Modeling The Desorption Of Antimony And Arsenic Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

This proposed semester project focuses on modeling the desorption of antimony (Sb) and arsenic (As) from contaminated streambed sediment using the computer program PHREEQC. This study is environmentally significant in that it will provide a tool which can be used by researchers and scientists to predict the desorption of these metals from contaminated sediment, thus giving insight into the fate and transport of these metals in the river systems. The modeling methodology will be to write the input file, which will model metal desorption based on the generalized two-layer model (GTLM) of Dzombak and Morel (1990), assuming that hydrous ferric oxides (HFO) are the dominant sorbent in the contaminated sediment. Key model input parameters will be estimated by experimental determination and/or taken from literature values, and will be applied in the desorption model. Model simulations will be compared to experimental desorption data generated by Baeza (2009) using the Fisher's exact test. The results obtained from the model will be aqueous concentrations of Sb and As in solution (mol/kg water) at specific pH values. These will be converted into normalized aqueous concentrations using Microsoft Excel in order to compare experimental and modeling results. It is expected that the model will accurately simulate batch equilibrium desorption data and that experimentally determined input parameters will yield improved model simulation of the desorption data.

Manadas Creek is an urban tributary of the Rio Grande, located near Laredo, Texas, in a highly industrialized area. An Sb smelter, which was decommissioned in 1999, is located on the bank of this creek, approximately 2 km upstream from the confluence of Manadas Creek and the Rio Grande. Recent studies (Baeza 2009; Baeza et al. 2010; 2011) have shown that at sites located in close proximity to the decommissioned Sb smelter, stream water and sediment are significantly contaminated with Sb and As. Antimony concentrations as high as 479 μg/L in stream water and 1103.2 mg/kg of dry sediment in streambed sediment, and arsenic concentrations as high as 22.46 μg/L in stream water and 14.2 mg/kg of dry sediment in streambed sediment have been observed (Baeza et al. 2010; 2011). Stream water and streambed sediment concentrations have consistently exceeded federal regulatory limits for human health consumption of water and organisms and soil screening levels (USEPA 2005; 2006).

The solubility and sorption kinetics of Sb and As are significantly influenced by changes in water chemistry, including changes in pH and ionic strength. In Manadas Creek, fluctuations in these parameters may occur as a result of rainfall events, which cause increased stream volume, changes in stream water composition, and changes in other hydrodynamic processes such as sediment resuspension, due to the addition of surface runoff to the creek. These types of changes in stream conditions can lead to the release, or desorption, of Sb and As from the sediment to the overlaying water. This release of Sb and As can adversely affect water quality of Manadas Creek and the Rio Grande, the latter of which serves as a primary drinking water source for 98% of the American and Mexican populations from Laredo, TX to the mouth of the Rio Grande (USIBWC 1994). In addition, Sb and As release from sediment will continue to compromise the health of the ecosystem, as both metals have been shown to have negative effects on aquatic flora and fauna. Developing an accurate model to simulate Sb and As desorption from contaminated sediment is necessary because it will provide researchers and scientists with a predictive tool to assess the effects of pH and ionic strength on Sb and As release.

2.2 Literature review

Several researchers have modeled As and Sb sorption/desorption from sediment, with less research conducted on Sb than As (Lumdson et al. 2001; Johnson et al. 2005; Frau et al. 2008). Most studies have modeled sorption/desorption based on the GTLM of Dzombak and Morel (1990), and have assumed that HFO are the dominant adsorbent.

Frau et al. (2008) conducted desorption experiments on three solid, natural samples collected from an inactive mine area in the presence of 6 different background electrolytes. Through XRD and chemical extraction analyses, they determined that As was mainly associated with the Fe fraction in the samples. The researchers then modeled desorption of arsenate from the solid samples in PHREEQC, using ferrihydrite as the main As-bearing phase. Desorption was modeled with the diffuse double layer model of Dzombak and Morel (1990), also known as the GTLM. Researchers were able to simulate As desorption fairly well, but had to modify the HFO site density for one of the three samples. Researchers indicated that this was a reasonable modification considering that some As may have been incorporated in the HFO structure through co-precipitation or bulk absorption. Model simulations were not accurate when carbonate surface complexations were not considered, which demonstrated that these complexations played an important role in As desorption.

Lumsdon et al. (2001) investigated the use of solid phase characterization by XRD and chemical extraction techniques for assessing As behavior in contaminated soils. Researchers then modeled As behavior by modeling the two basic processes of As solubility and adsorption/ desorption, individually. Adsorption/desorption of As to solid phases was modeled, using the GTLM and equilibrium constants of Dzombak and Morel (1990). It was assumed, based on solid phase characterization, that HFO was the dominant adsorbent of As in the soil samples. Adsorption/ desorption modeling results suggested that As would become increasingly mobile under increasing alkaline pH conditions, especially at pH values above 8. Arsenic desorption was sensitive to the ratio of As concentration to surface adsorption sites. Modeling simulations indicated that As adsorption would be greatest at low pH values.

