Arsenic Research Partnership
Adsorbent Treatment Technologies for Arsenic Removal
Subject Area: High-Quality Water
Adsorbent Treatment Technologies for Arsenic Removal
©2005 AwwaRF. All rights reserved.
About the Awwa Research Foundation The Awwa Research Foundation (AwwaRF) is a member-supported, international, nonprofit organization that sponsors research to enable water utilities, public health agencies, and other professionals to provide safe and affordable drinking water to consumers. The Foundation’s mission is to advance the science of water to improve the quality of life. To achieve this mission, the Foundation sponsors studies on all aspects of drinking water, including supply and resources, treatment, monitoring and analysis, distribution, management, and health effects. Funding for research is provided primarily by subscription payments from approximately 1,000 utilities, consulting firms, and manufacturers in North America and abroad. Additional funding comes from collaborative partnerships with other national and international organizations, allowing for resources to be leveraged, expertise to be shared, and broad-based knowledge to be developed and disseminated. Government funding serves as a third source of research dollars. From its headquarters in Denver, Colorado, the Foundation’s staff directs and supports the efforts of more than 800 volunteers who serve on the board of trustees and various committees. These volunteers represent many facets of the water industry, and contribute their expertise to select and monitor research studies that benefit the entire drinking water community. The results of research are disseminated through a number of channels, including reports, the Web site, conferences, and periodicals. For subscribers, the Foundation serves as a cooperative program in which water suppliers unite to pool their resources. By applying Foundation research findings, these water suppliers can save substantial costs and stay on the leading edge of drinking water science and technology. Since its inception, AwwaRF has supplied the water community with more than $300 million in applied research. More information about the Foundation and how to become a subscriber is available on the Web at www.awwarf.org.
©2005 AwwaRF. All rights reserved.
Adsorbent Treatment Technologies for Arsenic Removal Prepared by: Gary Amy, Hsiao-wen Chen, and Aleksandra Drizo University of Colorado (CU), Boulder, Colorado, United States Urs von Gunten Swiss Federal Institute of Aquatic Science and Technology (EAWAG), Dubendorf, Switzerland Phil Brandhuber and Ruth Hund McGuire Environmental Consultants (MEC), Denver, Colorado, United States Zaid Chowdhury, Sunil Kommineni, and Shahnawaz Sinha Malcolm Pirnie Inc. (MPI), Phoenix, Arizona, United States Martin Jekel Technical University of Berlin (TUB), Berlin, Germany and Kashi Banerjee Vivendi Water/U.S. Filter (USF), Pittsburgh, Pennsylvania, United States Sponsored by: Arsenic Research Partnership Jointly funded by: Awwa Research Foundation 6666 West Quincy Avenue, Denver, CO 80235-3098 and United States Environmental Protection Agency Washington, D.C., 20460 Published by:
and
©2005 AwwaRF. All rights reserved.
DISCLAIMER This study was sponsored by the Arsenic Research Partnership. The Arsenic Research Partnership consisted of the Awwa Research Foundation (AwwaRF), the U.S. Environmental Protection Agency (USEPA), and the Association of California Water Agencies (ACWA). The study was jointly funded by AwwaRF and USEPA under Cooperative Agreement No. CR 828216-01. AwwaRF, USEPA, and ACWA assume no responsibility for the content of the research study reported in this publication or for the opinions or statements of fact expressed in the report. The mention of trade names for commercial products does not represent or imply the approval or endorsement of AwwaRF, USEPA, or ACWA. This report is presented solely for informational purposes.
Copyright © 2005 by Awwa Research Foundation All Rights Reserved Printed in the U.S.A. ISBN 1-58321-399-6
©2005 AwwaRF. All rights reserved.
CONTENTS LIST OF TABLES....................................................................................................................
vii
LIST OF FIGURES ..................................................................................................................
ix
FOREWORD ............................................................................................................................ xiii ACKNOWLEDGMENTS ........................................................................................................
xv
EXECUTIVE SUMMARY ...................................................................................................... xvii CHAPTER 1: INTRODUCTION AND BACKGROUND ..................................................... Background ................................................................................................................... Adsorbents and Arsenic Removal................................................................................. Objectives ..................................................................................................................... Research Approach .......................................................................................................
1 1 1 2 2
CHAPTER 2: LITERATURE SURVEY SUMMARY...........................................................
5
CHAPTER 3: VENDOR AND MANUFACTURER SURVEY.............................................
15
CHAPTER 4: ARSENIC OCCURRENCE AND CO-OCCURRENCE ANALYSIS ............ Objectives of the Arsenic Co-Occurrence Analysis ..................................................... Source Data................................................................................................................... Creation of the Groundwater Arsenic Co-Occurrence Database.................................. Characteristics of Arsenic Occurrence.......................................................................... Characteristics of Arsenic Speciation ........................................................................... Characteristics of Arsenic Co-Occurrence ................................................................... General Water Quality Characteristics ............................................................. Are Waters That Contain Arsenic Different From Those That Do Not?.......... Direct Correlation Between Arsenic and Other Parameters ............................. Alternative Water Quality Testing Matrices................................................................. Summary .......................................................................................................................
19 19 19 20 22 26 26 27 27 32 33 35
CHAPTER 5: EXPERIMENTAL PROTOCOLS AND ANALYTICAL METHODS .......... Introduction................................................................................................................... Batch Testing Protocols ................................................................................................ Preliminary Screening Tests ............................................................................. Intensive Testing of Select Adsorbents............................................................. Column Testing Protocols............................................................................................. TCLP/Wet Testing Protocols .......................................................................................
37 37 37 37 38 40 40
v ©2005 AwwaRF. All rights reserved.
Analytical Methods....................................................................................................... Total Arsenic and Interferant Measurements.................................................... Arsenic Speciation ............................................................................................ QA/QC for Arsenic Analysis ........................................................................................
41 41 41 41
CHAPTER 6: BENCH TESTING ........................................................................................... Batch Testing ................................................................................................................ Preliminary Screening Experiments ................................................................. Intensive Testing of Selected Adsorbents.........................................................
43 43 43 52
CHAPTER 7: ARSENIC ADSORBENT DESIGN AND COSTING TOOL......................... Introduction................................................................................................................... Single Parameter and Multiple Regression Models...................................................... Single Parameter Model.................................................................................... Multiple Regression Model............................................................................... Description of Tool Algorithms and Calculations ........................................................ Instructions for Using the Tool ..................................................................................... Basis of Cost Opinions .................................................................................................
81 81 86 86 86 90 94 95
CHAPTER 8: CONCLUSIONS ..............................................................................................
97
APPENDIX A: LITERATURE SURVEY ............................................................................... 101 APPENDIX B: ADSORBENTS TESTED FOR ARSENIC REMOVAL: SYNTHESIS OF LITERATURE AND VENDOR/MANUFACTURER SURVEYS ............ 117 REFERENCES ......................................................................................................................... 131 ABBREVIATIONS .................................................................................................................. 137
vi ©2005 AwwaRF. All rights reserved.
TABLES 2.1 Adsorbents tested: description, water quality parameters and references .................
6
2.2 Batch isotherm studies: reported adsorption capacities and protocols used ..............
10
3.1 Adsorbents tested for arsenic removal: synthesis of literature review and vendor/manufacturer survey.......................................................................................
16
4.1 Database parameters and data acceptance criteria .....................................................
21
4.2 Summary of arsenic occurrence by physiographic region .........................................
25
4.3 Summary of occurrence of co-occurring parameters in national groundwater arsenic database .........................................................................................................
28
4.4 Pearson correlation coefficients at 95% confidence level measuring the degree of correlation between arsenic and co-occurring parameters.........................
34
4.5 Alternate water quality testing matrixes for waters containing moderate or high levels of arsenic..................................................................................................
35
5.1 Summary of quality assurance and quality control procedures for arsenic analysis.......................................................................................................................
41
6.1 Physical/chemical properties of adsorbents ...............................................................
53
6.2 Concentrations of interferents based on the co-occurrence survey............................
55
6.3 Ionic composition of NSF challenge water (pH = 7.5)..............................................
55
6.4 Water quality summary of utility-supplied waters.....................................................
56
6.5 CV values (%) for isotherm constants and Freundlich predictions ...........................
59
6.6 Values of Freundlich isotherm fitting parameters (KF and 1/n) for adsorption of arsenic onto tested media and predicted adsorption capacity when As(V) concentration in solution is 10 µg/L or 50 µg/L ........................................................
60
6.7 Values of Langmuir isotherm fitting parameters (KL and Qmax) for adsorption of arsenic onto tested media.......................................................................................
61
6.8 Summary of SMI results ............................................................................................
75
vii ©2005 AwwaRF. All rights reserved.
6.9 Monomeric versus polymeric silica concentrations...................................................
77
6.10 Experimental conditions employed in column tests ..................................................
78
6.11 TCLP and WET results for spent arsenic sorptive media ..........................................
80
7.1 Design and operating criteria used in sizing and costing of arsenic adsorption systems .....................................................................................................
82
7.2 Single parameter and multiple regression models .....................................................
87
viii ©2005 AwwaRF. All rights reserved.
FIGURES 1.1 Research approach .....................................................................................................
3
4.1 Database development flow chart ..............................................................................
20
4.2 Distribution of groundwater arsenic sites by state .....................................................
23
4.3 Physiographic zones of the United States..................................................................
23
4.4 National cumulative probability distribution for arsenic occurrence for 9867 stations extracted from NWIS...........................................................................
24
4.5 Cumulative probability distribution for arsenic occurrence by physiographic zone. ...........................................................................................................................
25
4.6 Comparison of cumulative probability distributions of pH for waters containing low, moderate, and high concentrations of arsenic ..................................
30
4.7 Comparison of cumulative probability distributions of silica for waters containing low, moderate, and high concentrations of arsenic ..................................
30
4.8 Comparison of cumulative probability distributions of fluoride for waters containing low, moderate, and high concentrations of arsenic ..................................
31
4.9 Comparison of cumulative probability distributions of alkalinity for waters containing low, moderate, and high concentrations of arsenic ..................................
31
4.10 Comparison of cumulative probability distributions of phosphate for waters containing low, moderate, and high concentrations of arsenic ..................................
32
4.11 Comparison of cumulative probability distributions of iron for waters containing low, moderate, and high concentrations of arsenic ..................................
32
4.12 Comparison of cumulative probability distributions of sulfate for waters containing low, moderate, and high concentrations of arsenic ..................................
33
6.1 (a) pH fluctuation and (b) arsenic removal in the absence or presence of a pH buffer: initial pH of 7.0 with GFH and As(V) = 100 µg/L...................................
45
6.2 As(V) removal (100 µg/L) at pH 7 as a function of time by (a) AA-400G, (b) AA-FS50, (c) ARM 100, (d) Bayoxide E33, (e) GFH, (f) MetSorb, (g) MIEX, (h) Z33 Rev. B, (i) Geothite, and (j) pyrolusite........................................
46
6.3 Isotherms for candidate adsorbents: As(V) at a pH of 7.0 and 24 hours ...................
51
ix ©2005 AwwaRF. All rights reserved.
6.4 Percent removal of As(V) by candidate adsorbents: As(V) = 100 µg/L; pH = 7.0, 24 hours ....................................................................................................
51
6.5 Percent removal of As(III) by candidate adsorbents: media dose = 100 mg/L; As(III) = 100 µg/L; pH = 7.0, 24 hours .....................................................................
52
6.6 Zeta potential versus pH for adsorbents.....................................................................
54
6.7 Isotherm replicate (triplicate) series for Metsorb G with As(V) at a pH of 7.0 and 24 hours.........................................................................................................
58
6.8 Isotherm replicate (triplicate) series for SMI-III with As(V) at a pH of 7.0 and 24 hours.......................................................................................................
58
6.9 Single-solute isotherms for As(V) at pH 6.0: in the absence of interferents .............
62
6.10 Single-solute isotherms for As(V) at pH 7.0: in the absence of interferents .............
62
6.11 Single-solute isotherms for As(V) at pH 8.0: in the absence of interferents .............
63
6.12 Summary of interference factors: ratio of As(V) capacity to that at pH 6.0 in the absence of interferents .....................................................................................
63
6.13 Adsorption capacity, Q10, for As(V) as a function of pH: absence of interferants ....
64
6.14 Adsorption capacity, Q10, for As(V) as a function of pH: presence of phosphate at 50% occurrence level ............................................................................
64
6.15 Adsorption capacity, Q10, for As(V) as a function of pH: presence of phosphate at 75% occurrence level ............................................................................
65
6.16 Adsorption capacity, Q10, for As(V) as a function of pH: presence of silica at 50% occurrence level....................................................................................
65
6.17 Adsorption capacity, Q10, for As(V) as a function of pH: presence of silica at 75% occurrence level....................................................................................
66
6.18 Adsorption capacity, Q10, for As(V) as a function of pH: presence of vanadium at 75% occurrence level ............................................................................
66
6.19 Isotherms for As(III) at a pH of 7.0: absence of interferants .....................................
67
6.20 Adsorption capacity, Q10, for As(III) as a pH of 7.0: absence of interferants ...........
67
6.21 Isotherms for As(V) in NSF challenge water.............................................................
70
x ©2005 AwwaRF. All rights reserved.
6.22 Adsorption capacity, Q10, for As(V) in NSF challenge water ...................................
70
6.23
Isotherm for As(III) in NSF challenge water ............................................................
71
6.24 Adsorption capacity, Q10, for As(III) in NSF challenge water ..................................
71
6.25 Isotherms for utility-supplied water: Tucson, AZ groundwater.................................
72
6.26 Isotherms for utility-supplied water: LADWP, CA surface water .............................
72
6.27 Isotherms for utility-supplied water: El Paso, TX groundwater ................................
73
6.28 Isotherms for utility-supplied water: Alamosa, CO groundwater ..............................
73
6.29 Isotherms for GFH with As(V) at varying pH levels and silica concentrations.........
78
6.30 Breakthrough curves for column tests with NSF challenge water.............................
79
7.1 Cost flowchart ............................................................................................................
91
7.2 Media replacement flowchart.....................................................................................
92
7.3 Backwash calculation flowchart.................................................................................
92
7.4 Plant footprint calculation flowchart..........................................................................
93
xi ©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
FOREWORD The Awwa Research Foundation (AwwaRF) is a nonprofit corporation that is dedicated to the implementation of a research effort to help utilities respond to regulatory requirements and traditional high-priority concerns of the industry. The research agenda is developed through a process of consultation with subscribers and drinking water professionals. Under the umbrella of a Strategic Research Plan, the Research Advisory Council prioritizes the suggested projects based upon current and future needs, applicability, and past work; the recommendations are forwarded to the Board of Trustees for final selection. The foundation also sponsors research projects through an unsolicited proposal process; the Collaborative Research, Research Applications, and Tailored Collaboration programs; and various joint research efforts with organizations such as the U.S. Environmental Protection Agency, the U.S. Bureau of Reclamation, and the Association of California Water Agencies. This publication is a result of one of these sponsored studies, and it is hoped that its findings will be applied in communities throughout the world. The following report serves not only as a means of communicating the results of the water industry’s centralized research program but also as a tool to enlist the further support of the nonmember utilities and individuals. Projects are managed closely from their inception to the final report by the foundation’s staff and large cadre of volunteers who willingly contribute their time and expertise. The foundation serves a planning and management function and awards contracts to other institutions such as water utilities, universities, and engineering firms. The funding for this research effort comes primarily from the Subscription Program, through which water utilities subscribe to the research program and make an annual payment proportionate to the volume of water they deliver and consultants and manufacturers subscribe based on their annual billings. The program offers a costeffective and fair method for funding research in the public interest. A broad spectrum of water supply issues is addressed by the foundation’s research agenda: resources, treatment and operations, distribution and storage, water quality and analysis, toxicology, economics, and management. The ultimate purpose of the coordinated effort is to assist water suppliers to provide the highest possible quality of water economically and reliably. The true benefits are realized when the results are implemented at the utility level. The foundation’s trustees are pleased to offer this publication as a contribution toward that end. The focus of this study was a comparative assessment of different adsorbent treatment technologies for arsenic (As) removal from drinking water supplies. The studied evaluated a large number commercially available and developmental adsorbents of different properties. These adsorbents were tested under a wide range of water quality conditions, including inhibitory species (e.g., phosphate) reducing the adsorbent capacity for arsenic. The evaluation encompassed arsenic in its two principal oxidation states: arsenate, As(V); and arsenite, As(III). The (leaching) stability of spent adsorbents was assessed to promote their use as throwaway materials after highcapacity arsenic adsorption. A decision framework (e.g., an algorithm-like procedure), in the form of a software package, was developed for helping utilities determine the most appropriate adsorbent based on cost and performance in terms of known water quality parameters. Walter J. Bishop Chair, Board of Trustees Awwa Research Foundation
Robert C. Renner Executive Director Awwa Research Foundation xiii ©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
ACKNOWLEDGMENTS PROJECT ADVISORY COMMITTEE (PAC) The advice and help provided by the Project Advisory Committee (PAC) and AwwaRF project officer, Traci Case, are sincerely appreciated. The PAC consisted of Tom Sorg, David Hand, Michelle DeHaan, and Jason Wen. PARTICIPATING UTILITIES We acknowledge the participation of utilities that supplied water for testing and associated water quality data: Tucson (AZ), El Paso (TX), Los Angeles Department of Water and Power (LADWP), and Alamosa (CO).
xv ©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
EXECUTIVE SUMMARY STUDY OBJECTIVES The objective of this study were: (i) to identify the inhibitory impacts of water quality on the performance of a range of adsorbents (commercially available and experimental) for the removal of arsenate and arsenite; (ii) to define adsorbent properties that will minimize the effects of water quality and/or maximize arsenic adsorption capacity; (iii) to investigate the applicability of various adsorbents to arsenite (As(III)) removal as opposed to a need for pre-oxidation to arsenate (As(V)); (iv) to assess the (leaching) stability of spent adsorbents to promote their use as throwaway materials; and (v) to develop a decision framework (e.g., an algorithm-like procedure) for helping utilities determine the most appropriate adsorbent based on cost and performance in terms of known water quality parameters. RESEARCH APPROACH The research approach was performed in two phases comprising a total of five tasks. The first phase, embodying the first and second tasks, was intended to complete a preliminary identification of candidate media and to define the experimental water quality matrix. The second phase, embodying the last three tasks, represented the experimental efforts, data analysis, comparative assessment, and selection framework. LITERATURE AND VENDOR SURVEYS Extensive literature and vendor surveys revealed that there are a large number of commercial and experimental adsorbents, available for arsenic removal. Some of the commercially available materials are mature products (e.g., activated alumina) that have been widely tested, revealing both their attributes and limitations, while others are more recent products that have not been as rigorously tested (e.g., various iron oxides). Based on their predominant adsorption mechanism as well as material composition, these adsorbents fall into several categories: (i) ion exchange media (e.g., MIEX); (ii) metal oxides (e.g., activated alumina and iron oxides); and (iii) redox-reactive media (e.g., MnO2). In this study, we have chosen to highlight the second category in the forms of both pure minerals and amended/impregnated materials. The literature review also revealed that adsorption capacities were influenced by arsenic speciation, arsenate (As(V)) versus arsenite (As(III)), and water quality in terms of pH conditions and the presence on interfering species competing with arsenic for adsorption sites. Influential interferants were revealed to include silica, phosphate, fluoride, sulfate, carbonate, and others. Based on the literature and vendor surveys, a total of 12 adsorbents, 10 commercially available and 2 experimental, were identified for experimental evaluation. Besides pH as an important water quality condition, phosphate and silica were identified for intensive interferant testing, later supplemented by vanadium and fluoride for less intensive assessment.
xvii ©2005 AwwaRF. All rights reserved.
