Health & Medicine

Evaluation of the potential of benchmarking to facilitate the measurement of chemical persistence in lakes

Description
Evaluation of the potential of benchmarking to facilitate the measurement of chemical persistence in lakes
Published
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Evaluation of the potential of benchmarking to facilitatethe measurement of chemical persistence in lakes q Hongyan Zou ⇑ , Matthew MacLeod, Michael S. McLachlan Department of Applied Environmental Science, Stockholm University, SE 106 91 Stockholm, Sweden h i g h l i g h t s   The chemical persistence in real lakescan be quantified by benchmarkingapproach.   The ranges of transformation half-lives that can be measured wereexplored.   Dominant physical loss processes of chemicals rely on chemical and lakeproperties.   Benchmarking will open new chancefor slowly degraded chemicals. g r a p h i c a l a b s t r a c ta r t i c l e i n f o  Article history: Received 24 May 2013Received in revised form 19 August 2013Accepted 23 August 2013Available online 5 October 2013 Keywords: Benchmarking approachPersistenceAquatic environmentMulti-media model a b s t r a c t The persistence of chemicals in the environment is rarely measured in the field due to a paucity of suit-able methods. Here we explore the potential of chemical benchmarking to facilitate the measurement of persistence in lake systems using a multimedia chemical fate model. The model results show that persis-tence in a lake can be assessed by quantifying the ratio of test chemical and benchmark chemical at asfew as two locations: the point of emission and the outlet of the lake. Appropriate selection of benchmarkchemicals also allows pseudo-first-order rate constants for physical removal processes such as volatiliza-tion and sediment burial to be quantified. We use the model to explore how the maximum persistencethat can be measured in a particular lake depends on the partitioning properties of the test chemicalof interest and the characteristics of the lake. Our model experiments demonstrate that combiningbenchmarking techniques with good experimental design and sensitive environmental analytical chem-istry may open new opportunities for quantifying chemical persistence, particularly for relatively slowlydegradable chemicals for which current methods do not perform well.   2013 The Authors. Published by Elsevier Ltd. All rights reserved. 1. Introduction Persistence in the environment is an undesirable property forsynthetic chemicals that escape from the technosphere, and chem-ical persistence is enshrined as a hazard criterion in many regula-tory frameworks for chemical management. The unacceptablethresholds of persistence in water and sediment, expressed asdegradation half-lives, generally lie in the range of 1–6 months(Boethling et al., 2009; van Wijk et al., 2009; Hope et al., 2010;Moermond et al., 2012; UNEP, 2008). Therefore, the ability tomeasure chemical persistence on the time scale of months is acornerstone of chemical management. However, there are cur-rently very few studies thathave directly measured the persistenceof organic chemicals in the real environment. 0045-6535/$ - see front matter    2013 The Authors. Published by Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.chemosphere.2013.08.097 q This is an open-access article distributed under the terms of the CreativeCommons Attribution-NonCommercial-ShareAlike License, which permits non-commercial use, distribution, and reproduction in any medium, provided thesrcinal author and source are credited. ⇑ Corresponding author. Tel.: +46 8 674 7330; fax: +46 8 674 7638. E-mail address:  hongyan.zou@itm.su.se (H. Zou).Chemosphere 95 (2014) 301–309 Contents lists available at ScienceDirect Chemosphere journal homepage: www.elsevier.com/locate/chemosphere  One approach that has been used to measure persistence in theenvironment is to compile a complete contaminant mass balancefor a well-defined system. Some excellent studies using this ap-proachhave been conducted inSwiss lakes to studythe persistenceof chemicals used in consumer products (Buser et al., 1998; Stolland Giger, 1998; Poiger et al., 2004). While such studies are valu-able, the information obtained about chemical persistence isuncertain because it is directly linked to the accuracy of measure-ments of chemical fluxes in natural systems that are characterizedby