Johnson et al. (2005) modeled Sb(V) desorption from contaminated soil samples using the computer program MQV40TIT. In their modeling, they assumed that HFO was the main Sb adsorbent. They employed surface complexation constants from Dzombrak and Morel (1990), and made further assumptions concerning HFO properties (Johnson et al. 2005). The desorption model results underestimated experimental aqueous Sb concentrations in the pH range from 4 to 6, and overestimated Sb concentrations in the pH range from 7 to 10. The researchers asserted that due to the numerous assumptions made concerning HFO properties, the model and experimental desorption curves were in good agreement (Johnson et al. 2005).

3. Research objectives and hypothesis

The research objectives are to 1.) model experimental batch data of Sb and As release from contaminated sediment using PHREEQC, considering the effects of pH and ionic strength, and 2.) estimate and apply key model input parameters. The hypotheses are: 1.) Experimental desorption results will be accurately simulated using PHREEQC, showing similar effects of pH and ionic strength on metal desorption, and 2.) model input parameters determined experimentally or based on literature values will yield accurate model simulations of experimental data.

4. Methodology

Previous researchers (Baeza 2009) have conducted batch experiments to measure the release of As and Sb from contaminated streambed sediment collected from Manadas Creek. In these experiments, a small mass (0.275 g) of contaminated sediment (previously analyzed for initital Sb and As concentrations) and solution (50 mL) at particular conditions of pH (pH 6 - 8) and ionic strength (I, from NaCl = 0.01 or 0.1) were added to polypropylene test tubes (Baeza 2009). Aqueous samples were taken at 0 and 24 hours, and analyzed for total Sb and As concentrations via inductively coupled plasma-mass spectrometry (ICP-MS) (Baeza 2009).

In this study, these equilibrium batch experiment results of Baeza 2009 will be simulated using PHREEQC, considering the effects of pH and ionic strength on metal release from contaminated streambed sediment. Two databases (MINTEQ.V4 database for Sb, and WATEQ4F database for As) will be used for the desorption modeling. Surface complexation modeling in PHREEQC will be accomplished by applying the GTLM of Dzombak and Morel (1990). To model Sb and As desorption, the input file will be coded, and will include a list of HFO surface species, HFO surface parameters, concentrations of elements present in solution, solution pH, and coding which will tell the model to output aqueous equilibrium concentrations at pH values ranging from 3 to 10.

Key model input parameters include all element concentrations, chemical reactions, chemical reaction log K values, and surface parameters including mass of surface material, sorbent composition, area per mass of surface material, type of binding sites (strong and weak), and surface site density. Some of these parameters are experimental values taken from the batch experiments, including elemental concentrations and mass of surface material. Previously conducted sequential extraction experiments have shown that Sb and As are mainly associated with the iron (Fe) oxide fraction of the sediment (Ruiz, unpublished data). As such, HFO will be applied as the dominant sorbent material in model simulations, which has also been assumed by previous researchers (Lumdson et al. 2001; Johnson et al. 2005; Frau et al. 2008). Pertinent reactions will be determined by considering the possible reactions based on sediment elemental concentrations and solution composition. Log K values found in PHREEQC databases for considered reactions will be applied and fitted, if necessary. The area per mass of surface material will be determined experimentally, by analyzing sediment using a Nova 2200e high speed surface area and pore size analyzer (Quantachrome Instruments). Surface site density will be determined via acid/base titration of sediment, following the method of Wang et al. (1997).

5. Expected results

Modeling results generated in this study will be the aqueous concentrations of Sb and As in solution (mol/kg water) at specific pH values. These will be converted into normalized aqueous concentrations using Microsoft Excel in order to compare experimental and modeling results. It is expected that PHREEQC model simulations will show desorption trends similar to experimental data, with increased desorption at low and high pH values, lower desorption at circumneutral pH values, and somewhat higher desorption across the pH range of 3 to 10 at higher ionic strength. It is also expected that experimentally determined input parameters (i.e. surface area and surface site density) will yield accurate model simulations of experimental data.

6. Assessment

To compare the experimental data and model simulations, the Fisher's exact test will be applied, using either Microsoft Excel or MatLab as the software platform. This test will convert the differences between the experimental and simulated data sets into a probability of whether or not the two data sets occur by chance. This test is chosen because the small sample size, due to limited experimental data (3 data points per ionic strength). In addition, the Fisher's exact test gives the exact P value, as compared to the chi square test, which gives an approximate P value. Based on the P value, the null hypothesis will either be accepted or rejected.

7. Timeline

Figure 7.1 shows the timeline for this project. Project proposal will be written March 1 - 7. During March 7 - 21, model input parameters will be determined, and model input file will be written. Model simulations will be conducted March 28 - April 8. Results will be finalized, final report will be written, and final project presentation will be prepared throughout the rest of April. Final project presentation will be on April 25 and final project report will be turned in on May 9.

Figure 7.1. Timeline for semester project.