method of analysis.

Compounds identified by Y,C were detected but at very low concentration in that sample. The symbols N,A and Y,B are contradictory as explained above; however, if the compound is present in that sample, it is likely that is at a low concentration.

Quantitation limits are given for each compound for which calibration standards were run by GC/FID. Whenever two composites were run for the same sample, one was assigned A and the other B. Only composite A was analyzed by GC/MS. When a composite was found to contain either many compounds and/or a high concentration of some compounds, the individual waste components also were leached and analyzed.

Generally, laboratory results indicate that although a wide variety of organic compounds was detected, most were present at low concentrations. The core oil and phenolic urethane systems leached the largest number of compounds and the highest concentrations. None of the samples leached organic compounds at concentrations at or higher than the proposed regulatory levels of the TCLP.

Next Month

Complete discussion of laboratory study results, description of the field study, discussion of field data and conclusions will be presented in the conclusion of this timely work. [References] [1]F. J. Blaha, R. K. Ham, W. C. Boyle, T. J. Kunes, D. G. Nichols, R. R. Stanforth, "Leachate and Groundwater Quality In and Around Ferrous Foundry Landfills and Comparisons to Leach Test Results," American Foundrymen's Society, Des Plaines, IL (1985). [2]W. D. Scott, C. E. Feazel, "A Review of Organic Sand Binder Chemistry," modern casting Tech Report No. 765 (Jun 1976). [3]E. C. Engroff, R. L. Fero, R. K. Ham, W. C. Boyle, "Laboratory Leaching of Organics From Ferrous Foundry Process Wastes," American Foundrymen's Society, Des Plaines, IL (1989). [4]R. L. Fero, R. K. Ham, W. C. Boyle, "An Investigation of Groundwater Contamination by Organic Compounds Leached from Iron Foundry Solid Wastes," American Foundrymen's Society, Des Plaines, IL (1986). [5]C. Santa Maria, "A Study of Foundry Waste Materials," M.S. Thesis, Dept of Metallurgical and Mineral Engineering, Univ of Wisconsin-Madison (1974).