large spatial and temporal variability. Studies of chemical per-sistence in mesocosms, which are reconstructions of a small por-tion of the natural environment under controlled conditions,avoid most of the problems of temporal and spatial variability.There are guidelines for using mesocosms for higher-tier riskassessment of plant protection products, and they have been usedto measure the persistence of some pesticides and other organicpollutants (Knuth and Heinis, 1995; Lahti et al., 1997; Knuthet al., 2000; EEC, 2002; Weaver et al., 2005). However, the abilityof mesocosms to reproduce the full complexity of the environmentis limited.Chemical benchmarking is a technique that can be used to over-come the problems of spatial and temporal variability that areencountered with measuring persistence directly in real environ-mental systems. As typically applied, a benchmark chemical is asubstance that behaves in a similar manner to the test chemicalof interest, with the exception of the unknown property of interest.By comparing the behavior of the test chemical and the benchmarkchemical, one obtains information about the relative magnitude of the unknown property between the two substances. If this prop-erty is known for the benchmark chemical, then it can be calcu-lated for the test chemical. The benchmarking principle has beenused to obtain information about chemical removal in estuariesby comparing the spatial gradient in the concentration of chemi-cals srcinating from the river with the gradient in salinity, a con-servative tracer of dilution of the river water (Bester et al., 1998).Benchmarking is the basic principle behind other types of tracerexperiments, for instance when a persistent water-soluble dye tra-cer is added with a water-soluble test chemical to a river to assesschemical removal (Sabaliunas et al., 2003; Whelan et al., 2007).This methodology has been extended to use a persistent chemicalalready present in a river as a tracer to assess the removal of otherchemicals already present in the river (Radke et al., 2010; Kunkeland Radke, 2012).In this paper we explore the potential of the benchmarkingtechnique to quantify the persistence of chemicals in the realenvironment. In doing so, we use benchmarking to quantify sev-eral unknown characteristics of the environmental system, there-by broadening the definition of a benchmarking chemical to be asubstance which is used as a reference point for the behavior of another substance. We focus on lakes, as they offer the possibilityof studying persistence in water and sediment on time scales thatcorrespond to the regulatory thresholds for persistence. We pos-tulate that for some chemicals persistence can be quantifiedbased on the change in concentration ratio of a test chemicaland a benchmark chemical in (i) the medium that is the majorvector of chemical input to the lake (e.g., inflowing water or anemission source) and (ii) the water flowing out of the lake. Ourgoal is to delineate the limitations of this methodology, and spe-cifically to define the range of transformation half-lives thatcould be quantified with such a study and how this range de-pends on the partitioning properties of the test chemical, theproperties of the lake system, and the uncertainty in our determi-nation of the concentration ratios. We envisage that our assess-ment will provide a basis for designing field studies to quantifythe persistence of chemicals in the aquatic environment underreal conditions. 2. Theory   2.1. Model We use a one-box model that assumes steady state and a chem-ical partitioning equilibrium between water and sediment to as-sess the potential to measure the persistence of contaminants inlake systems. The model includes water, suspended sedimentsand surface sediment in the lake, where the surface sediment con-sists of that volume of sediment that readily exchanges chemicalswith the water column (i.e., non-buried sediment). It is assumedthat the system is well-mixed. Chemical input to the lake, whichcould be via inflowing surface water, inflowing groundwater,atmospheric deposition, or direct emissions, is treated as a singleterm. Four processes for chemical loss are considered: advection,volatilization, sediment burial, and transformation (see Fig. 1).The chemical mass balance is described by the followingequation: I   ¼ ð G W  þ k W  Af  D  þ k B  AK  SW  f  D  þ k 0 R  ð V  W  þ K  