ARSENIC OCCURRENCE AND CO-OCCURRENCE SURVEY An arsenic occurrence and co-occurrence survey was conducted to indicate national (USA) occurrence levels of total arsenic, arsenic species domination, and co-occurrence trends between arsenic and interferants. Based on statistical analysis of an integrated national database for groundwater, the mean total arsenic concentration was determined to be 4.5 µg/L, with a 90 percentile value of about 10 µg/L. Geographically, the highest national levels were found in the Intermontane region encompassing states west of the continental divide (e.g., Arizona and New Mexico) except for Washington, western Oregon, and northern California; lowest levels were found in the Atlantic Plain region. Binning of data showed that about 5% of all groundwater sites contained ≥20 µg/L, about 5% contained between 10 to 20 µg/L, and about 5% between 5 to 10 µg/L. While a rigorous study of arsenic speciation in groundwater was not found, several limited studies suggested that As(V) is the predominant species in U.S. groundwaters although, in other global settings such as Bangladesh, As(III) is known to dominate in more anoxic groundwaters. There were semi-quantitative trends to indicate that groundwaters with arsenic ≥5 µg/L tended to be higher in pH, silica, fluoride, alkalinity, and phosphate than those with GFH > AA-FS50 > Z33 ~ Metsorb G. The higher pHZPC of the Bayoxide E33 versus GFH helped explain the better performance of the former iron oxide at higher pH conditions. The order of surface area (high to low) was AA-FS50 ~ GFH > Metsorb G > Bayoxide E33 > Z33. The very low surface area of the Z33 was manifested in a lower number of adsorption sites per unit mass. The material composition differed significantly: aluminum and iron content for AA-FS50, titanium dioxide for Metsorb G, clay content for Z33, and iron oxides for GFH and Bayoxide E33. The high capacity exhibited by the Metsorb E33 may involve a more complicated mechanism than simply surface complexation or ligand exchange. Three of the adsorbents; Bayoxide E33, Metsorb G, and GFH; were tested for arsenite removal; while all three adsorbents demonstrated a significant potential for As(III) removal, their As(III) capacities were significantly less than their As(V) capacities; the order of performance was Bayoxide E33 > Metsorb G > GFH. Three adsorbents (Bayoxide E33, Metsorb G, and GFH) were tested in a NSF challenge water containing As(V) or As(III) with multiple interferants at a constant pH (7.5) as well as several utility-supplied natural waters containing multiple interferants and variable pH. The order of capacities for As(V) in the NSF challenge water was Metsorb G > Bayoxide E33 > GFH, while the order of capacities for As(III) was Metsorb G >GFH > Bayoxide E33. The general order of capacities for the natural waters was Metsorb G > Bayoxide E33 > GFH. Results from a special batch study on SMI revealed a higher As(V) capacity than that of the other adsorbents; however, its performance was substantially reduced at higher pH conditions and/or in the presence of interferants. Moreover, its As(III) capacity was much lower than its As(V) capacity. Another special batch study was performed to more closely study the effects different forms of silica, monomeric versus polymeric. It was found that at higher silica concentrations, polymeric silica can form and potentially foul porous adsorbents, thus reducing their capacity. Four of the adsorbents (Metsorb G, AA-FS50, Bayoxide E33, and GFH) were tested in dynamic column tests employing the NSF challenge water spiked with As(V) at concentrations of 250 µg/L (for Metsorb G) to 1,000 µg/L (for the other three adsorbents); the purpose of the high initial concentration was to facilitate breakthrough. Based on an empty bed contact time (EBCT) of 5 minutes, the AA-FS50 and GFH showed similar results with breakthrough at about 4,000 to 5,000 bed volumes (BVs). The Metsorb G and Bayoxide E33 were run for longer periods of about 8,000 and 12,000 BVs, respectively. The Metsorb G experiment was terminated before a clear breakthrough trend was observed, however, this result is consistent with the lower initial As(V)
xix ©2005 AwwaRF. All rights reserved.
concentration employed in this test. The residuals from these column tests were subjected to TCLP and WET tests which indicated that all would not be classified as hazardous, permitting their disposal as throwaway adsorbents. ARSENIC ADSORBENT DESIGN AND COSTING TOOL An arsenic adsorbent design and costing tool was developed based on seven adsorbents: AA-FS50, Bayoxide E-33, GFH, Metsorb G, Z-33, SMI, and AA-400G (conventional, granular activated alumina). The software is an interactive tool which contains input and output forms. The user enters information in the input form, and the tool calculates design and cost parameters and displays the results in an output form. The type of input parameters that the user is required to enter includes system parameters (e.g., average flow), water quality (e.g., influent arsenic, phosphate, and silica concentrations), target water quality (e.g., treated-water arsenic), operational preferences (e.g., pH adjustment) and cost parameters (e.g., interest). The user has the option instructing the use of either Freundlich or Langmuir isotherm models to predict adsorption capacity. The tool generates outputs that include the calculated Freundlich or Langmuir coefficients, adsorption capacities, capital costs, operations and maintenance (O&M) costs, plant footprint, residuals quantities and water quality interference warnings. The outputs are calculated based on the water quality and system information that the user enters in the input form.
xx ©2005 AwwaRF. All rights reserved.
CHAPTER 1 INTRODUCTION AND BACKGROUND BACKGROUND Arsenic is a metalloid or oxyanion found in both ground and surface water sources. It occurs in both dissolved and colloidal/particulate forms. In drinking water supplies, dissolved arsenic occurs as either arsenate, As(V), or arsenite, As(III). In the pH range of drinking waters, the anionic arsenate (pKa1, pKa2, pKa3 = 2.2, 7.0, 11.5, respectively) dominates as either a monoor di-valent oxyanion (H2AsO4– or HAsO42–) whereas arsenite, the reduced form, is a nonionic species (H3AsO3; pKa1, pKa2, pKa3 = 9.2, 12.1, 13.4, respectively). The colloidal/particulate forms of arsenic found in drinking water supplies are thought to occur primarily as arsenic species adsorbed to naturally-occurring metal (e.g., iron) oxide surfaces and, to a lesser extent, derived from arsenic-bearing minerals (e.g., arsenic trioxide, As2O3 and arsenopyrite, FeAsS) subjected to geochemical weathering reactions. ADSORBENTS AND ARSENIC REMOVAL The applicability of conventional adsorbents, including activated alumina (AA) and ion exchange (IX), for the removal of arsenic has been established along with associated water quality constraints (e.g., SO42– for IX) and regeneration requirements (Amy, et al., 2000). However, conventional adsorbents either require that arsenic be in the form of arsenate (in the case of IX) or exhibit a much higher capacity for arsenate over arsenite (in the case of AA). While oxidation of As(III) to As(V) can be easily achieved by oxidants such as ozone, chlorine, or permanganate, this practice adds operational complexity and may embody secondary impacts such as the formation of disinfection by-products and possibly biodegradable organic matter (BOM). Recent work has involved the development of new adsorbents with less water quality constraints, higher capacities, and/or increased removals of arsenite. Moreover, the development of these high capacity adsorbents (discussed in later literature and vendor surveys) has been accompanied by a change in philosophy in which many are now considered to be throwaway materials that can be disposed of in a landfill if an appropriate leaching (e.g., TCLP) test can be passed. This approach eliminates the need for regeneration and associated problems with the regenerant stream. Besides the form of arsenic, water quality affects arsenic adsorption. The main water quality constraints for adsorbents are pH, affecting the charge of arsenate (mono- vs. di-valent) as well as the surface charge of the adsorbent (pHZPC), and competing anions, some which are affected by pH (e.g., silicate, phosphate, and natural organic matter (NOM) in the form of humic/fulvic acids) and others which are unaffected by pH (sulfate, nitrate and fluoride). The beneficial presence of iron as a precipitating adsorption surface has been recognized, largely through assessment of enhancing the performance of Fe/Mn removal and ferric coagulation plants in removing arsenic by promoting the formation of amorphous ferric hydroxide, Fe(OH)3 (Amy et al., 2000). Given that arsenate (and arsenite) embody both metal (As) and ligand (O) properties, the mechanisms of arsenate (and arsenate) adsorption onto metal (e.g., iron) oxide surfaces can occur through surface complexation or ligand exchange:
1 ©2005 AwwaRF. All rights reserved.
Surface Complexation: Ligand (L) Exchange:
>X–OH + HAsO42– → X–OHAsO42– + H+ >X–OH + L– → >X–L + OH–
(L– = HAsO42–)
Adsorbents potentially applicable to arsenic removal can be classified into three general categories: (i) established commercially available adsorbents which are known to effectively remove arsenic and are well-characterized with respect to their operational characterisitics and behavior under various water quality conditions (e.g., activated alumina, AA); (ii) emerging commercially available adsorbents which have been developed and demonstrated to effectively remove arsenic but whose performance in various water quality matrices have not been studied in detail (e.g., granular ferric hydroxide, GFH); and (iii) experimental media that often represent pure mineral materials that have only been studied in more narrowly focused research projects (e.g., pyrolusite). While most adsorbents are available in a granular form applicable to use in fixed-bed columns, some are available in a powdered form for other uses such as hybrid (coupled) adsorbent-membrane systems (e.g., powdered AA + microfiltration (MF) or ultrafiltration (UF)). OBJECTIVES The objective of this study were •
• • • •
to identify the inhibitory impacts of water quality on the performance of a range of adsorbents (commercially available and experimental) for the removal of arsenate and arsenite; to define adsorbent properties that will minimize the effects of water quality and/or maximize arsenic adsorption capacity (e.g., specific surface area, pHZPC); to investigate the applicability of various adsorbents to As(III) removal as opposed to a need for pre-oxidation to As(V); to assess the (leaching) stability of spent adsorbents to promote their use as throwaway materials; to develop a procedure (e.g., an algorithm-like procedure) for helping utilities determine the most appropriate adsorbent based on cost and performance in terms of known water quality parameters.
RESEARCH APPROACH The research approach was divided into two phases comprising a total of five tasks. The first phase, embodying the first and second tasks, was intended to complete a preliminary identification of candidate media and to define the experimental water quality matrix. The second phase, embodying the last three tasks, represented the experimental efforts, data analysis, comparative assessment, and selection framework. Since the number of adsorbents that have been developed and potential water quality matrices in which to test these adsorbents are virtually unlimited, it was imperative that the experimental combinations of adsorbents and water quality matrices be selected to benefit the greatest number of drinking water utilities. Therefore, the first task was to review the international literature and to contact vendors as a basis for identifying a list of candidate adsorbents. The second
2 ©2005 AwwaRF. All rights reserved.
Figure 1.1 Research approach
task involved the use of arsenic occurrence databases to help determine which ions and other water quality conditions (e.g., phosphate, pH) most commonly occur with arsenic. This information helped the research team identify water quality matrices that would most likely affect drinking water utilities. By combining the information gathered from the first two tasks, the project team was able to proceed to the experimental phase of the project with the knowledge that the select media were representative of the range of materials available and the test water quality conditions were representative of conditions that utilities will encounter. After identification of available adsorbents appropriate water quality conditions, candidate adsorbents were first studied in a set of preliminary experiments (Task 3) to assess kinetics and capacities, with results used to screen potential adsorbents and help identify a reduced number of select adsorbents to be subjected to intensive bench-scale testing (Task 4). These select adsorbents were characterized according to important physical/chemical properties influencing arsenic and competing anion adsorption. Batch and continuous-flow column tests were performed in the fourth task to define arsenic removal trends which, upon data analysis and isotherm modeling, indicated performance within various water quality matrices. Finally, once influential parameters were defined for each adsorbent tested, the research team developed an algorithm-like procedure in the fifth task for helping utilities select the most appropriate adsorbent type based on cost, water quality, and other factors. A summary of our research approach is shown in Figure 1.1. 3 ©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
CHAPTER 2 LITERATURE SURVEY SUMMARY The literature survey was completed with a total of over 400 articles and papers assembled from numerous journals, of which the most valuable sources were Journal of American Water Works Association, Environmental Science and Technology, Water Research, Desalination, Journal of Environmental Engineering, and Water Science and Technology, containing about onefourth of the articles. Relevant papers were also found in a number of other journals including: Separation Science and Technology, Environmental Technology, Environmental Engineering Science, Water, Air, and Soil Pollution, Chemosphere, Journal of Hazardous Materials, Journal Water Pollution Control Federation, Water Quality Research Journal of Canada, Colloidal Surface, Journal of Analytical Atomic Spectrometry, Journal of Water Chemistry and Technology, Powder Technology, Journal of Environmental Science and Health, Journal of Radioanalytical and Nuclear Chemistry, Soil Science, Water Environment Research, Chemical Engineering and Technology, International Journal of Water, Waste Management, Environmental Technology, Environmental Geology, and Geochim. Cosmochim Acta. The detailed literature survey is presented in Appendix A. Based on the detailed literature survey, two summary tables have been derived and are presented to summarize adsorbents tested (Table 2.1) and reported isotherm capacities (Table 2.2). The literature survey largely confirmed our original notions about influential interferants and types of adsorbent materials that have or might have merit for adsorption of oxyanions and metalloids such as arsenate (or arsenite). While a significant number of commercially available materials have been tested, many of the adsorbents tested are experimental or pure minerals. Many fall within the general category of metal (hydro)xides, with various iron oxides being very common. Some function as simple ion exchange media. Another common group of materials is surface-modified natural materials. The major interferants have been shown to be phosphate, silica, carbonate, sulfate, and fluoride. pH has been shown in many cases to exhibit a major effect on media performance.
5 ©2005 AwwaRF. All rights reserved.
Table 2.1 Adsorbents tested: description, water quality parameters and references
Adsorbents tested
Basis (adsorbent description)
1
Magnetic ion-exchange (iron-oxide, magnetite)
Magnetically impregnated ion-exchange resins (MIEX)
Water quality parameter(s) studied Sulfate NOM pH
Hydrous iron oxide particles (HIOPs)
Non-compressible and semicrystalline structure (hematite)
3
Sulfur-modified iron (SMI)
Metallic iron and elemental sulfur comp.
4
Granular ferric hydroxide (GFH) (Prepared by neutralizing & precipitating FeCl3 with NaOH)
β-FeOOH Akaganeit
5
Ferrihydrite (FH)
6
Geothite (α-FeOOH)
7
Haematite
pH, temperature
8
Iron oxide coated sands (IOCS) or iron oxide coated microsands (IOC-M)
Copper, chromate
6
©2005 AwwaRF. All rights reserved.
2
Sulfate NOM
Limiting parameter/ limiting level
References
Increasing (0–100 mg/L) Increasing DOC (0–4 mg/L) more effective at lower values (5.5)
Chang et al. (2004); Amy et al. (2002)
DOC (0–4 mg/L) pH wider range
Chang et al. (2004); Amy et al. (2002)
Chang et al. (2004); Amy et al. (2002) Sulfate Phosphate
Phosphate
Chang et al. (2004); Driehaus et al. (1998); Pal (2001); Selvin et al. (2001); Norton et al. (2001)
pH (5–9.5) silica, sulfate, NOM
Silica NOM
Robins et al. (2001); Jain et al. (1999); Raven (1998); Gulledge and O’Connor (1973);Waychunas et al. (1993, 1995); Nishimura and Umetsu (2000); Wong et al. (1995); Chen et al. (2005); Thirunavukkarasu et al. (2001)
Product formed from crystalization of ferrihydrite
Belzile and Tessier (1990); Matis (1999); Matis et al. (1997); Xiaohua and Harvey (1998); Manning et al. (1998); O’Reilly et al. (2001) pH 7 max ads. ↑ temp. reduces ads.
Singh et al. (1988) Chang et al. (2004); Lombi et al. (1999); Thirunavukkarasu et al. (2001); Khaodhiar et al. (2000) (Continued)
Table 2.1 (Continued)
Adsorbents tested 9
Basis (adsorbent description)
Amorphous iron hydroxide Pyrite of 95% purity
Water quality parameter studied
Parameter limiting level
References
Higher pH reduces adsorption
Pierce and Moore (1982)
pH
pH < 4 required
Zouboulis et al. (1993); Han and Fyfe (2000)
Twice as effective in removing As(V) than As(III)
Lackovic et al. (2000) sulfate
Lackovic et al. (2000); Krishna et al. (2001)
10
Iron-sulfide minerals
11
Zero-valent iron (ZVI) (Column experiment data only) (Cost: $250–400/ton)
12
Ferruginous manganese ore (FMO)
Major mineral phases: pyrolusite and geothite
pH 2–8 Divalent Cation (Ni2+, Co2+, Mg2+) presence enhances adsorption
Chakravarty et al. (2002)
13
Manganese greensand (MGS)
A zeolite-type glauconite mineral
Fe/As molar ratio (20 optimal) Initial As, Fe(II) and Mn (IV) conc.
Subramanian et al. (1997); Viraraghavan et al. (1999); Thirunavukkarasu et al. (2001)
14
Manganese dioxide
15
Manganese dioxide coated sand (MDCS)
Formed by oxidation of Mn(II) with permanganate in the presence of sand
16
Manganese oxyhydroxide coated sand (MOCS)
pHZPC of 2.87
17
Iron-modified AA
7
©2005 AwwaRF. All rights reserved.
pH
Addition of Mg2+,Pb2+, Ni2+, Ag and Ca2+ significantly increases As adsorption
Tartrate, phosphate, carbonate reduce adsorption, as well as Zn2+
Chen et al. (1999)
pH
For As (V), pH < 7.5
Thomson et al. (1998) Norton et al. (2001) (Continued)
Table 2.1 (Continued)
Adsorbents tested 18
Basis (adsorbent description)
Activated Alumina (AA)
Water quality parameter studied Salinity pH 4-7 effective for removal of As(V); Effect of init. Conc.: As (V) not-sig., As (III) sig.
Parameter limiting level
References
pH > 7, ads. decreases
Gupta and Chen (1978); Chang et al. (2004); Norton et al. (2001)
Gupta and Chen (1978)
Activated carbon (AC)
Salinity pH 3–5 effective removal of As(V)
20
Coconut husk carbon (CHC)
Prepared by treating CH with sulf. acid; pHpzc 7.5; surf area 206 m2/g ; AEC 1.23 meq/g
pH 2–12
Contact time 1–6 h temp. 30–60°C ads ↑ with temp. pH 12 max uptake
Manju et al. (1998)
21
Quaternized rice husk (QRH)
150 g ground rice treated with 2L of 1% w/v Na2CO3
pH NO3– no effect SO42–, CrO42– (concs. 10–500 mg/L) reduce adsorption
pH > 10 ads. decr. sorbent dosage (0.02 to 0.5 g ↑ in uptake from 18 to 86%) temp. 25–80°C
Lee et al. (1999)
22
Fly ash
By product from coal power stations
pH
pH > 4 ads. Decr
Diamadopoulos et al. (1993)
23
Red mud
Waste from alumina production
pH
As(III) ads effective at pH 9.5 As(V) ads decr. at pH >3.2
Altundogan et al. (2000)
24
Aluminium–loaded coral limestone (AL-CL)
8
©2005 AwwaRF. All rights reserved.
19
pH H2PO4– significant effect Cl–, NO3–, SO42– none
Ohki et al. (1996)
(Continued)
Table 2.1 (Continued)
Basis (adsorbent description)
25
Aluminium loaded shirasu-zeolites
Shirasu—volcanic pile present in Kyushu area, Japan
26
Natural zeolites, volcanic stone and cactaceous powder
27
California soils
28
Clay minerals (halloysite, kaolinite, illite, chlorite montmorillonite
29
Zirconium oxide (Zr-resin)
9
©2005 AwwaRF. All rights reserved.
Adsorbents tested
Zr (IV) loaded phosphoric acid chelating resin
Water quality parameter studied
Parameter limiting level
References Xu et al. (1998) Elizalde-Gonzales (2001)
Arid-zone soils from California; fine-loamy and coarse- loamy
pH 4-11
As(V) ads decreases at pH >7; As(III) ads↑ at pH >7
Manning (1997)
Lin and Puls (1999); Manning and Goldberg (1997) Amberlite XAD-7 based on cross-linked polyacrylate copolymer Phosphoric acid resin (RGP) in the hydrogen form; sp.surf area 29.2 m2/g; acid cap. 7 meq/g; P content 3.75 mmol/g
H2PO4– , fluoride significant effect River water versus salt water pH presence of electrolytes enhances adsorption
pH >10
Suzuki et al. (2001)
Zhu and Jyo (2001)
Table 2.2 Batch isotherm studies: reported adsorption capacities and protocols used Adsorption capacities achieved Adsorbents tested
As(III)
As(V)
Initial As (III) conc. range
Initial As(V) conc. range
Dose added
Effect of water quality parameters (or water quality parameters interferences) pH
Phosphate
Silica
Fluoride
Sulfate
Magnetically impregnated ion-exchange resins (MIEX)
3.78 µg/mg
0, 25, 50, 250, 1000, 2000 µg/L
2 ml/L
Yes
For pH 5.5/7.5 Yes/No
For pH 9.5 Yes
For pH 5.5/7.5 No/No
For pH 5.5 Yes
2
Hydrous iron oxide particles (HIOPs)
1.0 µg/mg
0–2000 µg/L
8 ml/L
None
No
Yes
No
No
3
Sulfur-modified iron (SMI)
0.32 µg/mg
0–2000 µg/L
2,500 mg/L
Slightly
No
No
No
No
4
Granular ferric hydroxide (GFH) Chang et al. (2004)
2.51 µg/mg
0–2000 µg/L
500 mg/L
None pH (5–9)
Yes
Yes
No
No
Driehaus et al. (1998)
ads. density 1 mmol/g Fe
residual conc. 10–40 µg/L
10
©2005 AwwaRF. All rights reserved.
1
5
Yes
No
Ferrihydrite (FH) Raven (1998)
0.60 molAs/ molFe (at pH 4.6 and 9.2)
at pH 4.6: 0.25 molAs/molFe at pH 9.2: 0.16 molAs/molFe
Chen et al. (2005) Thirunavukkarasu et al. (2001)
0.267– 26.7 mmol/L
0.267– 26.7 mmol/L
40 ml of 2.5 g/L FH in 0.1 M NaCl
As(V) ads higher at pH 4.6
62.71 µg/L 285 µg/g
325 µg/L
0–50 mg SiO2/L 0.02– 0.09 g FH in 100 mL
pH 7.4
(Continued)
Table 2.2 (Continued) Adsorption capacities achieved Adsorbents tested 6
As(III)
As(V)
Initial As(V) conc. range
Dose added
Effect of water quality parameters (or water quality parameters interferences) pH
Phosphate
Silica
Fluoride
Sulfate
Geothite (α-FeOOH) 2.1 µmol/m2
O’Reilly et al. (2001) Manning et al. (1998) Haematite
8
Iron oxide coated sands (IOCS) or iron oxide coated microsands (IOC-M)
11
©2005 AwwaRF. All rights reserved.