SW  f  D V  S ÞÞ C  W  ð 1 Þ where  I   is the rate of chemical input into the lake from all sources,mol h  1 ;  G W  is the flow rate of water out of the system, m 3 h  1 ;  k W is the overall air–water mass transfer coefficient for the chemicalsreferenced to the water phase, m h  1 ;  A  is the surface area of thewater body, m 2 ;  f  D  is the fraction of the chemical in water that isfreely dissolved;  k B  is the burial rate of bulk sediment, m h  1 ;  K  SW is the sediment/water equilibrium partition coefficient, m 3 waterm  3 bulk sediment (i.e., the concentration of chemical in the bulksediment divided by the concentration of freely dissolved chemicalin the water); k 0 R   is the first orderrate constant for transformation of thechemical in the system, h  1 ;  V  W  is the volumeof water, m 3 ;  V  S  isthe volume of surface sediment, m 3 ;  C  W  is the total (freely dissolvedplus sorbed) concentration of chemical in the water, mol m  3 . The half-life of the chemical in the lake due to transformationprocesses,  t  0.5R  , is defined by  t  0.5R   = ln2 /  k 0 R  . Note that this defini-tion includes transformation in both the water and the sedimentcompartments. It therefore gives a measure of the persistence inthe lake as a whole, not media-specific persistence. The model doesnot specify where in the lake the transformation is occurring; itcould be in the water column, in the surface sediment, or in both.A version of the mass balance model coded in Excel was used toexplore the potential of the theoretical framework for determiningchemical persistence in lakes and to assess the sensitivity of thedetermined transformation rate constants to certain sources of uncertainty. In the model,  k W ,  K  SW  and  f  D  were defined according Fig. 1.  Schematic illustration of the lake model showing chemical inputs (mol h  1 )in green and four removal fluxes in purple (mol h  1 ), i.e., volatilization, advection,burial and transformation (the terms shown in the equations are defined in thetext); red and black dots represent suspended solids and bottom sediments,respectively. (For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)302  H. Zou et al./Chemosphere 95 (2014) 301–309  to Eqs. (2)–(4), while the other parameters were treated asconstants.According to the two-film theory,  k W  can be expressed as fol-lows (Mackay, 2001): 1 k W ¼  1 k water þ  1 k air K  AW ð 2 Þ where  k water  is the water-side mass transfer coefficient, m h  1 ; k air  is the air-side mass transfer coefficient, m h  1 ;  K  AW  is theair–water partition coefficient of the chemical, m 3 water m  3 air. K  SW  was expressed as the sum of partitioning into sediment sol-ids and sediment pore water, whereby the former was assumed tobe dominated by partitioning into organic matter: K  SW  ¼  U S K  DS  W  þ ð 1  U S Þ ¼  U S q S  f  OC ; Sediment K  OC  þ ð 1  U S Þ ð 3 Þ where  K  DS-W  is the dry sediment–water partition coefficient, m 3 water m  3 dry sediment;  U S  is the volume fraction of solids insediments;  q S  is the density of the dry sediment solids, kg L   1 ;  f  OC,Sediment  is the organic carbon content of the sediments,kg OC kg  1 dry sediment;  K  OC  is the organic carbon–water partitioncoefficient, L water kg  1 OC. The QSAR   K  OC  = 0.35  K  OW  was used toestimate  K  OC  (Seth et al., 1999).  f  D  was estimated considering the partitioning of the chemicalinto organic matter in the suspended solids:  f  D  ¼  11 þ U W K  PW ¼  11 þ U W q S  f  OC ; SS K  OC ð 4 Þ where U W  is the volume fraction of suspended solids in the water;  f  OC,SS  is the organic carbon content in suspended solids (SS),kg OC kg  1 SS;  K  PW  is the suspended solids–water partition coeffi-cient, m 3 water m  3 suspended solids.  