7
0–3 mM 133– 266 mM
0.20 µg/mg
Thirunavukkarasu et al. (2001) 9
Initial As (III) conc. range
pH 6 20 ml of 2.5 g/L FH
1–10 mg/L
18.3 µg/g
325 µg/L
50 mm/g
0.667– 667 µmol/L
pH 2.8–5 none
0.5– 1.2 g IOCS in 100 mL
pH 7.4
Amorphous iron hydroxide Pierce and Moore (1982)
10
Iron-sulfide minerals
11
Zero-valent iron (ZVI) Lackovic et al. (2000) (Column experiment data only)
50 mm/g
298 µgAs/gFe0 669 µgAs/gFe0 13.3 µmol after 1850 pore after 1850 pore volumes volumes
0.667– 667 µmol/L
13.3 µmol
pH 4–10
Empty bed vol. 14 mL
After As ads No
After As ads No
No
(Continued)
Table 2.2 (Continued) Adsorption capacities achieved Adsorbents tested 12
Ferruginous manganese ore (FMO)
14
Manganese dioxide
As(III)
As(V)
At 0.040.18ppm 100% efficient
0.18ppm 100% efficient
Initial As (III) conc. range 0.04–0.18 ppm
Initial As(V) conc. range
50 mg MnO2 in 50 mL
16
Fluoride
Sulfate
pH 1–10
Yes
0.5 mg/L
10 g/L
Manganese oxyhydroxide coated sand (MOCS) at pH 7.5 1.6 µg/mg
0.5 mg/L
pH 6–9
Iron-modified AA Norton et al. (2001) (column exp.)
18
Silica
Manganese dioxide coated sand (MDCS)
Thomson et al. (1998) 17
Phosphate
44.04 mg/g
0.5 mg/L
12
©2005 AwwaRF. All rights reserved.
15
pH
0.04–0.18 ppm
Mean sorption energy 15.5 kJ/mol Chen et al. (1999)
Dose added
Effect of water quality parameters (or water quality parameters interferences)
31–41 µg/L
pH 8.7– 9.2
Yes
Yes
Activated Alumina (AA) Gupta and Chen (1978)
Norton et al. (2001)
4.10 mg/g
12.4 µM
53.4 µM
1 mg/L
1 mg/L
2 g/L of AA in 100 mL Yes, at conc >10 mg/L
Yes, at conc >2 mg/L (Continued)
Table 2.2 (Continued) Adsorption capacities achieved Adsorbents tested 19
As(III)
Initial As(V) conc. range
Dose added
Effect of water quality parameters (or water quality parameters interferences) pH
Phosphate
Silica
Fluoride
Sulfate
Activated carbon (AC) Gupta and Chen (1978)
20
As(V)
Initial As (III) conc. range
0.34 mg/g
Coconut husk carbon (CHC)
13
©2005 AwwaRF. All rights reserved.
copper-impregnated coconut husk carbon (CICHC)
146 µg/g
3 g/L of AC in 100 mL
50–600 mg/L
0.1 g of CHC in 50 mL
pH 2–12
2 g/L
21
Quaternized rice husk (QRH)
19 mg/g
22
Fly ash
27.8 mg/g
23
Red mud
24
Aluminium –loaded coral limestone (AL-CL)
25
Aluminium loaded shirasu-zeolites
26
Natural zeolites, volcanic stone and cactaceous powder
4.31 µmol/L at pH 9.5
26.4 µM
100–600 mg/L
Yes pH 4, 7, 10 80% reduction at pH 4
5.07 µmol/L at pH 3.2
2.5–30 mg/L
20 g/L
Yes
150 µg/g
2.5–50 mg/L
1 g of ALCL in 20 mL
pH 2–11 No
400 mg/L
0.05-2 g in 20 mL
Yes
No
(Continued)
Table 2.2 (Continued) Adsorption capacities achieved Adsorbents tested
14
©2005 AwwaRF. All rights reserved.
27
California soils
28
Clay minerals (halloysite, kaolinite, illite, chlorite montmorillonite
29
Zirconium oxide (Zr-resin) Zr (IV) loaded phosphoric acid chelating resin (column exp.)
As(III) 37–279 µmol/kg
1.06 µmol/g pH 8.5
As(V)
Initial As (III) conc. range
Initial As(V) conc. range
Dose added
Effect of water quality parameters (or water quality parameters interferences) pH
235–300 µmol/kg
0.27–13.3 µmol As
0.27–13.3 µmol As
1:10 (2 g soil in 20 mL)
pH 4–11 As(V) Yes As(III) Yes
Halloy site: 71.6 µg/g chlorite: 60.9 µg/g
30 µmol
35 µmol
1 g in 30 mL, 24 h
5.5–7.5
0.72 µmol/g pH 4.5 0.613 µmol/g pH 1.14 0.453 µmol/g pH 8.55
0.5 g in 50 mL
Phosphate
Silica
Fluoride
Yes
Yes
Sulfate
Yes
Yes
CHAPTER 3 VENDOR AND MANUFACTURER SURVEY Additional vendors and manufacturers, beyond those preliminarily identified in the original proposal submitted to the Awwa Research Foundation, were contacted. For media originally identified, we made inquiries to manufacturers about their availability in different physical forms (i.e., granular versus powdered) for use in different treatment configurations (i.e., packed beds versus slurry and/or membrane reactors); moreover, we inquired about the availability of new generations of existing products (e.g., sulfur modified iron (SMI)). The results of the vendor survey were compiled in an access database (not included). Table 3.1 represents a synthesis of the literature and vendor surveys; a longer version of this table appears in the appendix (Appendix B). The references cited in Table 3.1 are summarized in the References chapter. Through a critical examination based on the literature (i.e., Tables 2.1 and 2.2) and vendor surveys (i.e., Table 3.1) as well as input from the Project Advisory Committee (PAC) during a PAC meeting and through PAC comments in response to progress reports, a series of twelve candidate adsorbents were identified (their identities are revealed in Chapter 6) for testing. These included mostly commercially available media, both well developed and emerging, but also a couple of experimental media corresponding to pure minerals.
15 ©2005 AwwaRF. All rights reserved.
Table 3.1 Adsorbents tested for arsenic removal: synthesis of literature review and vendor/manufacturer survey Technology
Removal Arsenic capacity
Modified zeolite (Z-33)
Metsorb G > GFH. Based on the various figures showing individual interferant effects as well as the summaries in Tables 6.6 and 6.7 and Figure 6.12, a number of trends can be discerned from the work with
59 ©2005 AwwaRF. All rights reserved.
Table 6.6 Values of Freundlich isotherm fitting parameters (KF and 1/n) for adsorption of arsenic onto tested media and predicted adsorption capacity when As(V) concentration in solution is 10 mg/L or 50 mg/L Media
AA-FS50
60
©2005 AwwaRF. All rights reserved.
Bayoxide E33
GFH
MetSorb G
SMI-III Z33 Rev. B
As(V) (µg/L)
As(III) (µg/L)
PO4 (µg/L)
SiO2 (mg/L)
V (µg/L)
F (µg/L)
NSF challenge water
KF
1/n
Q10
Q50
KF
1/n
Q10
Q50
KF
1/n
Q10
Q50
100
0
0
0
0
0
0
0.300
0.548
1.06
2.56
0.472
0.426
1.26
2.50
0.273
0.446
0.763
1.56
100
0
125
0
0
0
0
00392
0.432
1.06
2.13
0.451
0.352
1.01
1.79
0.334
0.304
0.673
1.10
100
0
250
0
0
0
0
0.424
0.346
0.940
1.64
0.424
0.315
0.876
1.45
0.304
0.282
0.582
0.916
100
0
0
0
0
700
0
—
—
—
—
0.192
0.592
0.753
1.95
—
—
—
—
100
0
0
0
0
0
0
0.707
0.285
1.36
2.15
0.620
0.261
1.13
1.72
0.606
0.219
1.00
1.43
100
0
125
0
0
0
0
0.585
0.236
1.01
1.47
0.454
0.218
0.750
1.07
0.277
0.333
0.595
1.02
100
0
250
0
0
0
0
0.441
0.253
0.790
1.19
0.210
0.411
0.542
1.05
0.162
0.443
0.449
0.916
100
0
0
13.5
0
0
0
0.615
0.250
1.09
1.64
0.510
0.205
0.819
1.14
0.136
0.493
0.424
0.938
100
0
0
22
0
0
0
0.531
0.298
1.05
1.70
0.023
1.006
0.235
1.19
0.053
0.737
0.291
0.952
100
0
0
0
30
0
0
—
—
—
—
0.190
0.252
0.340
0.509
—
—
—
—
100
0
0
0
0
0
yes
—
—
—
—
0.131
0.591
0.513
1.33
—
—
—
—
0
100
0
0
0
0
0
—
—
—
—
0.330
0.421
0.872
1.72
—
—
—
—
0
100
0
0
0
0
yes
—
—
—
—
0.001
1.36
0.027
0.237
—
—
—
—
100
0
0
0
0
0
0
0.671
0.340
1.47
2.54
0.514
0.283
0.986
1.55
0.385
0.172
0.573
0.756
pH 6
pH 7 or 7.5 (NSF challenge water)
pH 8
100
0
125
0
0
0
0
0.663
0.249
1.18
1.75
0.513
0.227
0.866
1.25
0.487
0.056
0.554
0.606
100
0
250
0
0
0
0
0.602
0.188
0.928
1.26
0.418
0.144
0.582
0.734
0.177
0.312
0.364
0.601
100
0
0
13.5
0
0
0
0.570
0.186
0.875
1.18
0.276
0.343
0.607
1.05
0.409
0.020
0.427
0.441
100
0
0
22
0
0
0
0.522
0.244
0.916
1.36
0.263
0.325
0.556
0.937
0.112
0.309
0.227
0.374
100
0
0
0
30
0
0
—
—
—
—
0.023
0.709
0.116
0.364
—
—
—
—
100
0
0
0
0
0
yes
—
—
—
—
0.164
0.365
0.381
0.686
—
—
—
—
0
100
0
0
0
0
0
—
—
—
—
0.152
0.502
0.482
1.083
—
—
—
—
0
100
0
0
0
0
yes
—
—
—
—
0.029
0.714
0.152
0.480
—
—
—
—
100
0
0
0
0
0
0
0.661
0.301
1.32
2.15
0.589
0.276
1.11
1.73
0.584
0.220
0.968
1.38 1.09
100
0
125
0
0
0
0
0.653
0.245
1.15
1.71
0.689
0.171
1.02
1.35
0.313
0.318
0.651
100
0
250
0
0
0
0
0.615
0.305
1.24
2.03
0.379
0.366
0.879
1.59
0.252
0.390
0.620
1.16
100
0
0
13.5
0
0
0
0.554
0.339
1.21
2.08
0.091
0.669
0.426
1.25
0.218
0.318
0.453
0.756
100
0
0
22
0
0
0
0.365
0.431
0.984
1.97
0.115
0.579
0.436
1.11
0.091
0.501
0.290
0.649
100
0
0
0
30
0
0
—
—
—
—
0.277
0.210
0.449
0.630
—
—
—
—
100
0
0
0
0
0
yes
—
—
—
—
0.234
0.478
0.703
1.52
—
—
—
—
0
100
0
0
0
0
0
—
—
—
—
0.233
0.444
0.648
1.33
—
—
—
—
0
100
0
0
0
0
yes
—
—
—
—
0.052
0.565
0.191
0.474
—
—
—
—
100
0
0
0
0
0
0
—
—
—
—
1.74
0.799
11.0
39.7
—
—
—
—
100
0
0
0
0
0
0
0.352
0.241
0.614
0.906
0.251
0.235
0.431
0.629
0.103
0.389
0.253
0.473
100
0
125
0
0
0
0
0.239
0.245
0.420
0.622
0.226
0.182
0.345
0.462
0.221
0.098
0.277
0.324
100
0
250
0
0
0
0
0.314
0.123
0.417
0.509
0.209
0.155
0.298
0.383
0.068
0.345
0.150
0.262
Table 6.7 Values of Langmuir isotherm fitting parameters (KL and Qmax) for adsorption of arsenic onto tested media Media
AA-FS50
Bayoxide E33
61
©2005 AwwaRF. All rights reserved.
GFH
MetSorb G
SMI-III Z33 Rev. B
pH 7 or 7.5 (NSF challenge water)
As(V) (µg/L)
As(III) (µg/L)
PO4 (µg/L)
SiO2 (mg/L)
V (µg/L)
F (µg/L)
NSF challenge water
Qmax
KL
Qmax
KL
Qmax
KL
100
0
0
0
0
0
0
3.996
0.038
2.210
0.203
0.664
1.826
100
0
125
0
0
0
0
2.084
0.145
1.777
0.198
1.003
0.436
100
0
250
0
0
0
0
1.597
0.223
1.469
0.229
0.868
0.424
100
0
0
0
0
700
0
—
—
2.355
0.051
—
—
100
0
0
0
0
0
0
2.120
0.339
1.878
0.204
1.570
0.225
100
0
125
0
0
0
0
1.586
0.182
1.274
0.109
1.435
0.051
100
0
250
0
0
0
0
1.374
0.112
1.637
0.036
1.744
0.023
100
0
0
13.5
0
0
0
1.596
0.316
1.255
0.192
1.517
0.032
100
0
0
22
0
0
0
1.730
0.186
19.803
0.001
4.080
0.006
100
0
0
0
30
0
0
—
—
0.697
0.055
—
—
100
0
0
0
0
0
yes
—
—
2.252
0.027
—
—
0
100
0
0
0
0
0
—
—
2.106
0.076
—
—
0
100
0
0
0
0
yes
—
—
–0.843
-0.004
—
—
100
0
0
0
0
0
0
2.048
0.438
1.515
0.484
0.812
0.228
100
0
125
0
0
0
0
1.582
0.578
1.096
0.943
0.639
0.437
100
0
250
0
0
0
0
1.257
0.539
0.740
0.750
0.717
0.089
100
0
0
13.5
0
0
0
1.219
0.512
1.078
0.194
0.453
0.806
100
0
0
22
0
0
0
1.187
0.769
1.040
0.147
0.558
0.042
100
0
0
0
30
0
0
—
—
1.304
0.008
—
—
100
0
0
0
0
0
yes
—
—
0.951
0.056
—
—
0
100
0
0
0
0
0
—
—
5.521
0.015
—
—
0
100
0
0
0
0
yes
—
—
1.184
0.013
—
—
100
0
0
0
0
0
0
2.005
0.361
1.852
0.200
1.497
0.236
100
0
125
0
0
0
0
1.788
0.239
1.386
0.291
1.394
0.073
100
0
250
0
0
0
0
2.121
0.182
1.988
0.073
1.498
0.062
100
0
0
13.5
0
0
0
2.304
0.128
3.064
0.014
1.139
0.041
100
0
0
22
0
0
0
2.480
0.067
2.068
0.023
1.259
0.022
100
0
0
0
30
0
0
—
—
0.789
0.082
—
—
100
0
0
0
0
0
yes
—
—
2.321
0.038
—
—
0
100
0
0
0
0
0
—
—
3.600
0.026
—
—
0
100
0
0
0
0
yes
—
—
1.193
0.013
—
—
100
0
0
0
0
0
0
—
—
92.8
0.015
—
—
100
0
0
0
0
0
0
1.072
0.111
0.822
0.071
0.757
0.034
100
0
125
0
0
0
0
0.792
0.075
0.550
0.111
0.356
0.204
100
0
250
0
0
0
0
0.576
0.155
0.463
0.099
0.434
0.031
pH 6
pH 8
Figure 6.9 Single-solute isotherms for As(V) at pH 6.0: in the absence of interferents
Figure 6.10 Single-solute isotherms for As(V) at pH 7.0: in the absence of interferents
62 ©2005 AwwaRF. All rights reserved.
Figure 6.11 Single-solute isotherms for As(V) at pH 8.0: in the absence of interferents
Figure 6.12 Summary of interference factors: ratio of As(V) capacity to that at pH 6.0 in the absence of interferents
63 ©2005 AwwaRF. All rights reserved.
Figure 6.13 Adsorption capacity, Q10, for As(V) as a function of pH: absence of interferants
Figure 6.14 Adsorption capacity, Q10, for As(V) as a function of pH: presence of phosphate at 50% occurrence level
64 ©2005 AwwaRF. All rights reserved.
Figure 6.15 Adsorption capacity, Q10, for As(V) as a function of pH: presence of phosphate at 75% occurrence level
Figure 6.16 Adsorption capacity, Q10, for As(V) as a function of pH: presence of silica at 50% occurrence level
65 ©2005 AwwaRF. All rights reserved.
Figure 6.17 Adsorption capacity, Q10, for As(V) as a function of pH: presence of silica at 75% occurrence level
Figure 6.18 Adsorption capacity, Q10, for As(V) as a function of pH: presence of vanadium at 75% occurrence level
66 ©2005 AwwaRF. All rights reserved.
Figure 6.19 Isotherms for As(III) at a pH of 7.0: absence of interferants
Figure 6.20 Adsorption capacity, Q10, for As(III) as a pH of 7.0: absence of interferants
67 ©2005 AwwaRF. All rights reserved.
synthetic waters. First, in addressing the performance of each adsorbent independently, the following observations can be made. The AA-FS50 showed a moderate deterioration in performance as pH increased, and phosphate exerted only a small effect. An experiment with AA-FS50 in the presence of fluoride (Tables 6.6 and 6.7) showed a significant decrease in Q10 and Q50 but curiously little effect on Qmax. This result may perhaps be attributable to a mathematical anomaly in least-squares estimation of the two Langmuir constants. The Bayoxide E33 media showed only a slight pH effect but exhibited significant impacts by phosphate and vanadium with some silica effects. GFH showed a decrease in performance as pH increased, with phosphate, silica, and vanadium exhibiting adverse effects. MetSorb G showed a decrease in performance as pH increases, with phosphate, silica, and vanadium exhibiting effects, although the interferent effects were less at lower pH. Z33 performance decreased as pH increased and phosphate had a significant effect. Differences in physical/chemical properties of the adsorbents can help explain some of the differences. The order of pHZPC (high to low) was Bayoxide E33 > GFH > AA-FS50 > Z33 ~ Metsorb G. The higher pHZPC of the Bayoxide E33 versus GFH helps explain the better performance of the former iron oxide at higher pH conditions. The order of surface area (high to low) was AA-FS50 ~ GFH > Metsorb G > Bayoxide E33 > Z33. The very low surface area of the Z33 is manifested in a lower number of adsorption sites per unit mass. The material composition differ significantly; aluminum and iron content for AA-FS50, titanium dioxide for Metsorb G, clay content for Z33, and iron oxides for GFH and Bayoxide E33. The high capacity exhibited by the Metsorb E33 may involve a more complicated mechanism than simply surface complexation or ligand exchange. Second, in addressing adsorbent comparisons, the following observations can be made. The two iron oxide media (Bayoxide E33 and GFH) and Metsorb G generally showed comparable weight-based capacities although GFH performance dropped sharply at the highest pH; interferant effects were lowest for Metsorb G. The iron-modified activated alumina (AA-FS50) showed a higher capacity than the two iron oxide media and Metsorb G at lower pH, but a lower capacity at higher pH, indicating more of a pH dependency. The iron-modified zeolite (Z33) generally exhibited the lowest capacities. Based on the one triplicated experiment with a specific synthetic water (As(V) at pH 7.0 in the absence of interferants), the sulfur-modified iron (SMI) clearly exhibited the highest capacity for As(V) compared to the other five adsorbents in terms of Q10, Q50, and Qmax. As indicated earlier, the SMI was particularly problematical to evaluate because it was not possible to administer small amounts (doses) in batch tests because of its high density and large size (difficult to reduce with a simple mortar and pestle). At a pH of 7.0 in the absence of interferants, the Q10, Q50, and Qmax values for SMI were found to be 11.0, 39.6, and 92.8 µg/mg, respectively. While a significant experimental error occurred (Table 6.5), these parameter values nevertheless indicate a higher capacity for SMI compared to the other five adsorbents in the absence of interferants and at a pH of 7.0. The capacities discussed above are based on a dry weight basis. Given that the GFH was tested in its wet state containing 50% water by weight, its reported capacities would double if expressed on a dry weight basis. Moreover, the capacities reported herein are based on mass. These capacities could also be expressed in an alternative format based on surface area determinations (i.e., µg/cm2). For example, the wet weight based capacity for GFH at pH 7.0 in the absence of interferants was 1.55 µg/mg, based on Q50. On a dry weight basis, this corresponds to 3.1 µg/mg and, on a surface area basis, this corresponds to 0.063 µg/cm2 (wet) or 0.126 µg/cm2 (dry). One can also express capacities based on unit cost; this is indirectly addressed in Chapter 7.
68 ©2005 AwwaRF. All rights reserved.
For the vast majority of results, the Freundlich parameter exhibited values of 1/n < 1, indicating favorable adsorption. However, given the mathematical (i.e., curve fitting) linkage between 1/n and KF, the presence of an interferant did not result in clear increases in 1/n that would indicate a reduction in favorability. Hence, the capacity factors, Q10, Q50, and Qmax, are considered to be more insightful into interferant effects. Another point to be made is that, in the graphical representations of Freundlich relationships, the graphs are shown forced through the zero-zero intercept. While such an approach is consistent with theory (i.e., no (zero) adsorption in the presence of no (zero) driving force), this represents an extrapolation of experimental data. The results based on the NSF challenge water are shown in Tables 6.6 and 6.7 and are highlighted in Figures 6.21 through 6.24. The isotherm constants were determined by regression analysis of the linear forms of the Freundlich and Langmuir equation. In the vast majority of cases, the coefficient of determination (r2) was greater than 0.8, corresponding to a correlation coefficient (r) of about 0.9. The As(V) results are shown in Figure 6.21, portraying isotherms for three adsorbents (Bayoxide E33, GFH, and Metsorb G), and Figure 6.22, portraying Q10 values for these same three adsorbents. The order of capacities is Metsorb G > Bayoxide E33 > GFH. The higher pH of the NSF challenge water (pH 7.5) was likely the most influential factor on the GFH. The As(III) results are shown in Figure 6.23, portraying isotherms for three adsorbents (Bayoxide E33, GFH, and Metsorb G), and Figure 6.24, portraying Q10 values for these same three adsorbents. The order of capacities is Metsorb G >GFH > Bayoxide E33. The Bayoxide E33 was clearly less effective than the other two adsorbents whose performance was generally comparable. Compared to the single interferant tests previously discussed, the As(V) capacities with the NSF challenge water, based on Q10, Q50, and Qmax, were in some cases lower and in some cases higher; this can be attributed to the higher levels used for certain individual interferants (e.g., phosphate). On the other hand, the As(III) capacities with the NSF challenge water were uniformly lower that that observed for As(III) at pH 7.0 in the absence of interferants. Utility-Supplied Waters. Three adsorbents (Bayoxide E33, GFH, and Metsorb G) were tested in four different water quality matrices corresponding to the four natural waters provided by utilities. All of the waters contained ambient levels of arsenic and were spiked up to a total of 100 µg/L. The Tucson AZ and LADWP CA waters contained only As(V) and thus were spiked up to a total of 100 µg/L of As(V). The El Paso TX and Alamosa CO waters contained both As(V) and As(III) and thus were spiked with both As(V) and As(III), based on their ratios of As(V):As(III) (~1:1 for El Paso TX and ~ 2.6:1 for Alamosa CO), up to a total of 100 µg/L of total arsenic. The measurements of As(V) and As(III) shown in Table 6.4 were made immediately before experiments and it was assumed that these levels did not change over the 24-hour time frame of the experiments. The results are shown as isotherms in Figures 6.25, 6.26, 6.27, and 6.28 for the Tucson AZ, LADWP CA, El Paso TX, and Alamosa CO waters, respectively. For the Tucson AZ and the LADWP CA waters, the capacities followed the order: Metsorb G > Bayoxide E33 > GFH. For the El Paso TX water, Metsorb G and Bayoxide E33 provided significantly higher capacities than the GFH. Very low capacities were observed for all of the adsorbents with the Alamosa CO water. This latter water is characterized by very high levels of silica and vanadium. The observed capacities for the El Paso TX and Alamosa CO waters reflects their As(III) content. Some of the isotherms for utility-supplied waters show unfavorable adsorption, with corresponding 1/n values of greater than 1.0. In some cases, the isotherms shown encompass a significant extrapolation of data toward the zero-zero intercept.
69 ©2005 AwwaRF. All rights reserved.
Figure 6.21 Isotherms for As(V) in NSF challenge water
Figure 6.22 Adsorption capacity, Q10, for As(V) in NSF challenge water
70 ©2005 AwwaRF. All rights reserved.
Figure 6.23 Isotherm for As(III) in NSF challenge water
Figure 6.24 Adsorption capacity, Q10, for As(III) in NSF challenge water
71 ©2005 AwwaRF. All rights reserved.
Figure 6.25 Isotherms for utility-supplied water: Tucson, AZ groundwater
Figure 6.26 Isotherms for utility-supplied water: LADWP, CA surface water
72 ©2005 AwwaRF. All rights reserved.
Figure 6.27 Isotherms for utility-supplied water: El Paso, TX groundwater
Figure 6.28 Isotherms for utility-supplied water: Alamosa, CO groundwater
73 ©2005 AwwaRF. All rights reserved.
Special Study on Sulfur Modified Iron (SMI) While sulfur modified iron (SMI) performed well in the preliminary screening tests in terms of a high As(V) adsorption capacity, it was not formally included in the detailed intensive testing program because of concerns about experimental error and reproducibility (see previous section). Given its large grain size and its high density, accurate testing would have required a major revision to the general batch-testing protocol used for all of the other adsorbents such as using larger sample volumes and/or higher constituent concentrations. Nevertheless, a decision was made to perform additional experiments using the standard protocol to provide some semiquantitative insight into the effects of water quality on its adsorption capacity. These results are summarized in Table 6.8 in terms of Freundlich and Langmuir isotherm constants and isothermbased predictions of capacity (Q10, Q50, and Qmax). The isotherm constants were determined by regression analysis of the linear forms of the Freundlich and Langmuir equation. In a significant number of cases, the coefficient of determination (r2) was less than 0.8, corresponding to a correlation coefficient (r) of about 0.9. Table 6.8 reveals some trends but also some inconsistencies in the data. A clear result is that SMI exhibits a very high As(V) capacity at lower pH levels (6.0 and 7.0) in the absence of interferants. Moreover, the presence of interferants lowers its As(V) capacity, and the As(III) capacity is much lower than the As(V) capacity. However, other than these general trends, elucidation of other trends are convoluted by several inconsistencies. Data inconsistencies include: (i) the results of the QA/QC triplicate series do not compare well with the results of the corresponding single replicate performed in the special study; (ii) some values of Q50 are greater than corresponding values of Qmax which, in theory, is not possible; (iii) values of Q10 are greater than corresponding values of Q50 in a few cases which is not possible; (iv) Q values at a lower interferant are less than corresponding Q values at a higher interferant level in a few cases; and (v) capacities observed in the NSF challenge water (phosphate = 40 µg/L, silica = 20 mg/L, pH = 7.5) are not completely consistent with individual interferant levels. While SMI clearly exhibits a very high As(V) under lower pH levels in the absence of interferants, there have been concerns raised about operational problems in using SMI including: (i) a release of soluble iron, Fe(II); (ii) a release of reduced sulfur in the form of hydrogen sulfide, H2S; and (iii) column clogging as a consequence of redox reactions (e.g., release of soluble iron (Fe(II) and subsequent oxidation to insoluble iron (Fe(III)). The manufacturer has tried to address these problems by product improvements through various versions of SMI (SMI Version III was tested in this study) and through changes in operation of the media. In addition to operating the media in an upflow configuration, the manufacturer now recommends the addition of post treatment filtration to prevent iron carryover. While a precise mineral identity is not available, it is believed that the iron present in SMI is zero valent iron, Fe(0), that is oxidized to iron(III)(hydr)oxides with a high adsorption capacity through an initial corrosion or rust reaction followed by subsequent oxidation. However, the corrosion or rust reaction is influenced by redox (dissolved oxygen) and pH conditions, further contributing to reproducibility problems. This overall reaction pathway is quite complex and the corrosion rate will vary in different water qualities. The pathway for the corrosion or rust reaction
74 ©2005 AwwaRF. All rights reserved.
Table 6.8 Summary of SMI results
As(V) (µg/L)
As(III) (µg/L)
pH
Interferant
Interferant level
7.0
None
0
Freundlich constants* KF
1/n
Isotherm predictions (µg/mg)
Langmuir constants* KL
Qmax
Q10
Q50
Qmax
92.8
11.0
39.7
92.8
Triplicate Series 100
0
Single Replicates
75
©2005 AwwaRF. All rights reserved.
100
0
6.0
None
0
18.1
0.352
0.224
83.0
40.6
71.6
83.0
100
0
7.0
None
0
7.04
0.474
0.241
28.5
21.0
44.9
28.5
100
0
8.0
None
0
0.663
0.223
0.033
2.57
1.11
1.59
2.57
100
0
6.0
Phosphate
125 µg/L
9.89
0.350
0.991
19.4
22.1
38.9
19.4
100
0
6.0
Phosphate
250 µg/L
12.7
–0.168
–1.40
5.73
8.62
6.58
5.73
100
0
7.0
Phosphate
125 µg/L
6.82
0.198
0.194
15.7
10.8
14-8
15.7
100
0
7.0
Phosphate
250 µg/L
251
–0.726
–0.021
3.39
47.2
14.7
3.39
100
0
8.0
Phosphate
125 µg/L
0.105
0.483
0.008
2.35
0.32
0.70
2.35
100
0
8.0
Phosphate
250 µg/L
0.024
0.723
0.003
3.25
0.13
0.41
3.25
100
0
6.0
Silica
13.5 mg/L
13.7
0.061
0.152
19.6
15.8
17.4
19.6
100
0
6.0
Silica
22 mg/L
1.07
0.217
0.079
3.82
1.76
2.50
3.82
100
0
7.0
Silica
13.5 mg/L
0.863
0.169
0.034
2.47
1.27
1.67
2.47
100
0
7.0
Silica
22 mg/L
0.418
0.335
0.022
3.28
0.91
1.55
3.28
100
0
8.0
Silica
13.5 mg/L
0.012
0.726
0.002
1.97
0.06
0.21
1.97
100
0
8.0
Silica
22 mg/L
0.120
0.243
0.019
0.57
0.21
0.31
0.57
100
0
7.0
Vanadium
30 µg/L
0.103
0.538
0.006
3.49
0.35
0.84
3.49
100
0
7.5
NSF
challenge water
2.73
0.543
0.677
7.05
9.54
22.9
7.05
0
100
None
0
0.235
0.662
0.060
1.14
1.08
3.13
1.14
0
100
NSF
challenge water
0.504
0.376
0.045
3.16
1.20
2.20
3.16
*Based on Q and C expressed as µg/mg and µg/L, respectively.
followed by subsequent oxidation is shown below, with the resultant am-Fe(OH)3(s) having a high capacity for As(V) adsorption: 2 Fe(0) + O2 + 2 H2O → 2 Fe2+ + 4 OH– 4 Fe2+ + O2 → + 2 H+ → 4 Fe3+ + 2 OH– Fe3+ + 3 OH– → am-Fe(OH)3(s) Another hypothetical reaction that might occur would be the reduction of As(V) to As(III) by zero valent iron, Fe(0): Fe(0) + As(V) → Fe2+ + As(III) However, the resultant As(III) would be adsorbed to a lesser degree than As(V) by the am-Fe(OH)3(s) so, if this reaction occurs, it does appear to be dominant. The role of the sulfur present in SMI is less clear. If it is present as a sulfide, oxidation should release sulfate; if it is present as elemental S, reduction would yield sulfide. Even with the indicated QA/QC concerns, a decision was made to include SMI as part of the decision/cost framework presented in Chapter 7 with a warning given to the user about the degree of accuracy of the SMI algorithms. Given the potential merits of SMI, one of the recommendations for future research is that a separate focused study should be performed on SMI (see Chapter 8). Special Study on Silica Effects Based on results demonstrating that silica can be a significant interferant, a targeted set of experiments was performed with one adsorbent, GFH. The chemistry of silica can be complex with monomeric silica (silicic acid/silicate) dominating at lower concentrations and higher pH levels, while polymeric silica dominates at higher concentrations and lower pH. Monomeric silica can play a role as an interferant at higher pH levels above its pKa1 (9.5). Polymeric silica can foul adsorbents and potential block access to the pores of porous adsorbents by adsorbate molecules. First, a matrix of experiments was performed in which various solutions of silica were prepared at varying concentrations and pH conditions, with resultant measurements of monomeric and polymeric silica after equilibration (see Chapter 5). These results are shown in Table 6.9, demonstrating the expected trends of polymeric silica formation, calculated by the difference between total silica and monomeric silica. Next, a series of isotherms were performed with As(V) at 100 µg/L at each of two pH conditions (6.5 and 7.5) in the presence of each of three silica concentration ranges (19.3–21.8, 32.5–33.4, and 687–830 mg/L). The results are shown in Figure 6.29. These results indicate an inhibition of As(V) adsorption at both pH conditions but only at the highest silica concentration range (≥687 mg/L). However, interpolation between the intermediate silica concentration range and the highest concentration range, between 33.4 and 687 mg/L suggests that significant silica effects can be anticipated in natural waters such as the Alamosa CO water. While the results of this special study were based on GFH, similar fouling effects can be anticipated for other porous adsorbents (i.e., all of the other select adsorbents except for SMI).
76 ©2005 AwwaRF. All rights reserved.
Table 6.9 Monomeric versus polymeric silica concentrations Silica dose (mg/L)
pH
Total Si (mg/L)
Monomeric Si (mg/L)
Polymeric Si (mg/L)
30
4.5
34.0
29.4
4.6
80
4.5
89.7
79.2
10.5
200
4.5
227
189
38.0
500
4.5
514
448
66.0
30
6.5
34.0
29.5
4.5
80
6.5
91.4
78.8
12.6
200
6.5
214
189
25.0
500
6.5
496
188
308
30
10
38.7
29.9
8.8
80
10
99.7
77.8
21.9
200
10
207
185
21.7
500
10
494
430
64.3
30
13
35.1
8.5
pH > 8.5
pH > 8.5
pH > 8.5
6
6
6
6
6
pH Phosphate Silica Vanadium
• • • • •
pH Adjustment Recommended For Raw Water
pH > 7.5
pH > 8.5
pH > 8.5
Suggested Optimal pH
6
6
• • • •
pH Phosphate Silica Vanadium
pH Phosphate Silica? Vanadium?
• pH • Silica • Phosphate
• pH • Silica • Phosphate
• • • •
Significant Water Quality Parameters (in the Order of Significance)
Residuals Handling and Disposal Spent Media Disposal
Landfill
Landfill
Landfill
Landfill
Landfill
Landfill
Landfill
Spent Backwash Disposal
Sewer/Storm Drain/Wash or Recycled after Filtration
Sewer/Storm Drain/Wash or Recycled after Filtration
Sewer/Storm Drain/Wash or Recycled after Filtration
Sewer/Storm Drain/Wash or Recycled after Filtration
Sewer/Storm Drain/Wash or Recycled after Filtration
Sewer/Storm Drain/Wash or Recycled after Filtration
Not Applicable
(Continued)
Table 7.1 (Continued) Parameter
AA-400G
AA-FS50
Bayoxide E-33
GFH
Metsorb G
Z-33 Rev B
SMI
Unit Media Cost ($/lb or $/ft3)
For 2,000 lbs, $0.50/lb
For 2,000 lbs, $0.82/lb or $38.50/ft3
$5.50/lb-dry or $159.50/ft3
For 15,000 lbs, $2.50/lb or $187.50/ft3
$6/lb-dry or $294/ft3
$0.25/lb-dry
$4/lb-dry
Capital Cost Includes
• • • • • • •
Pressure Vessels Influent Strainer Media Transfer System Concrete, Valves, Pipes, Electrical and Instrumentation Site Work Permitting, Piloting, Operator Training pH Adjust System (where used)
O & M Cost Includes
• • • • • •
Purchase of New Media for Replacement Spent Media Transportation and Disposal Chemical (if pH is adjusted) Labor (to Operate and Maintain) Process Monitoring Analytics Energy
Cost Assumptions
85
©2005 AwwaRF. All rights reserved.
The Arsenic Adsorbent Design and Costing Tool is presented solely for informational purposes for utilities and individuals to research initial design and cost options for the adsorption treatment of arsenic using various adsorbents. By using the tool with project specific flow rates and water quality parameters, utilities can assess basic design options and costs associated with the various adsorption media available in the marketplace. SINGLE PARAMETER AND MULTIPLE REGRESSION MODELS The water quality effects on the adsorptive capacities for arsenic by the various adsorbents were modeled using single parameter and multiple regression models. These models were developed using the data from bench-scale testing generated during the bench testing summarized in Chapter 6. Single Parameter Model The single parameter model is used when the user enters only arsenic and pH information, and no other water quality information. In the single parameter model, the Freundlich and Langmuir coefficients were modeled based on raw water pH. The equations for the various adsorbents for the single parameter model are shown in Table 7.2. The R2 value indicates the correlation between measured and predicted adsorption capacities, with a higher R2 value indicating a better correlation. For example, the single parameter model equations for GFH are: •
•
Freundlich Model – KF = 0.853 – 0.084 (pH) – 1/n = 1.5243 – 0.143 (pH) Langmuir Model – KL = 1.1183 – 0.105 (pH) – Qmax = 5.783 – 0.618 (pH)
(7.1) (7.2) (7.3) (7.4)
Among the various water quality parameters, pH has the most significant effect on adsorbents performance. Therefore, pH was chosen as the modeling parameter for single parameter models. Table 7.2 identifies the pH limits (i.e., boundary conditions) over which the single parameter models are applicable. The pH boundary conditions shown on the table corresponds to the range of pH levels that were tested in bench-scale tests. Multiple Regression Model Multiple regression models were developed using Statistica® (StatSoft, Tulsa, OK), a statistical software package. Multiple regression models account for interferences from water quality parameters such phosphate, silica, vanadium and fluoride, in addition to the effect of pH. As discussed in earlier sections, performance of arsenic adsorbents is greatly influenced by cooccurring ions such as phosphate and silica. The effect of the co-occurring ions was modeled using linear multiple regression modeling. In linear multiple regression modeling, the independent variables (Xi; e.g., pH, phosphate, silica, vanadium, fluoride) are placed into equations to predict their partial correlation coefficients
86 ©2005 AwwaRF. All rights reserved.
Table 7.2 Single parameter and multiple regression models Sorbent
Models
87
©2005 AwwaRF. All rights reserved.
Equations KF = 0.853 – 0.084(pH) 1/n = 1.5243 – 0.143(pH) Single Parameter KL = 1.1183 – 0.105(pH) Qmax = 5.7843 – 0.618(pH) GFH KF = 0.6791 – 0.1289(pH–6.0) – 0.00029(PO4) – 0.010975(SiO2) – 0.017583(V ) 1/n = 0.2943 – 0.0575(pH–6.0) – 0.00042(PO4) + 0.000747(SiO2) + 0.15756(V) Multiple Regression KL = 0.5346 – 0.072339(pH–6.0) + 0.00839(PO4) – 0.00387(SiO2) – 0.015145(V) Qmax = 1.8584 – 0.449943(pH–6.0) – 0.00251(PO4) – 0.02582(SiO2) – 0.003492(V) KF = 0.9978 – 0.0505(pH) 1/n = 0.486 – 0.033(pH) Single Parameter KL = 0.655 – 0.057(pH) Qmax = 3.781 – 0.275(pH) KF = 0.813 – 0.16442(pH–6.0) – 0.001544(PO4) – 0.01936(SiO2) – 0.015276(V ) Bayoxide 1/n = 0.1066 + 0.090358(pH–6.0) + 0.000654(PO4) + 0.018384(SiO2) + 0.001825(V) Multiple Regression KL = 0.3269 – 0.079642(pH–6.0) – 0.000821(PO4) – 0.00619(SiO2) – 0.006403(V) Qmax = 1.3742 + 0.1940255(pH–6.0) – 0.000165(PO4) + 0.035823(SiO2) – 0.029029(V) KF = –0.732 + 0.172(pH) KF = 1.865 – 0.199(pH) 1/n = 0.8303 – 0.051(pH) Single Parameter KL = – 0.952 + 0.165(pH) KL = –11.158 + 1.623(pH) FS–50 Qmax = 13.952 – 1.666(pH) KF = 0.3913 – 0.0341(pH–6.0) + 0.00014(PO4) 1/n = 0.5122 – 0.0491(pH–6.0) – 0.00064(PO4) Multiple Regression KL = 0.232 + 0.379999(pH–6.0) – 0.00159(PO4) Qmax = 3.0874 – 0.8571(pH–6.0) – 0.00391(PO4)
R2 * Water quality limits 0.46 6.0 ≤ pH ≤ 8.0 0.62 6.0 ≤ pH ≤ 8.0 0.93 If pH > 8.0 and 125 µg/L ≤ PO4 ≤ 250 µg/L, Set PO4 = 125 µg/L 0.89 If pH > 8.0 and 125 µg/L ≤ PO4 ≤ 250 µg/L, Set PO4 = 125 µg/L 0.44 6.0 ≤ pH ≤ 8.0 0.00 6.0 ≤ pH ≤ 8.0 0.52 If 7.5 ≤ pH ≤ 8.5, adjust pH = 6.0 If 100 µg/L ≤ PO4 ≤ 250 µg/L, Set PO4 = 100 µg/L 0.38 If 10 mg/L ≤ SiO2 ≤ 22 mg/L, Set SiO2 = 10 mg/L If 5 mg/L ≤ V ≤ 30 mg/L, Set V = 5 mg/L
0.52 pH ≤ 7.0 pH > 7.0 6.0 ≤ pH ≤ 8.0 0.67 pH ≤ 7.0 pH > 7.0 6.0 ≤ pH ≤ 8.0 0.79 6.0 ≤ pH ≤ 8.0, PO4 ≤ 250 µg/L 0.73 6.0 ≤ pH ≤ 8.0, PO4 ≤ 250 µg/L (Continued)
Table 7.2 (Continued) Sorbent
88
©2005 AwwaRF. All rights reserved.