2.2. Principles of the benchmarking technique Chemical benchmarking entails comparing the behavior of twochemicals. Mathematically, this can be illustrated by dividing themass balance equation for the test chemical by the mass balanceequation for the benchmark chemical:orwhere the subscripts BM and Test refer to the benchmark and thetest chemicals, respectively. When there is a single dominantsource of chemical input to the lake (e.g., a wastewater treatmentplant (WWTP)) and this source is the same for both the benchmarkchemical and the test chemical, then the left hand side of Eq. (6) isa readily measurable parameter. Under steady state conditions it isthe ratio of the concentration of the test and benchmark chemicalsin the emission medium divided by the same ratio in the water.Eq. (6) can be used as a guide to quantify key properties of thelake system. Of the 4 chemical removal processes considered in themodel, advection can generally be reliably quantified withoutchemical measurements by directly measuring the water flow rate, G W . By selecting a suitable benchmark chemical for which all otherprocesses can be neglected, i.e., a persistent, hydrophilic, non-vol-atile substance, the denominator in Eq. (6) becomes simply  G W . Anexample of a potential benchmark of this kind is sucralose (Oppen-heimer et al., 2011).The benchmarking approach can now be used to quantify thesystem properties. This is done using diagnostic chemicals, where-by a diagnostic chemical is a test chemical used to quantify a sys-tem property. For instance, to quantify the burial rate of bulksediment  k B , we choose a diagnostic chemical for which volatiliza-tion and transformation can be neglected. Eq. (6) then reduces to I  Diag I  BM  =  C  W ; Diag I  W ; BM  ¼  G W  þ k B  AK  SW ; Diag  f  D ; Diag G W ð 7 Þ If one selects a hydrophobic chemical then the expression K  SW,Diag  f  D,Test  can be simplified to  v   1SS  , a variable independent of the properties of the chemical equal to the ratio of the volume nor-malized sorption capacity of the solids in the water column to thatin the sediment (see the Supplementary information for deriva-tion). Eq. (6) can then be further simplified and solved for the sed-iment burial rate  k B . k B  ¼  G W v  SS  AI  Diag I  BM  =  C  W ; Diag I  W ; BM   1   ð 8 Þ All of the variables on the right hand side of Eq. (8) can be mea-sured or are available. In this manner benchmarking can be used toquantify one system property, i.e., sediment burial rate. A potentialdiagnostic chemical to quantify  k B  is silver, which is emitted fromWWTPs and then sequestered into sediments (Blaser et al., 2008).The same approach can be used to quantify the overall air–water mass transfer coefficient,  k W . By choosing a persistent, vola-tile and hydrophilic diagnostic chemical for which sediment burialand transformation can be neglected. Eq. (6) can be simplified andsolved for  k W . k W ; Diag  ¼  G W  Af  D ; Diag I  Diag I  BM  =  C  W ; Diag C  W ; BM   1   ð 9 Þ Here  f  D,Diag , the fraction of the diagnostic chemical in water that isfreely dissolved, can be either measured in the field or estimatedfrom the organic carbon–water partition coefficient  K  OC  and the or-ganic carbon content and the concentration of particles suspendedin the water. One potential diagnostic chemical to quantify  k W  ismusk xylene. We note that  k W  can vary between chemicals, partic-ularly for chemicals with low air–water partition coefficients. Wereturn to this issue below. I  Test I  BM  ¼ð G W  þ k W ; Test  Af  D ; Test  þ k B  AK  SW ; Test  f  D ; Test  þ k 0 R  ; Test ð V  W  þ K  SW ; Test  f  D ; Test V  S ÞÞ C  W ; Test ð G W  þ k W ; BM  Af  D ; BM  þ k B  AK  SW ; BM  f  D ; BM  þ k 0 R  ; BM ð V  W  þ K  SW ; BM  f  D ; BM V  S ÞÞ C  W ; BM ð 5 Þ I  Test I  BM  =  C  W ; Test C  W ; BM  ¼ G W  þ k W ; Test  Af  D ; Test  þ k B  AK  SW ; Test  f  D ; Test  þ k 0 R  ; Test ð V  W  þ K  SW ; Test  f  D ; Test V  S Þ G W  þ k W ; BM  Af  D ; BM  þ k B  AK  SW ; BM  f  D ; BM  þ k 0 R  ; BM ð V  W  þ K  SW ; BM  f  D ; BM V  S Þð 6 Þ H. Zou et al./Chemosphere 95 (2014) 301–309  303  Having quantified the system parameters for advection,sediment burial and volatilization, we can now address thetransformation process. Choosing a persistent chemical as abenchmark, Eq. (6) can be solved for any test chemical for thetransformation rate constant in the system,  k 0 R  .Of the variables on the right hand side of the equation,  G W ,  A ,and  V  W  are system parameters that are generally known, while k W,BM ,  k W,Test  and  k B  can be determined using benchmarking exper-iments as described above.  f  D,BM ,  K  SW,BM ,  f  D,Test  and  K  SW,Test  arechemical parameters that can be estimated using QSPRs or mea-sured in the field, whereby the product  K  SW  f  D  is simplified to  v   1SS for chemicals that are present almost entirely in sorbed form inthe water.  