Metsorb
Z–33
Models
Equations KF = 0.8808 – 0.0385(pH) 1/n = 0.58492 – 0.0405(pH) KL = 1.327 – 0.161(pH) Single Parameter KL = –0.052 + 0.036(pH) Qmax = 3.5627 – 0.254(pH) KF = 0.7481 – 0.138961(pH–6.0) – 0.000712(PO4) – 0.02034(SiO2) – 0.011076(V ) 1/n = 0.2395 + 0.012615(pH–6.0) + 0.000314(PO4) + 0.01216(SiO2) – 0.001398(V) Multiple Regression KL = 0.3108 – 0.054318(pH–6.0) – 0.00057(PO4) – 0.01121(SiO2) – 0.005805(V) Qmax = 2.1453 – 0.391103(pH–6.0) – 0.000003(PO4) + 0.014399(SiO2) – 0.032191(V) KF 1.1068 – 0.1245(pH) 1/n = 0.277 – 0.006(pH) 1/n = – 0.843 + 0.154(pH) Single Parameter KL = 0.3415 – 0.0385(pH) Qmax = 1.9862 – 0.1575(pH) KF = 0.3251 – 0.085545(pH–6.0) – 0.000154(PO4) 1/n = 0.227 + 0.037154(pH–6.0) – 0.000323(PO4) KL = 0.0997 – 0.0122(pH–6.0) + 0.000091(PO4) Multiple Regression Qmax = 0.9923 – 0.148767(pH–6.0) – 0.001573(PO4) KL = 0.0997 – 0.0122(pH–6.0) + 0.000091(PO4) Qmax = 0.9923 – 0.148767(pH–6.0) – 0.001573(PO4)
R2 * Water quality limits 0.60 6.0 ≤ pH ≤ 8.0 0.31 pH ≤ 7.0 pH > 7.0 6.0 ≤ pH ≤ 8.0 0.84 If pH > 8, adjust pH = 6.0 If 125 µg/L ≤ PO4 ≤ 250 µg/L, Set PO4 = 125 µg/L
0.58 If 15 ≤ SiO2 ≤ 22 mg/L, set SiO2 = 15 mg/L If 5 ≤ V ≤ 30 mg/L, Set V = 5 mg/L
0.92 6.0 ≤ pH ≤ 8.0 pH ≤ 7.0 pH > 7.0 0.85 6.0 ≤ pH ≤ 8.0 0.95 6.0 ≤ pH ≤ 8.0, PO4 ≤ 250 µg/L 0.95 6.0 ≤ pH ≤ 8.0, PO4 ≤ 250 µg/L 0.86 6.0 ≤ pH ≤ 8.0, PO4 ≤ 250 µg/L
(Continued)
Table 7.2 (Continued) Sorbent
89
©2005 AwwaRF. All rights reserved.
SMI-III
AA400G
Models
Equations KF = 69.631 – 8.7185(pH) 1/n = – 0.38 + 0.122(pH) 1/n = 2.231 – 0.251(pH) Single Parameter KL = 0.122 + 0.017(pH) KL = 1.697 – 0.208(pH) Qmax = 319.66 – 40.233(pH) KF = 16.52 – 6.9093(pH–6.0) – 0.03769(PO4) – 0.5238(SiO2) – 0.31694(V ) 1/n = 0.3339 + 0.049108(pH–6.0) – 0.000674(PO4) – 0.00765(SiO2) – 0.005107(V) Multiple Regression KL = 0.3447 – 0.238751(pH–6.0) – 0.002934(PO4) – 0.00844(SiO2) – 0.003345(V) Qmax = 62.055 – 18.9939(pH–6.0) – 0.28(PO4) – 2.4591(SiO2) – 1.3192(V) Q (µg/mg)= 1.860271 + 0.001318(As) – 0.237217(pH) – 0.288581(PO4) Multiple Regression – 0.097215(F)
* Based on Qmeasured versus Qpredicted. Higher value indicates better correlations Input Units: PO4 in µg/L; SiO2 in mg/L, V in mg/L, F in mg/L Units for Freundlich constants: KF in µg-As/mg-media; 1/n no unit; Q = µg-As/mg-media Units for Langmuir constants: Qmax in µg-As/mg-media; KL µg/L; Q = µg-As/mg-media
R2 * Water quality limits 0.51 6.0 ≤ pH ≤ 8.0 pH ≤ 7.0 pH > 7.0 0.41 pH ≤ 7.0 pH > 7.0 6.0 ≤ pH ≤ 8.0 0.95 If 7.0 ≤ pH ≤ 8.5, adjust pH = 6.0 If 100 µg/L ≤ PO4 ≤ 250 µg/L, Set PO4 = 100 µg/L 0.78 If 10 mg/L ≤ SiO2 ≤ 22 mg/L, Set SiO2 = 10 mg/L If 5 ≤ V ≤ 30 mg/L, Set V = 5 mg/L 0.67 If 7.0 ≤ pH ≤ 8.5, adjust pH = 6.0 If 150 µg/L ≤ PO4 ≤ 250 µg/L, Set PO4 = 150 µg/L If 4 ≤ F ≤ 10 mg/L, Set F = 4 mg/L
with the dependent variables (Yj; e.g., Freundlich and Langmuir coefficients). The general form of linear multiple regression modeling is: Yj = Co + C1X1 + C2X2 + ………+ CnXn
(7.5)
where Yj is the predicted variable, Co is a constant, and Ci are coefficients for the various independent variables (Xi). The predictability of the model was evaluated through various statistical parameters including regression coefficient (R2), F statistics and p values (probability of null hypothesis). The models were tested by performing internal validation based on the data used in development of the models. Some of the models were also evaluated with natural water conditions based on limited pilot-scale data that were available. The multiple regression model equations for the seven adsorbents are shown in Table 7.2. The water quality limits over which the models are applicable and the R2 values are also shown in this table. Discussed below is an example of linear multiple regression model for GFH. In this example, the model predicts Freundlich constants, KF and 1/n, based on independent variables. These constants are used to predict (column mode) adsorptive capacity, Q (i.e., Equation 7.3) based on water quality and raw water arsenic (within this context, when Ce (equilibrium arsenic concentration) = C0 (initial raw concentration), the predicted Q = Q0 (column capacity)). KF = 0.6791 – 0.1289(pH – 6.0) – 0.00029(PO4) – 0.010975(SiO2) – 0.017583(V)
(7.6)
1/n = 0.2943 – 0.0575(pH – 6.0) – 0.00042(PO4) + 0.000747(SiO2) + 0.15756(V)
(7.7)
Q = KF Ce1/n (R2 = 0.93)
(7.8)
Equations 7.6 and 7.7 predicts the Freundlich constants, KF and 1/n, based on water quality inputs for pH, phosphate, silica and vanadium. Increasing pH (above pH of 6) will decrease the adsorption capacity and this effect is shown by the negative term in Equation 7.6: –0.1289(pH-6.0). The pH of 6.0 was assumed to be the baseline pH. Increasing concentrations of phosphate, silica and vanadium will decrease KF, 1/n and Q (Equation 7.8). The limits for the application of these equations are shown in Table 7.2. DESCRIPTION OF TOOL ALGORITHMS AND CALCULATIONS Various algorithms for the Arsenic Adsorbent Design and Costing Tool are shown in Figures 7.1 to 7.4. These algorithms include the decision logics for costing, media replacement, backwash calculations, and plant footprint calculations. The following is a brief description of the step-by-step calculations that are carried out within the tool: • •
If split-stream treatment is selected, the tool calculates the split-stream flows for design and operation; otherwise the inputted flows are used. Capital costs are calculated based on pre-determined cost equations that account for vessel size, number of vessels, influent Y-strainers, media transfer system, concrete,
90 ©2005 AwwaRF. All rights reserved.
Figure 7.1 Cost flowchart
91 ©2005 AwwaRF. All rights reserved.
Figure 7.2 Media replacement flowchart
Figure 7.3 Backwash calculation flowchart
92 ©2005 AwwaRF. All rights reserved.
Figure 7.4 Plant footprint calculation flowchart
valves, pipes, electrical and instrumentation, media staging area, media for first fill, media transportation, vessel transportation, housing, piloting, permitting, land, operator training, and pH adjustment (if selected or required). The cost equations were developed for each media over the flow range of 10–3500 gpm. The capital costs are then adjusted based on the current Engineering News Record (ENR) Construction Cost Index. •
•
The O&M costs are calculated within the tool based on: – Pre-determined media-specific equations for housing energy, process energy, increased well pumping, and analytical costs. – The other O&M costs (such as media replacement) are calculated based on associated project research that helps in identifying the Freundlich/Langmuir parameters for the user-entered water quality. Adsorption parameters are estimated for each media based on pH and concentrations of interfering ions (phosphate, silica, vanadium and fluoride). The estimated adsorption parameters were used to calculate the arsenic adsorption capacity for each media. Using the adsorption capacities, the annual media replacement costs are calculated. Finally, the costs for pH adjustment chemicals are calculated based on whether pH control has been selected, the raw water pH and alkalinity. The O&M costs are then adjusted based on the current ENR Skilled Labor Index. The total of capital and O&M costs are then presented in several formats including total present worth, total annual cost, and cost per 1000 gallons treated for comparison of the various adsorption alternatives. 93 ©2005 AwwaRF. All rights reserved.
Figure 7.2 illustrates an example algorithm using the Freundlich model that the tool would use to calculate the media replacement quantities. Depending on the water quality parameters inputted, the tool selects either a single parameter model or a multiple regression model. The multiple regression model will be used when the user enters water quality parameters other than pH and arsenic. The media run lengths are estimated using the calculated adsorption capacities. Figure 7.3 embodies the algorithm for calculating the volume of backwash water required per backwash cycle. The tool uses design parameters of suggested backwash rate, suggested loading rate, and suggested backwash time for each media to determine the overall volume of backwash water that will most likely need to be contained for one backwash cycle. Figure 7.4 embodies the flowchart that illustrates the calculations for treatment plant footprint. The tool calculates the required space for all adsorption vessels, backwash tanks, spent media storage, and pH adjustment system tanks (if pH adjustment is selected or required). A 30 percent (%) contingency is then added to the total footprint for items such as safety gear, maintenance systems, pumps, and work space. INSTRUCTIONS FOR USING THE TOOL To begin using the arsenic tool, the user must enter project specific parameters on the input form, choosing either the Freundlich or Langmuir adsorption models. It is recommended that the Freundlich model be used for initial model determinations, followed by the Langmuir model for comparison. By clicking on the input parameters button for either model, a page of input parameters will open for entering the key project specific parameters required for the model calculations. To assist with some of the parameters which may not be known, click the default values button to populate the input parameters page with a set of default parameters. Then, modify the input parameters with any data that is available for the specific project, such as flow rates and water quality parameters. Selecting “yes” for split stream treatment can lead to more cost effective treatment because only a portion of the total flow will be treated and then blended with the by-passed water to meet the treated water arsenic goal. Also, calculations for pH adjustment chemicals are considered when the pH adjustment box is checked. Generally, the optimum pH for arsenic adsorption is pH 6. For pH 6 to 8, pH adjustment may not be necessary, although it may still be economical because of higher adsorption capacity at adjusted pH conditions. If pH is between 6 and 8, results with/without pH adjustment should be compared for the most economical treatment. For pH values greater than 8, pH adjustment is recommended. For pH values greater than 8.5, the tool automatically assumes pH adjustment will be required. On the input page, there is also a button to enter optional water quality parameters. These parameters can affect the adsorption of arsenic onto specific media types and the model accounts for these water quality parameters during calculations. It is only necessary to enter values for those optional water quality parameters that are known for the specific project water. The model will use default values for any parameters that are not entered. In addition to the seven arsenic adsorbents built into the model, the ability to enter a user defined adsorbent has been built into the model to account for newer adsorbents. A user defined adsorbent may be entered by clicking on the corresponding button on the input form and entering the required data for the adsorbent.
94 ©2005 AwwaRF. All rights reserved.
Once the input form is completed, press the “Submit” button and the tool will calculate all of the output data and bring the user back to the main tool menu. Clicking on the output button will bring the user to the output page. Tool output consists of treatment costs, media replacement, backwash water requirements, and plant footprint estimates. The output page can be printed for comparison of several runs. The outputs can be saved by clicking on the “Save Output” button. From the output page, clicking on the “Return to Main Menu” will bring back the main menu from which new information can be entered to test other scenarios. In using the software to assess sulfur modified iron (SMI), the user is issued a “warning message,” advising of the lesser reliability of the SMI algorithms compared to the other adsorbents. The issue of SMI data reliability was discussed in Chapter 6. Through a critical evaluation of SMI data (see Table 6.8), only 9 of the 16 data sets were used in algorithm development. Thus, the tool can be used to assess SMI but the results can be considered only semi-quantitative. Nevertheless, they have some value in screening the merits and demerits of SMI compared to the various adsorbents. BASIS OF COST OPINIONS All cost data generated by the Arsenic Adsorbent Design and Cost Tool represent conceptual-level opinions of probable capital and operating and maintenance (O&M) costs for the specific facilities selected by the user. Capital cost opinions are expressed in February 2004 dollars, corresponding with an Engineering News Record (ENR) 20-Cities Average Construction Cost Index (CCI) of 6861.1. O&M cost opinions are expressed in February 2004 dollars, corresponding with an Engineering News Record (ENR) 20-Cities Average Skilled Labor Index (SLI) of 6659.7. The user can adjust capital and O&M cost opinions to another time period by entering different input values for the ENR CCI and SLI. The conceptual cost opinions generated by the Arsenic Adsorbent Design and Cost Tool are considered to fall within the range of Class 5 to Class 4 estimates as defined by the Association for the Advancement of Cost Engineering (AACE) International. These levels of engineering cost estimating are generally conducted on the basis of limited preliminary information and without detailed information such as process and instrumentation diagrams, engineering layouts, and equipment schedules. This level of cost estimating is appropriate for strategic planning purposes, assessment of initial viability, evaluation of alternative plans, project feasibility screening, and long range capital planning. Typical accuracy ranges recognized for Class 5 to Class 4 estimates are –30% to +50%.
95 ©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
CHAPTER 8 CONCLUSIONS Extensive literature and vendor surveys revealed that there are a large number of adsorbents, commercially available and experimental, available for arsenic removal. Some of the commercially available materials are mature products (e.g., activated alumina) that have been widely tested, revealing both their attributes and limitations, while others are more recent products that have not been as rigorously tested (e.g., various iron oxides). Based on their predominant adsorption mechanism as well as material composition, these adsorbents fall into several categories: (i) ion exchange media (e.g., MIEX); (ii) metal oxides (e.g., activated alumina and iron oxides); and (iii) redox-reactive media (e.g., MnO2). In this study, we have chosen to highlight the second category in the forms of both pure minerals and amended/impregnated materials. The literature review also revealed that adsorption capacities were influenced by arsenic speciation, arsenate (As(V)) versus arsenite (As(III)), and water quality in terms of pH conditions and the presence on interfering species competing with arsenic for adsorption sites. Influential interferants were revealed to include silica, phosphate, fluoride, sulfate, carbonate, and others. Based on the literature and vendor surveys, a total of 12 adsorbents, 10 commercially available and 2 experimental, were identified for experimental evaluation. Besides pH as an important water quality condition, phosphate and silica were identified for intensive interferant testing, later supplemented by vanadium and fluoride for less intensive assessment. An arsenic occurrence and co-occurrence survey was conducted to indicate national (USA) occurrence levels of total arsenic, arsenic species domination, and co-occurrence trends between arsenic and interferants. Based on statistical analysis of an integrated national database for groundwater, the mean total arsenic concentration was determined to be 4.5 µg/L, with a 90 percentile value of about 10 µg/L. Geographically, the highest national levels were found in the Intermontane region encompassing states west of the continental divide (e.g., Arizona and New Mexico) except for Washington, western Oregon, and northern California; lowest levels were found in the Atlantic Plain region. Binning of data showed that about 5% of all groundwater sites contained ≥20 µg/L, about 5% contained between 10 to 20 µg/L, and about 5% between 5 to 10 µg/L. While a rigorous study of arsenic speciation in groundwater was not found, several limited studies suggested that As(V) is the predominant species in U.S. groundwaters although, in other global settings such as Bangladesh, As(III) is known to dominate in more anoxic groundwaters. There were semi-quantitative trends to indicate that groundwaters with arsenic ≥5 µg/L tended to be higher in pH, silica, fluoride, alkalinity, and phosphate than those with GFH > AA-FS50 > Z33 ~ Metsorb G. The higher pHZPC of the Bayoxide E33 versus GFH helped explain the better performance of the former iron oxide at higher pH conditions. The order of surface area (high to low) was AA-FS50 ~ GFH > Metsorb G > Bayoxide E33 > Z33. The very low surface area of the Z33 was manifested in a lower number of adsorption sites per unit mass. The material composition differed significantly: aluminum and iron content for AA-FS50, titanium dioxide for Metsorb G, clay content for Z33, and iron oxides for GFH and Bayoxide E33. The high capacity exhibited by the Metsorb E33 may involve a more complicated mechanism than simply surface complexation or ligand exchange. Three of the adsorbents; Bayoxide E33, Metsorb G, and GFH; were tested for arsenite removal; while all three adsorbents demonstrated a significant potential for As(III) removal, their As(III) capacities were significantly less than their As(V) capacities; the order of performance was Bayoxide E33 > Metsorb G > GFH. Three adsorbents (Bayoxide E33, Metsorb G, and GFH) were tested in a NSF challenge water containing As(V) or As(III) with multiple interferants at a constant pH (7.5) as well as several utility-supplied natural waters containing multiple interferants and variable pH. The order of capacities for As(V) in the NSF challenge water was Metsorb G > Bayoxide E33 > GFH, while
98 ©2005 AwwaRF. All rights reserved.
the order of capacities for As(III) was Metsorb G >GFH > Bayoxide E33. The general order of capacities for the natural waters was Metsorb G > Bayoxide E33 > GFH. Results from a special batch study on SMI revealed a higher As(V) capacity than that of the other adsorbents; however, its performance was substantially reduced at higher pH conditions and/or in the presence of interferants. Moreover, its As(III) capacity was much lower than its As(V) capacity. Another special batch study was performed to more closely study the effects different forms of silica, monomeric versus polymeric. It was found that at higher silica concentrations, polymeric silica can form and potentially foul porous adsorbents, thus reducing their capacity. Four of the adsorbents (Metsorb G, AA-FS50, Bayoxide E33, and GFH) were tested in dynamic column tests employing the NSF challenge water spiked with As(V) at concentrations of 250 µg/L (for Metsorb G) to 1,000 µg/L (for the other three adsorbents); the purpose of the high initial concentration was to facilitate breakthrough. Based on an empty bed contact time (EBCT) of 5 minutes, the AA-FS50 and GFH showed similar results with breakthrough at about 4,000 to 5,000 bed volumes (BVs). The Metsorb G and Bayoxide E33 were run for longer periods of about 8,000 and 12,000 BVs, respectively. The Metsorb G experiment was terminated before a clear breakthrough trend was observed, however, this result is consistent with the lower initial As(V) concentration employed in this test. Based on the bench-testing results and synthesis of information from other sources, it was possible to develop an arsenic adsorbent design and costing tool based on seven adsorbents: AA-FS50, Bayoxide E-33, GFH, Metsorb G, Z-33, SMI, and AA-400G (conventional, granular activated alumina). Based on user-provided input; system parameters (e.g., average flow), water quality (e.g., influent arsenic, phosphate, and silica concentrations), target water quality (e.g., treated-water arsenic), operational preferences (e.g., pH adjustment), and cost parameters (e.g., interest); the software tool provides outputs that include: adsorption capacities, capital costs, operations and maintenance (O&M) costs, plant footprint, residuals quantities, and water quality interference warnings. The study revealed several areas that warrant additional research. These include: (i) the need for a more focused study on SMI, a promising adsorbent in terms of its high As(V) capacity, but potentially problematic in terms of reproducibility in performance; (ii) a need for additional study on whether silica is truly a competitive interferant or an adsorbent foulant; and (iii) an investigation to resolve discrepancies between isotherm- versus column-based adsorption capacities.
99 ©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
APPENDIX A LITERATURE SURVEY
101 ©2005 AwwaRF. All rights reserved.
BACKGROUND Various treatment methods have been tested for their potential to remove arsenic from drinking water under both laboratory and field conditions: coagulation-precipitation using iron and aluminium salts; adsorption onto activated alumina/activated carbon/activated bauxite; reverse osmosis; ion exchange; and oxidation followed by filtration (Viraraghavan et al., 1994; Amy et al., 2000). The applicability of conventional adsorbents, including activated alumina (AA) and ion exchange (IX) for the removal of arsenic, the associated water quality constraints (e.g., SO42– for IX) and their regeneration requirements have been previously reported (Amy et al., 2000). Conventional treatment technologies including coagulation, Fe/Mn removal and softening; sorption processes onto AA, IX and iron-oxide-coated sand (IOCs); membrane processes (nanofiltration (NF), ultrafiltration (UF) and microfiltration (MF)); and preoxidation of arsenite to arsenate were discussed and summarized earlier (Edwards, 1994; Cheng et al., 1994; Jekel, 1994; Scott and Morgan 1995; Hering and Elimelech, 1996; Amy et al., 2000; Brandhuber and Amy, 2001; Korngold et al., 2002). Amy et al. (2000) made a comparative assessment of existing arsenic removal technologies. They discussed the selection of arsenic treatment options based on costs, water quality, and a number of other considerations (e.g., residuals). In addition, they developed a decision tree to be used as a guide in selecting arsenic treatment options at a given utility based on estimated costs (Amy et al., 2000). The overall conclusion of their assessment was that the selection of a most appropriate treatment technology depends on water quality, initial arsenic concentration and form, treatment (MCL) objectives, treatment system capacity, and residuals handling costs. In addition, they stated that the decision tree was produced to define overall trends in the selection process and therefore, site-specific work should be carried out for each particular case (Amy et al., 2000). LIMITATIONS ASSOCIATED WITH THE APPLICATION OF CONVENTIONAL TECHNOLOGIES One of the main disadvantages in the application of conventional technologies is that they produce large quantities of sludge containing arsenic (hazardous in nature) that pose serious problems for safe disposal (Simeonova, 2000; Thirunavukkarasu et al., 2001). Another distinctive disadvantage is in the fact that an oxidation step (transformation of arsenite to arsenate) is a prerequisite to achieve the highest efficiency. The research to date has demonstrated that activated alum, for example, exhibits selectivity for arsenate species, and therefore only partial recovery and regeneration of adsorbed arsenic are attainable. In addition, the adsorption onto As on alum floc is very sensitive to pH. Another example is the application of ferric salts, which require the oxidation of As (III) for effective removal (Subramanian et al., 1997). Ion exchange, reverse osmosis and electrodialysis use anionic resins or specific membranes, but have limitations in terms of market availability, residuals (brines), and high costs. Regeneration creates residuals problems, and final pH readjustment and residuals management are often required (Simeonova, 2000). Therefore, conventional treatment technologies require increased maintenance and costs (due to the complexity, investment costs of the plants etc.), cost of chemical reagents (not always available in developing countries), reagents quantity and volume of the toxic residues produced in operation (transport, treatment, etc). (Saha et al., 2001; Amy et al., 2000; Simeonova, 2000;
102 ©2005 AwwaRF. All rights reserved.
Thirunavukkarasu et al., 2001). In addition, they do not always ensure the established MCL for drinking water. International concern has also focused on the problem of arsenic pollution of groundwater in Bagladesh and Bengal, India (Jiang, 2001; Thirunavukkarasu et al., 2001). Over the past 20 years, a significant amount of research has been conducted to develop an inovative technology to achieve low levels of As in drinking water (Gupta and Chen, 1978; Huxstep and Sorg, 1988; Edwards, 1994; Jekel, 1994; Hering and Elimelech, 1996, Amy et al, 2000; Robins et al, 2001). ADSORBENTS TESTED Arsenic can be removed by adsorption onto many adsorbent materials whose potential and suitability of application depend on a number of factors, such as removal efficiency, longevity, sensitivity to water quality parameters, costs, etc. This paper will review general characteristics and performance of the materials that have been tested world-wide. In addition, suggestions for selection criteria employed for screening of the materials will be provided. Two summary tables have been prepared and are cited from the text in Chapter 2: Table 2.1. Adsorbents tested: description, water quality parameters and references; and Table 2.2. Batch Isotherm Studies: reported adsorption capacities and protocols used. Amy et al (2000; 2002) presented a brief overview of adsorbents applicable to arsenic removal, classifying them to three categories: i) conventional (e.g., activated alumina, AA, activated carbon); ii) novel and commercially available (e.g granular ferric hydroxide, GFH); and iii) novel but not commercially available (e.g., sulphur modified iron, SMI). Chang et al. (2004) investigated the adsorption potential of six newly developed innovative adsorbents: magnetically impregnated ion-exchange resins (MIEX); hydrous iron oxide particles (HIOPs); sulfur-modified iron (SMI), iron oxide coated microsands (IOC-M); activated alumina (AA) and granular ferric hydroxide (GFH). A detailed description and physico-chemical properties of those adsorbents are discussed by (Chang et al., 2004). The performance of the adsorbents was compared with an innovative coagulation technique (ballasted flocculation) and immersed membrane technologies coupled with innovative adsorbents in both synthetic (Milli-Q: carbon and ion free) and natural waters (Chang et al., 2004). The adsorption isotherm studies with MQ water were carried out at two different pH values (pH 5.5 and 7.5). The results revealed that while MIEX and AA were more effective at lower pH (5.5), HIOPs, GFH and SMI were efficient over a wider pH. When adsorptive capacities were compared on the weight basis (w/w) MIEX, GFH and HIOPs had superior adsorptive capacity to SMI and AA, with MIEX having the highest value of 3.78 µg/mg for a As(V) equilibrium concentration of 10 µg/L. On the other hand, when compared on a surface area basis (w/m2), the order changed to MIEX > SMI > GFH >AA >HIOPs (Chang et al., 2004). The adsorption isotherm studies with natural waters showed that there was no significant difference in the efficiency of the adsorbents, when compared on weight basis (mg/L). The conventional coagulants (both Al+3 and Fe+3) were equally effective in removing arsenic as innovative adsorbents (Chang et al., 2004). Among innovative adsorbents, HIOPs was more effective then MIEX, AA and IOC-M, which was in discrepancy with a previous finding of adsorptive capacities carried out in MQ water. The authors concluded that adsorptive capacities of both MIEX and AA were adversely affected by the presence of various anions in natural waters. In addition, they stated that although ferric chloride addition was more effective in arsenic removal then any of investigated adsorbents, the later one have a distinguished advantage in arsenic
103 ©2005 AwwaRF. All rights reserved.
removal applications due to the fact that they can be easily separated and regenerated (Chang et al., 2004). Driehaus et al. (1998) investigated the potential of the granular ferric hydroxide (GFH) for arsenic removal from natural waters in Germany. They conducted batch studies at five different pH values (between 5 and 9), with shaking provided for 96 h. The adsorption density was calculated from the arsenic residual concentrations (10 and 40 µg/g versus 0.13 and 0.5 µmol/L) and the known initial concentrations of As (V) and GFH. As (V) adsorption was tested both in model systems (synthetic water with no competing anions) and natural waters. For the adsorption tests in natural waters, the initial As(V) concentration was 1.3 µmol/L (10 µg /L) resulting in a molar ratio of phosphate to arsenate of nearly 5. The adsorption density in natural water represented only 10–20% of that in model systems. In addition, fixed bed adsorber tests, consisting of two in-line columns were performed with natural waters. The height of the columns was 0.12 m and the filtration rate 6–10 m/h. The GFH was screened to a grain size of 0.2–0.4 mm and tap water was spiked with arsenate to a concentration of 100–180 µg/L. The average amount of As (V) sorbed to GFH was 8.4 g/kg (which was in accordance with the batch tests), after 34000 bed volumes were treated (Driehaus et al., 1998). The authors concluded that GFH is an effective adsorbent for arsenate removal. The adsorption density in model systems without competing water constituents was in the range of 1mmol As/g Fe at an As(V) equilibrium concentration of 10 µg/L and a pH of 7. However, the adsorption density decreased with increasing pH and phosphate content. The adsorption density in natural waters was only 0.1–0.2 mmol As/g Fe, depending on the pH and the amount of phosphate present in the water. The specific capacity of fixed bed reactors with GFH depended on pH, phosphate content, and on the raw water concentration of arsenate. It has been demonstrated that more then 50 000 bed volumes could be treated at filtration rates up to 15 m/h, which corresponds to lifetimes of GFH of at least several months. Pal (2001) showed that the arsenate adsorption density on GFH is very close to that of freshly prepared ferric hydroxide. He concluded that GFH can be applied successfully over the pH range between 5.5 to 9. The author also stated that several As removal plants using GFH in fixed bed reactors are being successfully operated in Germany and the UK. In addition, the research results revealed that GFH is capable of removing As from an initial concentration of 5 mg/L to a level much below the permissible limit of 0.01 mg/L (Pal, 2001). Selvin et al. (2001) presented further developments of granular ferric hydroxide media by Severn Trent Company (UK). The adsorptive and hydraulic properties of a number of new variants of GFH were referred to as Ap, Bg and Cg (characteristics not presented). An experimental feed As concentration of 20 µg/L was supplied to 1 L of media at an empty bed contact time (EBCT) of 45 seconds. The average As outlet concentration was 10 µg/L, demonstrating As removal of 50% (for 100,000 bed volumes). Of the four investigated materials, Cg demonstrated the strongest potential for As removal (As outlet concentrations 0–5 µg/L, for 120,000 bed volumes, rising to 10 µg/L between 120 and 150 000 bed volumes). In the hydraulic properties tests, Cg showed again a superior performance when compared to other variants of GFH (Selvin et al., 2001). The potential of ferrihydrite (FH) for arsenic removal has been recognised for three decades (Thirunavukkarasu et al., 2001; Robins et al., 2001, Jain et al., 1999; Raven 1998; Gulledge and O’Connor, 1973). Studies carried out by Waychunas et al. (1993; 1995) showed that at pH > 7, As (V) was adsorbed to ferrihydrite as a strongly bonded inner-sphere complex with either monodentate or bidentate attachment. Robins et al. (2001) reported that monodentate attachment predominates at pH values of 4–5. This result is in discrepancy with the findings by
104 ©2005 AwwaRF. All rights reserved.
Jain et al. (1999) who investigated surface charge reduction and net OH- release stoichiometry in As (III and V) adsorption on ferrihydrite. They compared the experimentally observed surface charge reduction and net optimal OH- release stoichiometry with the theoretical stoichiometry and provided evidence that a monodentate bonding mechanism might play an increasing role during As (V) adsorption on FH with increasing pH (at pH > 8). Adsorption of As (III) was reported to occur at pH 8–9 (Nishimura and Umetsu, 2000). Raven (1998) tested FH for As (III) and As (V) removal from synthetic waters, at two different pH values (4.6 and 9.2). The initial As (arsenite or arsenate) solution concentrations used in their experiments ranged from 0.267–26.7 mmol/L (20–2000 mg/L), corresponding to 0–13.3 molAs/kg ferrihydrite. The results revealed that at high concentrations, the amount of arsenite adsorbed on ferrihydrite was generally higher than the amount of arsenite. At lower pH (4.6) the amounts of adsorbed arsenate and arsenite were equal only at low initial As concentration (less then 1 molAs/ kg ferrihydrite). At high pH (9.2), more arsenite was adsorbed even at the lowest initial As concentrations. The adsorption maxima for arsenate on ferrihydrite corresponded to approximately 0.25 and 0.16 molAs/molFe at pH 4.6 and pH 9.2, respectively (Raven 1998). In contrast to arsenate, the maximum adsorption capacity of ferrihydrite for arsenite was not reached even for initial As concentrations as high as 13.3 molAs/ kgF and the highest observed adsorption density was approximately 0.60 molAs/ kgFe, regardless of the pH value. Maximum adsorption densities for arsenate reported in this study (0.25 and 0.11 molAs/molFe) are in accordance with the values reported by other researchers (Fuller et al., 1993; Ferguson and Anderson, 1974). However, in their study of As (III) and As (V) adsorption onto amorphous iron hydroxide, Pierce and Moore (1982) obtained significantly different values, with adsorption maxima as high as 5.0 molAs/ molFe for both arsenite and arsenate, which was attributed to the extremely high molar ratio of As to Fe in the solution (0.014-0.29) as opposed to 0.028 and 0.083 molAs/ molFe, employed by Raven (1998). Wong et al. (1995) also showed that at molar ratios of arsenic to iron greater then 3, a greater As (V) removal is achieved. The effect of Fe/As molar ratio in solution on arsenic removal was reported by several researchers (Krause and Ettel, 1988; Papassiopi et al., 1996). Thirunavukkarasu et al. (2001) conducted a batch study using various amounts of FH adsorbent and mixing it with raw water of high As concentration (325 µg/L) for 5 h at 125 rpm. The adsorption capacity of FH from their experiment was estimated at 285 µg/g.In their review of adsorption of arsenic on FH, Robins et al. (2001) pointed out that while extensive research has been conducted over the years to investigate the potential of this material for As removal, very little attention was given to the possibility of modifying the FH structure to improve its adsorptive capacity. Therefore, they investigated the coprecipitation of both Al (III) and Mn(III) with Fe (III) to form an aluminic ferrihydrite and a manganic ferrihydrite, respectively. They reported that both materials showed considerably better capacity for arsenic adsorption. The adsorption of As on geothite (α-FeOOH), a product formed from the crystallization of ferryhidrite, was investigated by Belzile and Tessier (1990), Matis (1999), Matis et al. (1997), Xiaohua and Harvey (1996; 1998), Manning et al. (1998) and O’Reilly et al. (2001). Belzile and Tessier (1990) compared the data (over the pH range 4–8) of Pierce and Moore (1980) for adsorption on ferrihydrite with existing data for adsorption on geothite and concluded that the adsorption on ferrihydrite was superior to adsorption by geothite. Xiaohua and Harvey (1996) investigated As (V) and As (III) bonding structures using Transmission-Fourier Transform Infrared (T-FTIR) and Attenuated Total Reflectance – FTIR (ATR-FTIR) spectroscopy. They showed that most of the arsenate and arsenite oxyanions replaced two singly coordinated surface OH groups (A-type)
105 ©2005 AwwaRF. All rights reserved.
to form binuclear bridging complexes: Fe-O-AsO, (OH)-O-Fe, and Fe-O-As(OH)-O-Fe. In addition, the authors used a sequential extraction method to evaluate different bonding strengths. They concluded that a strong chemical extraction was necessary to remove 75-80% of adsorbed arsenite or arsenate from geothite (Xiaohua and Harvey 1996). Xiaohua and Harvey (1998) conducted another study, where both indirect (FTIR) and direct (X-ray Absorption Near Edge Structure (XANES) spectroscopic techniques were applied to examine adsorption and oxidation of As(III) on geothite. They showed that after 20 days of experimentation, more then 20% of adsorbed As (III) was oxidized to As (V). Surface structures and stability of As (III) on geothite was also investigated by Manning et al. (1998), using a combination of standard batch techniques and X-ray absorption spectroscopy (XAS). For adsorption isotherm studies, 20 mL volumes of a 2.5 g L-1 α-FeOOH suspension containing 133 mM or 266 mM As (III) were equilibrated for 16 h. The suspension pH was adjusted from 3-11 using not more then 0.25 mL of 0.1 M NaOH or HCl. Experimental As (III) and As (V) adsorption envelopes on α-FeOOH were compared with the surface complexation model using the FITEQL computer program. The As (V) species displayed a distinct behavior from As (III), with the adsorption maximum at low pH (3), decreasing rapidly with an increase in pH. The authors postulated that this difference might have occurred due to differences in the structure of the As (III) and As (V) surface complexes, as it has been reported earlier that while As (III) forms a single bidentate binuclear surface structure, As (V) forms at least two complexes (Waychunas et al., 1993; Fendorf et al., 1997). Matis (1999) investigated sorption of As (V) by geothite particles and their flocculation. The main parameters affecting the sorption process were studied, such are the ratios of geothite and As (V), pH, contact time, temperature and ionic strength variations. The adsorption parameters were also determined (Table 2.2). O’Reilly et al. (2001) tested the effects of residence time on arsenate adsorption/desorption mechanisms on geothite. A batch sorption experiment was conducted employing an initial As concentration ranging from 0-3 mM (as sodium arsenate) for 24 h at 298 K on a reciprocating shaker. All solutions were pre-equilibrated at pH 6 and the pH was also measured at the end of experiment. The experimental conditions and shape of the isotherm obtained were similar to those found for arsenate on hydrous iron oxide and ferrihydrite, reported by Hsia et al. (1994) and Raven (1998). an adsorption capacity of 2.1 µmol/m2 was achieved. Desorption studies were conducted for times ranging from 45 min to 7 months. Data from the sorption kinetics study showed that initial arsenate sorption at pH 6 was very rapid, with over 93% being sorbed within a 24-h period, followed by the long period of a plateau phase. In India, Singh et al. (1988) studied the effect of different concentrations, pH and temperature on As (III) removal from aqueous solutions by haematite. Five different As (III) concentrations were used, ranging from 13.34–133.49 µmol/L (1–10 mg/L). The maximum adsorption capacity (2.63 µmol/g) (0.20 µgl/mg), calculated from the Langmuir equation, was achieved at 20°C, for pH 7.0. A two fold increase in temperature (from 20 to 40°C resulted in 1.2 fold decrease in adsorption capacity (Singh et al., 1988). pH had no significant effect on the amount of As (III) adsorbed by haematite over a lower pH range (2.8–5.0). However, the adsorption capacity started to increase abruptly at pH = 5, reaching a maximum at pH 7.0. Further increase in pH resulted in a sharp decrease of adsorption up to pH 11. The efficiency of iron oxide coated sand (IOCS) for As removal from drinking water was investigated by Thirunavukkarasu et al. (2001), Khaodhiar et al. (2000), Lombi et al. (1999), and Chang et al. (2004). Lombi et al. (1999) studied the kinetics and reversibility of As sorption on IOCS and soils with different chemical and physical characteristics. Soils and IOCS were equilibrated for five
106 ©2005 AwwaRF. All rights reserved.
different periods of time, with solutions containing As (III) and/or As (V). Samples of materials were sequentially extracted after 1, 10 and 30 days to access the effect of aging on the binding forms of As. Khaodhiar et al. (2000) found that arsenate species were strongly adsorbed on IOCS, forming inner-sphere surface complexes. They also studied the effect of the presence of copper and/or chromate in the solution on As removal and reported neither had an effect on the extent of adsorption. However, the presence of As (V) significantly decreased chromate adsorption, which was attributed to the competition for adsorption sites and electrostatic effects. Thirunavukkarasu et al. (2002) tested iron oxide coated sand (IOCS) and FH for arsenic removal from a natural water in Canada. Batch tests were conducted using various amounts of the adsorbents and mixing them with raw water of high As concentration (325 µg/L) for 5 h at 125 rpm. The adsorption capacities of IOCS and FH were estimated at 18.3 µg/g and 285 µg/g, respectively, with FH having a significantly higher affinity (a 15.6 fold higher As adsorption capacity then IOCS). This difference was attributed to the differences in the specific areas of those two adsorbents (5.1 and 141 m2/g, respectively). Speciation studies were also conducted with natural water containing arsenic, and particulate and soluble As concentrations were determined. The results revealed that particulate and soluble As contributed to 11.4 and 88.6% of total As present in the natural water, respectively. In the case of soluble arsenic, As (III) and As (V) were 47.3 and 52.7%, respectively. Several other iron based adsorbents were tested for their capacity to remove As (III) and As (V). Pierce and Moore (1982) studied amorphous iron hydroxide; Zouboulis et al. (1993) and Han and Fyfe (2000) studied iron-sulfide minerals, pyrite and pyrhotite. Lackovic et al. (2000), Krishna et al. (2001) and Ramaswami et al. (2001) investigated the use of zero-valent iron. The adsorbing capacities of these materials and experimental conditions employed are presented in Tables 2.1 and 2.2, respectively. Zeng (2001) developed a method for preparation of a granulated iron (III) based binary oxide adsorbent, which consisted mainly of amorphous hydrous ferric oxide (FeOOH) with silica as a binding agent. The key step in the method was the simultaneous generation of hydrous ferric oxide (FeOOH) and silica in one reactor and resulted in the formation of Fe-Si complexes. The author concluded that the addition of silica enhanced the granulated adsorbent strength but reduced the As adsorption capacity. A Si/Fe molar ratio of approximately 0.33 was found optimal for the balance of adsorbent strength and its adsorption capacity. Krishna et al. (2001) developed an approach based on the preoxidation of As (III) using hydrogen peroxide and the subsequent removal of total As using ferric (oxy) hydroxide coprecipitation. All experiments were performed at the natural pH of the water (6.8 – 7.1). It was found that the addition of Fe (II) salt and H2O2 was capable of removing As to 20 µg/L levels. Given that the levels of residual peroxide have to be acceptable for drinking water purposes, the authors carried out a set of optimisation experiments using lower amounts of reagents (ferrous ammonium sulfate + H2O2), which also resulted in high residual peroxide levels. In order to improve the efficiency of the treatment, the authors developed a two stage approach, where addition of Fenton’s reagent was used as a preliminary step, followed by passing the water through zero valent iron. This treatment, employing Fe(II) ammonium sulphate + H2O2 per litre for 10 min followed by passing the sample through iron scrap (at a relatively high flow rate of 150 ml/min) and filtering through sand, consistently yielded waters with arsenic concentrations of less then 10 ppb from a starting initial concentration of 2.5 mg/L of As (III). The scaling up of this procedure is currently being investigated by the authors.
107 ©2005 AwwaRF. All rights reserved.
Lackovic et al. (2000) investigated the As removal efficiency by zero-valent iron in laboratory and field column studies. The results from the laboratory experiments showed that zerovalent iron was twice as effective in removing arsenate (669 µg As/Fe) than arsenite (298 µg As/Fe) after 1850 pore volumes. Arsenic removal efficiencies of greater then 95% were observed in both laboratory and columns studies. The high removal efficiency was related to the surface area of the type of iron used. Spectroscopic analyses (SEM/EDX and XPS) provided evidence that surface precipitation was a predominant removal mechanism. The overall conclusion was that the use of iron fillings has a great potential for arsenic removal. Apart from the high efficiency, the material is also cost-effective ($250-400 per ton) and versatile (both in situ, large-scale units, and home-unit applications). Ramaswami et al. (2001) developed a batch-mixed treatment to test the appropriateness of zero-valent iron as a point-of use technology for arsenic removal in Bangladesh. A very high As removal efficiency (>93%) was achieved even for an initial As concentration as high as 2000 µg/L. These results are in accordance with those reported by Lackovic et al. (2000). The authors also provided evidence for the regenerative capacity of the iron filings. Chakravarty et al. (2002) studied a low cost ferruginous managanese ore (FMO) for As removal from groundwater. The results demonstrated that FMO could remove both As (III) and As (V) without any pre-treatment, the adsorption on As (III) being higher then for As (V). The optimal pH for As removal was 2-8 and once adsorbed, As did not desorb even when pH conditions were varied. The presence of bivalent cations (Ni2+, Co2+ and Mg2+) enhanced the adsorption capacity of FMO. As was removed from six groundwater samples with As concentrations ranging from 0.04-0.18 pm, with 100% efficiency. The efficiency of manganese greensand (MGS) for Arsenic removal from drinking water was tested by Subramanian et al. (1997), Viraraghavan et al., (1999) and Thirunavukkarasu et al. (2001). Subramanian et al. (1997) conducted laboratory scale batch and column studies to assess the effectiveness of KMnO4 oxidation followed by manganese greensand filtration for As removal below 25 µg/L. The physico-chemical characteristics of greensand, a zeolite-type glauconite mineral, were described previously (Subramanian et al., 1997). The capacity of MGS for As adsorption was tested by adding various masses of material (ranging from 2.5 to 30 g) to 250 mL beakers containing 100 mL of tap water supplemented with 200 µg/L As (III). The samples were shaken at 30 rpm for 24 h, decanted and analysed for residual As (III), As (V), As (total) and Mn. The results showed that adsorption capacity of MGS reached a plateau after 15 g of the adsorbent was added. The highest efficiency was only 62%, producing an effluent As concentration of 75 µg/L, which was 3 times higher then the value stated in objectives. Column studies consisted of two columns (diameter 10 cm; height 180 cm), fed with the tap water containing 200 µg/L As (III) at a filtration rate of 1-1.5 L/min/m2, resulting in an average residence time (empty bed contact time, EBCT) of 5.64 min.. The effect of Fe/As ratio was tested in a separate experiment by adding Fe(II) in the form of FeSO46H2O to the influent to produce Fe/As molar ratios of 10 [200 µg/L As (III) and 2 mg/L Fe(II)], 20 [100 µg/L As (III) and 2 mg/L Fe(II)] and 7 [50 µg/L As (III) and 0.35 mg/L Fe(II)]. The results from column studies showed that an As limiting concentration of 25 µg/L was reached after only 3 h of operation, corresponding to a throughput volume of 144 L. Testing was continued for another 21 h, before MGS filter was backwashed (1,003 L). As removal efficiency was found to be 41.3%. The investigation of the effect of Fe/As molar ratio showed, that at a Fe/As molar ratio of 10, and a KMnO4 dosage of 2.5 mg/L (as Mn) (for oxidation As (III) to As (II) and Fe(II) to Fe (III)), an As removal efficiency of 89% was achieved. After operating a column for additional 30 h
108 ©2005 AwwaRF. All rights reserved.
(volume passed during this time was approximately 1,440 L), the overall As removal efficiency was still high (83%). A Fe/As ratio of 10 contributed to a two-fold increase in As removal efficiency. As (III) concentration in the filtrate was low (3–13.4 µg/L, indicating almost quantitative oxidation to As (V) by KMnO4 and by the hydrated MnO2coating of MGS. Similar studies were conducted at Fe/As = 20 and Fe/As = 7. Overall, the highest As removal efficiency (81.8%) was achieved at the highest Fe/As ratio, producing an effluent consistently below 10 µg/L. The authors concluded that the addition of Fe at the molar ratio of 20 had the greatest performance in As (III) removal. This finding was in agreement with the work reported by other authors (Lauf, 1996; Magyar, 1992), who showed that MGS following KMnO4 oxidation was capable of removing As (III) from groundwater supplies, at >90% efficiency. Oxidation, adsorption and ion exchange represented the major removal mechanisms. Mn (IV) oxides coated on the greensand surface provide a redox active surface and serve as effective oxidants for As (III). The finding of this study, combined with the work reported by Scott and Morgan (1995) suggest that mechanisms of As removal involve four major steps: i) formation of an inner sphere complex where As (III) displaces the surface bound OH- and H2O species from the hydrated manganese oxide via ligand substitution and then binds directly to the oxide surface; ii) transfer of two electrons from As (III) to Mn (IV) of the oxide [i.e., surface oxidation of As (III) to As (V), and reduction of Mn (IV) to Mn (II)]; iii) adsorption of As (V) on the surface, and iv) release of the reduced Mn. Following this study, Viraraghavan et al. (1999) tested the effect of continuous vs. intermittent loading on the filter regeneration, reproducing the same experimental conditions described in the previous study Subramanian et al. (1997). The results showed that continuous regeneration gave better performance (Viraraghavan et al. 1999). A comparison of the MGS performance with IOCS and ion exchange resins activated with ferric ions (IX) was also made. The results revealed that, for the same initial As (III) concentration (200 µg/L), MGS with Fe addition at the ratio 20:1 achieved the highest As removal performance (83.3%) followed by IOCS (49.7%) and IX (37.6%). Adsorption studies were conducted to test manganese dioxide for As removal. Physicochemical characteristics of the material are presented in Table 2.2. The effect of various anions (nitrate, suphate, acetate, oxalate, sulphite, citrate, tartrate, phosphate and carbonate) and cations (Mg2=, Pb2+, Ni2+, Ag=, Ca2+, Ba2+, Sr2+, Li+, Al3=, Zn2+ on the adsorption of As on MnO2 was investigated. The anions were added in the form of their sodium salts, and the cations as their nitrate salts at concentrations of ~10–2 mol/L. The results revealed that in general, the presence of other anions in the solution had no significant effect on the distribution coefficient (Kd) of As on MnO2. The only significant change in Kd value occurred in the presence of tartrate, phosphate and carbonate ions which reduced adsorption by 90.5, 47.7 and 28.3%, respectively. In a case of cationic species, the addition of Mg2+, Pb2+, Ni2+, Ag+, Ca2+ led to a significant increase in As adsorption, while it was decreased in the presence of Zn2+ . Kinetics of As adsorption on MnO2 was investigated using the equations established by Weber and Morris (1963) as well as Langmuir and Freundlich equations (Barrow, 1973). The sorption capacity was estimated at 1.02 mmol/g, with a binding energy of 15.5 kJ/mol. Some researchers have studied removal of As from groundwater using MnO2.coated sand (MDCS), which prepared by forming manganese dioxide (δ -MnO2) by the oxidation of manganese ion with permanganate in the presence of sand. Both batch and column tests were carried out. Batch arsenic removal kinetics tests were conducted by mixing ground water spiked with 1.0 mg/L As (0.5 mg As (III); 0.5 mg of As (V)) with MDCS (10 g/L). Column tests were
109 ©2005 AwwaRF. All rights reserved.
operated in a downflow direction, using 182 g (125 mL) of the MDCS in a 200 mm ID glass column (bed depth 400 mm; porosity 0.36) and a flow rate of 1.7 mL/min (empty bed contact time, EBCT = 74 min). The initial As concentration was the same as in the batch tests. The column was run until the effluent As concentration exceeded 0.01 mg As/L. The MDCS was then regenerated in situ by backwashing with 2 L of a 0.2 N sodium hydroxide solution. In addition, a home arsenic removal unit was installed and its performance in removing As evaluated. The unit was comprised of two chambers, with the top one (200 mm ID × 380 mm) containing 6 kg of the MDCS medium up to the height of 125 mm and serving as a raw water reservoir. The bottom chamber (200 ID × 280 mm) served as storage for the treated water. The overall conclusion was that manganese dioxide-coated send (MDCS) has promise as a medium for use in small systems or home-treatment units. Thomson et al. (1998) investigated the use of manganese oxyhydroxide coated sand (MOCS) as a filter media for As removal, both in the presence and absence of a strong oxidizing agent (NaOCl). The results showed that surface charge of the MnOx coated sand and the charge of dissolved species were the major parameters influencing the extent of As (and other investigated contaminants, uranium and copper) removal. The authors concluded that as MnOx coating is negatively charged in the presence of OCl–, with a pHpzc of 2.87, anionic As(V) species are not attracted to the media at pH above that value. Overall, the results showed that while MnOx coated sand was efficient in removing cationic species (copper), but exhibited poor performance in removing As. In the absence of HOCl, only 0.0016 mg As/mg MnOx was removed. The flow rate, pH and the presence of an oxidant had the greatest effect on the results, while soluble Mn concentration did not eafect the removal. It was concluded that MOCS was not efficient for As removal. The performance of activated alumina, bauxite, and carbon for As removal are well documented in the literature; these materials belong to the group of convential adsorbents commonly used in the past. However, the study carried out by Gupta and Chen (1978) is noteworthy as it tested the effect of a number of water quality parameters (pH, salinity, silica) and type of adsorbent on As (III) and As (V) adsorption. The experiments were carried out using freshwater, seawater, seawater diluted 10 times and a 0.67 M NaCl solution, 2g/L of activated alumina and 3g/L of activated carbon were mixed in 100 ml of solution. Among the investigated adsorbents, activated alumina was most efficient in removing As (V). Within the first 10 minutes in fresh water, 50% and 40% of 53.4 µM As (V) was removed by activated alumina (2 g/L) and activated bauxite (2g/L), respectively, while activated carbon (3 g/L) removed only 23% of 26.4 µM As (V). Rates of As (III) adsorption on activated alumina and bauxite were much slower than those of As (V). Within the first 10 minutes, only 6% of 12.4 µM As (III) and 2% 12.6 µM As (III) were removed, respectively. An increase in salinity had a negative effect on the rate of adsorption and arsenic removal. The adsorption capacity for As (V) on activated alumina decreased 5-fold (from 4.11 (freshwater) to 0.81 (seawater) mg As(V)/g of adsorbent) with an increase in salinity. pH had the most prominent effect on arsenic adsorption. As (V) was effectively adsorbed by activated alumina and bauxite over the pH range 4–7, and then started to decrease with a further increase in pH values above 7. Activated carbon adsorbed As (V) most efficiently in the pH range of 3–5. For As(III) there was no significant difference in adsorption rate over the pH range 4–9; however further increase in pH values resulted in a sharp decrease of the adsorption capacity. The effect of the initial As (III) and As (V) concentrations was significant in a case of As (III), while As (V) removal was only slightly affected. As removal by aluminium and ferric hydroxides through coagulation, sedimentation and filtration also belongs to the group of conventional technologies, which is very well documented in
110 ©2005 AwwaRF. All rights reserved.
the general As literature. A study carried out by Gulledge and O’Connor (1973) is presented as an example of the technology efficiency. The results from this study showed that arsenic removal was influenced by pH and coagulant dose. Adsorption on ferric hydroxides was more efficient then on aluminium hydroxides. However, in both cases, increasing the coagulant dose resulted in an increase of As(V) adsorption. At lower pH values (5–7.5), there was no significant difference in the amount of As(V) adsorbed by any of the adsorbents. However, further increase in pH (8 and above) resulted in a sharp decrease of As(V) on both aluminium and ferric hydroxides. The authors postulated that a decrease in As(V) removal occurred either due to the change in the ionic form of As(V) (from H2AsO4– to HAsO42–) or the competition between hydroxyl ion and the exchange sites on the ferric and aluminium hydroxide precipitates (Gulledge and O’Connor, 1973). Twidwell et al. (1999) reviewed the use of aluminium hydroxide/alum for arsenic removal. They reported that arsenic adsorption was 30–40 – fold greater on amorphous aluminium hydroxide (112 g/kg) than on the more crystalline gibbsite (2.6 g/kg) for the pH range 4–10. A number of researchers demonstrated the efficiency of activated alumina in removing arsenic from drinking water. The overall conclusion from these studies was that activated alumina is competitive in its efficiency for arsenate removal and relatively cheap when compared to other adsorbents. It has been shown that activated alum was very efficient in removing arsenic for point-of-use water treatment. (a pilot study showed arsenic [0.15 mg/L] and fluoride removal and estimated treatment effective cost at $0.19 cents/1,000 gal (3,785 L). Arsenate adsorption by aluminium hydroxide or by activated alum is a function of solution pH (optimal removal is achieved at pH 4–7). Norton et al. (2001) investigated throw-away iron and aluminum sorbents (as emergent adsorbing media) versus conventional activated alumina. The selection of emergent adsorbents was based on several criteria: i) the media should exhibit high adsorption capacity at neutral pH; ii) the media should be easily disposable without the requirement for an on-site regeneration; iii) once saturated, the media should be non-hazardous and suitable for disposal in a municipal landfill; iv) the media should be NSF 61 certified for drinking water applications. The media selected for investigation were: conventional activated alumina, iron-modified activated alumina, high porosity AA and granular ferric hydroxide (GFH). The objective of their study was to evaluate the feasibility of using the selected adsorbents to treat groundwater at two sites (Tuscon and Scottsdale) in the USA, to meet an arsenic MCL of 10 ppb. Four columns containing 25 gallons (95 L) were fed with 5 gallons per minute (19 L/min) of raw water each, with the exception of GFH media, which was fed at 2 gpm (7.6 L/min) and an EBCT of 12.5 minutes (due to problems experienced with head loss). The raw water pH (after chlorination) was between 8.7 and 9.2. The results showed that initial As breakthrough for all of the investigated media occurred at 1,000 bed volumes. The effluent As concentrations were 0.015–0.06 mg/L compared with the influent concentrations (0.031–0.041 mg/L). Poor As removal by AA was expected since the optimum water pH for this media is 5.5. However, the emerging aluminium based media showed promising removal of As at pH as high as 8 to 8.5. The GFH media was able to remove As with no pH adjustment. For the first 2,500 bed volumes, there were no detectable As levels with a raw water pH of 9. However, from 2,000– 4,000 bed volumes treated, the effluent As concentrations increased at or below 0.01 mg/L. At a raw water pH of 6.6, the As levels were again non-detectable. The effects of Si and fluoride concentrations were also studied. It was found that while all media were successful in removing Si, the breakthrough of fluoride occurred sooner than it was observed for Si. All four columns reached complete breakthrough of fluoride after 2000 volumes treated.
111 ©2005 AwwaRF. All rights reserved.
The authors concluded that at ambient pH conditions (8.7–9.3) GFH was the only media that demonstrated high As removal efficiency. High Si concentrations could decrease its removal capability, but when raw water pH is lowered to 6.5, the Si interference may be diminished. The relatively poor performance of the alumina based media was attributed to the ambient pH condition and high Si concentration of the raw water. A number of additional alternative adsorbing materials have been tested in different parts of the world: Manju et al. (1998) tested the efficiency of coconut husk carbon (CHC) and amine modified coconut coir (AMCC),while Lee et al. (1999) studied quaternized rice husk (QRH) as a substitute for activated carbon (AC) which is not readily available and therefore not suitable for use in developing countries; Diamadopoulos et al. (1993) tested the efficiency of fly ash, Altundogan et al. (2000) studied adsorption on red mud, Ohki et al. (1996) and Xu et al. (1998) investigated removal by aluminium–loaded coral limestone and aluminium loaded shirasu-zeolites, Elizalde-Gonzalez et al. (2001) studied natural zeolites, volcanic stone and cactaceous powder, Manning (1997) tested adsorbing potential of three California soils. Dikshit et al. (2000) investigated As adsorption on kimberlite tailings, the mineral waste from diamond mining. The summary of the performance of these materials and experimental conditions are presented in Tables 2.1 and 2.2, respectively. Saha et al. (2001) performed a comprehensive study of arsenic removal efficiency of different adsorbent materials to select the most appropriate adsorbent. They conducted batch experiments on 18 different materials (kimberlite tailing, water hyacinth, wood characoal, banana pith, coal fly ash, spent tea leaf, mushroom, saw dust, rice husk ash, sand, activated carbon, bauxite, hematite, laterite, iron oxide coated sand (IOCS), actrivated alumina (AA), CalSiCo, and hydrous granular ferric oxide (GFO). As (III) removal efficiency was 0.01 ppm) for three weeks. The authors concluded that Zr-resin shows promise for the removal of low As (V) concentrations from actual tap water. Zr (IV)-loaded phosphoric acid chelating resin (RGP) described by Zhu and Jyo (2001) was effective in the removal of As (V) from natural water (including sea water), with a higher affinity for As (III) than As (V). The performance of this adsorbent was tested in column experiments, at an initial As (V) concentration of 2.5 mmol/L, pH 2 and flow rate of 10 mL/h, for 120 bed volumes. The effect of pH was also studied over a range of pH values (1.14; 2.03; 3,05; 4.03; 7.53 and 8.55). Breakthrough capacities of As (V) were 0.132 mmol/mL of wet resin at pH 1.14, decreasing to 0.077 mmol/L at pH 8.55 (0.25–0.44 mmol/g) per unit weight of dry resin. SELECTION CRITERIA FOR ADSORBING MATERIALS The general criteria include the cost of the material, the ease of operation and handling, the cost of transport (if the material is not locally available), the potential of regeneration and reuse and longevity (estimated life-time of the system) (Amy et al., 2000; Saha et al., 2001; Drizo et al., 1999). Drizo et al. (1999; 2000) and Forget (2001) conducted a significant amount of research to establish selection criteria for adsorptive materials to be used for phosphorus removal from wastewater. The same criteria are applicable for fixed bed adsorbent treatment of arsenic, or any other pollutant, and therefore will be stated here. The first criteria in selecting the materials are their physico-chemical properties such are: elemental composition, specific surface area, porosity, particle size distribution, hydraulic conductivity, because they all affect the rate of pollutant adsorption and/or pollutant adsorption capacity (Drizo et al., 1999). The adsorption capacity of candidate materials is then measured in the laboratory employing “pseudo-equlibrium” batch experiments, a technique established as a method of measuring P (phosphorus) retention characteristics of soils and sediments for several decades (Olsen and Watanabe, 1957; Barrow, 1978; Nair et al., 1984). The adsorption capacity of materials is estimated by fitting the Langmuir or Freundlich isotherm equations to experimental data (e.g., Drizo et al., 1999 and all of the references reported in this review). However, it has been shown recently that using the Langmuir equation with such experimental data can lead to biased
114 ©2005 AwwaRF. All rights reserved.
and unrealistic estimates of the adsorption parameters and should be used with caution for adsorption studies for fixed bed adsorption applications (Kinniburg, 1986; Drizo et al., 2000; Forget et al., 2001). Langmuir and Freundlisch equations have been and still are repeatedly used in the estimation of both P and As adsorption capacity of potential media often neglecting very important facts such as: i) the linear form of the Langmuir equation which enables the calculation of P adsorption maxima and binding energy only describes adsorption over a limited range of concentration; ii) the “maximum adsorption capacity” calculated from the observations at low concentrations is exceeded at higher concentrations; iii) equations for which the affinity for adsorption is constant, such as the Langmuir equation, are not consistent with the theoretical scientific knowledge of the adsorption process; iv) fitting the data to the linear form of the Langmuir equation where concentration/adsorption is plotted as a function of concentration produces variability and a low correlation coefficient; and v) if the Langmuir equation is used for data which does not approach the maximum adsorption plateau, the calculated adsorption maximum could be in error by 50% or more (Veith and Sposito, 1977; Harter, 1984; Kinniburg, 1986; Drizo et al., 2000). The problems associated with the lack of uniform methodology for estimation of materials adsorptive capacities was also confirmed in this review. Even though the performance of 30 different materials have been presented, it was not possible to make a real comparison of the most efficienct material. The adsorption capacity of any material used for water purification purposes will depend on the experimental procedure employed. The extent and rate of As adsorption in batch experiments is affected by a number of parameters: the range of initial As concentrations, material to solution ratio (in a case of ferryhydrite, the Fe/As molar ratio), the rotational speed of the shaker, time for equilibration, pH, ionic strength of the solution (Kinniburg et al., 1986; Drizo et al., 2000). In addition, the presence of other aqueous species (e.g., phosphate, silicate, fluoride, sulfate and organics) block the adsorption active sites and cause rapid saturation of the medium, reducing arsenic removal efficiency (Twidwell et al., 1999; Simeonova, 2000; Amy et al., 2000) A need for establishing a new procedure for P batch experiments which would provide a more realistic comparison of the materials has been recognised and several criteria for this purpose were suggested (Drizo et al., 2000; Forget et al., 2001). In addition, it is recommended that batch experiments should be coupled with a longer-term investigation of materials performance in columns for the estimation of removal efficiencies and retention capacities by different materials.
115 ©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
APPENDIX B ADSORBENTS TESTED FOR ARSENIC REMOVAL: SYNTHESIS OF LITERATURE AND VENDOR/MANUFACTURER SURVEYS
117 ©2005 AwwaRF. All rights reserved.
118 ©2005 AwwaRF. All rights reserved.
119 ©2005 AwwaRF. All rights reserved.
120 ©2005 AwwaRF. All rights reserved.
121 ©2005 AwwaRF. All rights reserved.
122 ©2005 AwwaRF. All rights reserved.
123 ©2005 AwwaRF. All rights reserved.
124 ©2005 AwwaRF. All rights reserved.
125 ©2005 AwwaRF. All rights reserved.
126 ©2005 AwwaRF. All rights reserved.
127 ©2005 AwwaRF. All rights reserved.
128 ©2005 AwwaRF. All rights reserved.