V  S , the volume of surface sediment in the lake, can beestimated from the spatial extent and the mixed layer depth of sediments. The equation thus demonstrates that  k 0 R   can be deter-mined by measuring the ratios  I  Test / I  BM  and  C  W,Test / C  W,BM . 3. Model experiments We performed model experiments to assess the range of degra-dation half-lives that can be practically assessed with the proposedbenchmarking technique, and how these depend on the partition-ing properties of the chemical and the physical properties of theenvironmental system. Our base-case model experiments wereconducted for Lake Boren, a lake in southern Sweden with a surfacearea of 27.7 km 2 , an average depth of 5.3 m, suspended solid con-centration of 7 g m  3 and a hydraulic residence time of 60 d. A mu-nicipal waste water treatment plant (WWTP) with a capacity of 40000 population equivalents discharges its effluent into Lake Bo-ren. For the purposeof the case study it is assumed that this WWTPis the primary source of both the test chemical and the benchmarkchemicals to the lake. Thermal stratification is limited to brief peri-ods during summer. Table S1lists all of the system variables. Fourvariationsof this base-case lake systemwere then definedin whichone of the following system properties was changed for each vari-ation: hydraulic residence time, depth, sediment burial rate ( k B ),and suspended solids concentration. For each system the modelwas run for a series of hypothetical chemicals with varying parti-tioning properties:   3 < log K  OW  < 10 and   10 < log K  AW  < 4.First the model was used to identify the dominant physical re-moval process (i.e., advection, burial, or volatilization) for all hypo-thetical chemicals in the specified lake system. A default emissionof 100 mol h  1 was chosen and chemical transformation was con-sidered to be negligible ( k 0 R   ¼  0). A removal process was defined asdominant if it contributed P 80% to the total loss of the chemicalfrom the system. When no single process contributed P 80%, thenthe two strongest processes were considered dominant providedthey contributed P 80%; otherwise all three processes were consid-ered important.The second application of the model was to calculate the mini-mum value of   k 0 R   that could be determined for each hypotheticalchemical in each system. The ability to measure chemical transfor-mation in the field using the benchmarking approach is con-strained by the properties of the chemical, the characteristics of the lake, and the precision with which the ratios of test and bench-mark chemicals can be measured in inputs and outflowing water.To this end, the model was run in benchmarking mode (i.e., tostudy the behavior of two chemicals relative to each other).It was assumed that the benchmarking chemical was chosen suchthat it has the same physical removal processes as the test chem-ical. In this case Eq. (10)can be simplified to k 0 R  ; Test  ¼  ln ð 2 Þ t  0 : 5R  ¼ð G W  þ k W  Af  D  þ k B  AK  SW  f  D Þ  I  Test I  BM  =  C  W ; Test C  W ; BM   1  V  W  þ K  SW  f  D V  S ð 11 Þ Of the variables on the right-hand side of the equation, only( I  Test / I  BM )/( C  W,Test / C  W,BM ) is unknown. This is the parameter thatmust be measured in the field if the benchmarking technique is ap-plied in the real world. Its value will be subject to several uncer-tainties, including the precision of the analytical methods todeterminethe ratio of the concentrations of the testand the bench-mark chemicals, and the variability of this ratio in both the emis-sions and the water. For illustrative purposes, we assumed thatthe uncertainty in ( I  Test / I  BM )/( C  W,Test / C  W,BM ) was such that the min-imum value that could be significantly distinguished from 1 was 2,i.e., that the ratio of the test chemical to the benchmark in thewater was two times lower than in the emissions. This value wasthen used to estimate the maximum value of   t  0.5R   that could bedetermined (Eq. (12)). t  0 : 5R  ; Max  ¼  ln ð 2 Þ   V  W  þ K  SW  f  D V  S ð G W  þ k W  Af  D  þ k B  AK  SW  f  D Þ I  Test I  BM C  W ; Test C  W ; BM  1 ! ¼  ln ð 2 Þ   V  W  þ K  SW  f  D V  S G W  þ k W  Af  D  þ k B  AK  SW  f  D ð 12 Þ An important source of model uncertainty was the use of aQSAR to estimate  K  OC  from  K  OW . To assess the possible impact of this model uncertainty on modeled  t  0.5R  , a sensitivity analysiswas conducted for  K  OC  using the full set of hypothetical chemicalsin the base-case lake system.The dependence of   k W  on  K  AW  at low values of   K  AW  is anotherpotential complication. Should volatilization be a dominant physi-cal removal process for chemicals in the low  K  AW  range, thenuncertainty in  t  0.5R   could be dominated by uncertainty in  K  AW ,which would complicate the selection of benchmark chemicals.To address this issue, the parameter sensitivity of the model’s pre-diction of   t  0.5R   to  K  AW  was assessed. 