129 ©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
REFERENCES Altundogan, S.H., S. Altundogan, F. Tumen, and M. Bildik. 2000. Arsenic Removal from Aqueous Solutions by Adsorption on Red Mud. Waste Management, 20(8): 761–767. Amy, G., M. Edwards, P. Brandhuber, L. McNeill, M. Benjamin, K. Vagliasindi, J. Carlson, and J. Chwirka. 2000. Arsenic Treatability Options and Evaluation of Residuals Management Issues. Denver, Colo.: Awwa Research Foundation and American Water Works Association. APHA, AWWA, WEF (American Public Health Association, American Water Works Association and Water Environment Federation). 1998. Standard Methods for the Examination of Water and Wastewater. 20th ed. Washington, D.C.: APHA. Barrow, G.M. 1973. Physical Chemistry. 3rd ed. London: McGraw-Hill. Barrow, N.J. 1978. The Description of Phosphate Adsorption Curves. Jour. Soil Science, 29: 447–462. Belzile, N., and A. Tessier. 1990. Interactions Between Arsenic and Iron Oxyhydroxides in Lacustrine Sediments. Geochim. Cosmochim. Acta, 54: 103–110. Brandhuber, P., and G. Amy. 2001. Arsenic Removal by a Charged Ultrafiltration Membrane— Influences of Membrane Operating Conditions and Water Quality on Arsenic Rejection. Desalination, 140: 1–14. Chakravarty, S., V. Dureja, G. Bhattacharyya, S. Maity, and S. Bhattacharjee. 2002. Removal of Arsenic from Groundwater Using Low Cost Ferruginous Manganese Ore. Water Research, 36(3): 625–633. Chang, Y., et al. 2004. Demonstration of Emerging Technologies for Arsenic Removal. Volume 1: Bench-Scale Testing. Denver, Colo.: Awwa Research Foundation and American Water Works Association. Chen, H.-W., C.C. Davis, and M. Edwards. 2005. Using Surface Complexation Modeling to Assess the Role of Silica in Arsenate Adsorption onto Metal Oxides. Jour. Water Supply: Res. and Technol.—AQUA. (in press) Chen, H.-W., M.M. Frey, D. Clifford, L.S. McNeill, and M. Edwards. 1999. Arsenic Treatment Considerations. Jour. AWWA, 91(3): 74–85. Cheng, R.C., S. Liang, H.-C. Wang, and M.D. Buehler. 1994. Enhanced Coagulation for Arsenic Removal. Jour. AWWA, 86(9): 79–91. Clifford, D., L. Ceber, and S. Chow. 1983. Arsenic(III)/Arsenic(V) Separation by Chloride-form Ion-Exchange Resins. In Proc. of the Water Quality Technology Conference. Denver, Colo.: American Water Works Association. Diamadopoulos, E., S. Ioannidis, and G.P. Sakellaropoulos. 1993. As(V) Removal from Aqueous Solutions by Fly Ash. Water Research, 27(12): 1773–1777. Dikshit, A.K., K. Pallamreddy, L.V. Praveen Reddy, and J.C. Saha. 2000. Arsenic in Groundwater and Its Sorption by Kimberlite Tailings. Jour. Environ. Science and Health, Part A: Toxic Hazardous Substances, 35(1): 65–87. Driehaus, W., M. Jekel, and U. Hildebrandt. 1998. Granular Ferric Hydroxide—A New Adsorbent for the Removal of Arsenic from Natural Water. Jour. Water SRT- Aqua, 47(1): 30–36. Drizo, A., A.C. Frost, K.A. Smith, and J. Grace. 1999. Physico-chemical Screening of Phosphateremoving Substrates for Use in Constructed Wetland Systems. Water Research, 33(17): 3595–3602.
131 ©2005 AwwaRF. All rights reserved.
Drizo, A., C. Forget, R.P. Chapuis, and Y. Comeau, Y. 2000. How Realistic Are the Linear Langmuir Predictions of Phosphate Retention by Adsorbing Materials? In Proc. of the First World Congress of the International Water Association. Paris: International Water Association. Edwards, M. 1994. Chemistry of Arsenic Removal During Coagulation and Fe-Mn Oxidation. Jour. AWWA, 86(9): 64–78. Edwards, M., S. Patel, L. McNeill, H. Chen, A.D. Eaton, R.C. Antweiler, and H. Taylor. 1998. Considerations in Arsenic Analysis and Speciation. Jour. AWWA, 90(3): 103. Elizalde-Gonzalez, M.P., J. Mattusch, W.-D. Einicke, and R. Wennrich. 2001. Sorption on Natural Solids for Arsenic Removal. Chem. Eng. Jour., 81(1–3): 187–195. Fendorf, S., M.J. Eick, P. Grossl, and D.L. Sparks. 1997. Arsenate and Chromate Retention on Goethite: I. Surface Structure. Environ. Science Technol., 31: 315–320. Ferguson, J., and J. Gavis. 1972. A Review of the Arsenic Cycle in Natural Waters. Water Res. 6: 1259. Ferguson, J.F., and M.A. Anderson. 1974. Chemical Forms of Arsenic in Water Supplies and Their Removal. In Chemistry of Water Supply, Treatment and Distribution. Edited by A.J. Rubin. Ann Arbor, Mich.: Ann Arbor Science. Ficklin, W. 1983. Separation of Arsenic(III) and Arsenic(V) in Groundwater by Ion Exchange. Talanta, 30(5): 371–373. Forget, C. 2001. Élimination du phosphore dissous des effluents piscicoles à l’aide de matériaux granulaires réactifs. Master’s Thesis. École Polytechnique, Montréal. Forget, C., A. Drizo, Y. Comeau, and C.P. Chapuis. 2001. Élimination du phosphore d’effluents de pisciculture par marais artificiels à substrat absorbant. In Proc. of the Américana 2001 Conference. Montréal. Frey, M., and M. Edwards. 1997. Surveying Arsenic Occurrence. Jour. AWWA, 89(3): 105–117. Fuller, C.C., J.A. Davis, and G.A. Waychunas. 1993. Surface Chemistry of Ferrihydrite Part 2: Kinetics of Arsenate Adsorption and Coprecipitation. Geochim. Cosmochim. Acta, 57(10): 2271–2282. Gulledge, J.H., and J.T. O’Connor. 1973. Removal of As(V) from Water by Adsorption on Aluminium and Ferric Hydroxides. Jour. AWWA, 8: 548–552. Gupta, S.K., and K.Y. Chen. 1978. Arsenic Removal by Adsorption. Jour. WPFC, 3: 493–506. Han, J., and W.S. Fyfe. 2000. Arsenic Removal from Water by Iron-Sulfide Minerals. Chinese Science Bull., 45(15): 1430–1434. Harter, R.D. 1984. Curve Fit Errors in Langmuir Adsorption Maxima. Soil Science Soc. Am. Jour., 48: 749–752. Hering, J.G., and M. Elimelech. 1996. Arsenic Removal by Enhanced Coagulation and Membrane Processes. Denver, Colo.: Awwa Research Foundation and American Water Works Association. Hsia, T.H., S.L. Lo, C.F. Lin, and D.Y. Lee. 1994. Colloids Surf. A: Physicochemi. Eng. Aspects, 85: 1–7. Huxstep, M.R., and T.J. Sorg. 1988. Reverse Osmosis Treatment to Remove Inorganic Contaminants from Drinking Water. EPA-600/S 2-87/109. USEPA. Jain, A., K.P. Raven, and R.H. Loeppert. 1999. Arsenite and Arsenate Adsorption on Ferrihydrite: Surface Charge Reduction and Net OH– Release Stoichiometry. Environ. Sci. Technol., 33(8): 1179–1184.
132 ©2005 AwwaRF. All rights reserved.
Jekel, M.R. 1994. Removal of Arsenic in Drinking Water Treatment. In Arsenic in the Environment, Part I. Edited by J.O. Nriagu. New York: John Wiley. Jiang, J.-Q. 2001. Removing Arsenic from Groundwater for the Developing World—A Review. Water Science Technology, 44(6): 89–98. Khaodhiar S., M.F. Azizian, K. Osathaphan, and P.O. Nelson. 2000. Copper, Chromium, and Arsenic Adsorption and Equilibrium Modeling in an Iron-Oxide-Coated Sand, Background Electrolyte System. Water, Air, and Soil Pollution, 119(1–4): 105–120. Kinniburg, D.G. 1986. General Purpose Adsorption Isotherms. Environ. Science Technology, 20: 895–904. Korngold, E., N. Belayev, and L. Aronov. 2002. Removal of Arsenic from Drinking Water by Anion Exchangers. Desalination, 141(1): 81–84. Krause, E., and V.A. Ettel. 1988. Solubility and Stability of Scorodite FeAsO4•H2O: New Data and Further Discussion. Am. Mineralogist, 73: 850–854. Krishna, M.V.B., K. Chandrasekaran, D. Karunasagar, and J. Arunachalam. 2001. A Combined Treatment Approach Using Fenton’s Reagent and Zero-Valent Iron for the Removal of Arsenic from Drinking Water. Jour. Hazardous Materials, B84: 229–240. Lackovic, J.A., N.P. Nikolaidis, and G.M. Dobbs. 2000. Inorganic Arsenic Removal by ZeroValent Iron. Environ. Engineering Science, 17(1): 29–39. Lauf, G. 1996. Optimization of Manganese Greensand for Arsenic Removal. In Proc. of the Water Quality and Technology Conference. Denver, Colo.: American Water Works Association. Lee, C.K., K.S. Low, S.C. Liew, and C.S. Choo. 1999. Removal of Arsenic (V) from Aqueous Solution by Quaternized Rice Husk. Environ. Technology, 20(9): 971–979. Lin, Z., and R.W. Puls. 1999. Arsenic Adsorption on Kaolin Group Minerals. In Proc. of the 217th ACS National Meeting. Anaheim, Calif.: American Chemical Society. Lombi, E., W.W. Wenzel, and R.S. Sletten. 1999. Arsenic Adsorption by Soils and Iron-oxide-coated Sand: Kinetics and Reversibility. Jour. Plant Nutr. Soil Science, 162 (4): 451–456. Magyar, J. 1992. Kelliher Arsenic Removal Study. Report WQ-149. Regina, Sask., Canada: Saskatchewan Environment and Public Safety. Manju, G.N., C. Raji, and T.S. Anirudhan. 1998. Evaluation of Coconut Husk Carbon for the Removal of Arsenic from Water. Water Research, 32(10): 3062–3071. Manning, B. 1997. Arsenic (III) and Arsenic (V) Adsorption on to Three California Soils. Soil Sci., 162 (12): 886–895. Manning, B., and S. Goldberg. 1997. Adsorption and Stability of As(III) at the Clay Mineral– Water Interface. Environ. Science Technology, 31(7): 2005–2011. Manning, B.A., S.E. Fendorf, and S. Goldberg. 1998. Surface Structures and Stability of Arsenic (III) on Goethite: Spectroscopic Evidence for Inner-sphere Complexes. Environ. Science Technology, 32(16): 2383–2388. Matis, K.A. 1999. Sorption of As(V) by Geothite Particles and Study of Their Flocculation. Water, Air, Pollution, 111(1–4): 297–316. Matis, K.A., A.I. Zouboulis, F.B. Malamas, M.D.R. Afonso, and M. Hudson. 1997. Flotation Removal of As(V) onto Goethite. Environ. Pollution, 97(3): 238–247. McNeill, L., and M. Edwards. 1995. Soluble Arsenic Removal at Water Treatment Plants. Jour. AWWA, 87(4): 105–113. Nair, P.S., T.J. Logan, A.N. Sharpley, L.E. Sommers, M.A. Tabatabai, and T.L. Yuan. 1984. Interlaboratory Comparison of a Standardized Phosphorus Adsorption Procedure. Jour. Environ. Qual., 13(4): 591–595.
133 ©2005 AwwaRF. All rights reserved.
Nishimura, T., and Y. Umetsu. 2000. Chemistry on Elimination of Arsenic, Antimony and Selenium from Aqueous Solutions with Iron (III) Species. In Minor Metals. Edited by C.A. Young. Littleton, Colo.: Society for Mining, Metallurgy, and Exploration Inc. Norton, M.V., Y.-J. Chang, T. Galeziewski, S. Kommineni, and Z. Chowdhury. 2001. Throw-away Iron and Aluminum Sorbents versus Conventional Activated Alumina for Arsenic Removal—Pilot Testing Results. In Proc. of the AWWA Annual Conference. Denver, Colo.: American Water Works Association. Ohki, A., K. Nakayachigo, K. Naka, S. Maeda. 1996. Adsorption of Inorganic and Organic Arsenic Compounds by Aluminum-loaded Coral Limestone. Applied Organometal Chem., 10(9): 747–752. Olsen, S.R., and F.S. Watanabe. 1957. A Method to Determine Phosphorus Adsorption Maximum of Soils as Measured by the Langmuir Isotherm. Soil Sci. Soc. Am. Proc., 21: 144–149. O’Reilly, S.E., D.G. Strawn, and D.L. Sparks. 2001. Residence Time Effects on Arsenate Adsorption/ Desorption Mechanisms on Goethite. Soil Sci. Soc. Am. Jour., 65: 67–77. Pal, B.N. 2001. Granular Ferric Hydroxide for Elimination of Arsenic from Drinking Water. In Proc. of the BUET-UNU International Workshop on Technologies for As Removal from Drinking Water. Edited by M.F. Ahmed, M.A. Ali, and Z. Adeel. Dhaka, Bangladesh: Bangladesh University of Engineering and Technology and United Nations University. Papassiopi, N., E. Vircikova, V. Nenov, A. Kontopoulos, and L. Molnar. 1996. Removal and Fixation of Arsenic in the Form of Ferric Arsenates: Three Parallel Experimental Studies. Hydrometallurgy, 41(2): 243–255. Pierce, L.M., and B.C. Moore. 1982. Adsorption of Arsenite and Arsenate on Amorphous Iron Hydroxide. Water Research, 16: 1247–1253. Ramaswami, A., S. Tawachsupa, and M. Isleyen. 2001. Batch-Mixed Iron Treatment of High Arsenic Waters. Water Research, 35(18): 4474–4479. Raoof, M.A., S.M. Hasany, M. Ikram, and Najam-ud-Din. 1994. Adsorption Studies of Arsenic on Manganese Dioxide Using Radiometric Techniques. Raven, K. 1998. Arsenite and Arsenate Adsorption on Ferrihydrite: Kinetics, Equilibrium and Adsorption Envelopes. Environ. Science Technology, 32(3): 344–349. Robins, R.G., T. Nishimura, and P. Singh. 2001. Removal of Arsenic from Drinking Water by Precipitation, Adsorption and Cementation. In Proc. of the BUET-UNU International Workshop on Technologies for As Removal from Drinking Water. Edited by M.F. Ahmed, M.A. Ali, and Z. Adeel. Dhaka, Bangladesh: Bangladesh University of Engineering and Technology and United Nations University. Saha, J.C., K. Dikshit, and M. Bandyopadhyay. 2001. Comparative Studies for Selection of Technologies for Arsenic Removal from Drinking Water. In Proc. of the BUET-UNU International Workshop on Technologies for As Removal from Drinking Water. Edited by M.F. Ahmed, M.A. Ali, and Z. Adeel. Dhaka, Bangladesh: Bangladesh University of Engineering and Technology and United Nations University. Saskatchewan Environment and Public Safety. 1987. Conditioning a Manganese Greensand Filter. Clear Water Reporter, 1(4). Scott, M.J., and J.J. Morgan. 1995. Reactions at Oxide Surfaces Part I: Oxidation of As(III) by Synthetic Birnessite. Environ. Science Technology, 29(8): 1898–1905. Selvin, N., G. Messham, J. Simms, R. Dennis, and D. Wilde. 2001. Further Developments of Granular Ferric Media (The Arsenic Removal Technology!) In Proc. of the AWWA Annual Conference. Denver, Colo.: American Water Works Association.
134 ©2005 AwwaRF. All rights reserved.
Simeonova, V.P. 2000. Pilot Study for Arsenic Removal. Water Science and Technology, IWA Publishing, 636–640. Singh, D.B., G. Prasad, D.C. Rupainwar, and V.N. Singh. 1988. As(III) Removal from Aqueous Solution by Adsorption. Water, Air, and Soil Pollution, 42: 373–386. Subramanian, K., T. Viraraghavan, S. Tanjore, and T. Phommavong. 1997. Manganese Greensand Removal of Arsenic in Drinking Water Treatment. Water Quality Res. Jour. Canada, 32(3): 551–561. Suzuki, T.M., M.L. Tanco, D.A.P. Tanaka, H. Matsunaga, and T. Yokoyama. 2001. Adsorption Characteristics and Removal of Oxo-Anions of Arsenic and Selenium on the Porous Polymers Loaded with Monoclinic Hydrous Zirconium Oxide. Sep. Sci. Technol., 36(1): 103–111. Thirunavukkarasu, O.S., T. Viraraghavan, and K.S. Subramanian, K.S. 2001. Removal of Arsenic in Drinking Water by Iron Oxide-coated Sand and Ferrihydrite-batch Studies. Water Qual. Res. Jour. Canada, 36(1): 55–70. Thomson, B.M., E.E. Smith, and M.D. Siegel. 1998. Development of a Treatment Process for Removal of Metals and Radionucleides on Manganese Oxyhydroxide Surfaces. Technical Report No. WERC 1005. New Mexico Waste-Management Education and Research Consortium and US Department of Energy. Twidwell, L.G., J. McCloskey, P. Miranda, and M. Gale. 1999. Technologies and Potential Technologies for Removing Arsenic from Processing and Mine Wastewater. In Proc. of the Global Symposium on Recycling, Waste Treatment Clean Technology. Edited by I. Gaballah, J. Hager, and T.M. Solozabal. Warrendale, Pa.: The Minerals, Metals, and Materials Society. USGS (United States Geological Survey). 2004. National Water Informational System (NWIS) [Online]. Available: . Veith, J.A., and G. Sposito. 1977. On the Use of the Langmuir Equation in the Interpretation of “Adsorption” Phenomena. Soil Sci. Soc. Am. Jour., 41: 697–702. Viraraghavan, T., K.S. Subramanian, and J.A. Aruldoss. 1999. Arsenic in Drinking Water— Problems and Solutions. Water Sci. Technol., 40(2): 69–77. Viraraghavan, T., K.S. Subramanian, and T.V. Swaminathan. 1996. Drinking Water Without Arsenic: A Review of Treatment Technologies. Environmental Systems Reviews. ENSIC, A.I.T., Bangkok. Waychunas, G.A., B.A. Rea, C.C. Fuller, and J.A. Davis. 1993. Surface Chemistry of Ferrihydrite, Part I: EXAFS Studies of the Geometry of Coprecipitated and Adsorbed Arsenate. Geochim. Cosmochim. Acta, 57: 2251–2269. Waychunas, G.A., B.A. Rea, J.A. Davis, and C.C. Fuller. 1995. Geometry of Sorbed Arsenate on Ferrihydrite Crystalline FeOOH: Re-evaluation of EXAFS Results and Topological Factors in Predicting Geometry, and Evidence for Monodentate Complexes. Geochim. Cosmochim. Acta, 59: 3655–3661. Weber, W.J., and J.C. Morris. 1963. Kinetics of Adsorption on Carbon from Solution. Jour. Sanit. Eng. Div., ASCE, SA2: 899. Welsh, A., D. Westjohn, D. Helsel, and R. Wanty. 2000. Arsenic in Groundwater of the United States: Occurrence and Geochemistry. GroundWater, 38(4):589–604. Wong, P.L.M., J.C.Y. Huang, and T.W. Cheng. 1995. The Mechanisms of the Removal of Arsenic from Water by Adsorption and Coprecipitation with Ferrihydrite. Jour. Chinese Inst. Environ. Engineering, 5(3): 241–251
135 ©2005 AwwaRF. All rights reserved.
Xiaohua, S., and E. Harvey. 1996. An Investigation of Arsenate and Arsenite Bonding Structures on Goethite by FTIR. Soil Science, 161(12): 865–873. Xiaohua, S., and E. Harvey. 1998. Adsorption and Oxidation of Arsenite on Goethite. Soil Science, 163(4): 278–288. Xu, Y., A. Ohki, and S. Maeda. 1998. Adsorption of Arsenic (V) by Use of Aluminium-Loaded Shirasu-Zeolites. Chemistry Letters, 10: 1015–1017. Zeng, L. 2001. Preparation of an Iron (III) Oxide Adsorbent for Arsenic Removal [Online]. Available: . Zhu, X., and A. Jyo. 2001. Removal of Arsenic (V) by Zirconium (IV)-loaded Phosphoric Acid Chelating Resin. Sep. Science Technology, 36(14): 3175–3189. Zouboulis, A., K.A. Kydros, and K.A. Matis. 1993. Arsenic (III) and Arsenic (V) Removal from Solution by Pyrite Fines. Sep. Science Technology, 28(15–16): 2449–2463.
136 ©2005 AwwaRF. All rights reserved.
ABBREVIATIONS AA As As(III) As(V)
activated alumina arsenic arsenite arsenate
BTC BV
breakthrough curve bed volume
EBCT
empty bed contact time
GFH GFO
granular ferric hydroxide granular ferric oxide
HLR
hydraulic loading rate
IX
ion exchange
KF KL
a Freundlich constant: capacity parameter a Langmuir constant
MDL MIEX
minimum detection limit magnetically impregnated ion exchange (resin)
pHZPC
zero-point charge pH
Q Q10 Q50 Qmax
adsorption capacity Q in equilibrium with 10 µg/L As Q in equilibrium with 50 µg/L As a Langmuir constant: maximum monolayer adsorption capacity
Si SiO2 SMI
silicon silica sulfur modified iron
TCLP
toxicity characteristic leaching procedure
V
vanadium
WET
waste extraction test
XRD
x-ray diffraction
1/n
a Freundlich constant: index of favorability 137 ©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
©2005 AwwaRF. All rights reserved.
6666 West Quincy Avenue Denver, CO 80235-3098 USA P 303.347.6100 www.awwarf.org email:
[email protected]
Sponsors Research Develops Knowledge Promotes Collaboration
1P-4.75C-91084-09/05-CM