4. Results 4.1. Dominant physical removal processes Fig. 2 shows the dominant physical removal processes identi-fied by the model simulation of the four lakes. The results are pre-sented on a two-dimensional partitioning space plot of log K  AW versus log K  OW . Each of the physical elimination processes, i.e., vol-atilization, advection and burial, has a portion of the partitioningspace where it is dominant. Between the regions where one pro-cess is dominant are transition zones where two or three processes k 0 R  ; Test  ¼ð G W  þ k W ; BM  Af  D ; BM  þ k B  AK  SW ; BM  f  D ; BM Þ I  Test I  BM C  W ; Test C  W ; BM !  G W   k W ; Test  Af  D ; Test   k B  AK  SW ; Test  f  D ; Test V  W  þ K  SW ; Test  f  D ; Test V  S ð 10 Þ 304  H. Zou et al./Chemosphere 95 (2014) 301–309  are controlling. The model results conform to the intuitive expec-tation that burial will be important for hydrophobic chemicals thatsorb strongly to sediment, that volatilization will be important forchemicals that are volatile, and that advection will dominate whenburial and volatilization are slow processes.Fig. 2 provides a useful basis for identifying chemicals thatcould be suitable benchmarks and diagnostic chemicals in experi-ments for characterizing the system properties as discussed in Sec-tion 2. A persistent chemical in the lower left region of thepartitioning space would be a suitable benchmark chemical (i.e.,mainly removed by advection), while a persistent chemical in theupper left region would be a suitable diagnostic chemical to char-acterize the system’s volatilization properties  k W , and a persistentchemical in the right region would be a suitable diagnostic chem-ical to characterize the system’s sediment burial properties  k B .The four panels in Fig. 2 illustrate that the boundaries betweenthe regions shift depending on the properties of the lake system.For example, increasing hydraulic residence time from 60 to1000 d (compare the base-case Lake A with Lake B) reduced theimportance of advection, which resulted in a retreat of the bound-aries of the regions in which advection was a dominant removalprocess by up to 1.5 log units in  K  OW  or  K  AW  and as well as the cre-ation of a new region on the right hand side of the chemical spaceplot in which burial alone was the dominant elimination process.Comparing Lake A with Lake C shows the impact of increasingthe water depth by a factor of 2. Since the hydraulic residence timeremained unchanged, this corresponded to a doubling of theadvection flux. The volatilization and burial fluxes were unchangedand hence their relative importance decreased. Advection is thedominant elimination process even for high  K  OW  compounds be- Fig. 2.  Partitioning space plots showing the dominant physical removal processes for hypothetical persistent chemicals in the four imaginary lakes. V refers to volatilization,A to advection, and B to sediment burial. A and B means that both advection and burial are dominant; A and V means that both advection and volatilization are dominant; Vand B means that both volatilization and burial are dominant; V, A, B means that all three removal processes are important. Lake A is the the base-case lake with theproperties of Lake Boren. The other three lakes (B–D) are also based on Lake Boren, but one key system property was changed. The values assigned to the properties of thesystem that were varied in the simulations were: Lake A: water depth 5.3 m, hydraulic residence time 60 d, burial rate 50 g OC m  2 year  1 ; Lake B: water depth 5.3 m,hydraulic residence time 1000 d, burial rate 50 g OC m  2 year  1 ; Lake C: water depth 10 m, hydraulic residence time 60 d, burial rate 50 g OC m  2 year  1 ; Lake D: waterdepth 5.3 m, hydraulic residence time 60 d, burial rate 150 g OC m  2 year  1 . H. Zou et al./Chemosphere 95 (2014) 301–309  305
Search
Similar documents
View more...
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks